[
https://issues.apache.org/jira/browse/FLINK-1775?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14379329#comment-14379329
]
ASF GitHub Bot commented on FLINK-1775:
---------------------------------------
GitHub user gyfora opened a pull request:
https://github.com/apache/flink/pull/534
[FLINK-1775] BarrierBuffer fix to avoid end of stream deadlock
This fixes several corner-cases in the barrier buffer logic.
Still two problems remain:
- If the StreamRecordWriter is active (a thread is running to flush the
ouputs every so many milliseconds)
- On travis the RecordWriters seem to deadlock at the sources for some
reason on broadcastEvent calls. (I cannot reproduce this locally)
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/mbalassi/flink FLINK-1775
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/flink/pull/534.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 #534
----
commit 35d71c8b8461e689ec02e64ef1089adfe1f4d7cb
Author: Gyula Fora <[email protected]>
Date: 2015-03-24T23:14:21Z
[FLINK-1775] BarrierBuffer fix to avoid end of stream deadlock
----
> BarrierBuffers don't correctly handle end of stream events
> ----------------------------------------------------------
>
> Key: FLINK-1775
> URL: https://issues.apache.org/jira/browse/FLINK-1775
> Project: Flink
> Issue Type: Bug
> Components: Streaming
> Reporter: Gyula Fora
> Assignee: Gyula Fora
>
> The current implementation causes deadlocks when the end of stream event
> comes from a currently blocked channel.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
