[jira] [Created] (KAFKA-14424) Cancellation of an ongoing replica reassignment should have sanity checks

["Lucas Wang (Jira)" <ji...@apache.org>]

Lucas Wang created KAFKA-14424:
----------------------------------

             Summary: Cancellation of an ongoing replica reassignment should have sanity checks
                 Key: KAFKA-14424
                 URL: https://issues.apache.org/jira/browse/KAFKA-14424
             Project: Kafka
          Issue Type: Improvement
            Reporter: Lucas Wang


When reassigning replicas, Kafka runs a sanity check to ensure all of the target replicas are alive before allowing the reassignment request to proceed.
However, for an AlterPartitionReassignments request that cancels an ongoing reassignment, there is no such check.
The result is that if the original replicas are offline, the cancellation may result in partitions
without any leaders. This problem has been observed in our clusters.

 

There should be some sanity check to ensure the cancellation would also land the partitions in valid states, e.g. by ensuring all of the original replicas are all alive.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)