2020-07-10 21:31:17 +08:00
|
|
|
%% This Source Code Form is subject to the terms of the Mozilla Public
|
|
|
|
|
%% License, v. 2.0. If a copy of the MPL was not distributed with this
|
|
|
|
|
%% file, You can obtain one at https://mozilla.org/MPL/2.0/.
|
2018-11-30 22:07:22 +08:00
|
|
|
%%
|
2023-01-02 12:17:36 +08:00
|
|
|
%% Copyright (c) 2007-2023 VMware, Inc. or its affiliates. All rights reserved.
|
2018-11-30 22:07:22 +08:00
|
|
|
%%
|
|
|
|
|
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
%% @doc Provides an easy to consume API for interacting with the {@link rabbit_fifo.}
|
|
|
|
|
%% state machine implementation running inside a `ra' raft system.
|
|
|
|
|
%%
|
|
|
|
|
%% Handles command tracking and other non-functional concerns.
|
|
|
|
|
-module(rabbit_fifo_client).
|
|
|
|
|
|
|
|
|
|
-export([
|
2023-01-19 22:33:28 +08:00
|
|
|
init/1,
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
init/2,
|
2018-12-14 18:10:29 +08:00
|
|
|
checkout/5,
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
cancel_checkout/2,
|
|
|
|
|
enqueue/3,
|
2023-01-19 22:33:28 +08:00
|
|
|
enqueue/4,
|
|
|
|
|
dequeue/4,
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
settle/3,
|
|
|
|
|
return/3,
|
|
|
|
|
discard/3,
|
|
|
|
|
credit/4,
|
2023-01-19 22:33:28 +08:00
|
|
|
handle_ra_event/4,
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
untracked_enqueue/2,
|
|
|
|
|
purge/1,
|
2019-01-04 15:46:15 +08:00
|
|
|
update_machine_state/2,
|
2019-06-04 18:51:25 +08:00
|
|
|
pending_size/1,
|
2019-10-25 13:36:48 +08:00
|
|
|
stat/1,
|
2021-11-04 00:26:40 +08:00
|
|
|
stat/2,
|
Message Containers (#5077)
This PR implements an approach for a "protocol (data format) agnostic core" where the format of the message isn't converted at point of reception.
Currently all non AMQP 0.9.1 originating messages are converted into a AMQP 0.9.1 flavoured basic_message record before sent to a queue. If the messages are then consumed by the originating protocol they are converted back from AMQP 0.9.1. For some protocols such as MQTT 3.1 this isn't too expensive as MQTT is mostly a fairly easily mapped subset of AMQP 0.9.1 but for others such as AMQP 1.0 the conversions are awkward and in some cases lossy even if consuming from the originating protocol.
This PR instead wraps all incoming messages in their originating form into a generic, extensible message container type (mc). The container module exposes an API to get common message details such as size and various properties (ttl, priority etc) directly from the source data type. Each protocol needs to implement the mc behaviour such that when a message originating form one protocol is consumed by another protocol we convert it to the target protocol at that point.
The message container also contains annotations, dead letter records and other meta data we need to record during the lifetime of a message. The original protocol message is never modified unless it is consumed.
This includes conversion modules to and from amqp, amqpl (AMQP 0.9.1) and mqtt.
COMMIT HISTORY:
* Refactor away from using the delivery{} record
In many places including exchange types. This should make it
easier to move towards using a message container type instead of
basic_message.
Add mc module and move direct replies outside of exchange
Lots of changes incl classic queues
Implement stream support incl amqp conversions
simplify mc state record
move mc.erl
mc dlx stuff
recent history exchange
Make tracking work
But doesn't take a protocol agnostic approach as we just convert
everything into AMQP legacy and back. Might be good enough for now.
Tracing as a whole may want a bit of a re-vamp at some point.
tidy
make quorum queue peek work by legacy conversion
dead lettering fixes
dead lettering fixes
CMQ fixes
rabbit_trace type fixes
fixes
fix
Fix classic queue props
test assertion fix
feature flag and backwards compat
Enable message_container feature flag in some SUITEs
Dialyzer fixes
fixes
fix
test fixes
Various
Manually update a gazelle generated file
until a gazelle enhancement can be made
https://github.com/rabbitmq/rules_erlang/issues/185
Add message_containers_SUITE to bazel
and regen bazel files with gazelle from rules_erlang@main
Simplify essential proprty access
Such as durable, ttl and priority by extracting them into annotations
at message container init time.
Move type
to remove dependenc on amqp10 stuff in mc.erl
mostly because I don't know how to make bazel do the right thing
add more stuff
Refine routing header stuff
wip
Cosmetics
Do not use "maybe" as type name as "maybe" is a keyword since OTP 25
which makes Erlang LS complain.
* Dedup death queue names
* Fix function clause crashes
Fix failing tests in the MQTT shared_SUITE:
A classic queue message ID can be undefined as set in
https://github.com/rabbitmq/rabbitmq-server/blob/fbe79ff47b4edbc0fd95457e623d6593161ad198/deps/rabbit/src/rabbit_classic_queue_index_v2.erl#L1048
Fix failing tests in the MQTT shared_SUITE-mixed:
When feature flag message_containers is disabled, the
message is not an #mc{} record, but a #basic_message{} record.
* Fix is_utf8_no_null crash
Prior to this commit, the function crashed if invalid UTF-8 was
provided, e.g.:
```
1> rabbit_misc:is_valid_shortstr(<<"😇"/utf16>>).
** exception error: no function clause matching rabbit_misc:is_utf8_no_null(<<216,61,222,7>>) (rabbit_misc.erl, line 1481)
```
* Implement mqtt mc behaviour
For now via amqp translation.
This is still work in progress, but the following SUITEs pass:
```
make -C deps/rabbitmq_mqtt ct-shared t=[mqtt,v5,cluster_size_1] FULL=1
make -C deps/rabbitmq_mqtt ct-v5 t=[mqtt,cluster_size_1] FULL=1
```
* Shorten mc file names
Module name length matters because for each persistent message the #mc{}
record is persisted to disk.
```
1> iolist_size(term_to_iovec({mc, rabbit_mc_amqp_legacy})).
30
2> iolist_size(term_to_iovec({mc, mc_amqpl})).
17
```
This commit renames the mc modules:
```
ag -l rabbit_mc_amqp_legacy | xargs sed -i 's/rabbit_mc_amqp_legacy/mc_amqpl/g'
ag -l rabbit_mc_amqp | xargs sed -i 's/rabbit_mc_amqp/mc_amqp/g'
ag -l rabbit_mqtt_mc | xargs sed -i 's/rabbit_mqtt_mc/mc_mqtt/g'
```
* mc: make deaths an annotation + fixes
* Fix mc_mqtt protocol_state callback
* Fix test will_delay_node_restart
```
make -C deps/rabbitmq_mqtt ct-v5 t=[mqtt,cluster_size_3]:will_delay_node_restart FULL=1
```
* Bazel run gazelle
* mix format rabbitmqctl.ex
* Ensure ttl annotation is refelected in amqp legacy protocol state
* Fix id access in message store
* Fix rabbit_message_interceptor_SUITE
* dializer fixes
* Fix rabbit:rabbit_message_interceptor_SUITE-mixed
set_annotation/3 should not result in duplicate keys
* Fix MQTT shared_SUITE-mixed
Up to 3.12 non-MQTT publishes were always QoS 1 regardless of delivery_mode.
https://github.com/rabbitmq/rabbitmq-server/blob/75a953ce286a10aca910c098805a4f545989af38/deps/rabbitmq_mqtt/src/rabbit_mqtt_processor.erl#L2075-L2076
From now on, non-MQTT publishes are QoS 1 if durable.
This makes more sense.
The MQTT plugin must send a #basic_message{} to an old node that does
not understand message containers.
* Field content of 'v1_0.data' can be binary
Fix
```
bazel test //deps/rabbitmq_mqtt:shared_SUITE-mixed \
--test_env FOCUS="-group [mqtt,v4,cluster_size_1] -case trace" \
-t- --test_sharding_strategy=disabled
```
* Remove route/2 and implement route/3 for all exchange types.
This removes the route/2 callback from rabbit_exchange_type and
makes route/3 mandatory instead. This is a breaking change and
will require all implementations of exchange types to update their
code, however this is necessary anyway for them to correctly handle
the mc type.
stream filtering fixes
* Translate directly from MQTT to AMQP 0.9.1
* handle undecoded properties in mc_compat
amqpl: put clause in right order
recover death deatails from amqp data
* Replace callback init_amqp with convert_from
* Fix return value of lists:keyfind/3
* Translate directly from AMQP 0.9.1 to MQTT
* Fix MQTT payload size
MQTT payload can be a list when converted from AMQP 0.9.1 for example
First conversions tests
Plus some other conversion related fixes.
bazel
bazel
translate amqp 1.0 null to undefined
mc: property/2 and correlation_id/message_id return type tagged values.
To ensure we can support a variety of types better.
The type type tags are AMQP 1.0 flavoured.
fix death recovery
mc_mqtt: impl new api
Add callbacks to allow protocols to compact data before storage
And make readable if needing to query things repeatedly.
bazel fix
* more decoding
* tracking mixed versions compat
* mc: flip default of `durable` annotation to save some data.
Assuming most messages are durable and that in memory messages suffer less
from persistence overhead it makes sense for a non existent `durable`
annotation to mean durable=true.
* mc conversion tests and tidy up
* mc make x_header unstrict again
* amqpl: death record fixes
* bazel
* amqp -> amqpl conversion test
* Fix crash in mc_amqp:size/1
Body can be a single amqp-value section (instead of
being a list) as shown by test
```
make -C deps/rabbitmq_amqp1_0/ ct-system t=java
```
on branch native-amqp.
* Fix crash in lists:flatten/1
Data can be a single amqp-value section (instead of
being a list) as shown by test
```
make -C deps/rabbitmq_amqp1_0 ct-system t=dotnet:roundtrip_to_amqp_091
```
on branch native-amqp.
* Fix crash in rabbit_writer
Running test
```
make -C deps/rabbitmq_amqp1_0 ct-system t=dotnet:roundtrip_to_amqp_091
```
on branch native-amqp resulted in the following crash:
```
crasher:
initial call: rabbit_writer:enter_mainloop/2
pid: <0.711.0>
registered_name: []
exception error: bad argument
in function size/1
called as size([<<0>>,<<"Sw">>,[<<160,2>>,<<"hi">>]])
*** argument 1: not tuple or binary
in call from rabbit_binary_generator:build_content_frames/7 (rabbit_binary_generator.erl, line 89)
in call from rabbit_binary_generator:build_simple_content_frames/4 (rabbit_binary_generator.erl, line 61)
in call from rabbit_writer:assemble_frames/5 (rabbit_writer.erl, line 334)
in call from rabbit_writer:internal_send_command_async/3 (rabbit_writer.erl, line 365)
in call from rabbit_writer:handle_message/2 (rabbit_writer.erl, line 265)
in call from rabbit_writer:handle_message/3 (rabbit_writer.erl, line 232)
in call from rabbit_writer:mainloop1/2 (rabbit_writer.erl, line 223)
```
because #content.payload_fragments_rev is currently supposed to
be a flat list of binaries instead of being an iolist.
This commit fixes this crash inefficiently by calling
iolist_to_binary/1. A better solution would be to allow AMQP legacy's #content.payload_fragments_rev
to be an iolist.
* Add accidentally deleted line back
* mc: optimise mc_amqp internal format
By removint the outer records for message and delivery annotations
as well as application properties and footers.
* mc: optimis mc_amqp map_add by using upsert
* mc: refactoring and bug fixes
* mc_SUITE routingheader assertions
* mc remove serialize/1 callback as only used by amqp
* mc_amqp: avoid returning a nested list from protocol_state
* test and bug fix
* move infer_type to mc_util
* mc fixes and additiona assertions
* Support headers exchange routing for MQTT messages
When a headers exchange is bound to the MQTT topic exchange, routing
will be performend based on both MQTT topic (by the topic exchange) and
MQTT User Property (by the headers exchange).
This combines the best worlds of both MQTT 5.0 and AMQP 0.9.1 and
enables powerful routing topologies.
When the User Property contains the same name multiple times, only the
last name (and value) will be considered by the headers exchange.
* Fix crash when sending from stream to amqpl
When publishing a message via the stream protocol and consuming it via
AMQP 0.9.1, the following crash occurred prior to this commit:
```
crasher:
initial call: rabbit_channel:init/1
pid: <0.818.0>
registered_name: []
exception exit: {{badmatch,undefined},
[{rabbit_channel,handle_deliver0,4,
[{file,"rabbit_channel.erl"},
{line,2728}]},
{lists,foldl,3,[{file,"lists.erl"},{line,1594}]},
{rabbit_channel,handle_cast,2,
[{file,"rabbit_channel.erl"},
{line,728}]},
{gen_server2,handle_msg,2,
[{file,"gen_server2.erl"},{line,1056}]},
{proc_lib,wake_up,3,
[{file,"proc_lib.erl"},{line,251}]}]}
```
This commit first gives `mc:init/3` the chance to set exchange and
routing_keys annotations.
If not set, `rabbit_stream_queue` will set these annotations assuming
the message was originally published via the stream protocol.
* Support consistent hash exchange routing for MQTT 5.0
When a consistent hash exchange is bound to the MQTT topic exchange,
MQTT 5.0 messages can be routed to queues consistently based on the
Correlation-Data in the PUBLISH packet.
* Convert MQTT 5.0 User Property
* to AMQP 0.9.1 headers
* from AMQP 0.9.1 headers
* to AMQP 1.0 application properties and message annotations
* from AMQP 1.0 application properties and message annotations
* Make use of Annotations in mc_mqtt:protocol_state/2
mc_mqtt:protocol_state/2 includes Annotations as parameter.
It's cleaner to make use of these Annotations when computing the
protocol state instead of relying on the caller (rabbitmq_mqtt_processor)
to compute the protocol state.
* Enforce AMQP 0.9.1 field name length limit
The AMQP 0.9.1 spec prohibits field names longer than 128 characters.
Therefore, when converting AMQP 1.0 message annotations, application
properties or MQTT 5.0 User Property to AMQP 0.9.1 headers, drop any
names longer than 128 characters.
* Fix type specs
Apply feedback from Michael Davis
Co-authored-by: Michael Davis <mcarsondavis@gmail.com>
* Add mc_mqtt unit test suite
Implement mc_mqtt:x_header/2
* Translate indicator that payload is UTF-8 encoded
when converting between MQTT 5.0 and AMQP 1.0
* Translate single amqp-value section from AMQP 1.0 to MQTT
Convert to a text representation, if possible, and indicate to MQTT
client that the payload is UTF-8 encoded. This way, the MQTT client will
be able to parse the payload.
If conversion to text representation is not possible, encode the payload
using the AMQP 1.0 type system and indiate the encoding via Content-Type
message/vnd.rabbitmq.amqp.
This Content-Type is not registered.
Type "message" makes sense since it's a message.
Vendor tree "vnd.rabbitmq.amqp" makes sense since merely subtype "amqp" is not
registered.
* Fix payload conversion
* Translate Response Topic between MQTT and AMQP
Translate MQTT 5.0 Response Topic to AMQP 1.0 reply-to address and vice
versa.
The Response Topic must be a UTF-8 encoded string.
