MetadataShell should take an advisory lock on the .lock file of the directory it is reading from.
Add an integration test of this functionality in MetadataShellIntegrationTest.java.
Note: in build.gradle, I had to add some dependencies on server-common's test files in order to use
MockFaultHandler, etc.
MetadataBatchLoader.java: fix a case where a log message was incorrect. The intention was to print
the number equivalent to (offset + index). Instead it was printing the offset, followed by the
index. So if the offset was 100 and the index was 1, 1001 would be printed rather than 101.
Co-authored-by: Igor Soarez <i@soarez.me>
Reviewers: David Arthur <mumrah@gmail.com>, José Armando García Sancio <jsancio@apache.org>
Handle AssignReplicasToDirs requests, persist metadata changes
with new directory assignments and possible leader elections.
Reviewers: Proven Provenzano <pprovenzano@confluent.io>, Ron Dagostino <rndgstn@gmail.com>, Colin P. McCabe <cmccabe@apache.org>
Part of KIP-714.
Adds support to expose main consumer client instance id.
Reviewers: Walker Carlson <wcarlson@confluent.io>, Lucas Brutschy <lbrutschy@confluent.io>
This patch implements the TxnOffsetCommit API. When a transactional offset commit is received, it is stored in the pending transactional offsets structure and waits there until the transaction is committed or aborted. Note that the handling of the transaction completion is not implemented in this patch.
Reviewers: Justine Olshan <jolshan@confluent.io>
The PR adds changes for the client APIs to register ClientTelemetryReporter, if enabled, and periodically report client metrics. The changes include front facing API changes with NetworkCLient issuing telemetry APIs.
The PR build is dependent on: #14620, #14724
Reviewers: Philip Nee <pnee@confluent.io>, Andrew Schofield <aschofield@confluent.io>, Kirk True <ktrue@confluent.io>, Matthias J. Sax <matthias@confluent.io>, Walker Carlson <wcarlson@apache.org>
Part of KIP-714.
Implements ClientTelemetryReporter which manages the lifecycle for client metrics collection. The reporter also defines TelemetrySender which will be used by Network clients to send API calls to broker.
Reviewers: Andrew Schofield <aschofield@confluent.io>, Philip Nee <pnee@confluent.io>, Matthias J. Sax <matthias@confluent.io>
This PR fixes some details of the interface to KafkaConsumer.committed which were different between the existing consumer and the new consumer.
Adds a unit test that validates the behaviour is the same for both consumer implementations.
Reviewers: Kirk True <ktrue@confluent.io>, Bruno Cadonna <cadonna@apache.org>
Implements KIP-992.
Adds TimestampedKeyQuery and TimestampedRangeQuery (IQv2) for ts-ks-store, plus changes semantics of existing KeyQuery and RangeQuery if issues against a ts-kv-store, now unwrapping value-and-timestamp and only returning the plain value.
Reviewers: Matthias J. Sax <matthias@confluent.io>
The PR provide implementation for client metrics manager along with other classes. Manager is responsible to support 3 operations:
UpdateSubscription - From kafka-configs.sh and reload from metadata cache.
Process Get Telemetry Request - From KafkaApis.scala
Process Push Telemetry Request - From KafkaApis.scala
Manager maintains an in-memory cache to keep track of client instances against their instance id.
Reviewers: Andrew Schofield <aschofield@confluent.io>, Jun Rao <junrao@gmail.com>
The PR outlines classes to collect metrics for client by KafkaMetricsCollector implementation. The MetricsCollector defines mechanism to collect client metrics in sum and gauge metrics format. This requires to define cumulative and delta telemetry metrics while collecting raw metrics.
Singl point metric class helps creating OTLP format Metric object wrapped over Single point metric class itself.
Reviewers: Andrew Schofield <aschofield@confluent.io>, Xavier Léauté <xavier@confluent.io>, Philip Nee <pnee@confluent.io>, Matthias J. Sax <matthias@confluent.io>
The PR adds support of alter/describe configs for client-metrics as defined in KIP-714
Reviewers: Andrew Schofield <aschofield@confluent.io>, Jun Rao <junrao@gmail.com>
This patch adds the second part of the Uniform Assignor, used when the subscriptions of each member in a consumer group are different.
Reviewers: Jeff Kim <jeff.kim@confluent.io>, David Jacot <djacot@confluent.io>
This patch copy over existing metrics and add new consumer group metrics to the new GroupCoordinatorService.
Now that each coordinator is responsible for a topic partition, this patch introduces a GroupCoordinatorMetrics that records gauges for global metrics such as the number of generic groups in PreparingRebalance state, etc. For GroupCoordinatorShard specific metrics, GroupCoordinatorMetrics will activate new GroupCoordinatorMetricsShards that will be responsible for incrementing/decrementing TimelineLong objects and then aggregate the total amount across all shards.
As the CoordinatorRuntime/CoordinatorShard does not care about group metadata, we have introduced a CoordinatorMetrics.java/CoordinatorMetricsShard.java so that in the future transaction coordinator metrics can also be onboarded in a similar fashion.
Main files to look at:
GroupCoordinatorMetrics.java
GroupCoordinatorMetricsShard.java
CoordinatorMetrics.java
CoordinatorMetricsShard.java
CoordinatorRuntime.java
Metrics to add after #14408 is merged:
offset deletions sensor (OffsetDeletions); Meter(offset-deletion-rate, offset-deletion-count)
Metrics to add after https://issues.apache.org/jira/browse/KAFKA-14987 is merged:
offset expired sensor (OffsetExpired); Meter(offset-expiration-rate, offset-expiration-count)
Reviewers: Justine Olshan <jolshan@confluent.io>
We only move Java classes that have minimal or no dependencies on Scala classes in this PR.
Details:
* Configured `server` module in build files.
* Changed `ControllerRequestCompletionHandler` to be an interface since it has no implementations.
* Cleaned up various import control files.
* Minor build clean-ups for `server-common`.
* Disabled `testAssignmentAggregation` when executed with Java 8, this is an existing issue (see #14794).
For broader context on this change, please check:
* KAFKA-15852: Move server code from `core` to `server` module
Reviewers: Divij Vaidya <diviv@amazon.com>
A new AssignmentsManager accumulates, batches, and sends KIP-858
assignment events to the Controller. Assignments are sent via
AssignReplicasToDirs requests.
Move QuorumTestHarness.formatDirectories into TestUtils so it can be
used in other test contexts.
Fix a bug in ControllerRegistration.java where the wrong version of the
record was being generated in ControllerRegistration.toRecord.
Reviewers: Colin P. McCabe <cmccabe@apache.org>, Proven Provenzano <pprovenzano@confluent.io>, Omnia G H Ibrahim <o.g.h.ibrahim@gmail.com>
meta.properties files are used by Kafka to identify log directories within the filesystem.
Previously, the code for handling them was in BrokerMetadataCheckpoint.scala. This PR rewrites the
code for handling them as Java and moves it to the apache.kafka.metadata.properties namespace. It
also gets rid of the separate types for v0 and v1 meta.properties objects. Having separate types
wasn't so bad back when we had a strict rule that zk clusters used v0 and kraft clusters used v1.
But ZK migration has blurred the lines. Now, a zk cluster may have either v0 or v1, if it is
migrating, and a kraft cluster may have either v0 or v1, at any time.
The new code distinguishes between an individual meta.properties file, which is represented by
MetaProperties, and a collection of meta.properties files, which is represented by
MetaPropertiesEnsemble. It is useful to have this distinction, because in JBOD mode, even if some
log directories are inaccessible, we can still use the ensemble to extract needed information like
the cluster ID. (Of course, even when not in JBOD mode, KRaft servers have always been able to
configure a metadata log directory separate from the main log directory.)
Since we recently added a unique directory.id to each meta.properties file, the previous convention
of passing a "canonical" MetaProperties object for the cluster around to various places in the code
needs to be revisited. After all, we can no longer assume all of the meta.properties files are the
same. This PR fixes these parts of the code. For example, it fixes the constructors of
ControllerApis and RaftManager to just take a cluster ID, rather than a MetaProperties object. It
fixes some other parts of the code, like the constructor of SharedServer, to take a
MetaPropertiesEnsemble object.
Another goal of this PR was to centralize meta.properties validation a bit more and make it
unit-testable. For this purpose, the PR adds MetaPropertiesEnsemble.verify, and a few other
verification methods. These enforce invariants like "the metadata directory must be readable," and
so on.
Reviewers: Igor Soarez <soarez@apple.com>, David Arthur <mumrah@gmail.com>, Divij Vaidya <diviv@amazon.com>, Proven Provenzano <pprovenzano@confluent.io>
The patch includes the following changes as part of KIP-966
* Allow ISR shrink to empty
* Allow leader election with ELR members
* Allow electing the last known leader
Reviewers: Artem Livshits <alivshits@confluent.io>, David Arthur <mumrah@gmail.com>
Reviewers: Christo Lolov <lolovc@amazon.com>, Colin P. McCabe <cmccabe@apache.org>, Proven Provenzano <pprovenzano@confluent.io>, Ron Dagostino <rdagostino@confluent.io>
This PR for KIP-714 - KAFKA-1564 lays out interfaces and classes for capturing client telemetry metrics.
Below image defines interaction of different classes among them interfaces have been included in the PR.
Reviewers: Walker Carlson <wcarlson@apache.org>, Matthias J. Sax <matthias@confluent.io>, Andrew Schofield <andrew_schofield@uk.ibm.com>, Kirk True <ktrue@confluent.io>, Philip Nee <pnee@confluent.io>, Jun Rao <junrao@gmail.com>,
Changes:
1. Introduces FetchRequestManager that implements the RequestManager
API for fetching messages from brokers. Unlike Fetcher, record
decompression and deserialization is performed on the application
thread inside CompletedFetch.
2. Restructured the code so that objects owned by the background thread
are not instantiated until the background thread runs (via Supplier)
to ensure that there are no references available to the
application thread.
3. Ensuring resources are properly using Closeable and using
IdempotentCloser to ensure they're only closed once.
4. Introduces ConsumerTestBuilder to reduce a lot of inconsistency in
the way the objects were built up for tests.
Reviewers: Philip Nee <pnee@confluent.io>, Lianet Magrans <lianetmr@gmail.com>, Jun Rao<junrao@gmail.com>
This patch introduces preliminary changes for Eligible Leader Replicas (KIP-966)
* New MetadataVersion 16 (3.7-IV1)
* New record versions for PartitionRecord and PartitionChangeRecord
* New tagged fields on PartitionRecord and PartitionChangeRecord
* New static config "eligible.leader.replicas.enable" to gate the whole feature
Reviewers: Artem Livshits <alivshits@confluent.io>, David Arthur <mumrah@gmail.com>, Colin P. McCabe <cmccabe@apache.org>
The PR includes:
* Added a new class of CleanShutdownFile which helps write and read from a clean shutdown file.
* Updated the BrokerRegistration API.
* Client side handling for the broker epoch.
* Minimum work on the controller side.
Reviewers: Jun Rao <junrao@gmail.com>
Implements the following metrics:
kafka.server:type=group-coordinator-metrics,name=num-partitions,state=loading
kafka.server:type=group-coordinator-metrics,name=num-partitions,state=active
kafka.server:type=group-coordinator-metrics,name=num-partitions,state=failed
kafka.server:type=group-coordinator-metrics,name=event-queue-size
kafka.server:type=group-coordinator-metrics,name=partition-load-time-max
kafka.server:type=group-coordinator-metrics,name=partition-load-time-avg
kafka.server:type=group-coordinator-metrics,name=thread-idle-ratio-min
kafka.server:type=group-coordinator-metrics,name=thread-idle-ratio-avg
The PR makes these metrics generic so that in the future the transaction coordinator runtime can implement the same metrics in a similar fashion.
Also, CoordinatorLoaderImpl#load will now return LoadSummary which encapsulates the start time, end time, number of records/bytes.
Co-authored-by: David Jacot <djacot@confluent.io>
Reviewers: Ritika Reddy <rreddy@confluent.io>, Calvin Liu <caliu@confluent.io>, David Jacot <djacot@confluent.io>, Justine Olshan <jolshan@confluent.io>
This is now possible since `InterBrokerSend` was moved from `core` to `server-common`.
Also rewrite/move `KafkaNetworkChannelTest`.
The scala version of `KafkaNetworkChannelTest` passed with the changes here (before I
deleted it).
Reviewers: Justine Olshan <jolshan@confluent.io>, José Armando García Sancio <jsancio@users.noreply.github.com>
The current KafkaConsumer offsetsForTimes fails with IllegalArgumentException if negative target timestamps are provided as arguments. This change includes the same validation and tests for the new consumer implementation (and some improved comments for the updateFetchPositions)
Reviewer: Lucas Brutschy <lbrutschy@confluent.io>
This patch implements the groups and offsets expiration in the new group coordinator.
Reviewers: Ritika Reddy <rreddy@confluent.io>, David Jacot <djacot@confluent.io>
When producer-batch is being retried, new-leader is known for the partition Vs the leader used in last attempt, then it is worthwhile to retry immediately to this new leader. A partition-leader is considered to be newer, if the epoch has advanced.
Reviewers: Walker Carlson <wcarlson@apache.org>, Kirk True <kirk@kirktrue.pro>, Andrew Schofield <andrew_schofield@uk.ibm.com
This patch implements DeleteGroups and OffsetDelete API in the new group coordinator.
Reviewers: yangy0000, Ritika Reddy <rreddy@confluent.io>, Jeff Kim <jeff.kim@confluent.io>, David Jacot <djacot@confluent.io>
This PR is part of #13247
It contains ReassignPartitionsIntegrationTest rewritten in java.
Goal of PR is reduce changes size in main PR.
