[
https://issues.apache.org/jira/browse/FLINK-20653?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17253973#comment-17253973
]
weiyunqing commented on FLINK-20653:
------------------------------------
[~yunta] "storing metadata first and then clean up the unreferenced incremental
state" There is no mistake in this statement, but the current implementation
method is to clear the unused file blocks immediately after registering the
incremental status. The code logic is in
[IncrementalRemoteKeyedStateHandle#registerSharedStates]
> SharedState should be registered after PendingCheckpoint storeCheckpoint
> ------------------------------------------------------------------------
>
> Key: FLINK-20653
> URL: https://issues.apache.org/jira/browse/FLINK-20653
> Project: Flink
> Issue Type: Bug
> Components: Runtime / Checkpointing
> Affects Versions: 1.12.0
> Reporter: weiyunqing
> Priority: Major
> Attachments: 123.png, image-2020-12-17-20-20-21-246.png
>
>
> SharedState should be registered after PendingCheckpoint storeCheckpoint，If
> the metadata storage fails, the deleted incremental state cannot be recovered。
> !image-2020-12-17-20-20-21-246.png!
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
