> Do we need to take this refactoring into account? Because this PR is actually 
> a supplement to the checkpoint exception handler.

Yes, I think we can have the same semantics as in the previous only-TM 
implementation (`setFailOnCheckpointingErrors` means `tolerableFailures == 0`) 
but now everything should happen in `CheckpointCoordinator`. That is why I also 
mentioned this point:

> Consider having only DecliningCheckpointExceptionHandler on TaskExecutor side 
> and letting now to handle all failure cases only in CheckpointCoordinator


[ Full content available at: https://github.com/apache/flink/pull/6567 ]
This message was relayed via gitbox.apache.org for [email protected]

Reply via email to