cshuo commented on issue #8276: URL: https://github.com/apache/hudi/issues/8276#issuecomment-3044064936
Hi all, I've reproduced the problem locally, and figured out the root cause: 1) the `stuck` happens in `instantToWrite` method w/ write function, where write tasks waiting for coordinator generating a new instant to write until timeout. 2) Before https://github.com/apache/hudi/pull/13285, flink write coordinator generates instant in synchronous mode, i.e., new instant will be generated until previous instant is committed. 3) For flink ingestion, commiting happens during `notifyCheckpointComplete` in write coordinator. 4) `notifyCheckpointComplete` will not be called for Savepoint. So when Savepoint is called asynchronously without stopping job, commit will not happen and new instant will not be generated, which leads to job stuck until checkpoint timeout exception. The possible solution is: 1) Using https://github.com/apache/hudi/pull/13285 as @danny0405 mentioned above, which is introduced in release 1.1. 2) Disable asynchronous savepoint or using 'Stop/Resume job with savepoint'. -- 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]
