Several Kafka log configurations in have synonyms. For example, log retention can be configured
either by log.retention.ms, or by log.retention.minutes, or by log.retention.hours. There is also
a faculty in Kafka to dynamically change broker configurations without restarting the broker. These
dynamically set configurations are stored in the metadata log and override what is in the broker
properties file.
Unfortunately, these two features interacted poorly; there was a bug where the dynamic log
configuration update code ignored synonyms. For example, if you set log.retention.minutes and then
reconfigured something unrelated that triggered the LogConfig update path, the retention value that
you had configured was overwritten.
The reason for this was incorrect handling of synonyms. The code tried to treat the Kafka broker
configuration as a bag of key/value entities rather than extracting the correct retention time (or
other setting with overrides) from the KafkaConfig object.
Separately from the above bug, the code did not honor the value of dynamically configured synonyms:
setting log.retention.minutes had no effect; only log.retention.ms was honored.
ServerSideAssignorBenchmark and TargetAssignmentBuilderBenchmark have
the same topic and member subscription setup for the most part. Factor
out the commonality so that it's easier to share new setups between both
benchmarks.
Reviewers: David Jacot <djacot@confluent.io>
TestUtils.tempDirectory already registers a shutdown hook for deleting the temp directory. There's no reason to also call File.deleteOnExit, since that just registers another hook to do the same thing.
Reviewers: TengYao Chi <kitingiao@gmail.com>, Ken Huang <s7133700@gmail.com>, Chia-Ping Tsai <chia7712@gmail.com>
On trunk, our CI runs in response to "push" events. The change in #17227 causes the workflow template to be invalid, which prevents the build from starting. This patch fixes that by defaulting to `false`
Reviewers: Justine Olshan <jolshan@confluent.io>
Fix the CI workflow to treat the `is-public-fork` input as a string.
Also add some docs on composite actions.
Reviewers: Chia-Ping Tsai <chia7712@gmail.com>
Found two issues in the flaky tests: (Put the log analysis under Jira comments.)
1) The error "java.nio.file.DirectoryNotEmptyException" occurs if the flush() of kafkaStreams.close() and purgeLocalStreamsState() are triggered in the same time. (The current timeout is 5 sec, which is too short since the CI is unstable and slow).
2) Racing issue: Task to-be restored in ks-1 are rebalanced to ks-2 before entering active restoring state. So no onRestoreSuspend() was triggered.
To solve the issues:
1) Remove the timeout in kafkaStreams.close()
2) Ensure all tasks in ks-1 are active restoring before start second KafkaStreams(ks-2)
Reviewers: Chia-Ping Tsai <chia7712@gmail.com>
This PR tries to improve the error message when broker.id is set to -1 and ZK migration is enabled. It is not
needed to disable the broker.id.generation.enable option. It is sufficient to just not use it (by not setting
the broker.id to -1).
Reviewers: Chia-Ping Tsai <chia7712@gmail.com>, Luke Chen <showuon@gmail.com>
Change the configurations under config/kraft to use controller.quorum.bootstrap.servers instead of controller.quorum.voters. Add comments explaining how to use the older static quorum configuration where appropriate.
In docs/ops.html, remove the reference to "tentative timelines for ZooKeeper removal" and "Tiered storage is considered as an early access feature" since they are no longer up-to-date. Add KIP-853 information.
In docs/quickstart.html, move the ZK instructions to be after the KRaft instructions. Update the KRaft instructions to use KIP-853.
In docs/security.html, add an explanation of --bootstrap-controller and document controller.quorum.bootstrap.servers instead of controller.quorum.voters.
Reviewers: Mickael Maison <mickael.maison@gmail.com>, Alyssa Huang <ahuang@confluent.io>, Colin P. McCabe <cmccabe@apache.org>
I am still chasing KAFKA-17493. I was able to narrow it down to an issue with the pending join members. This patch logs them in order to help me troubleshooting it further. I will revert this change when the issue is root caused.
Reviewers: David Arthur <mumrah@gmail.com>
This is the part-2 of the KIP-1075
To find the offset for a given timestamp, ListOffsets API is used by the client. When the topic is enabled with remote storage, then we have to fetch the remote indexes such as offset-index and time-index to serve the query. Also, the ListOffsets request can contain the query for multiple topics/partitions.
The time taken to read the indexes from remote storage is non-deterministic and the query is handled by the request-handler threads. If there are multiple LIST_OFFSETS queries and most of the request-handler threads are busy in reading the data from remote storage, then the other high-priority requests such as FETCH and PRODUCE might starve and be queued. This can lead to higher latency in producing/consuming messages.
In this patch, we have introduced a delayed operation for remote list-offsets call. If the timestamp need to be searched in the remote-storage, then the request-handler threads will pass-on the request to the remote-log-reader threads. And, the request gets handled in asynchronous fashion.
Covered the patch with unit and integration tests.
Reviewers: Satish Duggana <satishd@apache.org>, Luke Chen <showuon@gmail.com>, Chia-Ping Tsai <chia7712@gmail.com>
Kafka Streams system tests were failing with this error:
Failed to parse host name from entry 3001@d for the configuration controller.quorum.voters. Each entry should be in the form `{id}@{host}:{port}`.
The cause is that in kafka.py line 876, we create a delimited string from a list comprehension, but the input is a string itself, so each character gets appended vs. the bootstrap server string of host:port. To fix this, this PR adds split(',') to controller_quorum_bootstrap_servers. Note that this only applies when dynamicRaftQuorum=False
Reviewers: Alyssa Huang <ahuang@confluent.io>, Chia-Ping Tsai <chia7712@gmail.com>
Fix for the known issue that the logic for updating fetch positions in the new consumer was being performed partly in the app thread, party in the background thread, potentially leading to race conditions on the subscription state.
This PR moves the logic for updateFetchPositions to the background thread as a single event (instead of triggering separate events to validate, fetchOffsets, listOffsets). A new UpdateFetchPositionsEvent is triggered from the app thread and processed in the background, where it performs those same operations and updates the subscription state accordingly, without blocking the background thread.
This PR maintains the existing logic for keeping a pendingOffsetFetchRequest that does not complete within the lifetime of the updateFetchPositions attempt, and may be used on the next call to updateFetchPositions.
Reviewers: Andrew Schofield <aschofield@confluent.io>, Chia-Ping Tsai <chia7712@gmail.com>
This patch bring the PR and trunk builds closer in line. Rather than switching between `--scan` and `--no-scan`,
both scenarios now use `--no-scan` and rely on the CI Complete workflow to publish the scans.
Reviewers: Chia-Ping Tsai <chia7712@gmail.com>
This patch increases the verbosity of the logging in Connect's integration tests. This is to better understand the causes of the flaky tests described in KAFKA-17493.
Reviewers: Chris Egerton <chrise@aiven.io>
When brokers undergoing ZK migration register with the controller, it should verify that they have
provided a way to contact them via their inter.broker.listener. Otherwise the migration will fail
later on with a more confusing error message.
Reviewers: David Arthur <mumrah@gmail.com>
Colin P. McCabe
Sean Quah
Stig Døssing
Matthias J. Sax
Chia-Chuan Yu
bboyleonp666
Ken Huang
TaiJuWu
David Arthur
Kuan-Po Tseng
xijiu
Jason Taylor
陳昱霖(Yu-Lin Chen)
Jakub Scholz
José Armando García Sancio
Kamal Chandraprakash
David Jacot
Mickael Maison
Ivan Yurchenko
Bill Bejeck
Lianet Magrans
TengYao Chi
tkuramoto33
Clay Johnson
PoAn Yang