[
https://issues.apache.org/jira/browse/FLINK-26528?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17503631#comment-17503631
]
Aitozi edited comment on FLINK-26528 at 3/9/22, 3:07 PM:
---------------------------------------------------------
Yes, my meaning is to eliminate the unchanged update towards kubernetes. IMO,
The {{UpdateControl}} have four abilities
* update status
* update resource
* not update not reschedule
* not update but reschedule
So if we have change the status or resource (not recommended and not used now),
we should apply the update. If the status have not achieved the desired state,
and is not changed since last reconcile loop, we should just reschedule and
wait the next turn.
was (Author: aitozi):
Yes, my meaning is to eliminate the unchanged update to toward kubernetes. IMO,
The {{UpdateControl}} have four abilities
* update status
* update resource
* not update not reschedule
* not update but reschedule
So if we have change the status or resource (not recommended and not used now),
we should apply the update. If the status have not achieved the desired state,
and is not changed since last reconcile loop, we should just reschedule and
wait the next turn.
> Trigger the updateControl when the FlinkDeployment have changed
> ---------------------------------------------------------------
>
> Key: FLINK-26528
> URL: https://issues.apache.org/jira/browse/FLINK-26528
> Project: Flink
> Issue Type: Sub-task
> Components: Kubernetes Operator
> Reporter: Aitozi
> Priority: Major
>
> If the CR has not changed since last reconcile, we could create a
> UpdateControl with {{UpdateControl#noUpdate}} , this is meant to reduce the
> unnecessary update for resource
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
