[
https://issues.apache.org/jira/browse/KAFKA-19508?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18019391#comment-18019391
]
Jainil Rana commented on KAFKA-19508:
-------------------------------------
Hi [~lazymooner] , I set up a 3-node KRaft cluster on WSL and tested leader
failover. When I kill the leader broker, Raft elects a new leader and ISR
updates correctly. The producer pauses briefly but does *not* show any
{{{}NotLeaderOrFollowerException{}}}, messages eventually succeed without
error output.
Is this the expected behavior (i.e. client-side retries fully masking the
transient error), or should the producer normally expose a
{{NotLeaderOrFollowerException}} during leader transitions?
> Producer keeps reporting a NotLeaderOrFollowerException
> -------------------------------------------------------
>
> Key: KAFKA-19508
> URL: https://issues.apache.org/jira/browse/KAFKA-19508
> Project: Kafka
> Issue Type: Bug
> Components: clients, core, producer
> Affects Versions: 3.5.0
> Reporter: mooner
> Assignee: Jainil Rana
> Priority: Blocker
> Attachments: image-2025-07-15-15-55-01-596.png,
> image-2025-07-15-15-59-02-855.png, image-2025-07-15-16-01-59-877.png,
> image-2025-07-15-16-03-13-043.png
>
>
> 1.After the server restarts, the producer keeps reporting a
> NotLeaderOrFollowerException when producing messages.
> !image-2025-07-15-15-55-01-596.png!
> 2.Check the zknode data; the leader is functioning normally.
> !image-2025-07-15-15-59-02-855.png!
> 3.Strangely, at 04:34:01, the leader was 12, and at 04:34:16, the leader was
> 11, but the leaderEpoch did not change.
> !image-2025-07-15-16-03-13-043.png!
> !image-2025-07-15-16-01-59-877.png!
> 4.state-change logs
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
