[
https://issues.apache.org/jira/browse/KAFKA-7284?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16581572#comment-16581572
]
ASF GitHub Bot commented on KAFKA-7284:
---------------------------------------
vvcephei opened a new pull request #5513: KAFKA-7284: streams should unwrap
fenced exception
URL: https://github.com/apache/kafka/pull/5513
Unwrap the ProducerFencedException in RecordCollectorImpl
so it can be caught and converted to a TaskMigratedException
instead of triggering a shutdown.
See also #5499
### Committer Checklist (excluded from commit message)
- [ ] Verify design and implementation
- [ ] Verify test coverage and CI build status
- [ ] Verify documentation (including upgrade notes)
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
> Producer getting fenced may cause Streams to shut down
> ------------------------------------------------------
>
> Key: KAFKA-7284
> URL: https://issues.apache.org/jira/browse/KAFKA-7284
> Project: Kafka
> Issue Type: Bug
> Components: streams
> Affects Versions: 0.11.0.3, 1.0.2, 1.1.1, 2.0.0
> Reporter: John Roesler
> Assignee: John Roesler
> Priority: Critical
> Fix For: 2.0.1, 2.1.0
>
>
> As part of the investigation, I will determine what other versions are
> affected.
>
> In StreamTask, we catch a `ProducerFencedException` and throw a
> `TaskMigratedException`. However, in this case, the `RecordCollectorImpl` is
> throwing a `StreamsException`, caused by `KafkaException` caused by
> `ProducerFencedException`.
> In response to a TaskMigratedException, we would rebalance, but when we get a
> StreamsException, streams shuts itself down.
> In other words, we intended to do a rebalance in response to a producer
> fence, but actually, we shut down (when the fence happens inside the record
> collector).
> Coincidentally, Guozhang noticed and fixed this in a recent PR:
> [https://github.com/apache/kafka/pull/5428/files#diff-4e5612eeba09dabf30d0b8430f269ff6]
>
> The scope of this ticket is to extract that fix and associated tests, and
> send a separate PR to trunk and 2.0, and also to determine what other
> versions, if any, are affected.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