This commit re-uses the already defined RabbitMQ target addresses:
```
"/topic/" RK Publish to amq.topic with routing key RK
"/exchange/" X "/" RK Publish to exchange X with routing key RK
```
By default, the MQTT topic exchange is configure dto be amq.topic using
the 1st target address.
When an operator modifies the mqtt.exchange, the 2nd target address is
used.
* Apply PR feedback
and fix formatting
Co-authored-by: Michael Davis <mcarsondavis@gmail.com>
* tidy up
* Add MQTT message_containers test
* consistent hash exchange: avoid amqp legacy conversion
When hashing on a header value.
* Avoid converting to amqp legacy when using exchange federation
* Fix test flake
* test and dialyzer fixes
* dialyzer fix
* Add MQTT protocol interoperability tests
Test receiving from and sending to MQTT 5.0 and
* AMQP 0.9.1
* AMQP 1.0
* STOMP
* Streams
* Regenerate portions of deps/rabbit/app.bzl with gazelle
I'm not exactly sure how this happened, but gazell seems to have been
run with an older version of the rules_erlang gazelle extension at
some point. This caused generation of a structure that is no longer
used. This commit updates the structure to the current pattern.
* mc: refactoring
* mc_amqpl: handle delivery annotations
Just in case they are included.
Also use iolist_to_iovec to create flat list of binaries when
converting from amqp with amqp encoded payload.
---------
Co-authored-by: David Ansari <david.ansari@gmx.de>
Co-authored-by: Michael Davis <mcarsondavis@gmail.com>
Co-authored-by: Rin Kuryloski <kuryloskip@vmware.com>
2023-08-31 18:27:13 +08:00
|
|
|
query_single_active_consumer/1,
|
|
|
|
|
cluster_name/1
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
]).
|
|
|
|
|
|
2020-05-01 18:43:42 +08:00
|
|
|
-define(SOFT_LIMIT, 32).
|
2019-02-14 18:49:25 +08:00
|
|
|
-define(TIMER_TIME, 10000).
|
2021-11-04 00:26:40 +08:00
|
|
|
-define(COMMAND_TIMEOUT, 30000).
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
|
|
|
|
|
-type seq() :: non_neg_integer().
|
|
|
|
|
-type action() :: {send_credit_reply, Available :: non_neg_integer()} |
|
|
|
|
|
{send_drained, CTagCredit ::
|
2022-11-02 03:52:40 +08:00
|
|
|
{rabbit_fifo:consumer_tag(), non_neg_integer()}} |
|
|
|
|
|
rabbit_queue_type:action().
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
-type actions() :: [action()].
|
|
|
|
|
|
2023-08-23 17:56:30 +08:00
|
|
|
-record(consumer, {last_msg_id :: seq() | -1 | undefined,
|
2020-09-29 18:43:24 +08:00
|
|
|
ack = false :: boolean(),
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
delivery_count = 0 :: non_neg_integer()}).
|
|
|
|
|
|
2023-01-19 22:33:28 +08:00
|
|
|
-record(cfg, {servers = [] :: [ra:server_id()],
|
2020-07-10 17:20:25 +08:00
|
|
|
soft_limit = ?SOFT_LIMIT :: non_neg_integer(),
|
2023-01-19 22:33:28 +08:00
|
|
|
timeout :: non_neg_integer()
|
|
|
|
|
}).
|
2020-07-10 17:20:25 +08:00
|
|
|
|
|
|
|
|
-record(state, {cfg :: #cfg{},
|
2019-08-30 17:45:15 +08:00
|
|
|
leader :: undefined | ra:server_id(),
|
2020-07-10 17:20:25 +08:00
|
|
|
queue_status :: undefined | go | reject_publish,
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
next_seq = 0 :: seq(),
|
|
|
|
|
next_enqueue_seq = 1 :: seq(),
|
|
|
|
|
%% indicates that we've exceeded the soft limit
|
|
|
|
|
slow = false :: boolean(),
|
|
|
|
|
unsent_commands = #{} :: #{rabbit_fifo:consumer_id() =>
|
|
|
|
|
{[seq()], [seq()], [seq()]}},
|
|
|
|
|
pending = #{} :: #{seq() =>
|
2019-08-30 17:45:15 +08:00
|
|
|
{term(), rabbit_fifo:command()}},
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
consumer_deliveries = #{} :: #{rabbit_fifo:consumer_tag() =>
|
|
|
|
|
#consumer{}},
|
2020-07-10 17:20:25 +08:00
|
|
|
timer_state :: term()
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
}).
|
|
|
|
|
|
|
|
|
|
-opaque state() :: #state{}.
|
|
|
|
|
|
|
|
|
|
-export_type([
|
2019-02-14 18:49:25 +08:00
|
|
|
state/0,
|
|
|
|
|
actions/0
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
]).
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
%% @doc Create the initial state for a new rabbit_fifo sessions. A state is needed
|
|
|
|
|
%% to interact with a rabbit_fifo queue using @module.
|
|
|
|
|
%% @param Servers The known servers of the queue. If the current leader is known
|
|
|
|
|
%% ensure the leader node is at the head of the list.
|
2023-01-19 22:33:28 +08:00
|
|
|
-spec init([ra:server_id()]) -> state().
|
|
|
|
|
init(Servers) ->
|
|
|
|
|
init(Servers, ?SOFT_LIMIT).
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
|
|
|
|
|
%% @doc Create the initial state for a new rabbit_fifo sessions. A state is needed
|
|
|
|
|
%% to interact with a rabbit_fifo queue using @module.
|
|
|
|
|
%% @param Servers The known servers of the queue. If the current leader is known
|
|
|
|
|
%% ensure the leader node is at the head of the list.
|
|
|
|
|
%% @param MaxPending size defining the max number of pending commands.
|
2023-01-19 22:33:28 +08:00
|
|
|
-spec init([ra:server_id()], non_neg_integer()) -> state().
|
|
|
|
|
init(Servers, SoftLimit) ->
|
2020-07-14 19:13:55 +08:00
|
|
|
Timeout = application:get_env(kernel, net_ticktime, 60) + 5,
|
2023-01-19 22:33:28 +08:00
|
|
|
#state{cfg = #cfg{servers = Servers,
|
2020-07-10 17:20:25 +08:00
|
|
|
soft_limit = SoftLimit,
|
2020-07-14 19:13:55 +08:00
|
|
|
timeout = Timeout * 1000}}.
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
|
|
|
|
|
%% @doc Enqueues a message.
|
2023-01-19 22:33:28 +08:00
|
|
|
%% @param QueueName Name of the queue.
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
%% @param Correlation an arbitrary erlang term used to correlate this
|
|
|
|
|
%% command when it has been applied.
|
|
|
|
|
%% @param Msg an arbitrary erlang term representing the message.
|
|
|
|
|
%% @param State the current {@module} state.
|
|
|
|
|
%% @returns
|
2022-11-02 03:52:40 +08:00
|
|
|
%% `{ok, State, Actions}' if the command was successfully sent
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
%% {@module} assigns a sequence number to every raft command it issues. The
|
|
|
|
|
%% SequenceNumber can be correlated to the applied sequence numbers returned
|
|
|
|
|
%% by the {@link handle_ra_event/2. handle_ra_event/2} function.
|
2023-01-19 22:33:28 +08:00
|
|
|
-spec enqueue(rabbit_amqqueue:name(), Correlation :: term(),
|
|
|
|
|
Msg :: term(), State :: state()) ->
|
2022-11-02 03:52:40 +08:00
|
|
|
{ok, state(), actions()} | {reject_publish, state()}.
|
2023-01-19 22:33:28 +08:00
|
|
|
enqueue(QName, Correlation, Msg,
|
2020-07-10 17:20:25 +08:00
|
|
|
#state{queue_status = undefined,
|
|
|
|
|
next_enqueue_seq = 1,
|
2021-11-02 18:39:58 +08:00
|
|
|
cfg = #cfg{servers = Servers,
|
|
|
|
|
timeout = Timeout}} = State0) ->
|
2023-08-30 22:12:09 +08:00
|
|
|
%% the first publish, register and enqueuer for this process.
|
|
|
|
|
Reg = rabbit_fifo:make_register_enqueuer(self()),
|
|
|
|
|
case ra:process_command(Servers, Reg, Timeout) of
|
|
|
|
|
{ok, reject_publish, Leader} ->
|
|
|
|
|
{reject_publish, State0#state{leader = Leader,
|
|
|
|
|
queue_status = reject_publish}};
|
|
|
|
|
{ok, ok, Leader} ->
|
|
|
|
|
enqueue(QName, Correlation, Msg, State0#state{leader = Leader,
|
|
|
|
|
queue_status = go});
|
|
|
|
|
{error, {no_more_servers_to_try, _Errs}} ->
|
|
|
|
|
%% if we are not able to process the register command
|
|
|
|
|
%% it is safe to reject the message as we never attempted
|
|
|
|
|
%% to send it
|
|
|
|
|
{reject_publish, State0};
|
|
|
|
|
{timeout, _} ->
|
|
|
|
|
{reject_publish, State0};
|
|
|
|
|
Err ->
|
|
|
|
|
exit(Err)
|
2020-07-13 23:24:10 +08:00
|
|
|
end;
|
2023-01-19 22:33:28 +08:00
|
|
|
enqueue(_QName, _Correlation, _Msg,
|
2020-07-10 17:20:25 +08:00
|
|
|
#state{queue_status = reject_publish,
|
|
|
|
|
cfg = #cfg{}} = State) ->
|
|
|
|
|
{reject_publish, State};
|
2023-01-19 22:33:28 +08:00
|
|
|
enqueue(QName, Correlation, Msg,
|
2020-07-10 17:20:25 +08:00
|
|
|
#state{slow = Slow,
|
2022-01-26 00:52:01 +08:00
|
|
|
pending = Pending,
|
2020-07-14 19:13:55 +08:00
|
|
|
queue_status = go,
|
2022-02-18 23:38:57 +08:00
|
|
|
next_seq = Seq,
|
|
|
|
|
next_enqueue_seq = EnqueueSeq,
|
2022-11-02 03:52:40 +08:00
|
|
|
cfg = #cfg{soft_limit = SftLmt}} = State0) ->
|
2022-01-26 00:52:01 +08:00
|
|
|
Server = pick_server(State0),
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
% by default there is no correlation id
|
2022-02-18 23:38:57 +08:00
|
|
|
Cmd = rabbit_fifo:make_enqueue(self(), EnqueueSeq, Msg),
|
2022-01-26 00:52:01 +08:00
|
|
|
ok = ra:pipeline_command(Server, Cmd, Seq, low),
|
|
|
|
|
Tag = case map_size(Pending) >= SftLmt of
|
|
|
|
|
true -> slow;
|
|
|
|
|
false -> ok
|
|
|
|
|
end,
|
|
|
|
|
State = State0#state{pending = Pending#{Seq => {Correlation, Cmd}},
|
2022-02-18 23:38:57 +08:00
|
|
|
next_seq = Seq + 1,
|
|
|
|
|
next_enqueue_seq = EnqueueSeq + 1,
|
2022-01-26 00:52:01 +08:00
|
|
|
slow = Tag == slow},
|
|
|
|
|
case Tag of
|
|
|
|
|
slow when not Slow ->
|
2023-01-19 22:33:28 +08:00
|
|
|
{ok, set_timer(QName, State), [{block, cluster_name(State)}]};
|
2022-01-26 00:52:01 +08:00
|
|
|
_ ->
|
2022-11-02 03:52:40 +08:00
|
|
|
{ok, State, []}
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
end.
|
|
|
|
|
|
|
|
|
|
%% @doc Enqueues a message.
|
2023-01-19 22:33:28 +08:00
|
|
|
%% @param QueueName Name of the queue.
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
%% @param Msg an arbitrary erlang term representing the message.
|
|
|
|
|
%% @param State the current {@module} state.
|
|
|
|
|
%% @returns
|
2022-11-02 03:52:40 +08:00
|
|
|
%% `{ok, State, Actions}' if the command was successfully sent.
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
%% {@module} assigns a sequence number to every raft command it issues. The
|
|
|
|
|
%% SequenceNumber can be correlated to the applied sequence numbers returned
|
|
|
|
|
%% by the {@link handle_ra_event/2. handle_ra_event/2} function.
|
|
|
|
|
%%
|
2023-01-19 22:33:28 +08:00
|
|
|
-spec enqueue(rabbit_amqqueue:name(), Msg :: term(), State :: state()) ->
|
2022-11-02 03:52:40 +08:00
|
|
|
{ok, state(), actions()} | {reject_publish, state()}.
|
2023-01-19 22:33:28 +08:00
|
|
|
enqueue(QName, Msg, State) ->
|
|
|
|
|
enqueue(QName, undefined, Msg, State).
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
|
|
|
|
|
%% @doc Dequeue a message from the queue.
|
|
|
|
|
%%
|
2019-02-13 01:28:22 +08:00
|
|
|
%% This is a synchronous call. I.e. the call will block until the command
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
%% has been accepted by the ra process or it times out.
|
|
|
|
|
%%
|
2023-01-19 22:33:28 +08:00
|
|
|
%% @param QueueName Name of the queue.
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
%% @param ConsumerTag a unique tag to identify this particular consumer.
|
|
|
|
|
%% @param Settlement either `settled' or `unsettled'. When `settled' no
|
|
|
|
|
%% further settlement needs to be done.
|
|
|
|
|
%% @param State The {@module} state.
|
|
|
|
|
%%
|
|
|
|
|
%% @returns `{ok, IdMsg, State}' or `{error | timeout, term()}'
|
2023-01-19 22:33:28 +08:00
|
|
|
-spec dequeue(rabbit_amqqueue:name(), rabbit_fifo:consumer_tag(),
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
Settlement :: settled | unsettled, state()) ->
|
2020-09-29 18:43:24 +08:00
|
|
|
{ok, non_neg_integer(), term(), non_neg_integer()}
|
|
|
|
|
| {empty, state()} | {error | timeout, term()}.
|
2023-01-19 22:33:28 +08:00
|
|
|
dequeue(QueueName, ConsumerTag, Settlement,
|
|
|
|
|
#state{cfg = #cfg{timeout = Timeout}} = State0) ->
|
2020-07-10 17:20:25 +08:00
|
|
|
Node = pick_server(State0),
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
ConsumerId = consumer_id(ConsumerTag),
|
2018-12-14 18:10:29 +08:00
|
|
|
case ra:process_command(Node,
|
|
|
|
|
rabbit_fifo:make_checkout(ConsumerId,
|
|
|
|
|
{dequeue, Settlement},
|
|
|
|
|
#{}),
|
|
|
|
|
Timeout) of
|
2019-01-29 00:00:07 +08:00
|
|
|
{ok, {dequeue, empty}, Leader} ->
|
2020-09-29 18:43:24 +08:00
|
|
|
{empty, State0#state{leader = Leader}};
|
|
|
|
|
{ok, {dequeue, {MsgId, {MsgHeader, Msg0}}, MsgsReady}, Leader} ->
|
|
|
|
|
Count = case MsgHeader of
|
|
|
|
|
#{delivery_count := C} -> C;
|
|
|
|
|
_ -> 0
|
|
|
|
|
end,
|
|
|
|
|
IsDelivered = Count > 0,
|
|
|
|
|
Msg = add_delivery_count_header(Msg0, Count),
|
|
|
|
|
{ok, MsgsReady,
|
2023-01-19 22:33:28 +08:00
|
|
|
{QueueName, qref(Leader), MsgId, IsDelivered, Msg},
|
2020-07-10 17:20:25 +08:00
|
|
|
State0#state{leader = Leader}};
|
2019-11-20 18:24:12 +08:00
|
|
|
{ok, {error, _} = Err, _Leader} ->
|
|
|
|
|
Err;
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
Err ->
|
|
|
|
|
Err
|
|
|
|
|
end.
|
|
|
|
|
|
Message Containers (#5077)
This PR implements an approach for a "protocol (data format) agnostic core" where the format of the message isn't converted at point of reception.
