[
https://issues.apache.org/jira/browse/FLINK-30792?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17686621#comment-17686621
]
Roman Khachatryan commented on FLINK-30792:
-------------------------------------------
Do I understand correctly that the issue is the same as in FLINK-28440?
Couldn't we rely on the same solution then (ref counting state changes per
state handle)?
Per my understanding, if materialization fails or is too slow; then the next
checkpoint will have to upload those state changes (before activeSqn), causing
performance regression, won't it?
> clean up not uploaded state changes after materialization complete
> ------------------------------------------------------------------
>
> Key: FLINK-30792
> URL: https://issues.apache.org/jira/browse/FLINK-30792
> Project: Flink
> Issue Type: Bug
> Components: Runtime / State Backends
> Affects Versions: 1.16.0
> Reporter: Feifan Wang
> Priority: Major
> Labels: pull-request-available
> Attachments: image-2023-02-03-11-30-40-198.png
>
>
> We should clean up not uploaded state changes after materialization
> completed, otherwise it may cause FileNotFoundException.
> Since state changes before completed materialization in
> FsStateChangelogWriter#notUploaded will not be used in any subsequent
> checkpoint, I suggest clean up it while handle materialization result.
> How do you think about this ? [~ym] , [~roman]
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
