[
https://issues.apache.org/jira/browse/GIRAPH-871?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13939472#comment-13939472
]
Pavan Kumar commented on GIRAPH-871:
------------------------------------
+1 makes total sense.
Oversight on myside.
Since I tested on Corona (facebook hadoop) which does not have this issue. (as
pointed by Avery in the mailing list) we never got into the weird situation.
> Map task jvm never exits since netty 4 upgrade
> ----------------------------------------------
>
> Key: GIRAPH-871
> URL: https://issues.apache.org/jira/browse/GIRAPH-871
> Project: Giraph
> Issue Type: Bug
> Components: bsp
> Affects Versions: 1.1.0
> Reporter: Craig Muchinsky
> Fix For: 1.1.0
>
> Attachments: GIRAPH-871.patch
>
>
> Some changes made to the NettyClient.stop() logic related to netty 4 are
> causing the final worker thread to hang up indefinitely, preventing the map
> task jvm from exiting. In particular, the call to
> ProgressableUtils.awaitTerminationFuture() is blocking inside the
> ChannelFutureListener callback and deadlocking. Simply moving the
> ProgressableUtils.awaitTerminationFuture() calls outside of the
> ChannelFutureListener callback and placing them at the end of the
> NettyClient.stop() method seems to cure the problem.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