Currently all non AMQP 0.9.1 originating messages are converted into a AMQP 0.9.1 flavoured basic_message record before sent to a queue. If the messages are then consumed by the originating protocol they are converted back from AMQP 0.9.1. For some protocols such as MQTT 3.1 this isn't too expensive as MQTT is mostly a fairly easily mapped subset of AMQP 0.9.1 but for others such as AMQP 1.0 the conversions are awkward and in some cases lossy even if consuming from the originating protocol.
This PR instead wraps all incoming messages in their originating form into a generic, extensible message container type (mc). The container module exposes an API to get common message details such as size and various properties (ttl, priority etc) directly from the source data type. Each protocol needs to implement the mc behaviour such that when a message originating form one protocol is consumed by another protocol we convert it to the target protocol at that point.
The message container also contains annotations, dead letter records and other meta data we need to record during the lifetime of a message. The original protocol message is never modified unless it is consumed.
This includes conversion modules to and from amqp, amqpl (AMQP 0.9.1) and mqtt.
COMMIT HISTORY:
* Refactor away from using the delivery{} record
In many places including exchange types. This should make it
easier to move towards using a message container type instead of
basic_message.
Add mc module and move direct replies outside of exchange
Lots of changes incl classic queues
Implement stream support incl amqp conversions
simplify mc state record
move mc.erl
mc dlx stuff
recent history exchange
Make tracking work
But doesn't take a protocol agnostic approach as we just convert
everything into AMQP legacy and back. Might be good enough for now.
Tracing as a whole may want a bit of a re-vamp at some point.
tidy
make quorum queue peek work by legacy conversion
dead lettering fixes
dead lettering fixes
CMQ fixes
rabbit_trace type fixes
fixes
fix
Fix classic queue props
test assertion fix
feature flag and backwards compat
Enable message_container feature flag in some SUITEs
Dialyzer fixes
fixes
fix
test fixes
Various
Manually update a gazelle generated file
until a gazelle enhancement can be made
https://github.com/rabbitmq/rules_erlang/issues/185
Add message_containers_SUITE to bazel
and regen bazel files with gazelle from rules_erlang@main
Simplify essential proprty access
Such as durable, ttl and priority by extracting them into annotations
at message container init time.
Move type
to remove dependenc on amqp10 stuff in mc.erl
mostly because I don't know how to make bazel do the right thing
add more stuff
Refine routing header stuff
wip
Cosmetics
Do not use "maybe" as type name as "maybe" is a keyword since OTP 25
which makes Erlang LS complain.
* Dedup death queue names
* Fix function clause crashes
Fix failing tests in the MQTT shared_SUITE:
A classic queue message ID can be undefined as set in
https://github.com/rabbitmq/rabbitmq-server/blob/fbe79ff47b4edbc0fd95457e623d6593161ad198/deps/rabbit/src/rabbit_classic_queue_index_v2.erl#L1048
Fix failing tests in the MQTT shared_SUITE-mixed:
When feature flag message_containers is disabled, the
message is not an #mc{} record, but a #basic_message{} record.
* Fix is_utf8_no_null crash
Prior to this commit, the function crashed if invalid UTF-8 was
provided, e.g.:
```
1> rabbit_misc:is_valid_shortstr(<<"😇"/utf16>>).
** exception error: no function clause matching rabbit_misc:is_utf8_no_null(<<216,61,222,7>>) (rabbit_misc.erl, line 1481)
```
* Implement mqtt mc behaviour
For now via amqp translation.
This is still work in progress, but the following SUITEs pass:
```
make -C deps/rabbitmq_mqtt ct-shared t=[mqtt,v5,cluster_size_1] FULL=1
make -C deps/rabbitmq_mqtt ct-v5 t=[mqtt,cluster_size_1] FULL=1
```
* Shorten mc file names
Module name length matters because for each persistent message the #mc{}
record is persisted to disk.
```
1> iolist_size(term_to_iovec({mc, rabbit_mc_amqp_legacy})).
30
2> iolist_size(term_to_iovec({mc, mc_amqpl})).
17
```
This commit renames the mc modules:
```
ag -l rabbit_mc_amqp_legacy | xargs sed -i 's/rabbit_mc_amqp_legacy/mc_amqpl/g'
ag -l rabbit_mc_amqp | xargs sed -i 's/rabbit_mc_amqp/mc_amqp/g'
ag -l rabbit_mqtt_mc | xargs sed -i 's/rabbit_mqtt_mc/mc_mqtt/g'
```
* mc: make deaths an annotation + fixes
* Fix mc_mqtt protocol_state callback
* Fix test will_delay_node_restart
```
make -C deps/rabbitmq_mqtt ct-v5 t=[mqtt,cluster_size_3]:will_delay_node_restart FULL=1
```
* Bazel run gazelle
* mix format rabbitmqctl.ex
* Ensure ttl annotation is refelected in amqp legacy protocol state
* Fix id access in message store
* Fix rabbit_message_interceptor_SUITE
* dializer fixes
* Fix rabbit:rabbit_message_interceptor_SUITE-mixed
set_annotation/3 should not result in duplicate keys
* Fix MQTT shared_SUITE-mixed
Up to 3.12 non-MQTT publishes were always QoS 1 regardless of delivery_mode.
https://github.com/rabbitmq/rabbitmq-server/blob/75a953ce286a10aca910c098805a4f545989af38/deps/rabbitmq_mqtt/src/rabbit_mqtt_processor.erl#L2075-L2076
From now on, non-MQTT publishes are QoS 1 if durable.
This makes more sense.
The MQTT plugin must send a #basic_message{} to an old node that does
not understand message containers.
* Field content of 'v1_0.data' can be binary
Fix
```
bazel test //deps/rabbitmq_mqtt:shared_SUITE-mixed \
--test_env FOCUS="-group [mqtt,v4,cluster_size_1] -case trace" \
-t- --test_sharding_strategy=disabled
```
* Remove route/2 and implement route/3 for all exchange types.
This removes the route/2 callback from rabbit_exchange_type and
makes route/3 mandatory instead. This is a breaking change and
will require all implementations of exchange types to update their
code, however this is necessary anyway for them to correctly handle
the mc type.
stream filtering fixes
* Translate directly from MQTT to AMQP 0.9.1
* handle undecoded properties in mc_compat
amqpl: put clause in right order
recover death deatails from amqp data
* Replace callback init_amqp with convert_from
* Fix return value of lists:keyfind/3
* Translate directly from AMQP 0.9.1 to MQTT
* Fix MQTT payload size
MQTT payload can be a list when converted from AMQP 0.9.1 for example
First conversions tests
Plus some other conversion related fixes.
bazel
bazel
translate amqp 1.0 null to undefined
mc: property/2 and correlation_id/message_id return type tagged values.
To ensure we can support a variety of types better.
The type type tags are AMQP 1.0 flavoured.
fix death recovery
mc_mqtt: impl new api
Add callbacks to allow protocols to compact data before storage
And make readable if needing to query things repeatedly.
bazel fix
* more decoding
* tracking mixed versions compat
* mc: flip default of `durable` annotation to save some data.
Assuming most messages are durable and that in memory messages suffer less
from persistence overhead it makes sense for a non existent `durable`
annotation to mean durable=true.
* mc conversion tests and tidy up
* mc make x_header unstrict again
* amqpl: death record fixes
* bazel
* amqp -> amqpl conversion test
* Fix crash in mc_amqp:size/1
Body can be a single amqp-value section (instead of
being a list) as shown by test
```
make -C deps/rabbitmq_amqp1_0/ ct-system t=java
```
on branch native-amqp.
* Fix crash in lists:flatten/1
Data can be a single amqp-value section (instead of
being a list) as shown by test
```
make -C deps/rabbitmq_amqp1_0 ct-system t=dotnet:roundtrip_to_amqp_091
```
on branch native-amqp.
* Fix crash in rabbit_writer
Running test
```
make -C deps/rabbitmq_amqp1_0 ct-system t=dotnet:roundtrip_to_amqp_091
```
on branch native-amqp resulted in the following crash:
```
crasher:
initial call: rabbit_writer:enter_mainloop/2
pid: <0.711.0>
registered_name: []
exception error: bad argument
in function size/1
called as size([<<0>>,<<"Sw">>,[<<160,2>>,<<"hi">>]])
*** argument 1: not tuple or binary
in call from rabbit_binary_generator:build_content_frames/7 (rabbit_binary_generator.erl, line 89)
in call from rabbit_binary_generator:build_simple_content_frames/4 (rabbit_binary_generator.erl, line 61)
in call from rabbit_writer:assemble_frames/5 (rabbit_writer.erl, line 334)
in call from rabbit_writer:internal_send_command_async/3 (rabbit_writer.erl, line 365)
in call from rabbit_writer:handle_message/2 (rabbit_writer.erl, line 265)
in call from rabbit_writer:handle_message/3 (rabbit_writer.erl, line 232)
in call from rabbit_writer:mainloop1/2 (rabbit_writer.erl, line 223)
```
because #content.payload_fragments_rev is currently supposed to
be a flat list of binaries instead of being an iolist.
This commit fixes this crash inefficiently by calling
iolist_to_binary/1. A better solution would be to allow AMQP legacy's #content.payload_fragments_rev
to be an iolist.
* Add accidentally deleted line back
* mc: optimise mc_amqp internal format
By removint the outer records for message and delivery annotations
as well as application properties and footers.
* mc: optimis mc_amqp map_add by using upsert
* mc: refactoring and bug fixes
* mc_SUITE routingheader assertions
* mc remove serialize/1 callback as only used by amqp
* mc_amqp: avoid returning a nested list from protocol_state
* test and bug fix
* move infer_type to mc_util
* mc fixes and additiona assertions
* Support headers exchange routing for MQTT messages
When a headers exchange is bound to the MQTT topic exchange, routing
will be performend based on both MQTT topic (by the topic exchange) and
MQTT User Property (by the headers exchange).
This combines the best worlds of both MQTT 5.0 and AMQP 0.9.1 and
enables powerful routing topologies.
When the User Property contains the same name multiple times, only the
last name (and value) will be considered by the headers exchange.
* Fix crash when sending from stream to amqpl
When publishing a message via the stream protocol and consuming it via
AMQP 0.9.1, the following crash occurred prior to this commit:
```
crasher:
initial call: rabbit_channel:init/1
pid: <0.818.0>
registered_name: []
exception exit: {{badmatch,undefined},
[{rabbit_channel,handle_deliver0,4,
[{file,"rabbit_channel.erl"},
{line,2728}]},
{lists,foldl,3,[{file,"lists.erl"},{line,1594}]},
{rabbit_channel,handle_cast,2,
[{file,"rabbit_channel.erl"},
{line,728}]},
{gen_server2,handle_msg,2,
[{file,"gen_server2.erl"},{line,1056}]},
{proc_lib,wake_up,3,
[{file,"proc_lib.erl"},{line,251}]}]}
```
This commit first gives `mc:init/3` the chance to set exchange and
routing_keys annotations.
If not set, `rabbit_stream_queue` will set these annotations assuming
the message was originally published via the stream protocol.
* Support consistent hash exchange routing for MQTT 5.0
When a consistent hash exchange is bound to the MQTT topic exchange,
MQTT 5.0 messages can be routed to queues consistently based on the
Correlation-Data in the PUBLISH packet.
* Convert MQTT 5.0 User Property
* to AMQP 0.9.1 headers
* from AMQP 0.9.1 headers
* to AMQP 1.0 application properties and message annotations
* from AMQP 1.0 application properties and message annotations
* Make use of Annotations in mc_mqtt:protocol_state/2
mc_mqtt:protocol_state/2 includes Annotations as parameter.
It's cleaner to make use of these Annotations when computing the
protocol state instead of relying on the caller (rabbitmq_mqtt_processor)
to compute the protocol state.
* Enforce AMQP 0.9.1 field name length limit
The AMQP 0.9.1 spec prohibits field names longer than 128 characters.
Therefore, when converting AMQP 1.0 message annotations, application
properties or MQTT 5.0 User Property to AMQP 0.9.1 headers, drop any
names longer than 128 characters.
* Fix type specs
Apply feedback from Michael Davis
Co-authored-by: Michael Davis <mcarsondavis@gmail.com>
* Add mc_mqtt unit test suite
Implement mc_mqtt:x_header/2
* Translate indicator that payload is UTF-8 encoded
when converting between MQTT 5.0 and AMQP 1.0
* Translate single amqp-value section from AMQP 1.0 to MQTT
Convert to a text representation, if possible, and indicate to MQTT
client that the payload is UTF-8 encoded. This way, the MQTT client will
be able to parse the payload.
If conversion to text representation is not possible, encode the payload
using the AMQP 1.0 type system and indiate the encoding via Content-Type
message/vnd.rabbitmq.amqp.
This Content-Type is not registered.
Type "message" makes sense since it's a message.
Vendor tree "vnd.rabbitmq.amqp" makes sense since merely subtype "amqp" is not
registered.
* Fix payload conversion
* Translate Response Topic between MQTT and AMQP
Translate MQTT 5.0 Response Topic to AMQP 1.0 reply-to address and vice
versa.
The Response Topic must be a UTF-8 encoded string.
This commit re-uses the already defined RabbitMQ target addresses:
```
"/topic/" RK Publish to amq.topic with routing key RK
"/exchange/" X "/" RK Publish to exchange X with routing key RK
```
By default, the MQTT topic exchange is configure dto be amq.topic using
the 1st target address.
When an operator modifies the mqtt.exchange, the 2nd target address is
used.
* Apply PR feedback
and fix formatting
Co-authored-by: Michael Davis <mcarsondavis@gmail.com>
* tidy up
* Add MQTT message_containers test
* consistent hash exchange: avoid amqp legacy conversion
When hashing on a header value.
* Avoid converting to amqp legacy when using exchange federation
* Fix test flake
* test and dialyzer fixes
* dialyzer fix
* Add MQTT protocol interoperability tests
Test receiving from and sending to MQTT 5.0 and
* AMQP 0.9.1
* AMQP 1.0
* STOMP
* Streams
* Regenerate portions of deps/rabbit/app.bzl with gazelle
I'm not exactly sure how this happened, but gazell seems to have been
run with an older version of the rules_erlang gazelle extension at
some point. This caused generation of a structure that is no longer
used. This commit updates the structure to the current pattern.
* mc: refactoring
* mc_amqpl: handle delivery annotations
Just in case they are included.
Also use iolist_to_iovec to create flat list of binaries when
converting from amqp with amqp encoded payload.
