[
https://issues.apache.org/jira/browse/FLINK-24903?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Roman Khachatryan updated FLINK-24903:
--------------------------------------
Description:
[https://dev.azure.com/khachatryanroman/flink/_build/results?buildId=1225&view=logs&j=9dc1b5dc-bcfa-5f83-eaa7-0cb181ddc267&t=511d2595-ec54-5ab7-86ce-92f328796f20&l=7753]
{code}
2021-11-14T20:22:23.1142812Z Nov 14 20:22:23 [ERROR] Failures:
2021-11-14T20:22:23.1149388Z Nov 14 20:22:23 [ERROR]
AdaptiveSchedulerTest.testJobStatusListenerNotifiedOfJobStatusChanges:684
2021-11-14T20:22:23.1150058Z Nov 14 20:22:23 Expected: (a collection
containing <RUNNING> and a collection containing <FINISHED>)
2021-11-14T20:22:23.1150581Z Nov 14 20:22:23 but: a collection containing
<FINISHED> was <RUNNING>
2021-11-14T20:22:23.1152966Z Nov 14 20:22:23 [INFO]
2021-11-14T20:22:23.1156414Z Nov 14 20:22:23 [ERROR] Tests run: 6048,
Failures: 1, Errors: 0, Skipped: 97
{code}
Locally, it fails ~14 runs out of 100 (when running only
testJobStatusListenerNotifiedOfJobStatusChanges in a loop).
It looks like job termination future is always completed before the
jobStatusChangeListener is notified (AdaptiveScheduler.transitionToState,
targetState.getState() completes the future).
Sleeping for 1ms before checking the assertion prevents the failure.
cc: [~trohrmann]
was:
[https://dev.azure.com/khachatryanroman/flink/_build/results?buildId=1225&view=logs&j=9dc1b5dc-bcfa-5f83-eaa7-0cb181ddc267&t=511d2595-ec54-5ab7-86ce-92f328796f20&l=7753]
Locally, it fails ~14 runs out of 100 (when running only
testJobStatusListenerNotifiedOfJobStatusChanges in a loop).
It looks like job termination future is always completed before the
jobStatusChangeListener is notified (AdaptiveScheduler.transitionToState,
targetState.getState() completes the future).
Sleeping for 1ms before checking the assertion prevents the failure.
cc: [~trohrmann]
> AdaptiveSchedulerTest.testJobStatusListenerNotifiedOfJobStatusChanges unstable
> ------------------------------------------------------------------------------
>
> Key: FLINK-24903
> URL: https://issues.apache.org/jira/browse/FLINK-24903
> Project: Flink
> Issue Type: Bug
> Components: Runtime / Coordination
> Affects Versions: 1.15.0
> Reporter: Roman Khachatryan
> Priority: Major
> Fix For: 1.15.0
>
>
> [https://dev.azure.com/khachatryanroman/flink/_build/results?buildId=1225&view=logs&j=9dc1b5dc-bcfa-5f83-eaa7-0cb181ddc267&t=511d2595-ec54-5ab7-86ce-92f328796f20&l=7753]
> {code}
> 2021-11-14T20:22:23.1142812Z Nov 14 20:22:23 [ERROR] Failures:
> 2021-11-14T20:22:23.1149388Z Nov 14 20:22:23 [ERROR]
> AdaptiveSchedulerTest.testJobStatusListenerNotifiedOfJobStatusChanges:684
> 2021-11-14T20:22:23.1150058Z Nov 14 20:22:23 Expected: (a collection
> containing <RUNNING> and a collection containing <FINISHED>)
> 2021-11-14T20:22:23.1150581Z Nov 14 20:22:23 but: a collection
> containing <FINISHED> was <RUNNING>
> 2021-11-14T20:22:23.1152966Z Nov 14 20:22:23 [INFO]
> 2021-11-14T20:22:23.1156414Z Nov 14 20:22:23 [ERROR] Tests run: 6048,
> Failures: 1, Errors: 0, Skipped: 97
> {code}
> Locally, it fails ~14 runs out of 100 (when running only
> testJobStatusListenerNotifiedOfJobStatusChanges in a loop).
> It looks like job termination future is always completed before the
> jobStatusChangeListener is notified (AdaptiveScheduler.transitionToState,
> targetState.getState() completes the future).
> Sleeping for 1ms before checking the assertion prevents the failure.
>
> cc: [~trohrmann]
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
