lhotari commented on PR #25044: URL: https://github.com/apache/pulsar/pull/25044#issuecomment-3628526303
> > es continue to be produced while snapshotting is happening but not all messages are acknowledged. It should be possible to reproduce this bug in a test. > > After all, this bug does exist. I had made a previous attempt in #16651 to fix it and now I've revisited the solution so that it makes more sense and is now correct. > > #16651 fixes 2 separate but related issues. One where there's a race condition between the snapshot creation and markdelete updates and another one where the wrong position is used for the snapshot position. The snapshot's position should be the position when the snapshot request was created since that's the point which is guaranteed to be in sync. Any further position could contain messages from other clusters etc.. Another problem that it solves is that a snapshot couldn't be used when the snapshot position was farther ahead that it actually should have been. I revisited PR 16651 once again. The description and the changes contain the latest status of those changes. Besides the race condition issue fix, there's a fix for the wrong position issue. The issue was slightly different than what I thought before. The wrong position issue was about responding to the request with a response which uses the position where the message id is the latest message id in the topic. That didn't make sense to me and I changed it to use the position of the request marker message. -- 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: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