---------
Co-authored-by: David Ansari <david.ansari@gmx.de>
Co-authored-by: Michael Davis <mcarsondavis@gmail.com>
Co-authored-by: Rin Kuryloski <kuryloskip@vmware.com>
2023-08-31 18:27:13 +08:00
|
|
|
add_delivery_count_header(Msg, Count) ->
|
|
|
|
|
case mc:is(Msg) of
|
|
|
|
|
true when is_integer(Count) andalso
|
|
|
|
|
Count > 0 ->
|
|
|
|
|
mc:set_annotation(<<"x-delivery-count">>, Count, Msg);
|
|
|
|
|
_ ->
|
|
|
|
|
Msg
|
|
|
|
|
end.
|
2020-09-29 18:43:24 +08:00
|
|
|
|
|
|
|
|
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
%% @doc Settle a message. Permanently removes message from the queue.
|
|
|
|
|
%% @param ConsumerTag the tag uniquely identifying the consumer.
|
|
|
|
|
%% @param MsgIds the message ids received with the {@link rabbit_fifo:delivery/0.}
|
|
|
|
|
%% @param State the {@module} state
|
|
|
|
|
%%
|
|
|
|
|
-spec settle(rabbit_fifo:consumer_tag(), [rabbit_fifo:msg_id()], state()) ->
|
2020-09-29 18:43:24 +08:00
|
|
|
{state(), list()}.
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
settle(ConsumerTag, [_|_] = MsgIds, #state{slow = false} = State0) ->
|
2020-07-10 17:20:25 +08:00
|
|
|
Node = pick_server(State0),
|
2018-12-14 18:10:29 +08:00
|
|
|
Cmd = rabbit_fifo:make_settle(consumer_id(ConsumerTag), MsgIds),
|
2022-01-26 00:52:01 +08:00
|
|
|
{send_command(Node, undefined, Cmd, normal, State0), []};
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
settle(ConsumerTag, [_|_] = MsgIds,
|
|
|
|
|
#state{unsent_commands = Unsent0} = State0) ->
|
|
|
|
|
ConsumerId = consumer_id(ConsumerTag),
|
|
|
|
|
%% we've reached the soft limit so will stash the command to be
|
|
|
|
|
%% sent once we have seen enough notifications
|
|
|
|
|
Unsent = maps:update_with(ConsumerId,
|
|
|
|
|
fun ({Settles, Returns, Discards}) ->
|
2022-05-12 19:18:40 +08:00
|
|
|
%% MsgIds has fewer elements than Settles.
|
|
|
|
|
%% Therefore put it on the left side of the ++ operator.
|
|
|
|
|
%% The order in which messages are settled does not matter.
|
|
|
|
|
{MsgIds ++ Settles, Returns, Discards}
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
end, {MsgIds, [], []}, Unsent0),
|
2020-09-29 18:43:24 +08:00
|
|
|
{State0#state{unsent_commands = Unsent}, []}.
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
|
|
|
|
|
%% @doc Return a message to the queue.
|
|
|
|
|
%% @param ConsumerTag the tag uniquely identifying the consumer.
|
|
|
|
|
%% @param MsgIds the message ids to return received
|
|
|
|
|
%% from {@link rabbit_fifo:delivery/0.}
|
|
|
|
|
%% @param State the {@module} state
|
|
|
|
|
%% @returns
|
2022-11-02 03:52:40 +08:00
|
|
|
%% `{State, list()}' if the command was successfully sent.
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
%%
|
|
|
|
|
-spec return(rabbit_fifo:consumer_tag(), [rabbit_fifo:msg_id()], state()) ->
|
2020-09-29 18:43:24 +08:00
|
|
|
{state(), list()}.
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
return(ConsumerTag, [_|_] = MsgIds, #state{slow = false} = State0) ->
|
2020-07-10 17:20:25 +08:00
|
|
|
Node = pick_server(State0),
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
% TODO: make rabbit_fifo return support lists of message ids
|
2018-12-14 18:10:29 +08:00
|
|
|
Cmd = rabbit_fifo:make_return(consumer_id(ConsumerTag), MsgIds),
|
2022-01-26 00:52:01 +08:00
|
|
|
{send_command(Node, undefined, Cmd, normal, State0), []};
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
return(ConsumerTag, [_|_] = MsgIds,
|
|
|
|
|
#state{unsent_commands = Unsent0} = State0) ->
|
|
|
|
|
ConsumerId = consumer_id(ConsumerTag),
|
|
|
|
|
%% we've reached the soft limit so will stash the command to be
|
|
|
|
|
%% sent once we have seen enough notifications
|
|
|
|
|
Unsent = maps:update_with(ConsumerId,
|
|
|
|
|
fun ({Settles, Returns, Discards}) ->
|
|
|
|
|
{Settles, Returns ++ MsgIds, Discards}
|
|
|
|
|
end, {[], MsgIds, []}, Unsent0),
|
2021-11-17 03:29:22 +08:00
|
|
|
State1 = State0#state{unsent_commands = Unsent},
|
|
|
|
|
{State1, []}.
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
|
|
|
|
|
%% @doc Discards a checked out message.
|
|
|
|
|
%% If the queue has a dead_letter_handler configured this will be called.
|
|
|
|
|
%% @param ConsumerTag the tag uniquely identifying the consumer.
|
|
|
|
|
%% @param MsgIds the message ids to discard
|
|
|
|
|
%% from {@link rabbit_fifo:delivery/0.}
|
|
|
|
|
%% @param State the {@module} state
|
|
|
|
|
-spec discard(rabbit_fifo:consumer_tag(), [rabbit_fifo:msg_id()], state()) ->
|
2020-09-29 18:43:24 +08:00
|
|
|
{state(), list()}.
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
discard(ConsumerTag, [_|_] = MsgIds, #state{slow = false} = State0) ->
|
2020-07-10 17:20:25 +08:00
|
|
|
Node = pick_server(State0),
|
2018-12-14 18:10:29 +08:00
|
|
|
Cmd = rabbit_fifo:make_discard(consumer_id(ConsumerTag), MsgIds),
|
2022-01-26 00:52:01 +08:00
|
|
|
{send_command(Node, undefined, Cmd, normal, State0), []};
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
discard(ConsumerTag, [_|_] = MsgIds,
|
|
|
|
|
#state{unsent_commands = Unsent0} = State0) ->
|
|
|
|
|
ConsumerId = consumer_id(ConsumerTag),
|
|
|
|
|
%% we've reached the soft limit so will stash the command to be
|
|
|
|
|
%% sent once we have seen enough notifications
|
|
|
|
|
Unsent = maps:update_with(ConsumerId,
|
|
|
|
|
fun ({Settles, Returns, Discards}) ->
|
|
|
|
|
{Settles, Returns, Discards ++ MsgIds}
|
|
|
|
|
end, {[], [], MsgIds}, Unsent0),
|
2020-09-29 18:43:24 +08:00
|
|
|
{State0#state{unsent_commands = Unsent}, []}.
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
|
|
|
|
|
%% @doc Register with the rabbit_fifo queue to "checkout" messages as they
|
|
|
|
|
%% become available.
|
|
|
|
|
%%
|
2019-02-13 01:28:22 +08:00
|
|
|
%% This is a synchronous call. I.e. the call will block until the command
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
%% has been accepted by the ra process or it times out.
|
|
|
|
|
%%
|
|
|
|
|
%% @param ConsumerTag a unique tag to identify this particular consumer.
|
|
|
|
|
%% @param NumUnsettled the maximum number of in-flight messages. Once this
|
|
|
|
|
%% number of messages has been received but not settled no further messages
|
|
|
|
|
%% will be delivered to the consumer.
|
|
|
|
|
%% @param CreditMode The credit mode to use for the checkout.
|
|
|
|
|
%% simple_prefetch: credit is auto topped up as deliveries are settled
|
|
|
|
|
%% credited: credit is only increased by sending credit to the queue
|
|
|
|
|
%% @param State The {@module} state.
|
|
|
|
|
%%
|
|
|
|
|
%% @returns `{ok, State}' or `{error | timeout, term()}'
|
2018-12-14 18:10:29 +08:00
|
|
|
-spec checkout(rabbit_fifo:consumer_tag(),
|
|
|
|
|
NumUnsettled :: non_neg_integer(),
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
CreditMode :: rabbit_fifo:credit_mode(),
|
2018-12-14 18:10:29 +08:00
|
|
|
Meta :: rabbit_fifo:consumer_meta(),
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
state()) -> {ok, state()} | {error | timeout, term()}.
|
2020-09-29 18:43:24 +08:00
|
|
|
checkout(ConsumerTag, NumUnsettled, CreditMode, Meta,
|
|
|
|
|
#state{consumer_deliveries = CDels0} = State0) ->
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
Servers = sorted_servers(State0),
|
|
|
|
|
ConsumerId = {ConsumerTag, self()},
|
2018-12-14 18:10:29 +08:00
|
|
|
Cmd = rabbit_fifo:make_checkout(ConsumerId,
|
|
|
|
|
{auto, NumUnsettled, CreditMode},
|
|
|
|
|
Meta),
|
2020-09-29 18:43:24 +08:00
|
|
|
%% ???
|
|
|
|
|
Ack = maps:get(ack, Meta, true),
|
|
|
|
|
|
2022-09-30 23:20:50 +08:00
|
|
|
case try_process_command(Servers, Cmd, State0) of
|
|
|
|
|
{ok, Reply, Leader} ->
|
|
|
|
|
LastMsgId = case Reply of
|
|
|
|
|
ok ->
|
|
|
|
|
%% this is the pre 3.11.1 / 3.10.9
|
|
|
|
|
%% reply format
|
|
|
|
|
-1;
|
2023-08-23 17:56:30 +08:00
|
|
|
{ok, #{num_checked_out := NumChecked,
|
|
|
|
|
next_msg_id := NextMsgId}} ->
|
|
|
|
|
case NumChecked > 0 of
|
|
|
|
|
true ->
|
|
|
|
|
%% we cannot know if the pending messages
|
|
|
|
|
%% have been delivered to the client or they
|
|
|
|
|
%% are on their way to the current process.
|
|
|
|
|
%% We set `undefined' to signal this uncertainty
|
|
|
|
|
%% and will just accept the next arriving message
|
|
|
|
|
%% irrespective of message id
|
|
|
|
|
undefined;
|
|
|
|
|
false ->
|
|
|
|
|
NextMsgId - 1
|
|
|
|
|
end
|
2022-09-30 23:20:50 +08:00
|
|
|
end,
|
|
|
|
|
SDels = maps:update_with(
|
|
|
|
|
ConsumerTag, fun (C) -> C#consumer{ack = Ack} end,
|
|
|
|
|
#consumer{last_msg_id = LastMsgId,
|
|
|
|
|
ack = Ack}, CDels0),
|
|
|
|
|
{ok, State0#state{leader = Leader,
|
|
|
|
|
consumer_deliveries = SDels}};
|
|
|
|
|
Err ->
|
|
|
|
|
Err
|
|
|
|
|
end.
|
2021-11-04 00:26:40 +08:00
|
|
|
|
|
|
|
|
-spec query_single_active_consumer(state()) ->
|
|
|
|
|
{ok, term()} | {error, term()} | {timeout, term()}.
|
|
|
|
|
query_single_active_consumer(#state{leader = undefined}) ->
|
|
|
|
|
{error, leader_not_known};
|
|
|
|
|
query_single_active_consumer(#state{leader = Leader}) ->
|
|
|
|
|
case ra:local_query(Leader, fun rabbit_fifo:query_single_active_consumer/1,
|
|
|
|
|
?COMMAND_TIMEOUT) of
|
|
|
|
|
{ok, {_, Reply}, _} ->
|
|
|
|
|
{ok, Reply};
|
|
|
|
|
Err ->
|
|
|
|
|
Err
|
|
|
|
|
end.
|
|
|
|
|
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
%% @doc Provide credit to the queue
|
|
|
|
|
%%
|
|
|
|
|
%% This only has an effect if the consumer uses credit mode: credited
|
|
|
|
|
%% @param ConsumerTag a unique tag to identify this particular consumer.
|
|
|
|
|
%% @param Credit the amount of credit to provide to theq queue
|
|
|
|
|
%% @param Drain tells the queue to use up any credit that cannot be immediately
|
|
|
|
|
%% fulfilled. (i.e. there are not enough messages on queue to use up all the
|
|
|
|
|
%% provided credit).
|
|
|
|
|
-spec credit(rabbit_fifo:consumer_tag(),
|
|
|
|
|
Credit :: non_neg_integer(),
|
|
|
|
|
Drain :: boolean(),
|
|
|
|
|
state()) ->
|
2020-11-12 19:33:56 +08:00
|
|
|
{state(), actions()}.
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
credit(ConsumerTag, Credit, Drain,
|
|
|
|
|
#state{consumer_deliveries = CDels} = State0) ->
|
|
|
|
|
ConsumerId = consumer_id(ConsumerTag),
|
|
|
|
|
%% the last received msgid provides us with the delivery count if we
|
|
|
|
|
%% add one as it is 0 indexed
|
|
|
|
|
C = maps:get(ConsumerTag, CDels, #consumer{last_msg_id = -1}),
|
2020-07-10 17:20:25 +08:00
|
|
|
Node = pick_server(State0),
|
2018-12-14 18:10:29 +08:00
|
|
|
Cmd = rabbit_fifo:make_credit(ConsumerId, Credit,
|
|
|
|
|
C#consumer.last_msg_id + 1, Drain),
|
2022-01-26 00:52:01 +08:00
|
|
|
{send_command(Node, undefined, Cmd, normal, State0), []}.
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
|
|
|
|
|
%% @doc Cancels a checkout with the rabbit_fifo queue for the consumer tag
|
|
|
|
|
%%
|
2019-02-13 01:28:22 +08:00
|
|
|
%% This is a synchronous call. I.e. the call will block until the command
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
%% has been accepted by the ra process or it times out.
|
|
|
|
|
%%
|
|
|
|
|
%% @param ConsumerTag a unique tag to identify this particular consumer.
|
|
|
|
|
%% @param State The {@module} state.
|
|
|
|
|
%%
|
|
|
|
|
%% @returns `{ok, State}' or `{error | timeout, term()}'
|
|
|
|
|
-spec cancel_checkout(rabbit_fifo:consumer_tag(), state()) ->
|
|
|
|
|
{ok, state()} | {error | timeout, term()}.
|
|
|
|
|
cancel_checkout(ConsumerTag, #state{consumer_deliveries = CDels} = State0) ->
|
|
|
|
|
Servers = sorted_servers(State0),
|
|
|
|
|
ConsumerId = {ConsumerTag, self()},
|
2018-12-14 18:10:29 +08:00
|
|
|
Cmd = rabbit_fifo:make_checkout(ConsumerId, cancel, #{}),
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
State = State0#state{consumer_deliveries = maps:remove(ConsumerTag, CDels)},
|
2022-09-30 23:20:50 +08:00
|
|
|
case try_process_command(Servers, Cmd, State) of
|
|
|
|
|
{ok, _, Leader} ->
|
|
|
|
|
{ok, State#state{leader = Leader}};
|
|
|
|
|
Err ->
|
|
|
|
|
Err
|
|
|
|
|
end.