Reviewers: Taras Ledkov <tledkov@apache.org>, Justine Olshan <jolshan@confluent.io>
* Implements start and stop of task executors
* Introduce flush operation to keep consumer operations out of the processing threads
* Fixes corner case: handle requested unassignment during shutdown
* Fixes corner case: handle race between voluntary unassignment and requested unassigment
* Fixes corner case: task locking future completes for the empty set
* Fixes corner case: we should not reassign a task with an uncaught exception to a task executor
* Improved logging
* Number of threads controlled from outside, of the TaskManager
Reviewers: Bruno Cadonna <bruno@confluent.io>
Add support for --bootstrap-controller in the following command-line tools:
- kafka-cluster.sh
- kafka-configs.sh
- kafka-features.sh
- kafka-metadata-quorum.sh
To implement this, the following AdminClient APIs now support the new bootstrap.controllers
configuration:
- Admin.alterConfigs
- Admin.describeCluster
- Admin.describeConfigs
- Admin.describeFeatures
- Admin.describeMetadataQuorum
- Admin.incrementalAlterConfigs
- Admin.updateFeatures
Command-line tool changes:
- Add CommandLineUtils.initializeBootstrapProperties to handle parsing --bootstrap-controller
in addition to --bootstrap-server.
- Add --bootstrap-controller to ConfigCommand.scala, ClusterTool.java, FeatureCommand.java, and
MetadataQuorumCommand.java.
KafkaAdminClient changes:
- Add the AdminBootstrapAddresses class to handle extracting bootstrap.servers or
bootstrap.controllers from the config map for KafkaAdminClient.
- In AdminMetadataManager, store the new usingBootstrapControllers boolean. Generalize
authException to encompass the concept of fatal exceptions in general. (For example, the
fatal exception where we talked to the wrong node type.) Treat
MismatchedEndpointTypeException and UnsupportedEndpointTypeException as fatal exceptions.
- Extend NodeProvider to include information about whether bootstrap.controllers is supported.
- Modify the APIs described above to support bootstrap.controllers.
Server-side changes:
- Support DescribeConfigsRequest on kcontrollers.
- Add KRaftMetadataCache to the kcontroller to simplify implemeting describeConfigs (and
probably more APIs in the future). It's mainly a wrapper around MetadataImage, so there is
essentially no extra resource consumption.
- Split RuntimeLoggerManager out of ConfigAdminManager to handle the incrementalAlterConfigs
support for BROKER_LOGGER. This is now supported on kcontrollers as well as brokers.
- Fix bug in AuthHelper.computeDescribeClusterResponse that resulted in us always sending back
BROKER as the endpoint type, even on the kcontroller.
Miscellaneous:
- Fix a few places in exceptions and log messages where we wrote "broker" instead of "node".
For example, an exception in NodeApiVersions.java, and a log message in NetworkClient.java.
- Fix the slf4j log prefix used by KafkaRequestHandler logging so that request handlers on a
controller don't look like they're on a broker.
- Make the FinalizedVersionRange constructor public for the sake of a junit test.
- Add unit and integration tests for the above.
Reviewers: David Arthur <mumrah@gmail.com>, Doguscan Namal <namal.doguscan@gmail.com>
It offers a quickfix action for certain errors, includes a number of bug fixes and it
introduces a new warning by default (https://github.com/scala/scala/pull/10462).
In addition to the scala version bump, we also fix the new compiler warnings and
bump the scalafmt version (the previous version failed with the new scala version).
Release notes: https://github.com/scala/scala/releases/tag/v2.13.12
Reviewers: Divij Vaidya <diviv@amazon.com>, Satish Duggana <satishd@apache.org>
This PR is part of #13247
It contains changes to rewrite single test in java.
Intention is reduce changes in parent PR.
Reviewers: Luke Chen <showuon@gmail.com>, Taras Ledkov <tledkov@apache.org>
Resolves cache misses in checkstyle tasks due to absolute paths in configProperties.
Sets configDirectory extension property, which is made available by the checkstyle plugin as ${config_loc} in the checkstyle xml files, as shown in the Checkstyle Gradle docs. The absolute paths set in configProperties are then replaced by relative paths from configDirectory. Because the header and suppression config file names are static and only referenced once, these were removed from configProperties and the file names are given directly in checkstyle.xml
Reviewers: Divij Vaidya <diviv@amazon.com>
This continues the work of providing the groundwork for the fetch
refactoring work by introducing some new classes and refactoring the
existing code to use the new classes where applicable.
Changes:
* Minor clean up of the events classes to make data immutable,
private, and implement toString().
* Added IdempotentCloser which prevents a resource from being closed
more than once. It's general enough that it could be used elsewhere
in the project, but it's limited to the consumer internals for now.
* Split core Fetcher code into classes to buffer raw results
(FetchBuffer) and to collect raw results into ConsumerRecords
(FetchCollector). These can be tested and changed in isolation from
the core fetcher logic.
* Added NodeStatusDetector which abstracts methods from
ConsumerNetworkClient so that it and NetworkClientDelegate can be
used in AbstractFetch via the interface instead of using
ConsumerNetworkClient directly.
Reviewers: Jun Rao <junrao@gmail.com>
Implementation of KIP-580 to add exponential back-off to situations in which retry.backoff.ms
is used to delay backoff attempts. This KIP adds exponential backoff behavior with a maximum
controlled by a new config retry.backoff.max.ms, together with a +/- 20% of jitter to spread the
retry attempts of the client fleet.
Reviewers: Mayank Shekhar Narula <mayanks.narula@gmail.com>, Milind Luthra <i.milind.luthra@gmail.com>, Kirk True <kirk@mustardgrain.com>, Jun Rao<junrao@gmail.com>
This change introduces a remote log segment segment retention cleanup mechanism.
RemoteLogManager runs retention cleanup activity tasks on each leader replica. It assesses factors such as overall size and retention duration, subsequently removing qualified segments from remote storage. This process also involves adjusting the log-start-offset within the UnifiedLog accordingly. It also cleans up the segments which have epochs earlier than the earliest leader epoch in the current leader.
Co-authored-by: Satish Duggana <satishd@apache.org>
Co-authored-by: Kamal Chandraprakash <kamal.chandraprakash@gmail.com>
Reviewers: Jun Rao <junrao@gmail.com>, Divij Vaidya <diviv@amazon.com, Luke Chen <showuon@gmail.com>, Kamal Chandraprakash <kamal.chandraprakash@gmail.com>, Christo Lolov <lolovc@amazon.com>, Jorge Esteban Quilcate Otoya <quilcate.jorge@gmail.com>, Alexandre Dupriez <alexandre.dupriez@gmail.com>, Nikhil Ramakrishnan <ramakrishnan.nikhil@gmail.com>
`TieredStorageTestHarness` is a base class for integration tests exercising the tiered storage functionality. This uses `LocalTieredStorage` instance as the second-tier storage system and `TopicBasedRemoteLogMetadataManager` as the remote log metadata manager.
Co-authored-by: Alexandre Dupriez <alexandre.dupriez@gmail.com>
Co-authored-by: Kamal Chandraprakash <kamal.chandraprakash@gmail.com>
Reviewers: David Arthur <mumrah@gmail.com>, Ron Dagostino <rndgstn@gmail.com>, Manikumar Reddy <manikumar.reddy@gmail.com>, Viktor Somogyi <viktor.somogyi@cloudera.com>
Only initialize remote topic metrics when system-wise remote storage is enabled to avoid impacting performance for existing brokers. Also add tests.
Reviewers: Divij Vaidya <diviv@amazon.com>, Kamal Chandraprakash <kamal.chandraprakash@gmail.com>
* KAFKA-15107: Support custom metadata for remote log segment
This commit does the changes discussed in the KIP-917. Mainly, changes the `RemoteStorageManager` interface in order to return `CustomMetadata` and then ensures these custom metadata are stored, propagated, (de-)serialized correctly along with the standard metadata throughout the whole lifecycle. It introduces the `remote.log.metadata.custom.metadata.max.size` to limit the custom metadata size acceptable by the broker and stop uploading in case a piece of metadata exceeds this limit.
On testing:
1. `RemoteLogManagerTest` checks the case when a piece of custom metadata is larger than the configured limit.
2. `RemoteLogSegmentMetadataTest` checks if `createWithUpdates` works correctly, including custom metadata.
3. `RemoteLogSegmentMetadataTransformTest`, `RemoteLogSegmentMetadataSnapshotTransformTest`, and `RemoteLogSegmentMetadataUpdateTransformTest` were added to test the corresponding class (de-)serialization, including custom metadata.
4. `FileBasedRemoteLogMetadataCacheTest` checks if custom metadata are being correctly saved and loaded to a file (indirectly, via `equals`).
5. `RemoteLogManagerConfigTest` checks if the configuration setting is handled correctly.
Reviewers: Luke Chen <showuon@gmail.com>, Satish Duggana <satishd@apache.org>, Divij Vaidya <diviv@amazon.com>
Implement some of the metrics from KIP-938: Add more metrics for
measuring KRaft performance.
Add these metrics to QuorumControllerMetrics:
kafka.controller:type=KafkaController,name=TimedOutBrokerHeartbeatCount
kafka.controller:type=KafkaController,name=EventQueueOperationsStartedCount
kafka.controller:type=KafkaController,name=EventQueueOperationsTimedOutCount
kafka.controller:type=KafkaController,name=NewActiveControllersCount
Create LoaderMetrics with these new metrics:
kafka.server:type=MetadataLoader,name=CurrentMetadataVersion
kafka.server:type=MetadataLoader,name=HandleLoadSnapshotCount
Create SnapshotEmitterMetrics with these new metrics:
kafka.server:type=SnapshotEmitter,name=LatestSnapshotGeneratedBytes
kafka.server:type=SnapshotEmitter,name=LatestSnapshotGeneratedAgeMs
Reviewers: Ron Dagostino <rndgstn@gmail.com>
This patch does a few things:
1) It introduces the `OffsetAndMetadata` class which hold the committed offsets in the group coordinator.
2) It adds methods to deal with OffsetCommit records to `RecordHelpers`.
3) It adds `MetadataVersion#offsetCommitValueVersion` to get the version of the OffsetCommit value record that should be used.
Reviewers: Jeff Kim <jeff.kim@confluent.io>, David Arthur <mumrah@gmail.com>, Justine Olshan <jolshan@confluent.io>
This PR adds CommandDefaultOptions usage like in the other joptsimple based tools. It also moves the associated unit test class from streams to tools module as discussed in #13127 (comment)
Reviewers: Luke Chen <showuon@gmail.com>, Bruno Cadonna <cadonna@apache.org>, Sagar Rao <sagarmeansocean@gmail.com>
This patch implements the existing JoinGroup protocol within the new group coordinator.
Some notable differences:
* Methods return a CoordinatorResult to the runtime framework, which includes records to append to the log as well as a future to complete after the append succeeds/fails.
* The coordinator runtime ensures that only a single thread will be processing a group at any given time, therefore there is no more locking on groups.
* Instead of using on purgatories, we rely on the Timer interface to schedule/cancel delayed operations.
Reviewers: David Jacot <djacot@confluent.io>
* KAFKA-14953: Adding RemoteLogManager metrics
In this PR, I have added the following metrics that are related to tiered storage mentioned in[ KIP-405](https://cwiki.apache.org/confluence/display/KAFKA/KIP-405%3A+Kafka+Tiered+Storage).
|Metric|Description|
|-----------------------------------------|--------------------------------------------------------------|
| RemoteReadRequestsPerSec | Number of remote storage read requests per second |
| RemoteWriteRequestsPerSec | Number of remote storage write requests per second |
| RemoteBytesInPerSec | Number of bytes read from remote storage per second |
| RemoteReadErrorsPerSec | Number of remote storage read errors per second |
| RemoteBytesOutPerSec | Number of bytes copied to remote storage per second |
| RemoteWriteErrorsPerSec | Number of remote storage write errors per second |
| RemoteLogReaderTaskQueueSize | Number of remote storage read tasks pending for execution. |
| RemoteLogReaderAvgIdlePercent | Average idle percent of the remote storage reader thread pool|
| RemoteLogManagerTasksAvgIdlePercent | Average idle percent of RemoteLogManager thread pool |
Added unit tests for all the rate metrics.
Reviewers: Luke Chen <showuon@gmail.com>, Divij Vaidya <diviv@amazon.com>, Kamal Chandraprakash<kamal.chandraprakash@gmail.com>, Jorge Esteban Quilcate Otoya <quilcate.jorge@gmail.com>, Staniel Yao<yaolixinylx@gmail.com>, hudeqi<1217150961@qq.com>, Satish Duggana <satishd@apache.org>
This patch wires the new group coordinator in BrokerServer (KRaft only). With this, it is now possible to run a cluster with the new group coordinator and to use the ConsumerGroupHeartbeat API by specifying the following two properties:
- group.coordinator.new.enable = true (to enable the new group coordinator)
- unstable.api.versions.enable = true (to enable unreleased APIs)
Note that the new group coordinator does not support all the existing APIs yet.
Reviewers: Jeff Kim <jeff.kim@confluent.io>, Justine Olshan <jolshan@confluent.io>
KAFKA-14522 Rewrite and Move of RemoteIndexCache to storage module.
Cleanedup index file suffix usages and other minor cleanups
Reviewers: Jun Rao <junrao@gmail.com>, Ismael Juma <ismael@juma.me.uk>, Luke Chen <showuon@gmail.com>, Divij Vaidya <diviv@amazon.com>, Kamal Chandraprakash<kamal.chandraprakash@gmail.com>, Jorge Esteban Quilcate Otoya <quilcate.jorge@gmail.com>
This patch rewrites MockTimer in Java and moves it from core to server-common. This continues the work started in https://github.com/apache/kafka/pull/13820.
Reviewers: Divij Vaidya <diviv@amazon.com>
This patch adds (1) the logic to propagate a new MetadataImage to the running coordinators; and (2) the logic to ensure that all the consumer groups subscribed to topics with changes will refresh their subscriptions metadata on the next heartbeat. In the mean time, it ensures that freshly loaded consumer groups also refresh their subscriptions metadata on the next heartbeat.
Reviewers: Jeff Kim <jeff.kim@confluent.io>, Justine Olshan <jolshan@confluent.io>
Poison the transaction manager if we detect an illegal transition in the Sender thread. A ThreadLocal in is stored in TransactionManager so that the Sender can inform TransactionManager which thread it's using.
Reviewers: Daniel Urban <durban@cloudera.com>, Justine Olshan <jolshan@confluent.io>, Jason Gustafson <jason@confluent.io>
This patch introduces the GroupCoordinatorService. This is the new (incomplete) implementation of the group coordinator based on the coordinator runtime introduced in https://github.com/apache/kafka/pull/13795.
