[
https://issues.apache.org/jira/browse/FLINK-26331?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17497286#comment-17497286
]
Matthias Pohl commented on FLINK-26331:
---------------------------------------
Looking through different configuration parameters handling retries it looks
like I'm gonna rename the once that are already implemented:
{{jobmanager.cleanup.max-delay}} and {{jobmanager.cleanup.min-delay}}.
Relevant configuration parameters identified:
* {{rest.retry.delay}} / {{rest.retry.max-attempts}}
* {{high-availability.zookeeper.client.max-retry-attempts}} /
{{high-availability.zookeeper.client.retry-wait}}
* {{high-availability.kubernetes.leader-election.retry-period}} /
{{high-availability.kubernetes.leader-election.lease-duration}}
* {{resourcemanager.start-worker.max-failure-rate}} /
{{resourcemanager.start-worker.retry-interval}}
* {{taskmanager.network.retries}}
I'm gonna propose the following parameters as a consequence:
* {{jobmanager.cleanup.retry.min-delay}}
* {{jobmanager.cleanup.retry.max-delay}}
* {{jobmanager.cleanup.retry.max-attempts}}
> Make max retries configurable
> -----------------------------
>
> Key: FLINK-26331
> URL: https://issues.apache.org/jira/browse/FLINK-26331
> Project: Flink
> Issue Type: Sub-task
> Components: Runtime / Coordination
> Affects Versions: 1.15.0
> Reporter: Matthias Pohl
> Assignee: Matthias Pohl
> Priority: Major
>
> Right now, the retry mechanism is hard-coded to {{Integer.MAX_VALUE}}. We
> want to make that configurable as well and keep the default value at
> {{MAX_VALUE}}. This enables the user to disable the retry mechanism if
> necessary.
> We verify how retries strategies are configured in other places and align
> with that to have a consistent user experience.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