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
|
|
|
|
|
%% @doc Purges all the messages from a rabbit_fifo queue and returns the number
|
|
|
|
|
%% of messages purged.
|
2019-08-30 17:45:15 +08:00
|
|
|
-spec purge(ra:server_id()) -> {ok, non_neg_integer()} | {error | timeout, term()}.
|
2021-11-04 00:26:40 +08:00
|
|
|
purge(Server) ->
|
|
|
|
|
case ra:process_command(Server, rabbit_fifo:make_purge(), ?COMMAND_TIMEOUT) of
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
{ok, {purge, Reply}, _} ->
|
|
|
|
|
{ok, Reply};
|
|
|
|
|
Err ->
|
|
|
|
|
Err
|
|
|
|
|
end.
|
|
|
|
|
|
2019-06-04 18:51:25 +08:00
|
|
|
-spec pending_size(state()) -> non_neg_integer().
|
|
|
|
|
pending_size(#state{pending = Pend}) ->
|
|
|
|
|
maps:size(Pend).
|
|
|
|
|
|
2019-08-30 17:45:15 +08:00
|
|
|
-spec stat(ra:server_id()) ->
|
2019-01-29 00:00:07 +08:00
|
|
|
{ok, non_neg_integer(), non_neg_integer()}
|
|
|
|
|
| {error | timeout, term()}.
|
2019-01-10 16:59:41 +08:00
|
|
|
stat(Leader) ->
|
2019-01-29 00:00:07 +08:00
|
|
|
%% short timeout as we don't want to spend too long if it is going to
|
|
|
|
|
%% fail anyway
|
2019-10-25 13:36:48 +08:00
|
|
|
stat(Leader, 250).
|
|
|
|
|
|
|
|
|
|
-spec stat(ra:server_id(), non_neg_integer()) ->
|
|
|
|
|
{ok, non_neg_integer(), non_neg_integer()}
|
|
|
|
|
| {error | timeout, term()}.
|
|
|
|
|
stat(Leader, Timeout) ->
|
|
|
|
|
%% short timeout as we don't want to spend too long if it is going to
|
|
|
|
|
%% fail anyway
|
|
|
|
|
case ra:local_query(Leader, fun rabbit_fifo:query_stat/1, Timeout) of
|
|
|
|
|
{ok, {_, {R, C}}, _} -> {ok, R, C};
|
|
|
|
|
{error, _} = Error -> Error;
|
|
|
|
|
{timeout, _} = Error -> Error
|
|
|
|
|
end.
|
2019-01-04 15:46:15 +08:00
|
|
|
|
2020-07-10 17:20:25 +08:00
|
|
|
update_machine_state(Server, Conf) ->
|
2021-11-04 00:26:40 +08:00
|
|
|
case ra:process_command(Server, rabbit_fifo:make_update_config(Conf), ?COMMAND_TIMEOUT) of
|
2018-11-08 03:14:50 +08:00
|
|
|
{ok, ok, _} ->
|
|
|
|
|
ok;
|
|
|
|
|
Err ->
|
|
|
|
|
Err
|
|
|
|
|
end.
|
|
|
|
|
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
%% @doc Handles incoming `ra_events'. Events carry both internal "bookeeping"
|
|
|
|
|
%% events emitted by the `ra' leader as well as `rabbit_fifo' emitted events such
|
|
|
|
|
%% as message deliveries. All ra events need to be handled by {@module}
|
|
|
|
|
%% to ensure bookeeping, resends and flow control is correctly handled.
|
|
|
|
|
%%
|
|
|
|
|
%% If the `ra_event' contains a `rabbit_fifo' generated message it will be returned
|
|
|
|
|
%% for further processing.
|
|
|
|
|
%%
|
|
|
|
|
%% Example:
|
|
|
|
|
%%
|
|
|
|
|
%% ```
|
|
|
|
|
%% receive
|
|
|
|
|
%% {ra_event, From, Evt} ->
|
|
|
|
|
%% case rabbit_fifo_client:handle_ra_event(From, Evt, State0) of
|
|
|
|
|
%% {internal, _Seq, State} -> State;
|
|
|
|
|
%% {{delivery, _ConsumerTag, Msgs}, State} ->
|
|
|
|
|
%% handle_messages(Msgs),
|
|
|
|
|
%% ...
|
|
|
|
|
%% end
|
|
|
|
|
%% end
|
|
|
|
|
%% '''
|
|
|
|
|
%%
|
2023-01-19 22:33:28 +08:00
|
|
|
%% @param QName Name of the queue.
|
2019-08-30 17:45:15 +08:00
|
|
|
%% @param From the {@link ra:server_id().} of the sending process.
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
%% @param Event the body of the `ra_event'.
|
|
|
|
|
%% @param State the current {@module} state.
|
|
|
|
|
%%
|
|
|
|
|
%% @returns
|
|
|
|
|
%% `{internal, AppliedCorrelations, State}' if the event contained an internally
|
|
|
|
|
%% handled event such as a notification and a correlation was included with
|
|
|
|
|
%% the command (e.g. in a call to `enqueue/3' the correlation terms are returned
|
2021-10-08 23:02:19 +08:00
|
|
|
%% here).
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
%%
|
|
|
|
|
%% `{RaFifoEvent, State}' if the event contained a client message generated by
|
|
|
|
|
%% the `rabbit_fifo' state machine such as a delivery.
|
|
|
|
|
%%
|
|
|
|
|
%% The type of `rabbit_fifo' client messages that can be received are:
|
|
|
|
|
%%
|
|
|
|
|
%% `{delivery, ConsumerTag, [{MsgId, {MsgHeader, Msg}}]}'
|
|
|
|
|
%%
|
|
|
|
|
%% <li>`ConsumerTag' the binary tag passed to {@link checkout/3.}</li>
|
|
|
|
|
%% <li>`MsgId' is a consumer scoped monotonically incrementing id that can be
|
|
|
|
|
%% used to {@link settle/3.} (roughly: AMQP 0.9.1 ack) message once finished
|
|
|
|
|
%% with them.</li>
|
2023-01-19 22:33:28 +08:00
|
|
|
-spec handle_ra_event(rabbit_amqqueue:name(), ra:server_id(),
|
|
|
|
|
ra_server_proc:ra_event_body(), state()) ->
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
{internal, Correlators :: [term()], actions(), state()} |
|
2022-11-18 04:13:27 +08:00
|
|
|
{rabbit_fifo:client_msg(), state()} | {eol, actions()}.
|
2023-01-19 22:33:28 +08:00
|
|
|
handle_ra_event(QName, From, {applied, Seqs},
|
|
|
|
|
#state{cfg = #cfg{soft_limit = SftLmt}} = State0) ->
|
2020-09-29 18:43:24 +08:00
|
|
|
|
|
|
|
|
{Corrs, Actions0, State1} = lists:foldl(fun seq_applied/2,
|
|
|
|
|
{[], [], State0#state{leader = From}},
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
Seqs),
|
2020-09-29 18:43:24 +08:00
|
|
|
Actions = case Corrs of
|
|
|
|
|
[] ->
|
|
|
|
|
lists:reverse(Actions0);
|
|
|
|
|
_ ->
|
2022-11-12 01:56:32 +08:00
|
|
|
%%TODO consider using lists:foldr/3 above because
|
|
|
|
|
%% Corrs is returned in the wrong order here.
|
|
|
|
|
%% The wrong order does not matter much because the channel sorts the
|
|
|
|
|
%% sequence numbers before confirming to the client. But rabbit_fifo_client
|
|
|
|
|
%% is sequence numer agnostic: it handles any correlation terms.
|
2023-01-19 22:33:28 +08:00
|
|
|
[{settled, QName, Corrs}
|
2020-09-29 18:43:24 +08:00
|
|
|
| lists:reverse(Actions0)]
|
|
|
|
|
end,
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
case maps:size(State1#state.pending) < SftLmt of
|
|
|
|
|
true when State1#state.slow == true ->
|
|
|
|
|
% we have exited soft limit state
|
2019-02-14 18:49:25 +08:00
|
|
|
% send any unsent commands and cancel the time as
|
|
|
|
|
% TODO: really the timer should only be cancelled when the channel
|
|
|
|
|
% exits flow state (which depends on the state of all queues the
|
|
|
|
|
% channel is interacting with)
|
|
|
|
|
% but the fact the queue has just applied suggests
|
|
|
|
|
% it's ok to cancel here anyway
|
|
|
|
|
State2 = cancel_timer(State1#state{slow = false,
|
|
|
|
|
unsent_commands = #{}}),
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
% build up a list of commands to issue
|
|
|
|
|
Commands = maps:fold(
|
|
|
|
|
fun (Cid, {Settled, Returns, Discards}, Acc) ->
|
|
|
|
|
add_command(Cid, settle, Settled,
|
|
|
|
|
add_command(Cid, return, Returns,
|
2018-11-27 18:23:25 +08:00
|
|
|
add_command(Cid, discard,
|
|
|
|
|
Discards, Acc)))
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
end, [], State1#state.unsent_commands),
|
2020-07-10 17:20:25 +08:00
|
|
|
Node = pick_server(State2),
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
%% send all the settlements and returns
|
|
|
|
|
State = lists:foldl(fun (C, S0) ->
|
2022-01-26 00:52:01 +08:00
|
|
|
send_command(Node, undefined, C,
|
|
|
|
|
normal, S0)
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
end, State2, Commands),
|
2022-11-02 03:52:40 +08:00
|
|
|
{ok, State, [{unblock, cluster_name(State)} | Actions]};
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
_ ->
|
2020-09-29 18:43:24 +08:00
|
|
|
{ok, State1, Actions}
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
end;
|
2023-01-19 22:33:28 +08:00
|
|
|
handle_ra_event(QName, From, {machine, {delivery, _ConsumerTag, _} = Del}, State0) ->
|
|
|
|
|
handle_delivery(QName, From, Del, State0);
|
|
|
|
|
handle_ra_event(_QName, _, {machine, {queue_status, Status}},
|
2020-07-10 17:20:25 +08:00
|
|
|
#state{} = State) ->
|
|
|
|
|
%% just set the queue status
|
2020-10-02 20:37:25 +08:00
|
|
|
{ok, State#state{queue_status = Status}, []};
|
2023-01-19 22:33:28 +08:00
|
|
|
handle_ra_event(_QName, Leader, {machine, leader_change},
|
2021-11-29 21:19:58 +08:00
|
|
|
#state{leader = OldLeader} = State0) ->
|
2018-11-29 21:57:45 +08:00
|
|
|
%% we need to update leader
|
|
|
|
|
%% and resend any pending commands
|
2022-10-08 06:59:05 +08:00
|
|
|
rabbit_log:debug("~ts: Detected QQ leader change from ~w to ~w",
|
2021-11-29 21:19:58 +08:00
|
|
|
[?MODULE, OldLeader, Leader]),
|
|
|
|
|
State = resend_all_pending(State0#state{leader = Leader}),
|
|
|
|
|
{ok, State, []};
|
2023-01-19 22:33:28 +08:00
|
|
|
handle_ra_event(_QName, _From, {rejected, {not_leader, Leader, _Seq}},
|
2021-11-29 21:19:58 +08:00
|
|
|
#state{leader = Leader} = State) ->
|
|
|
|
|
{ok, State, []};
|
2023-01-19 22:33:28 +08:00
|
|
|
handle_ra_event(_QName, _From, {rejected, {not_leader, Leader, _Seq}},
|
2021-11-29 21:19:58 +08:00
|
|
|
#state{leader = OldLeader} = State0) ->
|
2022-10-08 06:59:05 +08:00
|
|
|
rabbit_log:debug("~ts: Detected QQ leader change (rejection) from ~w to ~w",
|
2021-11-29 21:19:58 +08:00
|
|
|
[?MODULE, OldLeader, Leader]),
|
2018-11-29 21:57:45 +08:00
|
|
|
State = resend_all_pending(State0#state{leader = Leader}),
|
2021-11-01 20:39:24 +08:00
|
|
|
{ok, cancel_timer(State), []};
|
2023-01-19 22:33:28 +08:00
|
|
|
handle_ra_event(_QName, _From, {rejected, {not_leader, _UndefinedMaybe, _Seq}}, State0) ->
|
2020-09-29 18:43:24 +08:00
|
|
|
% TODO: how should these be handled? re-sent on timer or try random
|
|
|
|
|
{ok, State0, []};
|
2023-01-19 22:33:28 +08:00
|
|
|
handle_ra_event(QName, _, timeout, #state{cfg = #cfg{servers = Servers}} = State0) ->
|
2019-02-14 18:49:25 +08:00
|
|
|
case find_leader(Servers) of
|
|
|
|
|
undefined ->
|
|
|
|
|
%% still no leader, set the timer again
|
2023-01-19 22:33:28 +08:00
|
|
|
{ok, set_timer(QName, State0), []};
|
2019-02-14 18:49:25 +08:00
|
|
|
Leader ->
|
|
|
|
|
State = resend_all_pending(State0#state{leader = Leader}),
|
2020-09-29 18:43:24 +08:00
|
|
|
{ok, State, []}
|
2019-02-14 18:49:25 +08:00
|
|
|
end;
|
2023-01-19 22:33:28 +08:00
|
|
|
handle_ra_event(_QName, _Leader, {machine, eol}, State) ->
|
2022-11-18 04:13:27 +08:00
|
|
|
{eol, [{unblock, cluster_name(State)}]}.
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
|
|
|
|
|
%% @doc Attempts to enqueue a message using cast semantics. This provides no
|
|
|
|
|
%% guarantees or retries if the message fails to achieve consensus or if the
|
|
|
|
|
%% servers sent to happens not to be available. If the message is sent to a
|
|
|
|
|
%% follower it will attempt the deliver it to the leader, if known. Else it will
|
|
|
|
|
%% drop the messages.
|
|
|
|
|
%%
|
|
|
|
|
%% NB: only use this for non-critical enqueues where a full rabbit_fifo_client state
|
|
|
|
|
%% cannot be maintained.
|
|
|
|
|
%%
|
|
|
|
|
%% @param CusterId the cluster id.
|
|
|
|
|
%% @param Servers the known servers in the cluster.
|
|
|
|
|
%% @param Msg the message to enqueue.
|
|
|
|
|
%%
|
|
|
|
|
%% @returns `ok'
|
2019-08-30 17:45:15 +08:00
|
|
|
-spec untracked_enqueue([ra:server_id()], term()) ->
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
ok.
|
|
|
|
|
untracked_enqueue([Node | _], Msg) ->
|
2018-12-14 18:10:29 +08:00
|
|
|
Cmd = rabbit_fifo:make_enqueue(undefined, undefined, Msg),
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
ok = ra:pipeline_command(Node, Cmd),
|
|
|
|
|
ok.
|
|
|
|
|
|
|
|
|
|
%% Internal
|
|
|
|
|
|
2021-11-04 00:26:40 +08:00
|
|
|
try_process_command([Server | Rem], Cmd,
|
|
|
|
|
#state{cfg = #cfg{timeout = Timeout}} = State) ->
|
|
|
|
|
case ra:process_command(Server, Cmd, Timeout) of
|
2022-09-30 23:20:50 +08:00
|
|
|
{ok, _, _} = Res ->
|
|
|
|
|
Res;
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
Err when length(Rem) =:= 0 ->
|
|
|
|
|
Err;
|
|
|
|
|
_ ->
|
|
|
|
|
try_process_command(Rem, Cmd, State)
|
|
|
|
|
end.
|
|
|
|
|
|
2021-11-29 21:19:58 +08:00
|
|
|
seq_applied({Seq, Response},
|
|
|
|
|
{Corrs, Actions0, #state{} = State0}) ->
|
|
|
|
|
%% sequences aren't guaranteed to be applied in order as enqueues are
|
|
|
|
|
%% low priority commands and may be overtaken by others with a normal priority.
