[
https://issues.apache.org/jira/browse/KAFKA-17186?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17868750#comment-17868750
]
Greg Harris commented on KAFKA-17186:
-------------------------------------
> Does this imply that I should introduce multiple workers to avoid triggering
> the rebalance delay?
If you have two workers and restart one, the rebalance delay will still happen,
the difference will be that it will only apply to ~half of the connectors and
tasks. Lowering the delay will still be necessary.
I was thinking about whether there might be an opportunity to improve the code,
not suggesting any action on your part.
> If {{scheduled.rebalance.max.delay.ms}} is configured in MM2, will it affect
>the old Kafka cluster in the two data center? Considering that Kafka
>consumer's configuration {{session.timeout.ms}} is tied to rebalancing,
The scheduled.rebalance.max.delay.ms does not affect consumer rebalances, it's
a connect worker configuration for connect rebalances, i.e. starting and
stopping connectors & tasks. I'm not sure what you mean by "affect the old
Kafka cluster".
> 'A.consumer.enable.auto.commit=true', the mirrorMaker.out shows the value is
> false, that is 'enable.auto.commit = false'.
That configuration in particular is hard-coded to false:
[https://github.com/apache/kafka/blob/ee03c5f98a622271f35f7d00776f9e8d3c8e0d39/connect/mirror/src/main/java/org/apache/kafka/connect/mirror/MirrorConnectorConfig.java#L182]
because offsets are managed by the connector
[https://github.com/apache/kafka/blob/ee03c5f98a622271f35f7d00776f9e8d3c8e0d39/connect/mirror/src/main/java/org/apache/kafka/connect/mirror/MirrorSourceTask.java#L218]
> if I want to keep only 1 day logs in the disk, and I try to set
> log4j.appender.RollingFile.strategy=delete
> log4j.appender.RollingFile.strategy.max=1
> However, the logs that earlier than yesterday still exist in the directory,
> that means no logs to be deleted. Any other configuration item should be set?
I am not familiar with log4j appender configuration, you'll need to consult
some documentation to figure that one out.
> Cannot receive message after stopping Source Mirror Maker 2
> -----------------------------------------------------------
>
> Key: KAFKA-17186
> URL: https://issues.apache.org/jira/browse/KAFKA-17186
> Project: Kafka
> Issue Type: Bug
> Components: mirrormaker
> Affects Versions: 3.7.1
> Environment: Source Kafka Cluster per Node:
> CPU(s): 32
> Memory: 32G/1.1G free
> Target Kafka Cluster standalone Node:
> CPU(s): 24
> Memory: 30G/909M free
> Kafka Version 3.7
> Mirrormaker Version 3.7.1
> Reporter: George Yang
> Priority: Major
> Attachments: image-2024-07-25-14-14-21-327.png, mirrorMaker.out
>
>
> Deploy nodes 1, 2, and 3 in Data Center A, with MM2 service deployed on node
> 1. Deploy node 1 in Data Center B, with MM2 service also deployed on node 1.
> Currently, a service on node 1 in Data Center A acts as a producer sending
> messages to the `myTest` topic. A service in Data Center B acts as a consumer
> listening to `A.myTest`.
> The issue arises when MM2 on node 1 in Data Center A is stopped: the consumer
> in Data Center B ceases to receive messages. Even after I restarting MM2 in
> Data Center A, the consumer in Data Center B still does not receive messages
> until approximately 5 minutes later when a rebalance occurs, at which point
> it begins receiving messages again.
