Hangleton commented on code in PR #14649:
URL: https://github.com/apache/kafka/pull/14649#discussion_r1377568323


##########
core/src/main/scala/kafka/log/UnifiedLog.scala:
##########
@@ -1754,6 +1754,7 @@ class UnifiedLog(@volatile var logStartOffset: Long,
         leaderEpochCache.foreach(_.clearAndFlush())
         producerStateManager.truncateFullyAndStartAt(newOffset)
         logStartOffset = logStartOffsetOpt.getOrElse(newOffset)
+        if (remoteLogEnabled()) maybeIncrementLocalLogStartOffset(newOffset, 
LogStartOffsetIncrementReason.SegmentDeletion)

Review Comment:
   The reason is indeed not always correct. In the case of truncation when a 
follower aligns with the leader, what was missing was the update of the local 
log start offset, which was unchanged after building the state on the follower 
from auxiliary files and the response from the initial follower fetcher.



-- 
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

Reply via email to