[jira] [Comment Edited] (FLINK-27925) Avoid to create watcher without the resourceVersion

["Aitozi (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/FLINK-27925?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17552554#comment-17552554 ] 

Aitozi edited comment on FLINK-27925 at 6/10/22 5:14 AM:
---------------------------------------------------------

I think it will only cache the current job's pod list, Right ? I think the memory consume is acceptable.

From the doc, it will trigger a request to ETCD if we create a watcher with the resource version unset. From our internal case, it's also observed.


was (Author: aitozi):
Will it cache extra pods of other jobs if we specify the current job label?   From the doc, it will trigger a request to ETCD if we create a watcher with the resource version unset. From out internal case, it's also observed

> Avoid to create watcher without the resourceVersion
> ---------------------------------------------------
>
>                 Key: FLINK-27925
>                 URL: https://issues.apache.org/jira/browse/FLINK-27925
>             Project: Flink
>          Issue Type: Improvement
>          Components: Deployment / Kubernetes
>            Reporter: Aitozi
>            Priority: Major
>
> Currently, we create the watcher in KubernetesResourceManager. But it do not pass the resourceVersion parameter, it will trigger a request to etcd. It will bring the burden to the etcd in large scale cluster (which have been seen in our internal k8s cluster). More detail can be found [here|https://kubernetes.io/docs/reference/using-api/api-concepts/#the-resourceversion-parameter] 
> I think we could use the informer to improve it (which will spawn a list-watch and maintain the resourceVersion internally)



--
This message was sent by Atlassian Jira
(v8.20.7#820007)