[
https://issues.apache.org/jira/browse/FLINK-13497?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16900764#comment-16900764
]
Till Rohrmann commented on FLINK-13497:
---------------------------------------
The reason why we don't run the {{CheckpointCoordinator}} from the
{{RpcEndpoint's}} main thread is that the {{CheckpointCoordinator}} contains
blocking calls. Hence, there is a possibility that we block the
{{RpcEndpoint's}} main thread which must never happen. This actually dates back
to FLINK-5960. Hence the proper solution as Piotr said is to encapsulate all
blocking operations in the {{ioExecutor}} of the {{CheckpointCoordinator}} and
then use the {{RpcEndpoint's}} main thread to run the {{CheckpointCoordinator}}.
> Checkpoints can complete after CheckpointFailureManager fails job
> -----------------------------------------------------------------
>
> Key: FLINK-13497
> URL: https://issues.apache.org/jira/browse/FLINK-13497
> Project: Flink
> Issue Type: Bug
> Components: Runtime / Checkpointing
> Affects Versions: 1.9.0, 1.10.0
> Reporter: Till Rohrmann
> Priority: Critical
> Fix For: 1.9.0
>
>
> I think that we introduced with FLINK-12364 an inconsistency wrt to job
> termination a checkpointing. In FLINK-9900 it was discovered that checkpoints
> can complete even after the {{CheckpointFailureManager}} decided to fail a
> job. I think the expected behaviour should be that we fail all pending
> checkpoints once the {{CheckpointFailureManager}} decides to fail the job.
--
This message was sent by Atlassian JIRA
(v7.6.14#76016)