Reviewers: Divij Vaidya <diviv@amazon.com>, Justine Olshan <jolshan@confluent.io>
The Java rewrite is kept relatively close to the Scala original
to minimize potential newly introduced bugs and to make reviewing
simpler. The following details might be of note:
- The `Logging` trait moved to InterBrokerSendThread with the
rewrite of ShutdownableThread has been similarly moved to any
subclasses that currently use it. InterBrokerSendThread's own
logging has been made to use ShutdownableThread's logger which
mimics the prefix/log identifier that the trait provided.
- The case RequestAndCompletionHandler class has been made a
separate POJO class and the internal-use UnsentRequests class
has been kept as a static nested class.
- The relatively commonly used but internal (not part of the
public API) clients classes that InterBrokerSendThread relies on
have been allowlisted in the server-common import control.
- The accompanying test class has also been moved and rewritten
with one new test added and most of the pre-existing tests made
stricter.
Reviewers: David Jacot <djacot@confluent.io>
This PR expands the scope of ApiVersionManager a bit to include returning the current
MetadataVersion and features that are in effect. This is useful in general because that information
needs to be returned in an ApiVersionsResponse. It also allows us to fix the ApiVersionManager
interface so that all subclasses implement all methods of the interface. Having subclasses that
don't implement some methods is dangerous because they could cause exceptions at runtime in
unexpected scenarios.
On the KRaft controller, we were previously performing a read operation in the QuorumController
thread to get the current metadata version and features. With this PR, we now read a volatile
variable maintained by a separate MetadataVersionContextPublisher object. This will improve
performance and simplify the code. It should not change the guarantees we are providing; in both
the old and new scenarios, we need to be robust against version skew scenarios during updates.
Add a Features class which just has a 3-tuple of metadata version, features, and feature epoch.
Remove MetadataCache.FinalizedFeaturesAndEpoch, since it just duplicates the Features class.
(There are some additional feature-related classes that can be consolidated in in a follow-on PR.)
Create a java class, EndpointReadyFutures, for managing the futures associated with individual
authorizer endpoints. This avoids code duplication between ControllerServer and BrokerServer and
makes this code unit-testable.
Reviewers: David Arthur <mumrah@gmail.com>, dengziming <dengziming1993@gmail.com>, Luke Chen <showuon@gmail.com>
This patch rewrite `MockTime` in Java and moves it to `server-common` module. This is a prerequisite to move `MockTimer` later on to `server-common` as well.
Reviewers: David Arthur <mumrah@gmail.com>
The OffsetFetcher is internally used by the KafkaConsumer to fetch offsets, validate and reset positions. For the new KafkaConsumer with a refactored threading model, similar functionality will be needed.
This is an initial refactoring for extracting logic from the OffsetFetcher, that will be reused by the new consumer implementation. No changes to the existing logic, just extracting classes, functions or pieces of logic.
All the functionality moved out of the OffsetFetcher is already covered by tests in OffsetFetcherTest and FetcherTest. There were no individual tests for the extracted functions, so no tests were migrated.
Reviewers: Jun Rao <junrao@gmail.com>
Adds CoordinatorEvent, CoordinatorEventProcessor, and MultiThreadedEventProcessor.
Reviewers: Kirk True <ktrue@confluent.io>, Jeff Kim <jeff.kim@confluent.io>, Justine Olshan <jolshan@confluent.io>
This patch adds the GroupMetadataManager to the group-coordinator module. This manager is responsible for handling the groups management, the members management and the entire reconciliation process. At this point, only the new consumer group type/protocol is implemented.
The new manager is based on an architecture inspired from the quorum controller. A request can access/read the state but can't mutate it directly. Instead, a list of records is generated together with the response and those records are applied to the state by the runtime framework. We use timeline data structures. Note that the runtime framework is not part of this patch. It will come in a following one.
Reviewers: Jeff Kim <jeff.kim@confluent.io>, Justine Olshan <jolshan@confluent.io>
Previously, if a user tried to perform an overly large batch operation on the KRaft controller
(such as creating a million topics), we would create a very large number of records in memory. Our
attempt to write these records to the Raft layer would fail, because there were too many to fit in
an atomic batch. This failure, in turn, would trigger a controller failover.
(Note: I am assuming here that no topic creation policy was in place that would prevent the
creation of a million topics. I am also assuming that the user operation must be done atomically,
which is true for all current user operations, since we have not implemented KIP-868 yet.)
With this PR, we fail immediately when the number of records we have generated exceeds the
threshold that we can apply. This failure does not generate a controller failover. We also now
fail with a PolicyViolationException rather than an UnknownServerException.
In order to implement this in a simple way, this PR adds the BoundedList class, which wraps any
list and adds a maximum length. Attempts to grow the list beyond this length cause an exception to
be thrown.
Reviewers: David Arthur <mumrah@gmail.com>, Ismael Juma <ijuma@apache.org>, Divij Vaidya <diviv@amazon.com>
Metadata image classes such as MetadataImage, ClusterImage, FeaturesImage, and so forth contain
numerous sub-images. This PR adds a structured way of traversing those sub-images. This is useful
for the metadata shell, and also for implementing toString functions.
In both cases, the previous solution was suboptimal. The metadata shell was previously implemented
in an ad-hoc way by mutating text-based tree nodes when records were replayed. This was difficult
to keep in sync with changes to the record types (for example, we forgot to do this for SCRAM). It
was also pretty low-level, being done at a level below that of the image classes. For toString, it
was difficult to keep the implementations consistent previously, and also support both redacted and
non-redacted output.
The metadata shell directory was getting crowded since we never had submodules for it. This PR
creates glob/, command/, node/, and state/ directories to keep things better organized.
Reviewers: David Arthur <mumrah@gmail.com>, Ron Dagostino <rdagostino@confluent.io>
This change includes
- Recognize the fetch requests with out of range local log offsets
- Add fetch implementation for the data lying in the range of [logStartOffset, localLogStartOffset]
- Add a new purgatory for async remote read requests which are served through a specific thread pool
We have an extended version of remote fetch that can fetch from multiple remote partitions in parallel, which we will raise as a followup PR.
A few tests for the newly introduced changes are added in this PR. There are some tests available for these scenarios in 2.8.x, refactoring with the trunk changes, will add them in followup PRs.
Other contributors:
Kamal Chandraprakash <kchandraprakash@uber.com> - Further improvements and adding a few tests
Luke Chen <showuon@gmail.com> - Added a few test cases for these changes.
PS: This functionality is pulled out from internal branches with other functionalities related to the feature in 2.8.x. The reason for not pulling all the changes as it makes the PR huge, and burdensome to review and it also needs other metrics, minor enhancements(including perf), and minor changes done for tests. So, we will try to have followup PRs to cover all those.
Reviewers: Jun Rao <junrao@gmail.com>, Alexandre Dupriez <alexandre.dupriez@gmail.com>, Divij Vaidya <diviv@amazon.com>, Jorge Esteban Quilcate Otoya <quilcate.jorge@gmail.com>
This patch introduces `GenericGroup` which rewrite the `GroupMetadata` in Java. The `GenericGroup` is basically a group using the current rebalance protocol in the new group coordinator.
Reviewers: Ritika Reddy <rreddy@confluent.io>, Christo Lolov <lolovc@amazon.com>, David Jacot <djacot@confluent.io>
This patch adds support for handling metadata snapshots while in dual-write mode. Prior to this change, if the active
controller loaded a snapshot, it would get out of sync with the ZK state.
In order to reconcile the snapshot state with ZK, several methods were added to scan through the metadata in ZK to
compute differences with the MetadataImage. Since this introduced a lot of code, I opted to split out a lot of methods
from ZkMigrationClient into their own client interfaces, such as TopicMigrationClient, ConfigMigrationClient, and
AclMigrationClient. Each of these has some iterator method that lets the caller examine the ZK state in a single pass
and without using too much memory.
Reviewers: Colin P. McCabe <cmccabe@apache.org>, Luke Chen <showuon@gmail.com>
Handle migrating SCRAM records in ZK when migrating from ZK to KRaft.
This includes handling writing back SCRAM records to ZK while in dual write mode where metadata updates are written to both the KRaft metadata log and to ZK. This allows for rollback of migration to include SCRAM metadata changes.
Reviewers: David Arthur <mumrah@gmail.com>
1. add ZkMigrationReady in apiVersionsResponse
2. check all nodes if ZkMigrationReady are ready before moving to next migration state
Reviewers: David Arthur <mumrah@gmail.com>, dengziming <dengziming1993@gmail.com>
This patch adds the concept of pre-migration mode to the KRaft controller. While in this mode,
the controller will only allow certain write operations. The purpose of this is to disallow metadata
changes when the controller is waiting for the ZK migration records to be committed.
The following ControllerWriteEvent operations are permitted in pre-migration mode
* completeActivation
* maybeFenceReplicas
* writeNoOpRecord
* processBrokerHeartbeat
* registerBroker (only for migrating ZK brokers)
* unregisterBroker
Raft events and other controller events do not follow the same code path as ControllerWriteEvent,
so they are not affected by this new behavior.
This patch also add a new metric as defined in KIP-868: kafka.controller:type=KafkaController,name=ZkMigrationState
In order to support upgrades from 3.4.0, this patch also redefines the enum value of value 1 to mean
MIGRATION rather than PRE_MIGRATION.
Reviewers: Chia-Ping Tsai <chia7712@gmail.com>, Colin P. McCabe <cmccabe@apache.org>
This patch adds ConsumerGroupMember.
Reviewers: Christo Lolov <lolovc@amazon.com>, Jeff Kim <jeff.kim@confluent.io>, Jason Gustafson <jason@confluent.io>
This patch renames from `ControllerPurgatory` to `DeferredEventQueue` and moves it from the `metadata` module to `server-common` module.
Reviewers: Alexandre Dupriez <alexandre.dupriez@gmail.com>, Ziming Deng <dengziming1993@gmail.com>, José Armando García Sancio <jsancio@apache.org>
Rework UserScramCredentialRecord to store serverKey and StoredKey rather than saltedPassword. This
is necessary to support migration from ZK, since those are the fields we stored in ZK. Update
latest MetadataVersion to IBP_3_5_IV2 and make SCRAM support conditional on this version. Moved
ScramCredentialData.java from org.apache.kafka.image to org.apache.kafka.metadata, which seems more
appropriate.
Reviewers: Colin P. McCabe <cmccabe@apache.org>
topic counts.
Introduces the use of persistent data structures in the KRaft metadata image to avoid copying the entire TopicsImage upon every change. Performance that was O(<number of topics in the cluster>) is now O(<number of topics changing>), which has dramatic time and GC improvements for the most common topic-related metadata events. We abstract away the chosen underlying persistent collection library via ImmutableMap<> and ImmutableSet<> interfaces and static factory methods.
Reviewers: Luke Chen <showuon@gmail.com>, Colin P. McCabe <cmccabe@apache.org>, Ismael Juma <ismael@juma.me.uk>, Purshotam Chauhan <pchauhan@confluent.io>
This PR updates foreign-key table-table join processors to ignore out-of-order records from versioned tables, as specified in KIP-914.
Reviewers: Matthias J. Sax <matthias@confluent.io>
This PR updates primary-key table-table join processors to ignore out-of-order records from versioned tables, as specified in KIP-914.
Reviewers: Matthias J. Sax <matthias@confluent.io>
This patch adds Group, Record and Result.
Reviewers: Jason Gustafson <jason@confluent.io>, Jeff Kim <jeff.kim@confluent.io>, Justine Olshan <jolshan@confluent.io>
Added functionality to copy log segments, indexes to the target remote storage for each topic partition enabled with tiered storage. This involves creating scheduled tasks for all leader partition replicas to copy their log segments in sequence to tiered storage.
Reviewers: Jun Rao <junrao@gmail.com>, Luke Chen <showuon@gmail.com>
In preparation for updating DSL join processors to have updated semantics when versioned stores are used (cf KIP-914), this PR adds test coverage for out-of-order data in joins to the existing integration tests for stream-table joins and primary-key table-table joins. Follow-up PRs will build on top of this change by adding new tests for versioned stores, and the out-of-order data will produce different results in those settings.
Reviewers: Matthias J. Sax <matthias@confluent.io>
The SnapshotReader exposes the "last contained log time". This is mainly used during snapshot cleanup. The previous implementation used the append time of the snapshot record. This is not accurate as this is the time when the snapshot was created and not the log append time of the last record included in the snapshot.
The log append time of the last record included in the snapshot is store in the header control record of the snapshot. The header control record is the first record of the snapshot.
To be able to read this record, this change extends the RecordsIterator to decode and expose the control records in the Records type.
Reviewers: Colin Patrick McCabe <cmccabe@apache.org>
Until this PR, all the code added for KIP-889 for introducing versioned stores to Kafka Streams has been accessible from internal packages only. This PR exposes the stores via public Stores.java methods, and also updates the TopologyTestDriver.
Reviewers: Matthias J. Sax <matthias@confluent.io>
This patch refactors the loadCache method in AclAuthorizer to make it reusable by ZkMigrationClient.
The loaded ACLs are converted to AccessControlEntryRecord. I noticed we still have the defunct
AccessControlRecord, so I've deleted it.
Also included here are the methods to write ACL changes back to ZK while in dual-write mode.
Reviewers: Rajini Sivaram <rajinisivaram@googlemail.com>, Colin P. McCabe <cmccabe@apache.org>
* KAFKA-14365: Extract common logic from Fetcher
Extract logic from Fetcher into AbstractFetcher.
Also introduce FetchConfig as a more concise way to delineate state from
incoming configuration.
Formalized the defaults in CommonClientConfigs and ConsumerConfig to be
accessible elsewhere.
* Removed overridden methods in favor of synchronizing where needed
Reviewers: Guozhang Wang <wangguoz@gmail.com>
Separate out KRaft controller metrics into two groups: metrics directly managed by the
QuorumController, and metrics handled by an external publisher. This separation of concerns makes
the code easier to reason about, by clarifying what metrics can be changed where.