>
> [Logs From Consumer on Data Center B]
> [2024-07-23 17:29:17,270] INFO [MirrorCheckpointConnector|worker] refreshing
> consumer groups took 185 ms (org.apache.kafka.connect.mirror.Scheduler:95)
> [2024-07-23 17:29:19,189] INFO [MirrorCheckpointConnector|worker] refreshing
> consumer groups took 365 ms (org.apache.kafka.connect.mirror.Scheduler:95)
> [2024-07-23 17:29:22,271] INFO [MirrorCheckpointConnector|worker] refreshing
> consumer groups took 186 ms (org.apache.kafka.connect.mirror.Scheduler:95)
> [2024-07-23 17:29:24,193] INFO [MirrorCheckpointConnector|worker] refreshing
> consumer groups took 369 ms (org.apache.kafka.connect.mirror.Scheduler:95)
> [2024-07-23 17:29:25,377] INFO [Worker clientId=B->A, groupId=B-mm2]
> Rebalance started
> (org.apache.kafka.connect.runtime.distributed.WorkerCoordinator:242)
> [2024-07-23 17:29:25,377] INFO [Worker clientId=B->A, groupId=B-mm2]
> (Re-)joining group
> (org.apache.kafka.connect.runtime.distributed.WorkerCoordinator:604)
> [2024-07-23 17:29:25,386] INFO [Worker clientId=B->A, groupId=B-mm2]
> Successfully joined group with generation Generation\{generationId=52,
> memberId='B->A-adc19038-a8b6-40fb-9bf6-249f866944ab', protocol='sessioned'}
> (org.apache.kafka.connect.runtime.distributed.WorkerCoordinator:665)
> [2024-07-23 17:29:25,390] INFO [Worker clientId=B->A, groupId=B-mm2]
> Successfully synced group in generation Generation\{generationId=52,
> memberId='B->A-adc19038-a8b6-40fb-9bf6-249f866944ab', protocol='sessioned'}
> (org.apache.kafka.connect.runtime.distributed.WorkerCoordinator:842)
> [2024-07-23 17:29:25,390] INFO [Worker clientId=B->A, groupId=B-mm2] Joined
> group at generation 52 with protocol version 2 and got assignment:
> Assignment\{error=0, leader='B->A-adc19038-a8b6-40fb-9bf6-249f866944ab',
> leaderUrl='NOTUSED', offset=1360, connectorIds=[MirrorCheckpointConnector],
> taskIds=[MirrorCheckpointConnector-0, MirrorCheckpointConnector-1,
> MirrorCheckpointConnector-2], revokedConnectorIds=[], revokedTaskIds=[],
> delay=0} with rebalance delay: 0
> (org.apache.kafka.connect.runtime.distributed.DistributedHerder:2580)
> [2024-07-23 17:29:25,390] INFO [Worker clientId=B->A, groupId=B-mm2] Starting
> connectors and tasks using config offset 1360
> (org.apache.kafka.connect.runtime.distributed.DistributedHerder:1921)
> [2024-07-23 17:29:25,390] INFO [Worker clientId=B->A, groupId=B-mm2] Finished
> starting connectors and tasks
> (org.apache.kafka.connect.runtime.distributed.DistributedHerder:1950)
> [2024-07-23 17:29:26,883] INFO [Worker clientId=A->B, groupId=A-mm2]
> Rebalance started
> (org.apache.kafka.connect.runtime.distributed.WorkerCoordinator:242)
> [2024-07-23 17:29:26,883] INFO [Worker clientId=A->B, groupId=A-mm2]
> (Re-)joining group
> (org.apache.kafka.connect.runtime.distributed.WorkerCoordinator:604)
> [2024-07-23 17:29:26,890] INFO [Worker clientId=A->B, groupId=A-mm2]
> Successfully joined group with generation Generation\{generationId=143,
> memberId='A->B-0d04e6c1-f12a-4121-89af-e9992a167a01', protocol='sessioned'}
> (org.apache.kafka.connect.runtime.distributed.WorkerCoordinator:665)
> [2024-07-23 17:29:26,893] INFO [Worker clientId=A->B, groupId=A-mm2]
> Successfully synced group in generation Generation\{generationId=143,
> memberId='A->B-0d04e6c1-f12a-4121-89af-e9992a167a01', protocol='sessioned'}
> (org.apache.kafka.connect.runtime.distributed.WorkerCoordinator:842)
> [Configuration]
> name=MCS-MM2
> clusters = A, B
> A.bootstrap.servers = [kafka1]:[port],[kafka2]:[port],[kafka3]:[port]
> B.bootstrap.servers = [kafka]:[port]
> # enable and configure individual replication flows
> A->B.enabled = true
> A->B.topics = .*
> B->A.enabled = true
> B->A.topics = .*
> replication.factor=2
> tasks.max=3
> emit.checkpoints.interval.seconds=5
> A.producer.acks=all
> A.producer.batch.size=50000
> A.consumer.auto.offset.reset=latest
> B.consumer.auto.offset.reset=latest
> A.consumer.enable.auto.commit=true
> B.consumer.enable.auto.commit=true
> A.consumer.max.poll.interval.ms=20000
> B.consumer.max.poll.interval.ms=20000
> checkpoints.topic.replication.factor=1
> heartbeats.topic.replication.factor=1
> offset-syncs.topic.replication.factor=1
> offset.storage.replication.factor=1
> status.storage.replication.factor=1
> config.storage.replication.factor=1
> refresh.topics.enabled=true
> refresh.topics.interval.seconds=5
> refresh.groups.enabled=true
> refresh.groups.interval.seconds=5
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
