dajac opened a new pull request, #15212:
URL: https://github.com/apache/kafka/pull/15212
We had a case where a partition got assigned to two members and we found a
bug in the partition epochs bookkeeping. Basically, when a member has a
partition pending revocation re-assigned to him before the revocation is
completed, the partition epoch is lost. Here is an example of such transition:
```
[2024-01-16 12:10:52,613] INFO [GroupCoordinator id=1
topic=__consumer_offsets partition=7] [GroupId
rdkafkatest_rnd53b4eb0c2de343_0113u] Member M2 transitioned from
CurrentAssignment(memberEpoch=11, previousMemberEpoch=9, targetMemberEpoch=14,
state=revoking, assignedPartitions={},
partitionsPendingRevocation={EnZMikZURKiUoxZf0rozaA=[0, 1, 2, 3, 4, 5, 6, 7]},
partitionsPendingAssignment={IKXGrFR1Rv-Qes7Ummas6A=[0, 5]}) to
CurrentAssignment(memberEpoch=15, previousMemberEpoch=11, targetMemberEpoch=15,
state=stable, assignedPartitions={EnZMikZURKiUoxZf0rozaA=[0, 1, 2, 3, 4, 5, 6,
7]}, partitionsPendingRevocation={}, partitionsPendingAssignment={}).
(org.apache.kafka.coordinator.group.GroupMetadataManager)
```
### 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 to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: jira-unsubscr...@kafka.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