The external publisher, ControllerServerMetricsPublisher, handles all metrics which are related to
the content of metadata. For example, metrics about number of topics or number of partitions, etc.
etc. It fits into the MetadataLoader metadata publishing framework as another publisher. Since
ControllerServerMetricsPublisher operates off of a MetadataImage, we don't have to create
(essentially) another copy of the metadata in memory, as ControllerMetricsManager. This reduces
memory consumption. Another benefit of operating off of the MetadataImage is that we don't have to
have special handling for each record type, like we do now in ControllerMetricsManager.
Reviewers: David Arthur <mumrah@gmail.com>
The new group coordinator needs to access cluster metadata (e.g. topics, partitions, etc.) and it needs a mechanism to be notified when the metadata changes (e.g. to trigger a rebalance). In KRaft clusters, the easiest is to subscribe to metadata changes via the MetadataPublisher.
Reviewers: Justine Olshan <jolshan@confluent.io>
This commit adds support to store the SCRAM credentials in a cluster with KRaft quorum servers and
no ZK cluster backing the metadata. This includes creating ScramControlManager in the controller,
and adding support for SCRAM to MetadataImage and MetadataDelta.
Change UserScramCredentialRecord to contain only a single tuple (name, mechanism, salt, pw, iter)
rather than a mapping between name and a list. This will avoid creating an excessively large record
if a single user has many entries. Because record ID 11 (UserScramCredentialRecord) has not been
used before, this is a compatible change. SCRAM will be supported in 3.5-IV0 and later.
This commit does not include KIP-900 SCRAM bootstrapping support, or updating the credential cache
on the controller (as opposed to broker). We will implement these in follow-on commits.
Reviewers: Manikumar Reddy <manikumar.reddy@gmail.com>, Colin P. McCabe <cmccabe@apache.org>
…g option
https://github.com/apache/kafka/pull/12951 accidentally changed the behavior of the `kafaka-metadata-quorum.sh` CLI by making it silently ignore a `--command-config <filename>` properties file that exists. This was an undetected regression in the 3.4.0 release. This patch fixes the issue such that any such specified file will be honored.
Reviewers: José Armando García Sancio <jsancio@apache.org>, Ismael Juma <ismael@juma.me.uk>
This patch adds the broker side `PartitionAssignor` interface as detailed in KIP-848. The interfaces differs a bit from the KIP in the following ways:
* The POJOs are not defined within the interface because the interface is to heavy like this.
* The interface is kept in the `group-coordinator` module for now. We don't want to have it out there until KIP-848 is ready to be released. We will move it to its final destination later.
Reviewers: Jeff Kim <jeff.kim@confluent.io>, Justine Olshan <jolshan@confluent.io>, Christo Lolov <lolovc@amazon.com>, Guozhang Wang <wangguoz@gmail.com>
Only send UMR to ZK brokers if the cluster metadata or topic metadata has changed.
Reviewers: Akhilesh C <akhileshchg@users.noreply.github.com>, Colin P. McCabe <cmccabe@apache.org>
Reviewers: Daniel Urban <durban@cloudera.com>, Greg Harris <greg.harris@aiven.io>, Viktor Somogyi-Vass <viktorsomogyi@gmail.com>, Mickael Maison <mickael.maison@gmail.com>
KAFKA-14551 Move/Rewrite LeaderEpochFileCache and its dependencies to the storage module.
For broader context on this change, you may want to look at KAFKA-14470: Move log layer to the storage module
Reviewers: Ismael Juma <ismael@juma.me.uk>, Jun Rao <junrao@gmail.com>, Alexandre Dupriez <alexandre.dupriez@gmail.com>
This patch moves the current `__consumer_offsets` records from the `core` module to the new `group-coordinator` module.
Reviewers: Christo Lolov <lolovc@amazon.com>, Mickael Maison <mickael.maison@gmail.com>
This patch migrates all the internal APIs of the current group coordinator to the new `GroupCoordinator` interface. It also makes the current implementation package private to ensure that it is not used anymore.
Reviewers: Justine Olshan <jolshan@confluent.io>
If KafkaEventQueue gets an InterruptedException while waiting for a condition variable, it
currently exits immediately. Instead, it should complete the remaining events exceptionally and
then execute the cleanup event. This will allow us to finish any necessary cleanup steps.
In order to do this, we require the cleanup event to be provided when the queue is contructed,
rather than when it's being shut down.
Also, handle cases where Event#handleException itself throws an exception.
Remove timed shutdown from the event queue code since nobody was using it, and it adds complexity.
Add server-common/src/test/resources/test/log4j.properties since this gradle module somehow avoided
having a test log4j.properties up to this point.
Reviewers: David Arthur <mumrah@gmail.com>
This patch introduces a preliminary state machine that can be used by KRaft
controller to drive online migration from Zk to KRaft.
MigrationState -- Defines the states we can have while migration from Zk to
KRaft.
KRaftMigrationDriver -- Defines the state transitions, and events to handle
actions like controller change, metadata change, broker change and have
interfaces through which it claims Zk controllership, performs zk writes and
sends RPCs to ZkBrokers.
MigrationClient -- Interface that defines the functions used to claim and
relinquish Zk controllership, read to and write from Zk.
Co-authored-by: David Arthur <mumrah@gmail.com>
Reviewers: Colin P. McCabe <cmccabe@apache.org>
Additional notable changes to fix multiple dependency ordering issues:
* Moved `ConfigSynonym` to `server-common`
* Moved synonyms from `LogConfig` to `ServerTopicConfigSynonyms `
* Removed `LogConfigDef` `define` overrides and rely on
`ServerTopicConfigSynonyms` instead.
* Moved `LogConfig.extractLogConfigMap` to `KafkaConfig`
* Consolidated relevant defaults from `KafkaConfig`/`LogConfig` in the latter
* Consolidate relevant config name definitions in `TopicConfig`
* Move `ThrottledReplicaListValidator` to `storage`
Reviewers: Satish Duggana <satishd@apache.org>, Mickael Maison <mickael.maison@gmail.com>
Also improved `LogValidatorTest` to cover a bug that was originally
only caught by `LogAppendTimeTest`.
For broader context on this change, please check:
* KAFKA-14470: Move log layer to storage module
Reviewers: Jun Rao <junrao@gmail.com>
The controller metrics in the controllers has three problems. 1) the active controller exposes uncommitted data in the metrics. 2) the active controller doesn't update the metrics when the uncommitted data gets aborted. 3) the controller doesn't update the metrics when the entire state gets reset.
We fix these issues by only updating the metrics when processing committed metadata records and reset the metrics when the metadata state is reset.
This change adds a new type `ControllerMetricsManager` which processes committed metadata records and updates the metrics accordingly. This change also removes metrics updating responsibilities from the rest of the controller managers.
Reviewers: Ron Dagostino <rdagostino@confluent.io>
This PR introduces the new metadata loader and snapshot generator. For the time being, they are
only used by the controller, but a PR for the broker will come soon.
The new metadata loader supports adding and removing publishers dynamically. (In contrast, the old
loader only supported adding a single publisher.) It also passes along more information about each
new image that is published. This information can be found in the LogDeltaManifest and
SnapshotManifest classes.
The new snapshot generator replaces the previous logic for generating snapshots in
QuorumController.java and associated classes. The new generator is intended to be shared between
the broker and the controller, so it is decoupled from both.
There are a few small changes to the old snapshot generator in this PR. Specifically, we move the
batch processing time and batch size metrics out of BrokerMetadataListener.scala and into
BrokerServerMetrics.scala.
Finally, fix a case where we are using 'is' rather than '==' for a numeric comparison in
snapshot_test.py.
Reviewers: David Arthur <mumrah@gmail.com>
`core` should only be used for legacy cli tools and tools that require
access to `core` classes instead of communicating via the kafka protocol
(typically by using the client classes).
Summary of changes:
1. Convert the command implementation and tests to Java and move it to
the `tools` module.
2. Introduce mechanism to capture stdout and stderr from tests.
3. Change `kafka-metadata-quorum.sh` to point to the new command class.
4. Adjusted the test classpath of the `tools` module so that it supports tests
that rely on the `@ClusterTests` annotation.
5. Improved error handling when an exception different from `TerseFailure` is
thrown.
6. Changed `ToolsUtils` to avoid usage of arrays in favor of `List`.
Reviewers: dengziming <dengziming1993@gmail.com>
This patch adds support for reading and writing ZooKeeper metadata during a KIP-866 migration.
For reading metadata from ZK, methods from KafkaZkClient and ZkData are reused to ensure we are decoding the JSON consistently.
For writing metadata, we use a new multi-op transaction that ensures only a single controller is writing to ZK. This is similar to the existing multi-op transaction that KafkaController uses, but it also includes a check on the new "/migration" ZNode. The transaction consists of three operations:
* CheckOp on /controller_epoch
* SetDataOp on /migration with zkVersion
* CreateOp/SetDataOp/DeleteOp (the actual operation being applied)
In the case of a batch of operations (such as topic creation), only the final MultiOp has a SetDataOp on /migration while the other requests use a CheckOp (similar to /controller_epoch).
Reviewers: Colin Patrick McCabe <cmccabe@apache.org>, dengziming <dengziming1993@gmail.com>
Split out the logic for applying dynamic configurations to a KafkaConfig object from
BrokerMetadataPublisher into a new class, DynamicConfigPublisher. This will allow the
ControllerServer to also run this code, in a follow-up change.
Create separate KafkaConfig objects in BrokerServer versus ControllerServer. This is necessary
because the controller will apply configuration changes as soon as its raft client catches up to
the high water mark, whereas the broker will wait for the active controller to acknowledge it has
caught up in a heartbeat response. So when running in combined mode, we want two separate
KafkaConfig objects that are changed at different times.
Minor changes: improve the error message when catching up broker metadata fails. Fix incorrect
indentation in checkstyle/import-control.xml. Invoke AppInfoParser.unregisterAppInfo from
SharedServer.stop so that it happens only when both the controller and broker have shut down.
Reviewers: David Arthur <mumrah@gmail.com>
This patch adds `joinGroup` to the new `GroupCoordinator` interface and updates `KafkaApis` to use it.
For the context, I will do the same for all the other interactions with the current group coordinator. In order to limit the changes, I have chosen to introduce the `GroupCoordinatorAdapter` that translates the new interface to the old one. It is basically a wrapper. This allows keeping the current group coordinator untouched for now and focus on the `KafkaApis` changes. Eventually, we can remove `GroupCoordinatorAdapter`.
Reviewers: Justine Olshan <jolshan@confluent.io>, Jeff Kim <jeff.kim@confluent.io>, Luke Chen <showuon@gmail.com>, Jason Gustafson <jason@confluent.io>
PR implementing KIP-770 (#11424) was reverted as it brought in a regression wrt pausing/resuming the consumer. That KIP also introduced a change to deprecate config CACHE_MAX_BYTES_BUFFERING_CONFIG and replace it with STATESTORE_CACHE_MAX_BYTES_CONFIG.
Reviewers: Anna Sophie Blee-Goldman <ableegoldman@apache.org>
This PR adds a new ImageWriter interface which replaces the generic Consumer interface which
accepted lists of records. It is better to do batching in the ImageWriter than to try to deal with
that complexity in the MetadataImage#write functions, especially since batching is not semantically
meaningful in KRaft snapshots. The new ImageWriter interface also supports freeze and close, which
more closely matches the semantics of the underlying Raft classes.
The PR also adds an ImageWriterOptions class which we can use to pass parameters to control how the
new image is written. Right now, the parameters that we are interested in are the target metadata
version (which may be more or less than the original image's version) and a handler function which
is invoked whenever metadata is lost due to the target version.
Convert over the MetadataImage#write function (and associated functions) to use the new ImageWriter
and ImageWriterOptions. In particular, we now have a way to handle metadata losses by invoking
ImageWriterOptions#handleLoss. This allows us to handle writing an image at a lower version, for
the first time. This support is still not enabled externally by this PR, though. That will come in
a future PR.
Get rid of the use of SOME_RECORD_TYPE.highestSupportedVersion() in several places. In general, we
do not want to "silently" change the version of a record that we output, just because a new version
was added. We should be explicit about what record version numbers we are outputting.
Implement ProducerIdsDelta#toString, to make debug logs look better.
Move MockRandom to the server-common package so that other internal broker packages can use it.
Reviewers: José Armando García Sancio <jsancio@apache.org>
Also moves the Streams LogCaptureAppender class into the clients module so that it can be used by both Streams and Connect.
Reviewers: Nigel Liang <nigel@nigelliang.com>, Kalpesh Patel <kpatel@confluent.io>, John Roesler <vvcephei@apache.org>, Tom Bentley <tbentley@redhat.com>
In https://github.com/apache/kafka/pull/12695, we discovered a gap in our testing of `StandardAuthorizer`. We addressed the specific case that was failing, but I think we need to establish a better methodology for testing which incorporates randomized inputs. This patch is a start in that direction. We implement a few basic property tests using jqwik which focus on prefix searching. It catches the case from https://github.com/apache/kafka/pull/12695 prior to the fix. In the future, we can extend this to cover additional operation types, principal matching, etc.
Reviewers: David Arthur <mumrah@gmail.com>
* KAFKA-13725: KIP-768 OAuth code mixes public and internal classes in same package
Move classes into a sub-package of "internal" named "secured" that
matches the layout more closely of the "unsecured" package.
Replaces the concrete implementations in the former packages with
sub-classes of the new package layout and marks them as deprecated. If
anyone is already using the newer OAuth code, this should still work.
* Fix checkstyle and spotbugs violations
Co-authored-by: Kirk True <kirk@mustardgrain.com>
Reviewers: Manikumar Reddy <manikumar.reddy@gmail.com>
This adds a new configuration `sasl.server.max.receive.size` that sets the maximum receive size for requests before and during authentication.
Reviewers: Tom Bentley <tbentley@redhat.com>, Mickael Maison <mickael.maison@gmail.com>
Co-authored-by: Manikumar Reddy <manikumar.reddy@gmail.com>
Co-authored-by: Mickael Maison <mickael.maison@gmail.com>
When deserializing KRPC (which is used for RPCs sent to Kafka, Kafka Metadata records, and some
other things), check that we have at least N bytes remaining before allocating an array of size N.
