[
https://issues.apache.org/jira/browse/FLINK-20672?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17784035#comment-17784035
]
Yun Tang commented on FLINK-20672:
----------------------------------
[~Zakelly] Thanks for the information. If so, I have another question: do we
really need the {{io-executor}} to work with {{FatalExitExceptionHandler}}?
From my point of view, if we do not delete the Savepoint correctly (as this is
also executed on the {{io-executor}}), shall we need to fail the whole
JobManager?
If the correct behavior of the exception handler of {{io-executor}} is not
fatal exiting, I think we shall correct that behavior first.
[~Zakelly], [~roman], [~srichter] WDYT?
> notifyCheckpointAborted RPC failure can fail JM
> -----------------------------------------------
>
> Key: FLINK-20672
> URL: https://issues.apache.org/jira/browse/FLINK-20672
> Project: Flink
> Issue Type: Bug
> Components: Runtime / Checkpointing
> Affects Versions: 1.11.3, 1.12.0
> Reporter: Roman Khachatryan
> Assignee: Zakelly Lan
> Priority: Not a Priority
> Labels: auto-deprioritized-major, auto-deprioritized-minor,
> pull-request-available
>
> Introduced in FLINK-8871, aborted RPC notifications are done asynchonously:
>
> {code}
> private void sendAbortedMessages(long checkpointId, long timeStamp) {
> // send notification of aborted checkpoints asynchronously.
> executor.execute(() -> {
> // send the "abort checkpoint" messages to necessary
> vertices.
> // ..
> });
> }
> {code}
> However, the executor that eventually executes this request is created as
> follows
> {code}
> final ScheduledExecutorService futureExecutor =
> Executors.newScheduledThreadPool(
> Hardware.getNumberCPUCores(),
> new ExecutorThreadFactory("jobmanager-future"));
> {code}
> ExecutorThreadFactory uses UncaughtExceptionHandler that exits JVM on error.
> cc: [~yunta]
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
