gaurav-narula commented on PR #17425: URL: https://github.com/apache/kafka/pull/17425#issuecomment-2462146077
> Have we considered the other alternative of making flush() in Indices and FileRecords check for existence? I'm not quite sure if I quite understand the "check for existence" bit. A couple of points for clarification Are you proposing an explicit check in `FileRecords#flush` and `Indices#flush`? If so, I think that would be racey - consider thread preemption right after the check. We can protect it by acquiring a lock but then the lock needs to be the same as the one acquired when [deleting the log](https://github.com/apache/kafka/blob/0181073d49b5a23ecdd493cd72c880a2cfc9c416/core/src/main/scala/kafka/log/UnifiedLog.scala#L1786). It seems a bit overkill, considering the file to be flushed may be due for deletion anyway. On the other hand, perhaps we can have a `maybeFlush()` in `Indices` and `FileRecords` which does a flush but swallows the exception, similar to what we do in `FileRecords.deleteIfExists` and its caller? I think it's also closer to how we handle things currently and avoids dealing with locks. -- 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