Remove DataInputStreamReadable since it was hard to make this class aware of how many bytes were
remaining. Instead, when reading an individual record in the Raft layer, simply create a
ByteBufferAccessor with a ByteBuffer containing just the bytes we're interested in.
Add SimpleArraysMessageTest and ByteBufferAccessorTest. Also add some additional tests in
RequestResponseTest.
Reviewers: Tom Bentley <tbentley@redhat.com>, Mickael Maison <mickael.maison@gmail.com>, Colin McCabe <colin@cmccabe.xyz>
Co-authored-by: Colin McCabe <colin@cmccabe.xyz>
Co-authored-by: Manikumar Reddy <manikumar.reddy@gmail.com>
Co-authored-by: Mickael Maison <mickael.maison@gmail.com>
The main changes here are ensuring that we always have a metadata.version record in the log, making
˘sure that the bootstrap file can be used for records other than the metadata.version record (for
example, we will want to put SCRAM initialization records there), and fixing some bugs.
If no feature level record is in the log and the IBP is less than 3.3IV0, then we assume the minimum KRaft
version for all records in the log.
Fix some issues related to initializing new clusters. If there are no records in the log at all,
then insert the bootstrap records in a single batch. If there are records, but no metadata version,
process the existing records as though they were metadata.version 3.3IV0 and then append a metadata
version record setting version 3.3IV0. Previously, we were not clearly distinguishing between the
case where the metadata log was empty, and the case where we just needed to add a metadata.version
record.
Refactor BootstrapMetadata into an immutable class which contains a 3-tuple of metadata version,
record list, and source. The source field is used to log where the bootstrap metadata was obtained
from. This could be a bootstrap file, the static configuration, or just the software defaults.
Move the logic for reading and writing bootstrap files into BootstrapDirectory.java.
Add LogReplayTracker, which tracks whether the log is empty.
Fix a bug in FeatureControlManager where it was possible to use a "downgrade" operation to
transition to a newer version. Do not store whether we have seen a metadata version or not in
FeatureControlManager, since that is now handled by LogReplayTracker.
Introduce BatchFileReader, which is a simple way of reading a file containing batches of snapshots
that does not require spawning a thread. Rename SnapshotFileWriter to BatchFileWriter to be
consistent, and to reflect the fact that bootstrap files aren't snapshots.
QuorumController#processBrokerHeartbeat: add an explanatory comment.
Reviewers: David Arthur <mumrah@gmail.com>, Jason Gustafson <jason@confluent.io>
AccessControlEntryRecord and RemoveAccessControlEntryRecord are added in KIP-801, FeatureLevelRecord was added in KIP-778, and BrokerRegistrationChangeRecord was added in KIP-841, and NoOpRecord was added in KIP-835, I added these 5 record types in MetadataShell.
Reviewers: Luke Chen <showuon@gmail.com>
Add `MetadataQuorumCommand` to describe quorum status, I'm trying to use arg4j style command format, currently, we only support one sub-command which is "describe" and we can specify 2 arguments which are --status and --replication.
```
# describe quorum status
kafka-metadata-quorum.sh --bootstrap-server localhost:9092 describe --replication
ReplicaId LogEndOffset Lag LastFetchTimeMs LastCaughtUpTimeMs Status
0 10 0 -1 -1 Leader
1 10 0 -1 -1 Follower
2 10 0 -1 -1 Follower
kafka-metadata-quorum.sh --bootstrap-server localhost:9092 describe --status
ClusterId: fMCL8kv1SWm87L_Md-I2hg
LeaderId: 3002
LeaderEpoch: 2
HighWatermark: 10
MaxFollowerLag: 0
MaxFollowerLagTimeMs: -1
CurrentVoters: [3000,3001,3002]
CurrentObservers: [0,1,2]
# specify AdminClient properties
kafka-metadata-quorum.sh --bootstrap-server localhost:9092 --command-config config.properties describe --status
```
Reviewers: Jason Gustafson <jason@confluent.io>
Before trying to commit a batch of records to the __cluster_metadata log, the active controller
should try to apply them to its current in-memory state. If this application process fails, the
active controller process should exit, allowing another node to take leadership. This will prevent
most bad metadata records from ending up in the log and help to surface errors during testing.
Similarly, if the active controller attempts to renounce leadership, and the renunciation process
itself fails, the process should exit. This will help avoid bugs where the active controller
continues in an undefined state.
In contrast, standby controllers that experience metadata application errors should continue on, in
order to avoid a scenario where a bad record brings down the whole controller cluster. The
intended effect of these changes is to make it harder to commit a bad record to the metadata log,
but to continue to ride out the bad record as well as possible if such a record does get committed.
This PR introduces the FaultHandler interface to implement these concepts. In junit tests, we use a
FaultHandler implementation which does not exit the process. This allows us to avoid terminating
the gradle test runner, which would be very disruptive. It also allows us to ensure that the test
surfaces these exceptions, which we previously were not doing (the mock fault handler stores the
exception).
In addition to the above, this PR fixes a bug where RaftClient#resign was not being called from the
renounce() function. This bug could have resulted in the raft layer not being informed of an active
controller resigning.
Reviewers: David Arthur <mumrah@gmail.com>
KIP-770 introduced a performance regression and needs some re-design.
Needed to resolve some conflict while reverting.
This reverts commits 1317f3f77a and 0924fd3f9f.
Reviewers: Sagar Rao <sagarmeansocean@gmail.com>, Guozhang Wang <guozhang@confluent.io>
This PR adds the ability to pause and resume KafkaStreams instances as well as named/modular topologies (KIP-834).
Co-authored-by: Bruno Cadonna <cadonna@apache.org>
Reviewers: Bonnie Varghese <bvarghese@confluent.io>, Walker Carlson <wcarlson@confluent.io>, Guozhang Wang <guozhang@apache.org>, Bruno Cadonna <cadonna@apache.org>
New gradle task `connect:runtime:genConnectOpenAPIDocs` that generates `connect_rest.yaml` under `docs/generated`.
This task is executed when `siteDocsTar` runs.
Implementation of KIP-846: Source/sink node metrics for Consumed/Produced throughput in Streams
Adds the following INFO topic-level metrics for the total bytes/records consumed and produced:
bytes-consumed-total
records-consumed-total
bytes-produced-total
records-produced-total
Reviewers: Kvicii <Karonazaba@gmail.com>, Guozhang Wang <guozhang@apache.org>, Bruno Cadonna <cadonna@apache.org>
Implements the behavior described in KIP-618: using a transactional producer for writes to the config topic that should only be performed by the leader of the cluster.
Reviewers: Luke Chen <showuon@gmail.com>, Tom Bentley <tbentley@redhat.com>
This PR refactors the leader API access in the follower fetch path.
Added a LeaderEndPoint interface which serves all access to the leader.
Added a LocalLeaderEndPoint and a RemoteLeaderEndPoint which implements the LeaderEndPoint interface to handle fetches from leader in local & remote storage respectively.
Reviewers: David Jacot <djacot@confluent.io>, Kowshik Prakasam <kprakasam@confluent.io>, Jun Rao <junrao@gmail.com>
This patch builds on #12072 and adds controller support for metadata.version. The kafka-storage tool now allows a
user to specify a specific metadata.version to bootstrap into the cluster, otherwise the latest version is used.
Upon the first leader election of the KRaft quroum, this initial metadata.version is written into the metadata log. When
writing snapshots, a FeatureLevelRecord for metadata.version will be written out ahead of other records so we can
decode things at the correct version level.
This also includes additional validation in the controller when setting feature levels. It will now check that a given
metadata.version is supportable by the quroum, not just the brokers.
Reviewers: José Armando García Sancio <jsancio@gmail.com>, Colin P. McCabe <cmccabe@apache.org>, dengziming <dengziming1993@gmail.com>, Alyssa Huang <ahuang@confluent.io>
These tests belongs to ApiVersionsResponseTest, and accidentally copied them to MetadataVersionTest when working on #12072.
Reviewer: Luke Chen <showuon@gmail.com>
Refactoring ApiVersion to MetadataVersion to support both old IBP versioning and new KRaft versioning (feature flags)
for KIP-778.
IBP versions are now encoded as enum constants and explicitly prefixed w/ IBP_ instead of KAFKA_, and having a
LegacyApiVersion vs DefaultApiVersion was not necessary and replaced with appropriate parsing rules for extracting
the correct shortVersions/versions.
Co-authored-by: David Arthur <mumrah@gmail.com>
Reviewers: Ismael Juma <ismael@juma.me.uk>, David Arthur <mumrah@gmail.com>, dengziming <dengziming1993@gmail.com>, Colin P. McCabe <cmccabe@apache.org>
Updates the KStream process API to cover the use cases
of both process and transform, and deprecate the KStream transform API.
Implements KIP-820
Reviewer: John Roesler <vvcephei@apache.org>
Ensure that we can set log.flush.interval.ms at the broker or cluster level via
IncrementalAlterConfigs. This was broken by KAFKA-13749, which added log.flush.interval.ms as the
second synonym rather than the first. Add a regression test to DynamicConfigChangeTest.
Create ControllerRequestContext and pass it to every controller API. This gives us a uniform way to
pass through information like the deadline (if there is one) and the Kafka principal which is
making the request (in the future we will want to log this information).
In ControllerApis, enforce a timeout for broker heartbeat requests which is equal to the heartbeat
request interval, to avoid heartbeats piling up on the controller queue. This should have been done
previously, but we overlooked it.
Add a builder for ClusterControlManager and ReplicationControlManager to avoid the need to deal
with a lot of churn (especially in test code) whenever a new constructor parameter gets added for
one of these.
In ControllerConfigurationValidator, create a separate function for when we just want to validate
that a ConfigResource is a valid target for DescribeConfigs. Previously we had been re-using the
validation code for IncrementalAlterConfigs, but this was messy.
Split out the replica placement code into a separate package and reorganize it a bit.
Reviewers: David Arthur <mumrah@gmail.com
Prior to this commit FK response sink routed FK results to
SubscriptionResolverJoinProcessorSupplier using the primary key.
There are cases, where this behavior is incorrect. For example,
if KTable key serde differs from the data source serde which might
happen without a key changing operation.
Instead of determining the resolver partition by serializing the PK
this patch includes target partition in SubscriptionWrapper payloads.
Default FK response-sink partitioner extracts the correct partition
from the value and routes the message accordingly.
Reviewers: Matthias J. Sax <matthias@confluent.io>
This PR includes the changes to feature flags that were outlined in KIP-778. Specifically, it
changes UpdateFeatures and FeatureLevelRecord to remove the maximum version level. It also adds
dry-run to the RPC so the controller can actually attempt the upgrade (rather than the client). It
introduces an upgrade type enum, which supersedes the allowDowngrade boolean. Because
FeatureLevelRecord was unused previously, we do not need to introduce a new version.
The kafka-features.sh tool was overhauled in KIP-778 and now includes the describe, upgrade,
downgrade, and disable sub-commands. Refer to
[KIP-778](https://cwiki.apache.org/confluence/display/KAFKA/KIP-778%3A+KRaft+Upgrades) for more
details on the new command structure.
Reviewers: Colin P. McCabe <cmccabe@apache.org>, dengziming <dengziming1993@gmail.com>
Previously, when in KRaft mode, CreateTopics did not return the active configurations for the
topic(s) it had just created. This PR addresses that gap. We will now return these topic
configuration(s) when the user has DESCRIBE_CONFIGS permission. (In the case where the user does
not have this permission, we will omit the configurations and set TopicErrorCode. We will also omit
the number of partitions and replication factor data as well.)
For historical reasons, we use different names to refer to each topic configuration when it is set
in the broker context, as opposed to the topic context. For example, the topic configuration
"segment.ms" corresponds to the broker configuration "log.roll.ms". Additionally, some broker
configurations have synonyms. For example, the broker configuration "log.roll.hours" can be used to
set the log roll time instead of "log.roll.ms". In order to track all of this, this PR adds a
table in LogConfig.scala which maps each topic configuration to an ordered list of ConfigSynonym
classes. (This table is then passed to KafkaConfigSchema as a constructor argument.)
Some synonyms require transformations. For example, in order to convert from "log.roll.hours" to
"segment.ms", we must convert hours to milliseconds. (Note that our assumption right now is that
topic configurations do not have synonyms, only broker configurations. If this changes, we will
need to add some logic to handle it.)
This PR makes the 8-argument constructor for ConfigEntry public. We need this in order to make full
use of ConfigEntry outside of the admin namespace. This change is probably inevitable in general
since otherwise we cannot easily test the output from various admin APIs in junit tests outside the
admin package.
Testing:
This PR adds PlaintextAdminIntegrationTest#testCreateTopicsReturnsConfigs. This test validates
some of the configurations that it gets back from the call to CreateTopics, rather than just checking
if it got back a non-empty map like some of the existing tests. In order to test the
configuration override logic, testCreateDeleteTopics now sets up some custom static and dynamic
configurations.
In QuorumTestHarness, we now allow tests to configure what the ID of the controller should be. This
allows us to set dynamic configurations for the controller in testCreateDeleteTopics. We will have
a more complete fix for setting dynamic configuations on the controller later.
This PR changes ConfigurationControlManager so that it is created via a Builder. This will make it
easier to add more parameters to its constructor without having to update every piece of test code
that uses it. It will also make the test code easier to read.
Reviewers: David Arthur <mumrah@gmail.com>
Since the topology-level cache size config only controls whether we disable the caching layer entirely for that topology, setting it to anything other than 0 has no effect. The actual cache memory is still just split evenly between the threads, and shared by all topologies.
It's possible we'll want to change this in the future, but for now we should make sure to log a warning so that users who do try to set this override to some nonzero value are made aware that it doesn't work like this.
