mvanhorn commented on PR #21714: URL: https://github.com/apache/kafka/pull/21714#issuecomment-6434772407
Rebased onto trunk again (9f27eb6), and the PR is now smaller. #22836 fixed KAFKA-20253 for the consumer protocol in July, and trunk's `maximumTimeToWait` now waits a retry backoff while the coordinator is unknown. So I dropped this branch's changes to the async heartbeat managers and their test. On current trunk they also made a joining member wait out its first heartbeat interval, which broke trunk's own `testInitialHeartbeatIntervalUsesRequestTimeout`. What's left is the classic protocol side, which #22836 didn't touch. `AbstractCoordinator`'s heartbeat thread now calls `heartbeat.failHeartbeat()` in its `AuthenticationException` and `GroupAuthorizationException` handlers. Without it the heartbeat timer stays expired after the failure, so `timeToNextHeartbeat()` returns 0 and the thread spins. `HeartbeatTest.testFailHeartbeatResetsTimerWithBackoff` covers that. HeartbeatTest, AbstractCoordinatorTest, ConsumerHeartbeatRequestManagerTest and StreamsGroupHeartbeatRequestManagerTest pass on JDK 21 (402 tests), and `:clients` checkstyle is clean. @lianetm, this is ready whenever you have time. <!-- r:cec9f7dcedc8.cf18b83229e9 --> -- 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: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