|
|
|
|
|
{Actions, State} = maybe_add_action(Response, Actions0, State0),
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
case maps:take(Seq, State#state.pending) of
|
|
|
|
|
{{undefined, _}, Pending} ->
|
2021-11-29 21:19:58 +08:00
|
|
|
{Corrs, Actions, State#state{pending = Pending}};
|
|
|
|
|
{{Corr, _}, Pending}
|
|
|
|
|
when Response /= not_enqueued ->
|
|
|
|
|
{[Corr | Corrs], Actions, State#state{pending = Pending}};
|
|
|
|
|
_ ->
|
|
|
|
|
{Corrs, Actions, State#state{}}
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
end;
|
|
|
|
|
seq_applied(_Seq, Acc) ->
|
|
|
|
|
Acc.
|
|
|
|
|
|
|
|
|
|
maybe_add_action(ok, Acc, State) ->
|
|
|
|
|
{Acc, State};
|
2021-11-29 21:19:58 +08:00
|
|
|
maybe_add_action(not_enqueued, Acc, State) ->
|
|
|
|
|
{Acc, State};
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
maybe_add_action({multi, Actions}, Acc0, State0) ->
|
|
|
|
|
lists:foldl(fun (Act, {Acc, State}) ->
|
|
|
|
|
maybe_add_action(Act, Acc, State)
|
|
|
|
|
end, {Acc0, State0}, Actions);
|
|
|
|
|
maybe_add_action({send_drained, {Tag, Credit}} = Action, Acc,
|
|
|
|
|
#state{consumer_deliveries = CDels} = State) ->
|
|
|
|
|
%% add credit to consumer delivery_count
|
|
|
|
|
C = maps:get(Tag, CDels),
|
|
|
|
|
{[Action | Acc],
|
|
|
|
|
State#state{consumer_deliveries =
|
|
|
|
|
update_consumer(Tag, C#consumer.last_msg_id,
|
|
|
|
|
Credit, C, CDels)}};
|
|
|
|
|
maybe_add_action(Action, Acc, State) ->
|
|
|
|
|
%% anything else is assumed to be an action
|
|
|
|
|
{[Action | Acc], State}.
|
|
|
|
|
|
|
|
|
|
% resends a command with a new sequence number
|
|
|
|
|
resend(OldSeq, #state{pending = Pending0, leader = Leader} = State) ->
|
|
|
|
|
case maps:take(OldSeq, Pending0) of
|
|
|
|
|
{{Corr, Cmd}, Pending} ->
|
|
|
|
|
%% resends aren't subject to flow control here
|
|
|
|
|
resend_command(Leader, Corr, Cmd, State#state{pending = Pending});
|
|
|
|
|
error ->
|
|
|
|
|
State
|
|
|
|
|
end.
|
2020-07-10 21:25:17 +08:00
|
|
|
|
2018-11-29 21:57:45 +08:00
|
|
|
resend_all_pending(#state{pending = Pend} = State) ->
|
|
|
|
|
Seqs = lists:sort(maps:keys(Pend)),
|
|
|
|
|
lists:foldl(fun resend/2, State, Seqs).
|
|
|
|
|
|
2020-09-29 18:43:24 +08:00
|
|
|
maybe_auto_ack(true, Deliver, State0) ->
|
|
|
|
|
%% manual ack is enabled
|
|
|
|
|
{ok, State0, [Deliver]};
|
|
|
|
|
maybe_auto_ack(false, {deliver, Tag, _Ack, Msgs} = Deliver, State0) ->
|
|
|
|
|
%% we have to auto ack these deliveries
|
|
|
|
|
MsgIds = [I || {_, _, I, _, _} <- Msgs],
|
|
|
|
|
{State, Actions} = settle(Tag, MsgIds, State0),
|
|
|
|
|
{ok, State, [Deliver] ++ Actions}.
|
|
|
|
|
|
2023-01-19 22:33:28 +08:00
|
|
|
handle_delivery(QName, Leader, {delivery, Tag, [{FstId, _} | _] = IdMsgs},
|
|
|
|
|
#state{consumer_deliveries = CDels0} = State0)
|
2021-11-17 03:29:22 +08:00
|
|
|
when is_map_key(Tag, CDels0) ->
|
2020-09-29 18:43:24 +08:00
|
|
|
QRef = qref(Leader),
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
{LastId, _} = lists:last(IdMsgs),
|
2020-09-29 18:43:24 +08:00
|
|
|
Consumer = #consumer{ack = Ack} = maps:get(Tag, CDels0),
|
|
|
|
|
%% format as a deliver action
|
|
|
|
|
Del = {deliver, Tag, Ack, transform_msgs(QName, QRef, IdMsgs)},
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
%% TODO: remove potential default allocation
|
2020-09-29 18:43:24 +08:00
|
|
|
case Consumer of
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
#consumer{last_msg_id = Prev} = C
|
2023-08-23 17:56:30 +08:00
|
|
|
when Prev =:= undefined orelse FstId =:= Prev+1 ->
|
|
|
|
|
%% Prev =:= undefined is a special case where a checkout was done
|
|
|
|
|
%% for a previously cancelled consumer that still had pending messages
|
|
|
|
|
%% In this case we can't reliably know what the next expected message
|
|
|
|
|
%% id should be so have to accept whatever message comes next
|
2020-09-29 18:43:24 +08:00
|
|
|
maybe_auto_ack(Ack, Del,
|
|
|
|
|
State0#state{consumer_deliveries =
|
|
|
|
|
update_consumer(Tag, LastId,
|
|
|
|
|
length(IdMsgs), C,
|
|
|
|
|
CDels0)});
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
#consumer{last_msg_id = Prev} = C
|
|
|
|
|
when FstId > Prev+1 ->
|
2018-12-18 20:46:00 +08:00
|
|
|
NumMissing = FstId - Prev + 1,
|
|
|
|
|
%% there may actually be fewer missing messages returned than expected
|
|
|
|
|
%% This can happen when a node the channel is on gets disconnected
|
|
|
|
|
%% from the node the leader is on and then reconnected afterwards.
|
|
|
|
|
%% When the node is disconnected the leader will return all checked
|
|
|
|
|
%% out messages to the main queue to ensure they don't get stuck in
|
|
|
|
|
%% case the node never comes back.
|
2022-10-03 21:55:52 +08:00
|
|
|
case get_missing_deliveries(State0, Prev+1, FstId-1, Tag) of
|
2020-11-13 19:04:32 +08:00
|
|
|
{protocol_error, _, _, _} = Err ->
|
|
|
|
|
Err;
|
|
|
|
|
Missing ->
|
|
|
|
|
XDel = {deliver, Tag, Ack, transform_msgs(QName, QRef,
|
|
|
|
|
Missing ++ IdMsgs)},
|
|
|
|
|
maybe_auto_ack(Ack, XDel,
|
|
|
|
|
State0#state{consumer_deliveries =
|
|
|
|
|
update_consumer(Tag, LastId,
|
|
|
|
|
length(IdMsgs) + NumMissing,
|
|
|
|
|
C, CDels0)})
|
|
|
|
|
end;
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
#consumer{last_msg_id = Prev}
|
|
|
|
|
when FstId =< Prev ->
|
|
|
|
|
case lists:dropwhile(fun({Id, _}) -> Id =< Prev end, IdMsgs) of
|
|
|
|
|
[] ->
|
2020-09-29 18:43:24 +08:00
|
|
|
{ok, State0, []};
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
IdMsgs2 ->
|
2023-01-19 22:33:28 +08:00
|
|
|
handle_delivery(QName, Leader, {delivery, Tag, IdMsgs2}, State0)
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
end;
|
2020-09-29 18:43:24 +08:00
|
|
|
C when FstId =:= 0 ->
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
% the very first delivery
|
2020-09-29 18:43:24 +08:00
|
|
|
maybe_auto_ack(Ack, Del,
|
|
|
|
|
State0#state{consumer_deliveries =
|
|
|
|
|
update_consumer(Tag, LastId,
|
|
|
|
|
length(IdMsgs),
|
|
|
|
|
C#consumer{last_msg_id = LastId},
|
|
|
|
|
CDels0)})
|
2021-11-17 03:29:22 +08:00
|
|
|
end;
|
2023-01-19 22:33:28 +08:00
|
|
|
handle_delivery(_QName, _Leader, {delivery, Tag, [_ | _] = IdMsgs},
|
2021-11-17 03:29:22 +08:00
|
|
|
#state{consumer_deliveries = CDels0} = State0)
|
|
|
|
|
when not is_map_key(Tag, CDels0) ->
|
|
|
|
|
%% Note:
|
|
|
|
|
%% https://github.com/rabbitmq/rabbitmq-server/issues/3729
|
|
|
|
|
%% If the consumer is no longer in the deliveries map,
|
|
|
|
|
%% we should return all messages.
|
|
|
|
|
MsgIntIds = [Id || {Id, _} <- IdMsgs],
|
|
|
|
|
{State1, Deliveries} = return(Tag, MsgIntIds, State0),
|
|
|
|
|
{ok, State1, Deliveries}.
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
|
2020-09-29 18:43:24 +08:00
|
|
|
transform_msgs(QName, QRef, Msgs) ->
|
|
|
|
|
lists:map(
|
|
|
|
|
fun({MsgId, {MsgHeader, Msg0}}) ->
|
|
|
|
|
{Msg, Redelivered} = case MsgHeader of
|
|
|
|
|
#{delivery_count := C} ->
|
|
|
|
|
{add_delivery_count_header(Msg0, C), true};
|
|
|
|
|
_ ->
|
|
|
|
|
{Msg0, false}
|
|
|
|
|
end,
|
Message Containers (#5077)
This PR implements an approach for a "protocol (data format) agnostic core" where the format of the message isn't converted at point of reception.
Currently all non AMQP 0.9.1 originating messages are converted into a AMQP 0.9.1 flavoured basic_message record before sent to a queue. If the messages are then consumed by the originating protocol they are converted back from AMQP 0.9.1. For some protocols such as MQTT 3.1 this isn't too expensive as MQTT is mostly a fairly easily mapped subset of AMQP 0.9.1 but for others such as AMQP 1.0 the conversions are awkward and in some cases lossy even if consuming from the originating protocol.
This PR instead wraps all incoming messages in their originating form into a generic, extensible message container type (mc). The container module exposes an API to get common message details such as size and various properties (ttl, priority etc) directly from the source data type. Each protocol needs to implement the mc behaviour such that when a message originating form one protocol is consumed by another protocol we convert it to the target protocol at that point.
The message container also contains annotations, dead letter records and other meta data we need to record during the lifetime of a message. The original protocol message is never modified unless it is consumed.
This includes conversion modules to and from amqp, amqpl (AMQP 0.9.1) and mqtt.
COMMIT HISTORY:
* Refactor away from using the delivery{} record
In many places including exchange types. This should make it
easier to move towards using a message container type instead of
basic_message.
Add mc module and move direct replies outside of exchange
Lots of changes incl classic queues
Implement stream support incl amqp conversions
simplify mc state record
move mc.erl
mc dlx stuff
recent history exchange
Make tracking work
But doesn't take a protocol agnostic approach as we just convert
everything into AMQP legacy and back. Might be good enough for now.
Tracing as a whole may want a bit of a re-vamp at some point.
tidy
make quorum queue peek work by legacy conversion
dead lettering fixes
dead lettering fixes
CMQ fixes
rabbit_trace type fixes
fixes
fix
Fix classic queue props
test assertion fix
feature flag and backwards compat
Enable message_container feature flag in some SUITEs
Dialyzer fixes
fixes
fix
test fixes
Various
Manually update a gazelle generated file
until a gazelle enhancement can be made
https://github.com/rabbitmq/rules_erlang/issues/185
Add message_containers_SUITE to bazel
and regen bazel files with gazelle from rules_erlang@main
Simplify essential proprty access
Such as durable, ttl and priority by extracting them into annotations
at message container init time.
Move type
to remove dependenc on amqp10 stuff in mc.erl
mostly because I don't know how to make bazel do the right thing
add more stuff
Refine routing header stuff
wip
Cosmetics
Do not use "maybe" as type name as "maybe" is a keyword since OTP 25
which makes Erlang LS complain.
* Dedup death queue names
* Fix function clause crashes
Fix failing tests in the MQTT shared_SUITE:
A classic queue message ID can be undefined as set in
https://github.com/rabbitmq/rabbitmq-server/blob/fbe79ff47b4edbc0fd95457e623d6593161ad198/deps/rabbit/src/rabbit_classic_queue_index_v2.erl#L1048
Fix failing tests in the MQTT shared_SUITE-mixed:
When feature flag message_containers is disabled, the
message is not an #mc{} record, but a #basic_message{} record.
* Fix is_utf8_no_null crash
Prior to this commit, the function crashed if invalid UTF-8 was
provided, e.g.:
```
1> rabbit_misc:is_valid_shortstr(<<"😇"/utf16>>).
** exception error: no function clause matching rabbit_misc:is_utf8_no_null(<<216,61,222,7>>) (rabbit_misc.erl, line 1481)
```
* Implement mqtt mc behaviour
For now via amqp translation.
This is still work in progress, but the following SUITEs pass:
```
make -C deps/rabbitmq_mqtt ct-shared t=[mqtt,v5,cluster_size_1] FULL=1
make -C deps/rabbitmq_mqtt ct-v5 t=[mqtt,cluster_size_1] FULL=1
```
* Shorten mc file names
Module name length matters because for each persistent message the #mc{}
record is persisted to disk.
```
1> iolist_size(term_to_iovec({mc, rabbit_mc_amqp_legacy})).
30
2> iolist_size(term_to_iovec({mc, mc_amqpl})).
17
```
This commit renames the mc modules:
```
ag -l rabbit_mc_amqp_legacy | xargs sed -i 's/rabbit_mc_amqp_legacy/mc_amqpl/g'
ag -l rabbit_mc_amqp | xargs sed -i 's/rabbit_mc_amqp/mc_amqp/g'
ag -l rabbit_mqtt_mc | xargs sed -i 's/rabbit_mqtt_mc/mc_mqtt/g'
```
* mc: make deaths an annotation + fixes
* Fix mc_mqtt protocol_state callback
* Fix test will_delay_node_restart
```
make -C deps/rabbitmq_mqtt ct-v5 t=[mqtt,cluster_size_3]:will_delay_node_restart FULL=1
```
* Bazel run gazelle
* mix format rabbitmqctl.ex
* Ensure ttl annotation is refelected in amqp legacy protocol state
* Fix id access in message store
* Fix rabbit_message_interceptor_SUITE
* dializer fixes
* Fix rabbit:rabbit_message_interceptor_SUITE-mixed
set_annotation/3 should not result in duplicate keys
* Fix MQTT shared_SUITE-mixed
Up to 3.12 non-MQTT publishes were always QoS 1 regardless of delivery_mode.
https://github.com/rabbitmq/rabbitmq-server/blob/75a953ce286a10aca910c098805a4f545989af38/deps/rabbitmq_mqtt/src/rabbit_mqtt_processor.erl#L2075-L2076
From now on, non-MQTT publishes are QoS 1 if durable.
This makes more sense.
The MQTT plugin must send a #basic_message{} to an old node that does
not understand message containers.
