ableegoldman commented on pull request #10986: URL: https://github.com/apache/kafka/pull/10986#issuecomment-878748345
Ok I realize we actually do have a test that reproduces this already: `ConsumerCoordinatorTest.testRebalanceWithMetadataChange`. This test sets up a case where a change in topic metadata triggers a rebalance after a member had joined the group, after which the change is reverted so that the metadata is ultimately the same. Then a `NOT_COORDINATOR` response is sent to fail the initial JoinGroup, and the test just verifies that the member attempts to rejoin until successful. It also verifies things like the number of times each rebalance callback is invoked, and the set of partitions that the callbacks receive. This test actually only failed in the COOPERATIVE case, which confirms that the behavior remains correct for the EAGER case. When following the COOPERATIVE protocol, the test was formerly assuming that the member would retain all partitions despite actually having its generation and memberId cleared when the initial JoinGroup is failed. So it was technically asserting the wrong behavior beforehand; just fixing this gives us a unit test for this patch after all. -- 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