[
https://issues.apache.org/jira/browse/FLINK-24275?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17417720#comment-17417720
]
Austin Cawley-Edwards commented on FLINK-24275:
-----------------------------------------------
I see two options:
* Treat the endpoint imperatively ("Switch this Job to Cancelled")
** Fail on already cancelled (i.e. 409)
* Treat the endpoint declaratively ("Ensure this Job is Cancelled")
** Return a success message (i.e., 200, or 202 like is currently returned)
You know which fits better with the Flink API concepts than I do, but I would
lean towards the second option to simplify how users handle it. If duplicate
calls return an error, users have to classify this error as acceptable. I don't
see any case where a 409 error is handled any differently than a success.
> Allow idempotent job cancellation
> ---------------------------------
>
> Key: FLINK-24275
> URL: https://issues.apache.org/jira/browse/FLINK-24275
> Project: Flink
> Issue Type: Improvement
> Components: Runtime / REST
> Reporter: Austin Cawley-Edwards
> Priority: Major
>
> As a user of Flink, I want to be able to cancel a job from an external system
> in a fault-tolerant way without guessing if the job has already been
> cancelled.
>
> Currently, the cancel endpoint (PATCH /jobs/:jobid?mode=cancel) will return a
> 404 if the job is already cancelled. This makes it hard to detect if the job
> truly doesn't exist, or if it is already in the desired state.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
