lh0156 opened a new pull request, #23054: URL: https://github.com/apache/kafka/pull/23054
## Motivation MirrorCheckpointTask retries the same checkpoint forever when a target topic-partition has been deleted. The aggregate alter-offsets future turns a single missing partition into a recurring error, and the compacted checkpoints topic retains the stale record. ## Changes - Process `alterConsumerGroupOffsets` results per partition. - Stop retrying a partition after `UnknownTopicOrPartitionException`. - Emit a checkpoint tombstone so the stale record is removed from the compacted topic. - Apply checkpoint tombstones when loading `CheckpointStore`. - Synchronize checkpoint-store access used by poll and offset-sync scheduler. ## Tests - `./gradlew :connect:mirror-client:test :connect:mirror:test` - `./gradlew :connect:mirror-client:spotlessCheck :connect:mirror:spotlessCheck :connect:mirror-client:checkstyleMain :connect:mirror-client:checkstyleTest :connect:mirror:checkstyleMain :connect:mirror:checkstyleTest :connect:mirror-client:spotbugsMain :connect:mirror:spotbugsMain` - `git diff --check` Jira: https://issues.apache.org/jira/browse/KAFKA-19107 -- 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]
