[
https://issues.apache.org/jira/browse/FLINK-26583?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Matthias Pohl updated FLINK-26583:
----------------------------------
Description:
-We experience some unwanted behavior if a clean JobResult is listed in the
JobResultStore and a job with the same Job ID is submitted in Application Mode.
We would expect that the second submission fails with a
{{DuplicateJobSubmissionException}}. Instead, the submission succeeds with the
job not running anymore.-
Update:
We reiterated over the problem and decided that the Exception is not the
desired failure because it would cause a failover in k8s setups, for instance.
We rather want to inform the user through a warning because Flink still behaves
as expected.
was:
We experience some unwanted behavior if a clean JobResult is listed in the
JobResultStore and a job with the same Job ID is submitted in Application Mode.
We would expect that the second submission fails with a
{{DuplicateJobSubmissionException}}. Instead, the submission succeeds with the
job not running anymore.
> The user is not informed in any way when a job is resubmitted but already
> globally-terminated
> ---------------------------------------------------------------------------------------------
>
> Key: FLINK-26583
> URL: https://issues.apache.org/jira/browse/FLINK-26583
> Project: Flink
> Issue Type: Bug
> Components: Runtime / Coordination
> Affects Versions: 1.15.0
> Reporter: Matthias Pohl
> Assignee: Matthias Pohl
> Priority: Critical
> Labels: pull-request-available
> Fix For: 1.15.0
>
> Attachments: log.jm
>
>
> -We experience some unwanted behavior if a clean JobResult is listed in the
> JobResultStore and a job with the same Job ID is submitted in Application
> Mode.
> We would expect that the second submission fails with a
> {{DuplicateJobSubmissionException}}. Instead, the submission succeeds with
> the job not running anymore.-
> Update:
> We reiterated over the problem and decided that the Exception is not the
> desired failure because it would cause a failover in k8s setups, for
> instance. We rather want to inform the user through a warning because Flink
> still behaves as expected.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
