[
https://issues.apache.org/jira/browse/FLINK-25872?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17488851#comment-17488851
]
Piotr Nowojski commented on FLINK-25872:
----------------------------------------
I'm bumping this to release blocker, as this bug would result with a state
corruption.
We discussed this issue with [~ym] offline and we agreed that if we run out of
time for a proper fix for this bug, we should disable support for recovery in
`CLAIM` mode, by for example throwing `UnsupportedOperationException()` during
recovery. But we need this assertion or a proper fix before 1.15.0 release.
> Restoring from incremental non-changelog checkpoint with changelog
> state-backend enabled in CLAIM mode discards state in use
> ----------------------------------------------------------------------------------------------------------------------------
>
> Key: FLINK-25872
> URL: https://issues.apache.org/jira/browse/FLINK-25872
> Project: Flink
> Issue Type: Bug
> Components: Runtime / Checkpointing, Runtime / State Backends
> Reporter: Yun Tang
> Priority: Blocker
> Fix For: 1.15.0
>
>
> If we restore from incremental checkpoint with changelog state-backend
> enabled in snapshot CLAIM mode, the restored checkpoint would be discarded on
> subsume, which lead to the private state in Incremental state handle could be
> deleted by mistake. This bug is like FLINK-25478.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
