[
https://issues.apache.org/jira/browse/FLINK-26957?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17521663#comment-17521663
]
Matthias Pohl commented on FLINK-26957:
---------------------------------------
[~tanyuxin] did you do some testing with the fix?
> FileSystemJobResultStore calls flush on an already closed OutputStream
> -----------------------------------------------------------------------
>
> Key: FLINK-26957
> URL: https://issues.apache.org/jira/browse/FLINK-26957
> Project: Flink
> Issue Type: Bug
> Components: Runtime / Coordination
> Affects Versions: 1.15.0, 1.16.0
> Reporter: Matthias Pohl
> Assignee: Matthias Pohl
> Priority: Blocker
> Labels: pull-request-available
> Fix For: 1.15.0
>
>
> We experienced problems with some FileSystems when creating the dirty JRS
> entries (see initial discussion in FLINK-26555). The {{writeValue}} method
> closes the {{OutputStream}} by default which causes the subsequent {{flush}}
> call to fail.
> It didn't appear in the unit tests because {{LocalDataOutputStream.flush}} is
> a no-op operation. We still have to investigate why it didn't appear when
> doing the tests with the presto and hadoop S3 filesystems.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
