[
https://issues.apache.org/jira/browse/FLINK-21921?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17306879#comment-17306879
]
Yang Wang commented on FLINK-21921:
-----------------------------------
[~yittg] Thanks for raising this concern. I think maybe it is not a major issue
because we will always receive the MODIFIED event periodically.
But I agree with you that having the same logics in {{onAdded}} with
{{onModified}} is harmless.
> Unreasonable ADDED event ignoring in callback handler in
> KubernetesLeaderRetrievalDriver
> ----------------------------------------------------------------------------------------
>
> Key: FLINK-21921
> URL: https://issues.apache.org/jira/browse/FLINK-21921
> Project: Flink
> Issue Type: Bug
> Reporter: Yi Tang
> Priority: Major
>
> The comment is not exactly accurate in case of JM restarting or ConfigMap
> watching restarting(KubernetesTooOldResourceVersionException)
> {code:java}
> private class ConfigMapCallbackHandlerImpl
> implements
> FlinkKubeClient.WatchCallbackHandler<KubernetesConfigMap> {
> @Override
> public void onAdded(List<KubernetesConfigMap> configMaps) {
> // The ConfigMap is created by KubernetesLeaderElectionDriver
> with empty data. We do not
> // process this
> // useless event.
> }
> {code}
> Maybe we should handle it like onModified.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
