[
https://issues.apache.org/jira/browse/FLINK-14114?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16937901#comment-16937901
]
Zhu Zhu commented on FLINK-14114:
---------------------------------
Currently it's "akka.client.timeout" which is not a real Akka config and is
only used by Flink client.
And agreed that the usage of the {{timeout}} should be reconsidered, as it is
only used for {{cancel}} and {{getAccumulators}}, and does not apply some other
requests like {{listJobs}} or {{stopWithSavepoint}}. Let the caller to decide
the timeout looks good as some operations are not expected to follow a same
timeout.
> Shift down ClusterClient#timeout to RestClusterClient
> -----------------------------------------------------
>
> Key: FLINK-14114
> URL: https://issues.apache.org/jira/browse/FLINK-14114
> Project: Flink
> Issue Type: Sub-task
> Components: Client / Job Submission
> Reporter: tison
> Assignee: Zhu Zhu
> Priority: Major
> Labels: pull-request-available
> Fix For: 1.10.0
>
> Time Spent: 20m
> Remaining Estimate: 0h
>
> {{ClusterClient#timeout}} is only used in {{RestClusterClient}}, even without
> this prerequisite we can always shift down {{timeout}} field to subclasses of
> {{ClusterClient}}. It is towards an interface-ized {{ClusterClient}}. By side
> effect, we could reduce the dependency to parsing duration with Scala
> Duration on the fly.
> CC [~till.rohrmann] [~zhuzh]
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
