[
https://issues.apache.org/jira/browse/HIVE-14111?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15357921#comment-15357921
]
Siddharth Seth commented on HIVE-14111:
---------------------------------------
[~sershe] - I didn't meant to add complexity to the patch. The one I attached
contained comments from me walking through the code - some related to this
jira, other comments in general. We could check in the comments (which are
added in your second patch). It would be better to create jiras to track them
though - especially ones which you think are valid (I think some of the
comments in the file I uploaded were questions, and some may not have been
valid).
bq. What is session.destroy? Rather what object is this; do you mean
TezClient::stop?
In TezTask.submit ->
TezSessionPoolManager.getInstance().destroySession(sessionState);
I think that's the main bit which needs to be looked at from your original
patch, to get past the core problem here of a session getting lost in case of
an error.
> better concurrency handling for TezSessionState - part I
> --------------------------------------------------------
>
> Key: HIVE-14111
> URL: https://issues.apache.org/jira/browse/HIVE-14111
> Project: Hive
> Issue Type: Bug
> Reporter: Sergey Shelukhin
> Assignee: Sergey Shelukhin
> Attachments: HIVE-14111.01.patch, HIVE-14111.02.patch,
> HIVE-14111.patch, sessionPoolNotes.txt
>
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
