[
https://issues.apache.org/jira/browse/FLINK-29329?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17608037#comment-17608037
]
Yun Tang commented on FLINK-29329:
----------------------------------
I think the problem of not triggering the checkpoints anymore should be related
to the [schedule
timer|https://github.com/apache/flink/blob/b5cd9f34ab73fa69a3db5a09908c1aa954ed0597/flink-runtime/src/main/java/org/apache/flink/runtime/checkpoint/CheckpointCoordinator.java#L168].
If you could reproduce this problem, I think you could use jmap dump the job
manager to see what happened to CheckpointCoordinator#timer.
> Checkpoint can not be triggered if encountering OOM
> ---------------------------------------------------
>
> Key: FLINK-29329
> URL: https://issues.apache.org/jira/browse/FLINK-29329
> Project: Flink
> Issue Type: Bug
> Reporter: Yuxin Tan
> Priority: Major
> Fix For: 1.13.7
>
> Attachments: job-exceptions-1.txt
>
>
> When writing a checkpoint, an OOM error is thrown. But the JM is not failed
> and is restored because I found a log "No master state to restore".
> Then JM never makes checkpoints anymore. Currently, the root cause is not
> that clear, maybe this is a bug and we should deal with the OOM or other
> exceptions when making checkpoints.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