Also includes some minor refactoring plus a fix for an off-by-one error in #11796
Reviewers: Luke Chen <showuon@gmail.com>, Walker Carlson <wcarlson@confluent.io>, Sagar Rao <sagarmeansocean@gmail.com>
With major server components like the new quorum controller being moved outside of the `core` module, it is useful to have shared dependencies moved into `server-common`. An example of this is Yammer metrics which server components still rely heavily upon. All server components should have access to the default registry used by the broker so that new metrics can be registered and metric naming conventions should be standardized. This is particularly important in KRaft where we are attempting to recreate identically named metrics in the controller context. This patch takes a step in this direction. It moves `KafkaYammerMetrics` into `server-common` and it implements
standard metric naming utilities there.
Reviewers: Manikumar Reddy <manikumar.reddy@gmail.com>
This patch fixes a bug in the `AlterConfigPolicy.RequestMetadata.equals` method where we were not comparing the class correctly.
Co-authored-by: David Jacot <djacot@confluent.io>
Reviewers: David Jacot <djacot@confluent.io>
Create KafkaConfigSchema to encapsulate the concept of determining the types of configuration keys.
This is useful in the controller because we can't import KafkaConfig, which is part of core. Also
introduce the TimelineObject class, which is a more generic version of TimelineInteger /
TimelineLong.
Reviewers: David Arthur <mumrah@gmail.com>
Currently, when using KRaft mode, users still have to have an Apache ZooKeeper instance if they want to use AclAuthorizer. We should have a built-in Authorizer for KRaft mode that does not depend on ZooKeeper. This PR introduces such an authorizer, called StandardAuthorizer. See KIP-801 for a full description of the new Authorizer design.
Authorizer.java: add aclCount API as described in KIP-801. StandardAuthorizer is currently the only authorizer that implements it, but eventually we may implement it for AclAuthorizer and others as well.
ControllerApis.scala: fix a bug where createPartitions was authorized using CREATE on the topic resource rather than ALTER on the topic resource as it should have been.
QuorumTestHarness: rename the controller endpoint to CONTROLLER for consistency (the brokers already called it that). This is relevant in AuthorizerIntegrationTest where we are examining endpoint names. Also add the controllerServers call.
TestUtils.scala: adapt the ACL functions to be usable from KRaft, by ensuring that they use the Authorizer from the current active controller.
BrokerMetadataPublisher.scala: add broker-side ACL application logic.
Controller.java: add ACL APIs. Also add a findAllTopicIds API in order to make junit tests that use KafkaServerTestHarness#getTopicNames and KafkaServerTestHarness#getTopicIds work smoothly.
AuthorizerIntegrationTest.scala: convert over testAuthorizationWithTopicExisting (more to come soon)
QuorumController.java: add logic for replaying ACL-based records. This means storing them in the new AclControlManager object, and integrating them into controller snapshots. It also means applying the changes in the Authorizer, if one is configured. In renounce, when reverting to a snapshot, also set newBytesSinceLastSnapshot to 0.
Reviewers: YeonCheol Jang <YeonCheolGit@users.noreply.github.com>, Jason Gustafson <jason@confluent.io>
During some recent reviews, @mjsax pointed out that StateStore layers
are constructed differently the stores are added via the PAPI vs. the DSL.
This PR adds KeyValueStore PAPI construction to the
IQv2StoreIntegrationTest so that we can ensure IQv2 works on every
possible state store.
Reviewers: Patrick Stuedi <pstuedi@apache.org>, Guozhang Wang <guozhang@apache.org>
Require that topics exist before topic configurations can be created for them.
Merge the code from ConfigurationControlManager#checkConfigResource into
ControllerConfigurationValidator to avoid duplication.
Add KRaft support to DynamicConfigChangeTest.
Split out tests in DynamicConfigChangeTest that don't require a cluster into
DynamicConfigChangeUnitTest to save test time.
Reviewers: David Arthur <mumrah@gmail.com>
Implements the major part of the IQv2 framework as proposed in KIP-796.
Reviewers: Patrick Stuedi <pstuedi@apache.org>, Vicky Papavasileiou <vpapavasileiou@confluent.io>, Bruno Cadonnna <cadonna@apache.org>
With the changes for topic IDs, we have a different flow. When a broker receives a request, it uses a map to convert the topic ID to topic names. If the topic ID is not found in the map, we return a top level error and close the session. This decision was motivated by the difficulty to store “unresolved” partitions in the session. In earlier iterations we stored an “unresolved” partition object in the cache, but it was somewhat hard to reason about and required extra logic to try to resolve the topic ID on each incremental request and add to the session. It also required extra logic to forget the topic (either by topic ID if the topic name was never known or by topic name if it was finally resolved when we wanted to remove from the session.)
One helpful simplifying factor is that we only allow one type of request (uses topic ID or does not use topic ID) in the session. That means we can rely on a session continuing to have the same information. We don’t have to worry about converting topics only known by name to topic ID for a response and we won’t need to convert topics only known by ID to name for a response.
This PR introduces a change to store the "unresolved partitions" in the cached partition object. If a version 13+ request is sent with a topic ID that is unknown, a cached partition will be created with that fetch request data and a null topic name. On subsequent incremental requests, unresolved partitions may be resolved with the new IDs found in the metadata cache. When handling the request, getting all partitions will return a TopicIdPartition object that will be used to handle the request and build the response. Since we can rely on only one type of request (with IDs or without), the cached partitions map will have different keys depending on what fetch request version is being used.
This PR involves changes both in FetchSessionHandler and FetchSession. Some major changes are outlined below.
1. FetchSessionHandler: Forgetting a topic and adding a new topic with the same name - We may have a case where there is a topic foo with ID 1 in the session. Upon a subsequent metadata update, we may have topic foo with ID 2. This means that topic foo has been deleted and recreated. When sending fetch requests version 13+ we will send a request to add foo ID 2 to the session and remove foo ID 1. Otherwise, we will fall back to the same behavior for versions 12 and below
2. FetchSession: Resolving in Incremental Sessions - Incremental sessions contain two distinct sets of partitions. Partitions that are sent in the latest request that are new/updates/forgotten partitions and the partitions already in the session. If we want to resolve unknown topic IDs we will need to handle both cases.
* Partitions in the request - These partitions are either new or updating/forgetting previous partitions in the session. The new partitions are trivial. We either have a resolved partition or create a partition that is unresolved. For the other cases, we need to be a bit more careful.
* For updated partitions we have a few cases – keep in mind, we may not programmatically know if a partition is an update:
1. partition in session is resolved, update is resolved: trivial
2. partition in session is unresolved, update is unresolved: in code, this is equivalent to the case above, so trivial as well
3. partition in session is unresolved, update is resolved: this means the partition in the session does not have a name, but the metadata cache now contains the name – to fix this we can check if there exists a cached partition with the given ID and update it both with the partition update and with the topic name.
4. partition in session is resolved, update is unresolved: this means the partition in the session has a name, but the update was unable to be resolved (ie, the topic is deleted) – this is the odd case. We will look up the partition using the ID. We will find the old version with a name but will not replace the name. This will lead to an UNKNOWN_TOPIC_OR_PARTITION or INCONSISTENT_TOPIC_ID error which will be handled with a metadata update. Likely a future request will forget the partition, and we will be able to do so by ID.
5. Two partitions in the session have IDs, but they are different: only one topic ID should exist in the metadata at a time, so likely only one topic ID is in the fetch set. The other one should be in the toForget. We will be able to remove this partition from the session. If for some reason, we don't try to forget this partition — one of the partitions in the session will cause an inconsistent topic ID error and the metadata for this partition will be refreshed — this should result in the old ID being removed from the session. This should not happen if the FetchSessionHandler is correctly in sync.
* For the forgotten partitions we have the same cases:
1. partition in session is resolved, forgotten is resolved: trivial
2. partition in session is unresolved, forgotten is unresolved: in code, this is equivalent to the case above, so trivial as well
3. partition in session is unresolved, forgotten is resolved: this means the partition in the session does not have a name, but the metadata cache now contains the name – to fix this we can check if there exists a cached partition with the given ID and try to forget it before we check the resolved name case.
4. partition in session is resolved, update is unresolved: this means the partition in the session has a name, but the update was unable to be resolved (ie, the topic is deleted) We will look up the partition using the ID. We will find the old version with a name and be able to delete it.
5. both partitions in the session have IDs, but they are different: This should be the same case as described above. If we somehow do not have the ID in the session, no partition will be removed. This should not happen unless the Fetch Session Handler is out of sync.
* Partitions in the session - there may be some partitions in the session already that are unresolved. We can resolve them in forEachPartition using a method that checks if the partition is unresolved and tries to resolve it using a topicName map from the request. The partition will be resolved before the function using the cached partition is applied.
Reviewers: David Jacot <djacot@confluent.io>
This task is to provide a concrete implementation of the interfaces defined in KIP-255 to allow Kafka to connect to an OAuth/OIDC identity provider for authentication and token retrieval. While KIP-255 provides an unsecured JWT example for development, this will fill in the gap and provide a production-grade implementation.
The OAuth/OIDC work will allow out-of-the-box configuration by any Apache Kafka users to connect to an external identity provider service (e.g. Okta, Auth0, Azure, etc.). The code will implement the standard OAuth client credentials grant type.
The proposed change is largely composed of a pair of AuthenticateCallbackHandler implementations: one to login on the client and one to validate on the broker.
See the following for more detail:
KIP-768
KAFKA-13202
Reviewers: Yi Ding <dingyi.zj@gmail.com>, Ismael Juma <ismael@juma.me.uk>, Jun Rao <junrao@gmail.com>
When loading a snapshot the broker BrokerMetadataListener was using the batch's append time, offset
and epoch. These are not the same as the append time, offset and epoch from the log. This PR fixes
it to instead use the lastContainedLogTimeStamp, lastContainedLogOffset and lastContainedLogEpoch
from the SnapshotReader.
This PR refactors the MetadataImage and MetadataDelta to include an offset and epoch. It also swaps
the order of the arguments for ReplicaManager.applyDelta, in order to be more consistent with
MetadataPublisher.publish.
Reviewers: Colin P. McCabe <cmccabe@apache.org>
Added snapshots for consumed remote log metadata for each partition to avoid consuming again in case of broker restarts. These snapshots are stored in the respective topic partition log directories.
Reviewers: Kowshik Prakasam <kprakasam@confluent.io>, Cong Ding <cong@ccding.com>, Jun Rao <junrao@gmail.com>
Avoid O(N) behavior in KRaftMetadataCache#topicNamesToIds and
KRaftMetadataCache#topicIdsToNames by returning a map subclass that
exposes the TopicsImage data structures without copying them.
Reviewers: Jason Gustafson <jason@confluent.io>, Ismael Juma <ismael@juma.me.uk>
Add support for CreateTopicsPolicy and AlterConfigsPolicy when running in KRaft mode.
Reviewers: David Arthur <mumrah@gmail.com>, Niket Goel <ngoel@confluent.io>
The ReplicaManager, LogManager, and KafkaApis class all have many
constructor parameters. It is often difficult to add or remove a
parameter, since there are so many locations that need to be updated. In
order to address this problem, we should use named parameters when
constructing these objects from Scala code. This will make it easy to
add new optional parameters without modifying many test cases. It will
also make it easier to read git diffs and PRs, since the parameters will
have names next to them. Since Java does not support named paramters,
this PR adds several Builder classes which can be used to achieve the
same effect.
ReplicaManager also had a secondary constructor, which this PR removes.
The function of the secondary constructor was just to provide some
default parameters for the main constructor. However, it is simpler just
to actually use default parameters.
Reviewers: David Arthur <mumrah@gmail.com>
This PR aims to remove tombstones that persist indefinitely due to low throughput. Previously, deleteHorizon was calculated from the segment's last modified time.
In this PR, the deleteHorizon will now be tracked in the baseTimestamp of RecordBatches. After the first cleaning pass that finds a record batch with tombstones, the record batch is recopied with deleteHorizon flag and a new baseTimestamp that is the deleteHorizonMs. The records in the batch are rebuilt with relative timestamps based on the deleteHorizonMs that is recorded. Later cleaning passes will be able to remove tombstones more accurately on their deleteHorizon due to the individual time tracking on record batches.
KIP 534: https://cwiki.apache.org/confluence/display/KAFKA/KIP-534%3A+Retain+tombstones+and+transaction+markers+for+approximately+delete.retention.ms+milliseconds
Co-authored-by: Ted Yu <yuzhihong@gmail.com>
Co-authored-by: Richard Yu <yohan.richard.yu@gmail.com>
* Add the following producer metrics:
flush-time-total: cumulative sum of time elapsed during in flush.
txn-init-time-total: cumulative sum of time elapsed during in initTransactions.
txn-begin-time-total: cumulative sum of time elapsed during in beginTransaction.
txn-send-offsets-time-total: cumulative sum of time elapsed during in sendOffsetsToTransaction.
txn-commit-time-total: cumulative sum of time elapsed during in commitTransaction.
txn-abort-time-total: cumulative sum of time elapsed during in abortTransaction.
* Add the following consumer metrics:
commited-time-total: cumulative sum of time elapsed during in committed.
commit-sync-time-total: cumulative sum of time elapsed during in commitSync.
* Add a total-blocked-time metric to streams that is the sum of:
consumer’s io-waittime-total
consumer’s iotime-total
consumer’s committed-time-total
consumer’s commit-sync-time-total
restore consumer’s io-waittime-total
restore consumer’s iotime-total
admin client’s io-waittime-total
admin client’s iotime-total
producer’s bufferpool-wait-time-total
producer's flush-time-total
producer's txn-init-time-total
producer's txn-begin-time-total
producer's txn-send-offsets-time-total
producer's txn-commit-time-total
producer's txn-abort-time-total
Reviewers: Bruno Cadonna <cadonna@confluent.io>, Guozhang Wang <wangguoz@gmail.com>
Updates the scala fmt to the latest stable version.
Applies all the style fixes (all source code changes are done by scala
fmt).
Removes setting about dangling parentheses as `true` is already the
default.
Reviewer: John Roesler <john@confluent.io>
Pt. 1: #10609
Pt. 2: #10683
Pt. 3: #10788
In Pt. 3 we implement the addNamedTopology API. This can be used to update the processing topology of a running Kafka Streams application without resetting the app, or even pausing/restarting the process. It's up to the user to ensure that this API is called on every instance of an application to ensure all clients are able to run the newly added NamedTopology.
