dajac commented on PR #12674: URL: https://github.com/apache/kafka/pull/12674#issuecomment-1256427128
> This change makes sense. I guess the other interesting case is if fetch from follower _is_ enabled on the broker, but the client is not sending the epoch (KIP-320). What should we do in that case? Is it safe to let it through? We may need another JIRA to discuss that case, which is not as clear, but still potentially problematic. I think that one fundamental issue is that a follower is accessible even if it has not caught-up yet with the leader. The follower might be far behind for instance or even have wrong records if the truncation did not happen yet. It seems to me that we should make it inaccessible until the follower has at least got the updated high watermark from the leader. Once we have the HWM, we could perhaps also use OFFSET_NOT_AVAILABLE if the requested offset are not available yet. I will file a jira and put my thoughts in it. -- 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
