XComp opened a new pull request #14847: URL: https://github.com/apache/flink/pull/14847
## What is the purpose of the change In [FLINK-21030](https://issues.apache.org/jira/browse/FLINK-21030), we experienced a bug when calling stop-with-savepoint on an embarrassingly parallel job where one pipeline failed causing the job ending up in an inconsistent state. The solution was to trigger a global failover in such a case. ## Brief change log We couldn't rely on the `ExecutionGraph`'s `terminationFuture` to complete. The solution was to monitor the `terminationFutures` of all current executions, instead. We cause a failure if at least one of them switches to a non-FINISHED state indicating that something went wrong while stopping the job. ## Verifying this change Two tests were added: * testing the behavior if the failure happens during savepoint creation. No explicit restart needs to be triggered since the tasks are not stopped. * testing the behavior if the failure happens during stopping the job. In this case, second failover will be triggered that causes job to get back to RUNNING. ## Does this pull request potentially affect one of the following parts: - Dependencies (does it add or upgrade a dependency): no - The public API, i.e., is any changed class annotated with `@Public(Evolving)`: no - The serializers: no - The runtime per-record code paths (performance sensitive): no - Anything that affects deployment or recovery: JobManager (and its components), Checkpointing, Kubernetes/Yarn/Mesos, ZooKeeper: yes - The S3 file system connector: no ## Documentation - Does this pull request introduce a new feature? no - If yes, how is the feature documented? not applicable ---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org