Reviewers: Guozhang Wang <guozhang@confluent.io>, Walker Carlson <wcarlson@confluent.io>
Pt. 1: #10609
Pt. 2: #10683
Pt. 3: #10788
The TopologyMetadata is next up after Pt. 1 #10609. This PR sets up the basic architecture for running an app with multiple NamedTopologies, though the APIs to add/remove them dynamically are not implemented until Pt. 3
Reviewers: Guozhang Wang <guozhang@confluent.io>, Walker Carlson <wcarlson@confluent.io>
This patch adds support for unregistering listeners to `RaftClient`.
Reviewers: Colin P. McCabe <cmccabe@apache.org>, Jason Gustafson <jason@confluent.io>
KAFKA-9555 Added default RLMM implementation based on internal topic storage.
This is the initial version of the default RLMM implementation.
This includes changes containing default RLMM configs, RLMM implementation, producer/consumer managers.
Introduced TopicBasedRemoteLogMetadataManagerHarness which takes care of bringing up a Kafka cluster and create remote log metadata topic and initializes TopicBasedRemoteLogMetadataManager.
Refactored existing RemoteLogMetadataCacheTest to RemoteLogSegmentLifecycleTest to have parameterized tests to run both RemoteLogMetadataCache and also TopicBasedRemoteLogMetadataManager.
Refactored existing InmemoryRemoteLogMetadataManagerTest, RemoteLogMetadataManagerTest to have parameterized tests to run both InmemoryRemoteLogMetadataManager and also TopicBasedRemoteLogMetadataManager.
This is part of tiered storage KIP-405 efforts.
Reviewers: Kowshik Prakasam <kprakasam@confluent.io>, Cong Ding <cong@ccding.com>, Jun Rao <junrao@gmail.com>
Support the KIP-455 reassignment API when in KRaft mode. Reassignments
which merely rearrange partitions complete immediately. Those that only
remove a partition complete immediately if the ISR would be non-empty
after the specified removals. Reassignments that add one or more
partitions follow the KIP-455 pattern of adding all the adding replicas
to the replica set, and then waiting for the ISR to include all the new
partitions before completing. Changes to the partition sets are
accomplished via PartitionChangeRecord.
Reviewers: Jun Rao <junrao@gmail.com>
1) Bring the generation field back to the CooperativeStickyAssignor so we don't need to rely so heavily on the ConsumerCoordinator properly updating its SubscriptionState after eg falling out of the group. The plain StickyAssignor always used the generation since it had to, so we just make sure the CooperativeStickyAssignor has this tool as well
2) In case of unforeseen problems or further bugs that slip past the generation field safety net, the assignor will now explicitly look out for partitions that are being claimed by multiple consumers as owned in the same generation. Such a case should never occur, but if it does, we have to invalidate this partition from the ownedPartitions of both consumers, since we can't tell who, if anyone, has the valid claim to this partition.
3) Fix a subtle bug that I discovered while writing tests for the above two fixes: in the constrained algorithm, we compute the exact number of partitions each consumer should end up with, and keep track of the "unfilled" members who must -- or might -- require more partitions to hit their quota. The problem was that members at the minQuota were being considered as "unfilled" even after we had already hit the maximum number of consumers allowed to go up to the maxQuota, meaning those minQuota members could/should not accept any more partitions beyond that. I believe this was introduced in #10509, so it shouldn't be in any released versions and does not need to be backported.
Reviewers: Guozhang Wang <guozhang@apache.org>, Luke Chen <showuon@gmail.com>
Updated FetchRequest and FetchResponse to use topic IDs rather than topic names.
Some of the complicated code is found in FetchSession and FetchSessionHandler.
We need to be able to store topic IDs and maintain a cache on the broker for IDs that may not have been resolved. On incremental fetch requests, we will try to resolve them or remove them if in toForget.
Reviewers: Rajini Sivaram <rajinisivaram@googlemail.com>, Chia-Ping Tsai <chia7712@gmail.com>, Jun Rao <junrao@gmail.com>
This implements the request and response portion of KIP-709. It updates the OffsetFetch request and response to support fetching offsets for multiple consumer groups at a time. If the broker does not support the new OffsetFetch version, clients can revert to the previous behaviour and use a request for each coordinator.
Reviewers: Rajini Sivaram <rajinisivaram@googlemail.com>, Konstantine Karantasis <konstantine@confluent.io>
Create the image/ module for storing, reading, and writing broker metadata images.
Metadata images are immutable. New images are produced from existing images
using delta classes. Delta classes are mutable, and represent changes to a base
image.
MetadataImage objects can be converted to lists of KRaft metadata records. This
is essentially writing a KRaft snapshot. The resulting snapshot can be read
back into a MetadataDelta object. In practice, we will typically read the
snapshot, and then read a few more records to get fully up to date. After that,
the MetadataDelta can be converted to a MetadataImage as usual.
Sometimes, we have to load a snapshot even though we already have an existing
non-empty MetadataImage. We would do this if the broker fell too far behind and
needed to receive a snapshot to catch up. This is handled just like the normal
snapshot loading process. Anything that is not in the snapshot will be marked
as deleted in the MetadataDelta once finishSnapshot() is called.
In addition to being used for reading and writing snapshots, MetadataImage also
serves as a cache for broker information in memory. A follow-up PR will replace
MetadataCache, CachedConfigRepository, and the client quotas cache with the
corresponding Image classes. TopicsDelta also replaces the "deferred
partition" state that the RaftReplicaManager currently implements. (That change
is also in a follow-up PR.)
Reviewers: Jason Gustafson <jason@confluent.io>, David Arthur <mumrah@gmail.com>
Implements KIP-745 https://cwiki.apache.org/confluence/display/KAFKA/KIP-745%3A+Connect+API+to+restart+connector+and+tasks to change connector REST API to restart a connector and its tasks as a whole.
Testing strategy
- [x] Unit tests added for all possible combinations of onlyFailed and includeTasks
- [x] Integration tests added for all possible combinations of onlyFailed and includeTasks
- [x] System tests for happy path
Reviewers: Randall Hauch <rhauch@gmail.com>, Diego Erdody <erdody@gmail.com>, Konstantine Karantasis <k.karantasis@gmail.com>
Add header and footer records for raft snapshots. This helps identify when the snapshot
starts and ends. The header also contains a time. The time field is currently set to 0.
KAFKA-12997 will add in the necessary wiring to use the correct timestamp.
Reviewers: Jose Sancio <jsancio@gmail.com>, Colin P. McCabe <cmccabe@apache.org>
We had been using `RecordAccumulator.beginFlush` in order to force the `RecordAccumulator` to flush pending batches when a transaction was being completed. Internally, `RecordAccumulator` has a simple counter for the number of flushes in progress. The count gets incremented in `beginFlush` and it is expected to be decremented by `awaitFlushCompletion`. The second call to decrement the counter never happened in the transactional path, so the counter could get stuck at a positive value, which means that the linger time would effectively be ignored.
This patch fixes the problem by removing the use of `beginFlush` in `Sender`. Instead, we now add an additional condition in `RecordAccumulator` to explicitly check when a transaction is being completed.
Reviewers: Guozhang Wang <wangguoz@gmail.com>
Added tiered storage related configs including remote log manager configs.
Added local log retention configs to LogConfig.
Added tests for the added configs.
Reviewers: Kowshik Prakasam <kprakasam@confluent.io>, Jun Rao <junrao@gmail.com>
They have been deprecated since 0.10.0. Full list of removes configs:
* port
* host.name
* advertised.port
* advertised.host.name
Also adjust tests to take the removals into account. Some tests were
no longer relevant and have been removed.
Finally, took the chance to:
* Clean up unnecessary usage of `KafkaConfig$.MODULE$` in
related files.
* Add missing `Test` annotations to `AdvertiseBrokerTest` and
make necessary changes for the tests to pass.
Reviewers: David Jacot <djacot@confluent.io>, Luke Chen <showuon@gmail.com>
Directly use `RaftClient.Listener`, `SnapshotWriter` and `SnapshotReader` in the quorum controller.
1. Allow `RaftClient` users to create snapshots by specifying the last committed offset and last committed epoch. These values are validated against the log and leader epoch cache.
2. Remove duplicate classes in the metadata module for writing and reading snapshots.
3. Changed the logic for comparing snapshots. The old logic was assuming a certain batch grouping. This didn't match the implementation of the snapshot writer. The snapshot writer is free to merge batches before writing them.
4. Improve `LocalLogManager` to keep track of multiple snapshots.
5. Improve the documentation and API for the snapshot classes to highlight the distinction between the offset of batches in the snapshot vs the offset of batches in the log. These two offsets are independent of one another. `SnapshotWriter` and `SnapshotReader` expose a method called `lastOffsetFromLog` which represents the last inclusive offset from the log that is represented in the snapshot.
Reviewers: dengziming <swzmdeng@163.com>, Jason Gustafson <jason@confluent.io>
This PR includes adding the NamedTopology to the Subscription/AssignmentInfo, and to the StateDirectory so it can place NamedTopology tasks within the hierarchical structure with task directories under the NamedTopology parent dir.
Reviewers: Walker Carlson <wcarlson@confluent.io>, Guozhang Wang <guozhang@confluent.io>
This patch removes the temporary shim layer we added to bridge the interface
differences between MetaLogManager and RaftClient. Instead, we now use the
RaftClient directly from the metadata module. This also means that the
metadata gradle module now depends on raft, rather than the other way around.
Finally, this PR also consolidates the handleResign and handleNewLeader APIs
into a single handleLeaderChange API.
Co-authored-by: Jason Gustafson <jason@confluent.io>
Implement a striped replica placement algorithm for KRaft. This also
means implementing rack awareness. Previously, KRraft just chose
replicas randomly in a non-rack-aware fashion. Also, allow replicas to
be placed on fenced brokers if there are no other choices. This was
specified in KIP-631 but previously not implemented.
Reviewers: Jun Rao <junrao@gmail.com>
The QuorumController should honor the timeout for RPC requests
which feature a timeout. For electLeaders, attempt to trigger a leader
election for all partitions when the request specifies null for the topics
argument.
Reviewers: David Arthur <mumrah@gmail.com>
Introduce List serde for primitive types or custom serdes with a serializer and a deserializer according to KIP-466
Reviewers: Anna Sophie Blee-Goldman <ableegoldman@apache.org>, Matthias J. Sax <mjsax@conflunet.io>, John Roesler <roesler@confluent.io>, Michael Noll <michael@confluent.io>
Added server-common module to have server side common classes. Moved ApiMessageAndVersion, RecordSerde, AbstractApiMessageSerde, and BytesApiMessageSerde to server-common module.
Reivewers: Kowshik Prakasam <kprakasam@confluent.io>, Jun Rao <junrao@gmail.com>
KAFKA-12429: Added serdes for the default implementation of RLMM based on an internal topic as storage. This topic will receive events of RemoteLogSegmentMetadata, RemoteLogSegmentUpdate, and RemotePartitionDeleteMetadata. These events are serialized into Kafka protocol message format.
Added tests for all the event types for that topic.
This is part of the tiered storaqe implementation KIP-405.
Reivewers: Kowshik Prakasam <kprakasam@confluent.io>, Jun Rao <junrao@gmail.com>
Implement Raft Snapshot loading API.
1. Adds a new method `handleSnapshot` to `raft.Listener` which is called whenever the `RaftClient` determines that the `Listener` needs to load a new snapshot before reading the log. This happens when the `Listener`'s next offset is less than the log start offset also known as the earliest snapshot.
2. Adds a new type `SnapshotReader<T>` which provides a `Iterator<Batch<T>>` interface and de-serializes records in the `RawSnapshotReader` into `T`s
3. Adds a new type `RecordsIterator<T>` that implements an `Iterator<Batch<T>>` by scanning a `Records` object and deserializes the batches and records into `Batch<T>`. This type is used by both `SnapshotReader<T>` and `RecordsBatchReader<T>` internally to implement the `Iterator` interface that they expose.
4. Changes the `MockLog` implementation to read one or two batches at a time. The previous implementation always read from the given offset to the high-watermark. This made it impossible to test interesting snapshot loading scenarios.
5. Removed `throws IOException` from some methods. Some of types were inconsistently throwing `IOException` in some cases and throwing `RuntimeException(..., new IOException(...))` in others. This PR improves the consistent by wrapping `IOException` in `RuntimeException` in a few more places and replacing `Closeable` with `AutoCloseable`.
6. Updated the Kafka Raft simulation test to take into account snapshot. `ReplicatedCounter` was updated to generate snapshot after 10 records get committed. This means that the `ConsistentCommittedData` validation was extended to take snapshots into account. Also added a new invariant to ensure that the log start offset is consistently set with the earliest snapshot.
Reviewers: dengziming <swzmdeng@163.com>, David Arthur <mumrah@gmail.com>, Jason Gustafson <jason@confluent.io>
Fixes the issue with https://issues.apache.org/jira/browse/KAFKA-10847.
To fix the above problem, the left/outer stream-stream join processor uses a buffer to hold non-joined records for some time until the window closes, so they are not processed if a join is found during the join window time. If the window of a record closes and a join was not found, then this should be emitted and processed by the consequent topology processor.
A new time-ordered window store is used to temporary hold records that do not have a join and keep the records keys ordered by time. The KStreamStreamJoin has a reference to this new store . For every non-joined record seen, the processor writes it to this new state store without processing it. When a joined record is seen, the processor deletes the joined record from the new state store to prevent further processing.
Records that were never joined at the end of the window + grace period are emitted to the next topology processor. I use the stream time to check for the expiry time for determinism results . The KStreamStreamJoin checks for expired records and emit them every time a new record is processed in the join processor.
The new state store is shared with the left and right join nodes. The new store needs to serialize the record keys using a combined key of <joinSide-recordKey>. This key combination helps to delete the records from the other join if a joined record is found. Two new serdes are created for this, KeyAndJoinSideSerde which serializes a boolean value that specifies the side where the key is found, and ValueOrOtherValueSerde that serializes either V1 or V2 based on where the key was found.