* Field content of 'v1_0.data' can be binary
Fix
```
bazel test //deps/rabbitmq_mqtt:shared_SUITE-mixed \
--test_env FOCUS="-group [mqtt,v4,cluster_size_1] -case trace" \
-t- --test_sharding_strategy=disabled
```
* Remove route/2 and implement route/3 for all exchange types.
This removes the route/2 callback from rabbit_exchange_type and
makes route/3 mandatory instead. This is a breaking change and
will require all implementations of exchange types to update their
code, however this is necessary anyway for them to correctly handle
the mc type.
stream filtering fixes
* Translate directly from MQTT to AMQP 0.9.1
* handle undecoded properties in mc_compat
amqpl: put clause in right order
recover death deatails from amqp data
* Replace callback init_amqp with convert_from
* Fix return value of lists:keyfind/3
* Translate directly from AMQP 0.9.1 to MQTT
* Fix MQTT payload size
MQTT payload can be a list when converted from AMQP 0.9.1 for example
First conversions tests
Plus some other conversion related fixes.
bazel
bazel
translate amqp 1.0 null to undefined
mc: property/2 and correlation_id/message_id return type tagged values.
To ensure we can support a variety of types better.
The type type tags are AMQP 1.0 flavoured.
fix death recovery
mc_mqtt: impl new api
Add callbacks to allow protocols to compact data before storage
And make readable if needing to query things repeatedly.
bazel fix
* more decoding
* tracking mixed versions compat
* mc: flip default of `durable` annotation to save some data.
Assuming most messages are durable and that in memory messages suffer less
from persistence overhead it makes sense for a non existent `durable`
annotation to mean durable=true.
* mc conversion tests and tidy up
* mc make x_header unstrict again
* amqpl: death record fixes
* bazel
* amqp -> amqpl conversion test
* Fix crash in mc_amqp:size/1
Body can be a single amqp-value section (instead of
being a list) as shown by test
```
make -C deps/rabbitmq_amqp1_0/ ct-system t=java
```
on branch native-amqp.
* Fix crash in lists:flatten/1
Data can be a single amqp-value section (instead of
being a list) as shown by test
```
make -C deps/rabbitmq_amqp1_0 ct-system t=dotnet:roundtrip_to_amqp_091
```
on branch native-amqp.
* Fix crash in rabbit_writer
Running test
```
make -C deps/rabbitmq_amqp1_0 ct-system t=dotnet:roundtrip_to_amqp_091
```
on branch native-amqp resulted in the following crash:
```
crasher:
initial call: rabbit_writer:enter_mainloop/2
pid: <0.711.0>
registered_name: []
exception error: bad argument
in function size/1
called as size([<<0>>,<<"Sw">>,[<<160,2>>,<<"hi">>]])
*** argument 1: not tuple or binary
in call from rabbit_binary_generator:build_content_frames/7 (rabbit_binary_generator.erl, line 89)
in call from rabbit_binary_generator:build_simple_content_frames/4 (rabbit_binary_generator.erl, line 61)
in call from rabbit_writer:assemble_frames/5 (rabbit_writer.erl, line 334)
in call from rabbit_writer:internal_send_command_async/3 (rabbit_writer.erl, line 365)
in call from rabbit_writer:handle_message/2 (rabbit_writer.erl, line 265)
in call from rabbit_writer:handle_message/3 (rabbit_writer.erl, line 232)
in call from rabbit_writer:mainloop1/2 (rabbit_writer.erl, line 223)
```
because #content.payload_fragments_rev is currently supposed to
be a flat list of binaries instead of being an iolist.
This commit fixes this crash inefficiently by calling
iolist_to_binary/1. A better solution would be to allow AMQP legacy's #content.payload_fragments_rev
to be an iolist.
* Add accidentally deleted line back
* mc: optimise mc_amqp internal format
By removint the outer records for message and delivery annotations
as well as application properties and footers.
* mc: optimis mc_amqp map_add by using upsert
* mc: refactoring and bug fixes
* mc_SUITE routingheader assertions
* mc remove serialize/1 callback as only used by amqp
* mc_amqp: avoid returning a nested list from protocol_state
* test and bug fix
* move infer_type to mc_util
* mc fixes and additiona assertions
* Support headers exchange routing for MQTT messages
When a headers exchange is bound to the MQTT topic exchange, routing
will be performend based on both MQTT topic (by the topic exchange) and
MQTT User Property (by the headers exchange).
This combines the best worlds of both MQTT 5.0 and AMQP 0.9.1 and
enables powerful routing topologies.
When the User Property contains the same name multiple times, only the
last name (and value) will be considered by the headers exchange.
* Fix crash when sending from stream to amqpl
When publishing a message via the stream protocol and consuming it via
AMQP 0.9.1, the following crash occurred prior to this commit:
```
crasher:
initial call: rabbit_channel:init/1
pid: <0.818.0>
registered_name: []
exception exit: {{badmatch,undefined},
[{rabbit_channel,handle_deliver0,4,
[{file,"rabbit_channel.erl"},
{line,2728}]},
{lists,foldl,3,[{file,"lists.erl"},{line,1594}]},
{rabbit_channel,handle_cast,2,
[{file,"rabbit_channel.erl"},
{line,728}]},
{gen_server2,handle_msg,2,
[{file,"gen_server2.erl"},{line,1056}]},
{proc_lib,wake_up,3,
[{file,"proc_lib.erl"},{line,251}]}]}
```
This commit first gives `mc:init/3` the chance to set exchange and
routing_keys annotations.
If not set, `rabbit_stream_queue` will set these annotations assuming
the message was originally published via the stream protocol.
* Support consistent hash exchange routing for MQTT 5.0
When a consistent hash exchange is bound to the MQTT topic exchange,
MQTT 5.0 messages can be routed to queues consistently based on the
Correlation-Data in the PUBLISH packet.
* Convert MQTT 5.0 User Property
* to AMQP 0.9.1 headers
* from AMQP 0.9.1 headers
* to AMQP 1.0 application properties and message annotations
* from AMQP 1.0 application properties and message annotations
* Make use of Annotations in mc_mqtt:protocol_state/2
mc_mqtt:protocol_state/2 includes Annotations as parameter.
It's cleaner to make use of these Annotations when computing the
protocol state instead of relying on the caller (rabbitmq_mqtt_processor)
to compute the protocol state.
* Enforce AMQP 0.9.1 field name length limit
The AMQP 0.9.1 spec prohibits field names longer than 128 characters.
Therefore, when converting AMQP 1.0 message annotations, application
properties or MQTT 5.0 User Property to AMQP 0.9.1 headers, drop any
names longer than 128 characters.
* Fix type specs
Apply feedback from Michael Davis
Co-authored-by: Michael Davis <mcarsondavis@gmail.com>
* Add mc_mqtt unit test suite
Implement mc_mqtt:x_header/2
* Translate indicator that payload is UTF-8 encoded
when converting between MQTT 5.0 and AMQP 1.0
* Translate single amqp-value section from AMQP 1.0 to MQTT
Convert to a text representation, if possible, and indicate to MQTT
client that the payload is UTF-8 encoded. This way, the MQTT client will
be able to parse the payload.
If conversion to text representation is not possible, encode the payload
using the AMQP 1.0 type system and indiate the encoding via Content-Type
message/vnd.rabbitmq.amqp.
This Content-Type is not registered.
Type "message" makes sense since it's a message.
Vendor tree "vnd.rabbitmq.amqp" makes sense since merely subtype "amqp" is not
registered.
* Fix payload conversion
* Translate Response Topic between MQTT and AMQP
Translate MQTT 5.0 Response Topic to AMQP 1.0 reply-to address and vice
versa.
The Response Topic must be a UTF-8 encoded string.
This commit re-uses the already defined RabbitMQ target addresses:
```
"/topic/" RK Publish to amq.topic with routing key RK
"/exchange/" X "/" RK Publish to exchange X with routing key RK
```
By default, the MQTT topic exchange is configure dto be amq.topic using
the 1st target address.
When an operator modifies the mqtt.exchange, the 2nd target address is
used.
* Apply PR feedback
and fix formatting
Co-authored-by: Michael Davis <mcarsondavis@gmail.com>
* tidy up
* Add MQTT message_containers test
* consistent hash exchange: avoid amqp legacy conversion
When hashing on a header value.
* Avoid converting to amqp legacy when using exchange federation
* Fix test flake
* test and dialyzer fixes
* dialyzer fix
* Add MQTT protocol interoperability tests
Test receiving from and sending to MQTT 5.0 and
* AMQP 0.9.1
* AMQP 1.0
* STOMP
* Streams
* Regenerate portions of deps/rabbit/app.bzl with gazelle
I'm not exactly sure how this happened, but gazell seems to have been
run with an older version of the rules_erlang gazelle extension at
some point. This caused generation of a structure that is no longer
used. This commit updates the structure to the current pattern.
* mc: refactoring
* mc_amqpl: handle delivery annotations
Just in case they are included.
Also use iolist_to_iovec to create flat list of binaries when
converting from amqp with amqp encoded payload.
---------
Co-authored-by: David Ansari <david.ansari@gmx.de>
Co-authored-by: Michael Davis <mcarsondavis@gmail.com>
Co-authored-by: Rin Kuryloski <kuryloskip@vmware.com>
2023-08-31 18:27:13 +08:00
|
|
|
|
2020-09-29 18:43:24 +08:00
|
|
|
{QName, QRef, MsgId, Redelivered, Msg}
|
|
|
|
|
end, Msgs).
|
|
|
|
|
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
update_consumer(Tag, LastId, DelCntIncr,
|
2020-09-29 18:43:24 +08:00
|
|
|
#consumer{delivery_count = D} = C, Consumers) ->
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
maps:put(Tag,
|
2020-09-29 18:43:24 +08:00
|
|
|
C#consumer{last_msg_id = LastId,
|
|
|
|
|
delivery_count = D + DelCntIncr},
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
Consumers).
|
|
|
|
|
|
|
|
|
|
|
2022-10-03 21:55:52 +08:00
|
|
|
get_missing_deliveries(State, From, To, ConsumerTag) ->
|
|
|
|
|
%% find local server
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
ConsumerId = consumer_id(ConsumerTag),
|
2022-10-03 21:55:52 +08:00
|
|
|
rabbit_log:debug("get_missing_deliveries for ~w from ~b to ~b",
|
|
|
|
|
[ConsumerId, From, To]),
|
|
|
|
|
Cmd = {get_checked_out, ConsumerId, lists:seq(From, To)},
|
|
|
|
|
ServerId = find_local_or_leader(State),
|
|
|
|
|
case ra:aux_command(ServerId, Cmd) of
|
|
|
|
|
{ok, Missing} ->
|
2019-10-30 20:26:59 +08:00
|
|
|
Missing;
|
|
|
|
|
{error, Error} ->
|
2022-10-08 06:59:05 +08:00
|
|
|
{protocol_error, internal_error, "Cannot query missing deliveries from ~tp: ~tp",
|
2022-10-03 21:55:52 +08:00
|
|
|
[ServerId, Error]};
|
2019-10-30 20:26:59 +08:00
|
|
|
{timeout, _} ->
|
2022-10-08 06:59:05 +08:00
|
|
|
{protocol_error, internal_error, "Cannot query missing deliveries from ~tp: timeout",
|
2022-10-03 21:55:52 +08:00
|
|
|
[ServerId]}
|
2019-10-30 20:26:59 +08:00
|
|
|
end.
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
|
2020-07-10 17:20:25 +08:00
|
|
|
pick_server(#state{leader = undefined,
|
2020-07-10 21:25:17 +08:00
|
|
|
cfg = #cfg{servers = [N | _]}}) ->
|
2018-11-27 18:23:25 +08:00
|
|
|
%% TODO: pick random rather that first?
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
N;
|
2020-07-10 17:20:25 +08:00
|
|
|
pick_server(#state{leader = Leader}) ->
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
Leader.
|
|
|
|
|
|
|
|
|
|
% servers sorted by last known leader
|
2020-07-10 17:20:25 +08:00
|
|
|
sorted_servers(#state{leader = undefined,
|
|
|
|
|
cfg = #cfg{servers = Servers}}) ->
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
Servers;
|
2020-07-10 17:20:25 +08:00
|
|
|
sorted_servers(#state{leader = Leader,
|
|
|
|
|
cfg = #cfg{servers = Servers}}) ->
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
[Leader | lists:delete(Leader, Servers)].
|
|
|
|
|
|
|
|
|
|
consumer_id(ConsumerTag) ->
|
|
|
|
|
{ConsumerTag, self()}.
|
|
|
|
|
|
2021-11-12 23:59:51 +08:00
|
|
|
send_command(Server, Correlation, Command, _Priority,
|
|
|
|
|
#state{pending = Pending,
|
2022-02-18 23:38:57 +08:00
|
|
|
next_seq = Seq,
|
2022-01-26 00:52:01 +08:00
|
|
|
cfg = #cfg{soft_limit = SftLmt}} = State)
|
2021-11-12 23:59:51 +08:00
|
|
|
when element(1, Command) == return ->
|
|
|
|
|
%% returns are sent to the aux machine for pre-evaluation
|
|
|
|
|
ok = ra:cast_aux_command(Server, {Command, Seq, self()}),
|
2022-01-26 00:52:01 +08:00
|
|
|
Tag = case map_size(Pending) >= SftLmt of
|
2021-11-12 23:59:51 +08:00
|
|
|
true -> slow;
|
|
|
|
|
false -> ok
|
|
|
|
|
end,
|
2022-01-26 00:52:01 +08:00
|
|
|
State#state{pending = Pending#{Seq => {Correlation, Command}},
|
2022-02-18 23:38:57 +08:00
|
|
|
next_seq = Seq + 1,
|
2022-01-26 00:52:01 +08:00
|
|
|
slow = Tag == slow};
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
send_command(Server, Correlation, Command, Priority,
|
|
|
|
|
#state{pending = Pending,
|
2022-02-18 23:38:57 +08:00
|
|
|
next_seq = Seq,
|
2022-01-26 00:52:01 +08:00
|
|
|
cfg = #cfg{soft_limit = SftLmt}} = State) ->
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
ok = ra:pipeline_command(Server, Command, Seq, Priority),
|
2022-01-26 00:52:01 +08:00
|
|
|
Tag = case map_size(Pending) >= SftLmt of
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
true -> slow;
|
|
|
|
|
false -> ok
|
|
|
|
|
end,
|
2022-01-26 00:52:01 +08:00
|
|
|
State#state{pending = Pending#{Seq => {Correlation, Command}},
|
2022-02-18 23:38:57 +08:00
|
|
|
next_seq = Seq + 1,
|
2022-01-26 00:52:01 +08:00
|
|
|
slow = Tag == slow}.
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
|
|
|
|
|
resend_command(Node, Correlation, Command,
|
2022-02-18 23:38:57 +08:00
|
|
|
#state{pending = Pending,
|
|
|
|
|
next_seq = Seq} = State) ->
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
ok = ra:pipeline_command(Node, Command, Seq),
|
2022-02-18 23:38:57 +08:00
|
|
|
State#state{pending = Pending#{Seq => {Correlation, Command}},
|
|
|
|
|
next_seq = Seq + 1}.
