Andrew Schofield created KAFKA-20330:
----------------------------------------
Summary: Share partition leader restart without leadership change
gives wrong exception
Key: KAFKA-20330
URL: https://issues.apache.org/jira/browse/KAFKA-20330
Project: Kafka
Issue Type: Bug
Affects Versions: 4.2.0
Reporter: Andrew Schofield
Assignee: Andrew Schofield
Fix For: 4.3.0
If I have a single-broker cluster and a share consumer with implicit
acknowledgement, it is possible to get a share session error when records are
fetched before the broker restarts but acknowledged afterwards. The handling of
the exception looks good, but it's not the intended exception in this
situation. The code handles leadership change correctly, but it can be confused
by a leadership non-change in spite of a restart.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
