At the detailed level:
Let's consider topics that are enqueued for deletion but haven't started the 
actual deletion, maybe they have ongoing partition reassignments that would 
still take many hours to finish. And for those partitions during a controlled 
shutdown we won't try to switch leaderships because we no longer care about 
their availability.
And here even if the shutting down broker is still leader for those partitions, 
I think it's ok to shutdown the broker immediately because again we don't care 
about the partition availability anymore. Using !isTopicWithDeletionStarted 
will still return those partitions and block the controlled shutdown, which is 
probably not ideal.

At a higher level:
isTopicQueuedUpForDeletion is used in many other places so that for other types 
of operations, once a topic is enqueued, it effectively no longer exists. I 
think that also applies here, and using the same method consistently simplifies 
reasoning. Thoughts?

[ Full content available at: https://github.com/apache/kafka/pull/5388 ]
This message was relayed via gitbox.apache.org for [email protected]

Reply via email to