hutchiko commented on pull request #11283: URL: https://github.com/apache/kafka/pull/11283#issuecomment-918537415
@guozhangwang @ableegoldman unfortunately I could never reproduce the CI failures however I have pushed up a refactor of the method which I think was responsible for the flakiness. The original version of the the method was scanning backwards through the changelog topic searching for the top record so I could cross check that record's offset with the checkpointed offset. It had an implicit assumption that the consumer it was driving backwards would always get some records after a 50ms `poll` - thinking this through it's obviously a false assumption. I switched the logic around so it just consumes forwards until it finds the end of the topic there are no assumptions about timing in the new logic so I'm hoping that will fix the flakiness. -- 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
