hi @tillrohrmann I have refactored this PR and counted the failure number in the `CheckpointCoordinator`. I think I should push the implementation to let you estimate. I have tested the counter's run path, but I don't know if it is the [right way](https://github.com/apache/flink/pull/6567/files#diff-a38ea0fa799bdaa0b354d80cd8368c60R1010) of failing the `ExecutionGraph` . And maybe the test case I added have too much assert, I will reduce it.
[ Full content available at: https://github.com/apache/flink/pull/6567 ] This message was relayed via gitbox.apache.org for [email protected]
