ableegoldman commented on pull request #10342: URL: https://github.com/apache/kafka/pull/10342#issuecomment-804392576
Thanks @guozhangwang ! > If users mistakenly use the same stat dir, is that undefined behavior or do we fail fast in a deterministic way? If it's the former do we have test case covering it? Yes, we added the guard against that in PR #9978 > Do we have test case covering "thread is gone but the map still have it" case for the cleanup thread? Not yet, but that's what I was referring to when I said I wanted to add another test to this PR in my [earlier comment above yours](https://github.com/apache/kafka/pull/10342#issuecomment-803050846) > I suggest we also add something in docs emphasizing that we were trying to guard against multiple instances with the same state dir on same node, and we do not do that now. As noted above, we do still guard against this case but with a top-level state directory lock instead of individual task dir locks. This situation is documented in [this PR](https://github.com/apache/kafka/pull/10187) -- 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org
