Matthew Seal
Matthew Seal
Likely benign error, the background thread for queue processing is probably not being mutexed against the closing state and trying one last time to take some action while it is...
Thanks for reporting, I initially thought we were publishing ARM windows wheels but it looks like I was mistaken on second glance. I'll mark this as an enhancement request.
Closing as we're developing this within the `async` branch
Related ticket for API coverage: https://github.com/confluentinc/confluent-kafka-python/issues/1429. Will mark this ticket as high priority to get attention / see what's still not implemented into the unreleased version of the repo.
I wonder if this is related to the root cause of https://github.com/confluentinc/confluent-kafka-python/issues/1361
I agree this should be reflected similar to how you describe. We'll take a look at this as we start on adding proper type hints to this repository.
I've marked this ticket to look at in more depth. I want to consider if this is the shape for customizing the payload handling here, but I've similarly proposed making...
Appears something in librdkafka is using already freed memory when trying to cleanup a producer being deallocated. Likely there's some race or a missing GIL lock that's needed. I know...
Thanks, that helps. I've started some discussions on this topic internally. We're queuing looking into this though not sure exactly when we'll get it resolved.
Bumping on the thread so it's visible if someone has bandwidth to tackle (been over committed myself), it comes up periodically as an unexpected issue.