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
|
2018-12-17 20:58:51 +08:00
|
|
|
add_command(_, _, [], Acc) ->
|
Quorum queues (#1706)
* Test queue.declare method with quorum type
[#154472130]
* Cosmetics
[#154472130]
* Start quorum queue
Includes ra as a rabbit dependency
[#154472152]
* Update info and list operations to use quorum queues
Basic implementation. Might need an update when more functionality
is added to the quorum queues.
[#154472152]
* Stop quorum queue
[#154472158]
* Restart quorum queue
[#154472164]
* Introduce UId in ra config to support newer version of ra
Improved ra stop
[#154472158]
* Put data inside VHost specific subdirs
[#154472164]
* Include ra in rabbit deps to support stop_app/start_app command
[#154472164]
* Stop quorum queues in `rabbit_amqqueue:stop/1`
[#154472158]
* Revert creation of fifo ets table inside rabbit
Now supported by ra
[#154472158]
* Filter quorum queues
[#154472158]
* Test restart node with quorum queues
[#154472164]
* Publish to quorum queues
[#154472174]
* Use `ra:restart_node/1`
[#154472164]
* Wait for stats to be published when querying quorum queues
[#154472174]
* Test publish and queue length after restart
[#154472174]
* Consume messages from quorum queues with basic.get
[#154472211]
* Autoack messages from quorum queues on basic.get
[#154472211]
* Fix no_ack meaning
no_ack = true is equivalent to autoack
[#154472211]
* Use data_dir as provided in the config
If we modify the data_dir, ra is not able to delete the data
when a queue is deleted
[#154472158]
* Remove unused code/variables
[#154472158]
* Subscribe to a quorum queue
Supports auto-ack
[#154472215]
* Ack messages consumed from quorum queues
[#154472221]
* Nack messages consumed from quorum queues
[#154804608]
* Use delivery tag as consumer tag for basic.get in quorum queues
[#154472221]
* Support for publisher confirms in quorum queues
[#154472198]
* Integrate with ra_fifo_client
* Clear queue state on queue.delete
[#154472158]
* Fix quorum nack
[#154804608]
* Test redelivery after nack
[#154804608]
* Nack without requeueing
[#154472225]
* Test multiple acks
[#154804208]
* Test multiple nacks
[#154804314]
* Configure dead letter exchange with queue declare
[#155076661]
* Use a per-vhost process to handle dead-lettering
Needs to hold state for quorum queues
[#155401802]
* Implement dead-lettering on nack'ed messages
[#154804620]
* Use queue name as a resource on message delivery
Fixes a previously introduced bug
[#154804608]
* Handle ra events on dead letter process
[#155401802]
* Pass empty queue states to queue delete
Queue deletion on vhost deletion calls directly to rabbit_amqqueue.
Queue states are not available, but we can provide an empty map as
in deletion the states are only needed for cleanup.
* Generate quorum queue stats and events
Consumer delete events are still pending, as depend on basic.cancel
(not implemented yet), ra terminating or ra detecting channel down
[#154472241]
* Ensure quorum mapping entries are available before metric emission
[#154472241]
* Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var
[#154472152]
* Use untracked enqueues when sending wihtout channel
Updated several other calls missed during the quorum implementation
* Revert "Configure data_dir, uses new RABBITMQ_QUORUM_BASE env var"
This reverts commit f2261212410affecb238fcbd1fb451381aee4036.
* Configure data_dir, uses new RABBITMQ_QUORUM_DIR based on mnesia dir
[#154472152]
* Fix get_quorum_state
* Fix calculation of quorum pids
* Move all quorum queues code to its own module
[#154472241]
* Return an error when declaring a quorum queue with an incompatible argument
[#154521696]
* Cleanup of quorum queue state after queue delete
Also fixes some existing problems where the state wasn't properly
stored
[#155458625]
* Revert Revert "Declare a quorum queue using the queue.declare method"
* Remove duplicated state info
[#154472241]
* Start/stop multi-node quorum queue
[#154472231]
[#154472236]
* Restart nodes in a multi-node quorum cluster
[#154472238]
* Test restart and leadership takeover on multiple nodes
[#154472238]
* Wait for leader down after deleting a quorum cluster
It ensures an smooth delete-declare sequence without race
conditions. The test included here detected the situation before
the fix.
[#154472236]
* Populate quorum_mapping from mnesia when not available
Ensures that leader nodes that don't have direct requests can get
the mapping ra name -> queue name
* Cosmetics
* Do not emit core metrics if queue has just been deleted
* Use rabbit_mnesia:is_process_alive
Fixes bug introduced by cac9583e1bb2705be7f06c2ab7f416a75d11c875
[#154472231]
* Only try to report stats if quorum process is alive
* Implement cancel consumer callback
Deletes metrics and sends consumer deleted event
* Remove unnecessary trigger election call
ra:restart_node has already been called during the recovery
* Apply cancellation callback on node hosting the channel
* Cosmetics
* Read new fifo metrics which store directly total, ready and unack
* Implement basic.cancel for quorum queues
* Store leader in amqqueue record, report all in stats
[#154472407]
* Declare quorum queue in mnesia before starting the ra cluster
Record needs to be stored first to update the leader on ra effects
* Revert
* Purge quorum queues
[#154472182]
* Improve use of untracked_enqueue
Choose the persisted leader id instead of just using the id of the
leader at point of creation.
* Store quorum leader in the pid field of amqqueue record
Same as mirrored queues, no real need for an additional field
* Improve recovery
When a ra node has never been started on a rabbit node ensure it doesn't
fail but instead rebuilds the config and starts the node as a new node.
Also fix issue when a quorum queue is declared when one of it's rabbit
nodes are unavailable.
[#157054606]
* Cleanup core metrics after leader change
[#157054473]
* Return an error on sync_queue on quorum queues
[#154472334]
* Return an error on cancel_sync_queue on quorum queues
[#154472337]
* Fix basic_cancel and basic_consume return values
Ensure the quorum queue state is always returned by these functions.
* Restore arity of amqqeueu delete and purge functions.
This avoids some breaking changes in the cli.
* Fix bug returning consumers.
* remove rogue debug log
* Integrate ingress flow control with quorum queues
[#157000583]
* Configure commands soft limit
[#157000583]
* Support quorum pids on rabbit_mnesia:is_process_alive
* Publish consumers metric for quorum queues
* Whitelist quorum directory in is_virgin_node
Allow the quorum directoy to exist without affecting the status of the
Rabbit node.
* Delete queue_metrics on leader change.
Also run the become_leader handler in a separate process to avoid
blocking.
[#157424225]
* Report cluster status in quorum queue infos. New per node status command.
Related to
[#157146500]
* Remove quorum_mapping table
As we can store the full queue name resource as the cluster id of the
ra_fifo_client state we can avoid needed the quorum_mapping table.
* Fix xref issue
* Provide quorum members information in stats
[#157146500]
* fix unused variable
* quorum queue multiple declare handling
Extend rabbit_amqqueue:internal_declare/2 to indicate if the queue
record was created or exisiting. From this we can then provide a code
path that should handle concurrent queue declares of the same quorum
queue.
* Return an error when declaring exclusive/auto-delete quorum queue
[#157472160]
* Restore lost changes
from 79c9bd201e1eac006a42bd162e7c86df96496629
* recover another part of commit
* fixup cherry pick
* Ra io/file metrics handler and stats publishing
[#157193081]
* Revert "Ra io/file metrics handler and stats publishing"
This reverts commit 05d15c786540322583fc655709825db215b70952.
* Do not issue confirms on node down for quorum queues.
Only a ra_event should be used to issue positive confirms for a quorum
queue.
* Ra stats publishing
[#157193081]
* Pick consumer utilisation from ra data
[#155402726]
* Handle error when deleting a quorum queue and all nodes are already down
This is in fact a successful deletion as all raft nodes are already 'stopped'
[#158656366]
* Return an error when declaring non-durable quorum queues
[#158656454]
* Rename dirty_query to committed_query
* Delete stats on leader node
[#158661152]
* Give full list of nodes to fifo client
* Handle timeout in quorum basic_get
* Fix unused variable error
* Handle timeout in basic get
[#158656366]
* Force GC after purge
[#158789389]
* Increase `ra:delete_cluster` timeout to 120s
* Revert "Force GC after purge"
This reverts commit 5c98bf22994eb39004760799d3a2c5041d16e9d4.
* Add quorum member command
[#157481599]
* Delete quorum member command
[#157481599]
* Implement basic.recover for quorum queues
[#157597411]
* Change concumer utilisation
to use the new ra_fifo table and api.
* Set max quorum queue size limit
Defaults to 7, can be configured per queue on queue.declare
Nodes are selected randomly from the list of nodes, but the one
that is executing the queue.declare command
[#159338081]
* remove potentially unrelated changes to rabbit_networking
* Move ra_fifo to rabbit
Copied ra_fifo to rabbit and renamed it rabbit_fifo.
[#159338031]
* rabbit_fifo tidy up
* rabbit_fifo tidy up
* rabbit_fifo: customer -> consumer rename
* Move ra_fifo tests
[#159338031]
* Tweak quorum_queue defaults
* quorum_queue test reliability
* Optimise quorum_queue test suite.
By only starting a rabbit cluster per group rather than test.
[#160612638]
* Renamings in line with ra API changes
* rabbit_fifo fixes
* Update with ra API changes
Ra has consolidated and simplified it's api. These changes update to
confirm to that.
* Update rabbit_fifo with latest ra changes
* Clean up out of date comment
* Return map of states
* Add test case for basic.get on an empty queue
Before the previous patch, any subsequent basic.get would crash as
the map of states had been replaced by a single state.
* Clarify use of deliver tags on record_sent
* Clean up queues after testcase
* Remove erlang monitor of quorum queues in rabbit_channel
The eol event can be used instead
* Use macros to make clearer distinctions between quorum/classic queues
Cosmetic only
* Erase queue stats on 'eol' event
* Update to follow Ra's cluster_id -> cluster_name rename.
* Rename qourum-cluster-size
To quorum-initial-group-size
* Issue confirms on quorum queue eol
Also avoid creating quorum queue session state on queue operation
methods.
* Only classic queues should be notified on channel down
* Quorum queues do not support global qos
Exit with protocol error of a basic.consume for a quorum queue is issued
on a channel with global qos enabled.
* unused variable name
* Refactoring
Strictly enfornce that channels do not monitor quorum queues.
* Refactor foreach_per_queue in the channel.
To make it call classic and quorum queues the same way.
[#161314899]
* rename function
* Query classic and quorum queues separately
during recovery as they should not be marked as stopped during failed
vhost recovery.
* Remove force_event_refresh function
As the only user of this function, the management API no longer requires
it.
* fix errors
* Remove created_at from amqqueue record
[#161343680]
* rabbit_fifo: support AMQP 1.0 consumer credit
This change implements an alternative consumer credit mechanism similar
to AMQP 1.0 link credit where the credit (prefetch) isn't automatically
topped up as deliveries are settled and instead needs to be manually
increased using a credit command. This is to be integrated with the AMQP
1.0 plugin.
[#161256187]
* Add basic.credit support for quorum queues.
Added support for AMQP 1.0 transfer flow control.
[#161256187]
* Make quorum queue recover idempotent
So that if a vhost crashes and runs the recover steps it doesn't fail
because ra servers are still running.
[#161343651]
* Add tests for vhost deletion
To ensure quorum queues are cleaned up on vhost removal.
Also fix xref issue.
[#161343673]
* remove unused clause
* always return latest value of queue
* Add rabbitmq-queues scripts. Remove ra config from .bat scripts.
* Return error if trying to get quorum status of a classic queue.
2018-10-29 17:47:29 +08:00
|
|
|
Acc;
|
2018-12-17 20:58:51 +08:00
|
|
|
add_command(Cid, settle, MsgIds, Acc) ->
|
|
|
|
|
[rabbit_fifo:make_settle(Cid, MsgIds) | Acc];
|
|
|
|
|
add_command(Cid, return, MsgIds, Acc) ->
|
2019-10-01 00:02:19 +08:00
|
|
|
[rabbit_fifo:make_return(Cid, MsgIds) | Acc];
|
2018-12-17 20:58:51 +08:00
|
|
|
add_command(Cid, discard, MsgIds, Acc) ->
|
2019-10-01 00:02:19 +08:00
|
|
|
[rabbit_fifo:make_discard(Cid, MsgIds) | Acc].
|
2019-02-14 18:49:25 +08:00
|
|
|
|
2023-01-19 22:33:28 +08:00
|
|
|
set_timer(QName, #state{leader = Leader0,
|
|
|
|
|
cfg = #cfg{servers = [Server | _]}} = State) ->
|
2020-09-29 18:43:24 +08:00
|
|
|
Leader = case Leader0 of
|
|
|
|
|
undefined -> Server;
|
|
|
|
|
_ ->
|
|
|
|
|
Leader0
|
|
|
|
|
end,
|
2019-02-14 18:49:25 +08:00
|
|
|
Ref = erlang:send_after(?TIMER_TIME, self(),
|
2020-09-29 18:43:24 +08:00
|
|
|
{'$gen_cast',
|
|
|
|
|
{queue_event, QName, {Leader, timeout}}}),
|
2019-02-14 18:49:25 +08:00
|
|
|
State#state{timer_state = Ref}.
|
|
|
|
|
|
|
|
|
|
cancel_timer(#state{timer_state = undefined} = State) ->
|
|
|
|
|
State;
|
|
|
|
|
cancel_timer(#state{timer_state = Ref} = State) ->
|
2023-01-24 05:26:14 +08:00
|
|
|
_ = erlang:cancel_timer(Ref, [{async, true}, {info, false}]),
|
2019-02-14 18:49:25 +08:00
|
|
|
State#state{timer_state = undefined}.
|
|
|
|
|
|
2022-10-03 21:55:52 +08:00
|
|
|
find_local_or_leader(#state{leader = Leader,
|
|
|
|
|
cfg = #cfg{servers = Servers}}) ->
|
|
|
|
|
case find_local(Servers) of
|
|
|
|
|
undefined ->
|
|
|
|
|
Leader;
|
|
|
|
|
ServerId ->
|
|
|
|
|
ServerId
|
|
|
|
|
end.
|
|
|
|
|
|
|
|
|
|
find_local([{_, N} = ServerId | _]) when N == node() ->
|
|
|
|
|
ServerId;
|
|
|
|
|
find_local([_ | Rem]) ->
|
|
|
|
|
find_local(Rem);
|
|
|
|
|
find_local([]) ->
|
|
|
|
|
undefined.
|
|
|
|
|
|
|
|
|
|
|
2019-02-14 18:49:25 +08:00
|
|
|
find_leader([]) ->
|
|
|
|
|
undefined;
|
|
|
|
|
find_leader([Server | Servers]) ->
|
|
|
|
|
case ra:members(Server, 500) of
|
|
|
|
|
{ok, _, Leader} -> Leader;
|
|
|
|
|
_ ->
|
|
|
|
|
find_leader(Servers)
|
|
|
|
|
end.
|
2020-09-29 18:43:24 +08:00
|
|
|
|
|
|
|
|
qref({Ref, _}) -> Ref;
|
|
|
|
|
qref(Ref) -> Ref.
|
2022-11-02 03:52:40 +08:00
|
|
|
|
|
|
|
|
-spec cluster_name(state()) ->
|
|
|
|
|
atom().
|
|
|
|
|
cluster_name(#state{cfg = #cfg{servers = [{Name, _Node} | _]}}) ->
|
|
|
|
|
Name.
|