mjsax commented on pull request #10529: URL: https://github.com/apache/kafka/pull/10529#issuecomment-818392640
Updated the PR. > (although, why would the max poll interval impact a timeout related to the admin client? Or do I not understand where this timeout is ultimately coming from) We use `InternalTopicManager` in the consumer group leader `assign()` callback to create internal topics. If we get errors back from admin client, we retry. To bound the retries, we use `max.poll.interval.ms / 2` because `max.poll.interval.ms` defines an upper bound before `assign()` must finish (otherwise the leader gets kicked out of the group). (Cf. KIP-572). > But we should probably try to actually handle the null even so, and fail with a more useful error message in case we ever do hit this again. `admin.deleteTopic` would never return `null` in production, and thus handling `null` within `InternalTopicManager` for this case seems not to make sense. -- 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
