[
https://issues.apache.org/jira/browse/FLINK-12333?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16829049#comment-16829049
]
Till Rohrmann commented on FLINK-12333:
---------------------------------------
You are right [~yunta]. All subclasses of
{{AbstractAsynchronousOperationHandlers}} can potentially block the cluster
shut down if the results are not queried.
> Add documentation for all async operations through REST
> --------------------------------------------------------
>
> Key: FLINK-12333
> URL: https://issues.apache.org/jira/browse/FLINK-12333
> Project: Flink
> Issue Type: Improvement
> Components: Documentation, Runtime / REST
> Affects Versions: 1.8.0
> Reporter: Vishal Santoshi
> Assignee: Yun Tang
> Priority: Minor
>
> Exit code 2 seen ( after 5 minutes ) when
> {code:java}
> curl --header "Content-Type: application/json" --request POST --data
> '{"target-directory":"***********","cancel-job":true}'
> https://***************/jobs/00000000000000000000000000000000/savepoints{code}
> It seems that when a REST call is
> "Triggered the cancel with savepoint command from via the REST call. This
> command is an asynchronous operation which produces a result (the savepoint
> path). In order to deliver asynchronous results to the caller, Flink waits
> before shutting down until they are delivered or until it times out after 5
> minutes."
>
> That implies that one has to execute
> {code:java}
> curl --request GET
> https://**********/jobs/00000000000000000000000000000000/savepoints/[Request_id]
> {code}
> on the request_id returned by the first call ( within 5 minutes ) , for a
> clean exit ( code 0 )
>
> Please add this life cycle in flink documentation , in all probability here
> [https://ci.apache.org/projects/flink/flink-docs-stable/monitoring/rest_api.html#jobs-jobid-savepoints]
>
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
