[
https://issues.apache.org/jira/browse/FLINK-5660?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15839907#comment-15839907
]
ASF GitHub Bot commented on FLINK-5660:
---------------------------------------
GitHub user tillrohrmann opened a pull request:
https://github.com/apache/flink/pull/3220
[FLINK-5660] [state] Fix state cleanup of PendingCheckpoint
When calling PendingCheckpoint.dispose, the state contained in a pending
checkpoint
is discarded by an asynchronous task. Since this task accesses the
taskStates field
we must not clear it in PendingCheckpoint.dispose. Instead we will clear it
once
all state objects have been discarded from within the asynchronous task.
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/tillrohrmann/flink pendingCheckpointFixCleanup
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/flink/pull/3220.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #3220
----
----
> Not properly cleaning PendingCheckpoints up
> -------------------------------------------
>
> Key: FLINK-5660
> URL: https://issues.apache.org/jira/browse/FLINK-5660
> Project: Flink
> Issue Type: Bug
> Components: State Backends, Checkpointing
> Affects Versions: 1.2.0, 1.3.0
> Reporter: Till Rohrmann
> Assignee: Till Rohrmann
> Fix For: 1.3.0, 1.2.1
>
>
> When cleaning up a {{PendingCheckpoint}}, then it sometimes happens that not
> all state handles are properly discarded. The reason is that the discard
> operation is executed asynchronously but the synchronous {{discard}} call
> already cleans the task state collection.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