Reviewers: Matthias J. Sax <matthias@confluent.io>, Guozhang Wang <wangguoz@gmail.com>
Deprecates the following
1. StreamsConfig.EXACTLY_ONCE
2. StreamsConfig.EXACTLY_ONCE_BETA
3. Producer#sendOffsetsToTransaction(Map offsets, String consumerGroupId)
And introduces a new StreamsConfig.EXACTLY_ONCE_V2 config. Additionally, this PR replaces usages of the term "eos-beta" throughout the code with the term "eos-v2"
Reviewers: Matthias J. Sax <mjsax@confluent.io>
Implement the createPartitions RPC which adds more partitions to a topic
in the KIP-500 controller. Factor out some of the logic for validating
manual partition assignments, so that it can be shared between
createTopics and createPartitions. Add a startPartition argument to the
replica placer.
Reviewers: Jason Gustafson <jason@confluent.io>
KAFKA-12368: Added inmemory implementations for RemoteStorageManager and RemoteLogMetadataManager.
Added inmemory implementation for RemoteStorageManager and RemoteLogMetadataManager. A major part of inmemory RLMM will be used in the default RLMM implementation which will be based on topic storage. These will be used in unit tests for tiered storage.
Added tests for both the implementations and their supported classes.
This is part of tiered storage implementation, KIP-405.
Reivewers: Kowshik Prakasam <kprakasam@confluent.io>, Jun Rao <junrao@gmail.com>
Implement controller-side snapshot generation.Implement QuorumController snapshot
generation. Note that this PR does not handle KRaft integration, just the internal
snapshot record generation and consumption logic.
Reading a snapshot is relatively straightforward. When the QuorumController
starts up, it loads the most recent snapshot. This is just a series of records
that we replay, plus a log offset ("snapshot epoch") that we advance to.
Writing a snapshot is more complex. There are several components:
the SnapshotWriter which persists the snapshot, the SnapshotGenerator
which manages writing each batch of records, and the SnapshotGeneratorManager
which interfaces the preceding two classes with the event queue.
Controller snapshots are done incrementally. In order to avoid blocking the
controller thread for a long time, we pull a few record batches at a time from
our record batch iterators. These iterators are implemented by controller
manager classes such as ReplicationControlManager, ClusterControlManager, etc.
Finally, this PR adds ControllerTestUtils#deepSortRecords and
ControllerTestUtils#assertBatchIteratorContains, which make it easier to write
unit tests. Since records are often constructed from unsorted data structures,
it is often useful to sort them before comparing them.
Reviewers: David Arthur <mumrah@gmail.com>
1. Add `canGrantVote` to `EpochState`
2. Move the if-else in `KafkaRaftCllient.handleVoteRequest` to `EpochState`
3. Add unit tests for `canGrantVote`
Reviewers: Jason Gustafson <jason@confluent.io>
Introduce "testkit" package which includes KafkaClusterTestKit class for enabling integration tests of self-managed clusters. Also make use of this new integration test harness in the ClusterTestExtentions JUnit extension.
Adds RaftClusterTest for basic self-managed integration test.
Reviewers: Jason Gustafson <jason@confluent.io>, Colin P. McCabe <cmccabe@apache.org>
Co-authored-by: Colin P. McCabe <cmccabe@apache.org>
Previously we implemented ClusterId validation for the Fetch API in the Raft implementation. This patch adds ClusterId validation to the remaining Raft RPCs.
Reviewers: José Armando García Sancio <jsancio@users.noreply.github.com>, Jason Gustafson <jason@confluent.io>
This patch changes the raft simulation tests to use jqwik, which is a property testing library. This provides two main benefits:
- It simplifies the randomization of test parameters. Currently the tests use a fixed set of `Random` seeds, which means that most builds are doing redundant work. We get a bigger benefit from allowing each build to test different parameterizations.
- It makes it easier to reproduce failures. Whenever a test fails, jqwik will report the random seed that failed. A developer can then modify the `@Property` annotation to use that specific seed in order to reproduce the failure.
This patch also includes an optimization for `MockLog.earliestSnapshotId` which reduces the time to run the simulation tests dramatically.
Reviewers: Ismael Juma <ismael@juma.me.uk>, Chia-Ping Tsai <chia7712@gmail.com>, José Armando García Sancio <jsancio@gmail.com>, David Jacot <djacot@confluent.io>
1. Remove org.apache.log4j from allowed import list of shell, trogdor subpackage; they uses slf4j, not log4.
2. Remove org.slf4j from allowed import list of clients, server subpackage: org.slf4j is allowed globally.
3. Remove org.apache.log4j from streams subpackage's allowed import list
Reviewers: David Jacot <david.jacot@gmail.com>, Chia-Ping Tsai <chia7712@gmail.com>
This patch enables delete topic support for the new KIP-500 controller. Also fixes the following:
- Fix a bug where feature level records were not correctly replayed.
- Fix a bug in TimelineHashMap#remove where the wrong type was being returned.
Reviewers: Jason Gustafson <jason@confluent.io>, Justine Olshan <jolshan@confluent.io>, Ron Dagostino <rdagostino@confluent.io>, Chia-Ping Tsai <chia7712@gmail.com>, Jun Rao <junrao@gmail.com>
Co-authored-by: Jason Gustafson <jason@confluent.io>
Always grab a new thread.id and verify that a thread has fully shut down to DEAD before removing from the `threads` list and making that id available again
Reviewers: Walker Carlson <wcarlson@confluent.io>, Bruno Cadonna <cadonna@confluent.io>
Implements KIP-695
Reverts a previous behavior change to Consumer.poll and replaces
it with a new Consumer.currentLag API, which returns the client's
currently cached lag.
Uses this new API to implement the desired task idling semantics
improvement from KIP-695.
Reverts fdcf8fbf72 / KAFKA-10866: Add metadata to ConsumerRecords (#9836)
Reviewers: Chia-Ping Tsai <chia7712@gmail.com>, Guozhang Wang <guozhang@apache.org>
The quorum controller stores metadata in the KIP-500 metadata log, not in Apache
ZooKeeper. Each controller node is a voter in the metadata quorum. The leader of the
quorum is the active controller, which processes write requests. The followers are standby
controllers, which replay the operations written to the log. If the active controller goes away,
a standby controller can take its place.
Like the ZooKeeper-based controller, the quorum controller is based on an event queue
backed by a single-threaded executor. However, unlike the ZK-based controller, the quorum
controller can have multiple operations in flight-- it does not need to wait for one operation
to be finished before starting another. Therefore, calls into the QuorumController return
CompleteableFuture objects which are completed with either a result or an error when the
operation is done. The QuorumController will also time out operations that have been
sitting on the queue too long without being processed. In this case, the future is completed
with a TimeoutException.
The controller uses timeline data structures to store multiple "versions" of its in-memory
state simultaneously. "Read operations" read only committed state, which is slightly older
than the most up-to-date in-memory state. "Write operations" read and write the latest
in-memory state. However, we can not return a successful result for a write operation until
its state has been committed to the log. Therefore, if a client receives an RPC response, it
knows that the requested operation has been performed, and can not be undone by a
controller failover.
Reviewers: Jun Rao <junrao@gmail.com>, Ron Dagostino <rdagostino@confluent.io>
The Kafka Metadata shell is a new command which allows users to
interactively examine the metadata stored in a KIP-500 cluster.
It can examine snapshot files that are specified via --snapshot.
The metadata tool works by replaying the log and storing the state into
in-memory nodes. These nodes are presented in a fashion similar to
filesystem directories.
Reviewers: Jason Gustafson <jason@confluent.io>, David Arthur <mumrah@gmail.com>, Igor Soarez <soarez@apple.com>
Previously all APIs were accessible on every listener exposed by the broker, but
with KIP-500, that is no longer true. We now have more complex requirements for
API accessibility.
For example, the KIP-500 controller exposes some APIs which are not exposed by
brokers, such as BrokerHeartbeatRequest, and does not expose most client APIs,
such as JoinGroupRequest, etc. Similarly, the KIP-500 broker does not implement
some APIs that the ZK-based broker does, such as LeaderAndIsrRequest and
UpdateFeaturesRequest.
All of this means that we need more sophistication in how we expose APIs and
keep them consistent with the ApiVersions API. Up until now, we have been
working around this using the controllerOnly flag inside ApiKeys, but this is
not rich enough to support all of the cases listed above. This PR introduces a
new "listeners" field to the request schema definitions. This field is an array
of strings which indicate the listener types in which the API should be exposed.
We currently support "zkBroker", "broker", and "controller". ("broker"
indicates the KIP-500 broker, whereas zkBroker indicates the old broker).
This PR also creates ApiVersionManager to encapsulate the creation of the
ApiVersionsResponse based on the listener type. Additionally, it modifies
SocketServer to check the listener type of received requests before forwarding
them to the request handler.
Finally, this PR also fixes a bug in the handling of the ApiVersionsResponse
prior to authentication. Previously a static response was sent, which means that
changes to features would not get reflected. This also meant that the logic to
ensure that only the intersection of version ranges supported by the controller
would get exposed did not work. I think this is important because some clients
rely on the initial pre-authenticated ApiVersions response rather than doing a
second round after authentication as the Java client does.
One final cleanup note: I have removed the expectation that envelope requests
are only allowed on "privileged" listeners. This made sense initially because
we expected to use forwarding before the KIP-500 controller was available. That
is not the case anymore and we expect the Envelope API to only be exposed on the
controller listener. I have nevertheless preserved the existing workarounds to
allow verification of the forwarding behavior in integration testing.
Reviewers: Colin P. McCabe <cmccabe@apache.org>, Ismael Juma <ismael@juma.me.uk>
This PR adds the KIP-500 BrokerServer and ControllerServer classes and
makes some related changes to get them working. Note that the ControllerServer
does not instantiate a QuorumController object yet, since that will be added in
PR #10070.
* Add BrokerServer and ControllerServer
* Change ApiVersions#computeMaxUsableProduceMagic so that it can handle
endpoints which do not support PRODUCE (such as KIP-500 controller nodes)
* KafkaAdminClientTest: fix some lingering references to decommissionBroker
that should be references to unregisterBroker.
* Make some changes to allow SocketServer to be used by ControllerServer as
we as by the broker.
* We now return a random active Broker ID as the Controller ID in
MetadataResponse for the Raft-based case as per KIP-590.
* Add the RaftControllerNodeProvider
* Add EnvelopeUtils
* Add MetaLogRaftShim
* In ducktape, in config_property.py: use a KIP-500 compatible cluster ID.
Reviewers: Colin P. McCabe <cmccabe@apache.org>, David Arthur <mumrah@gmail.com>
We don't really need it and it causes problems in older Android versions
and GraalVM native image usage (there are workarounds for the latter).
Move the logic to separate classes that are only invoked when the
relevant compression library is actually used. Place such classes
in their own package and enforce via checkstyle that only these
classes refer to compression library packages.
To avoid cyclic dependencies, moved `BufferSupplier` to the `utils`
package.
Reviewers: Chia-Ping Tsai <chia7712@gmail.com>
Add MetaLogListener, LocalLogManager, and related classes. These
classes are used by the KIP-500 controller and broker to interface with the
Raft log.
Also add the Controller interface. The implementation will be added in a separate PR.
Reviewers: Ron Dagostino <rdagostino@confluent.io>, David Arthur <mumrah@gmail.com>
Add KafkaEventQueue, which is used by the KIP-500 controller to manage its event queue.
Compared to using an Executor, KafkaEventQueue has the following advantages:
* Events can be given "deadlines." If an event lingers in the queue beyond the deadline, it
will be completed with a timeout exception. This is useful for implementing timeouts for
controller RPCs.
* Events can be prepended to the queue as well as appended.
* Events can be given tags to make them easier to manage. This is especially useful for
rescheduling or cancelling events which were previously scheduled to execute in the future.
Reviewers: Jun Rao <junrao@gmail.com>, José Armando García Sancio <jsancio@gmail.com>
This patch adds a `RecordSerde` implementation for the metadata record format expected by KIP-631.
Reviewers: Colin McCabe <cmccabe@apache.org>, Ismael Juma <mlists@juma.me.uk>
Replace BrokerStates.scala with BrokerState.java, to make it easier to use from Java code if needed. This also makes it easier to go from a numeric type to an enum.
Reviewers: Chia-Ping Tsai <chia7712@gmail.com>
Add the metadata gradle module, which will contain the metadata record
definitions, and other metadata-related broker-side code.
Add MetadataParser, MetadataParseException, etc.
Reviewers: José Armando García Sancio <jsancio@gmail.com>, Ismael Juma <ismael@juma.me.uk>, David Arthur <mumrah@gmail.com>
Colin Patrick McCabe
Igor Soarez
Matthias J. Sax
David Jacot
Alieh Saeedi
Apoorv Mittal
Nikolay
Andrew Schofield
Hanyu Zheng
Ritika Reddy
Jeff Kim
Ismael Juma
Yash Mayya
Calvin Liu
Kirk True
Chris Egerton
Apoorv Mittal
Omnia G.H Ibrahim
Lianet Magrans
Mayank Shekhar Narula
Dongnuo Lyu
Lucas Brutschy
Tyler Bertrand
zhaohaidao
Kamal Chandraprakash
Satish Duggana
Proven Provenzano
Greg Harris
Luke Chen
Ivan Yurchenko
Hao Li
Federico Valeri
Abhijeet Kumar
vamossagar12
Dimitar Dimitrov
David Arthur
Gantigmaa Selenge
Ron Dagostino
Victoria Xia
José Armando García Sancio
Matthew Wong
Akhilesh C
Rajini Sivaram
Patrik Marton
Alexandre Garnier
Jason Gustafson
Manikumar Reddy
Matthew de Detrich
dengziming
James Hughes
Chris Egerton
Mickael Maison
A. Sophie Blee-Goldman
Rittika Adhikari
Alyssa Huang
Jorge Esteban Quilcate Otoya
Aleksandr Sorokoumov
Idan Kamara
Chris Egerton
Wenjun Ruan
Colin Patrick McCabe
John Roesler
Justine Olshan
Kirk True
dengziming
Rohan
Josep Prat
Sanjana Kaundinya
kpatelatwork
Niket
Daniyar Yeralin
Vito Jeng
Sergio Peña
Luke Chen
Lev Zemlyanov
Lee Dongjin
Ron Dagostino