[
https://issues.apache.org/jira/browse/HIVE-14739?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15485740#comment-15485740
]
Prasanth Jayachandran commented on HIVE-14739:
----------------------------------------------
Thanks [~cnauroth] for reviewing the patch!
I am not sure why ShutdownHookManager was forked initially (probably it was
private initially in hadoop not sure though). I looked at the history and the
initial commit didn't have any intent to fork it. But then later HIVE-11768
added delete on exit hook that manages the temp files that are created to be
cleared on shutdown.
I refactored the hive's ShutdownHookManger to retain the DELETE_ON_EXIT hook
and retain some methods that will just delegate to Hadoop's ShutdownHookManager.
Also replaced all Threads with Runnable.
> Replace runnables directly added to runtime shutdown hooks to avoid deadlock
> ----------------------------------------------------------------------------
>
> Key: HIVE-14739
> URL: https://issues.apache.org/jira/browse/HIVE-14739
> Project: Hive
> Issue Type: Bug
> Affects Versions: 2.2.0
> Reporter: Deepesh Khandelwal
> Assignee: Prasanth Jayachandran
> Attachments: HIVE-14739.1.patch, HIVE-14739.2.patch
>
>
> [~deepesh] reported that a deadlock can occur when running queries through
> hive cli. [~cnauroth] analyzed it and reported that hive adds shutdown hooks
> directly to java Runtime which may execute in non-deterministic order causing
> deadlocks with hadoop's shutdown hooks. In one case, hadoop shutdown locked
> FileSystem#Cache and FileSystem.close whereas hive shutdown hook locked
> FileSystem.close and FileSystem#Cache order causing a deadlock.
> Hive and Hadoop has ShutdownHookManager that runs the shutdown hooks in
> deterministic order based on priority. We should use that to avoid deadlock
> throughout the code.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)