[
https://issues.apache.org/jira/browse/HIVE-17401?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16153433#comment-16153433
]
Peter Vary commented on HIVE-17401:
-----------------------------------
Hi [~xuefuz],
I have taken a look at the patch. Looks logical, and much easier to read than
the previous version. No to mention that clears up a bug :).
Quick question? Do we need {{lastIdleTime}} as a long after this change? It
might be enough to have it as a boolean, since we only use it as a flag that
the operation lock is acquired, since otherwise lastIdleTime equals
lastAccessTime.
Thanks for the investigation and the patch!
Peter
> Hive session idle timeout doesn't function properly
> ---------------------------------------------------
>
> Key: HIVE-17401
> URL: https://issues.apache.org/jira/browse/HIVE-17401
> Project: Hive
> Issue Type: Bug
> Affects Versions: 1.1.0
> Reporter: Xuefu Zhang
> Assignee: Xuefu Zhang
> Attachments: HIVE-17401.1.patch, HIVE-17401.patch
>
>
> It's apparent in our production environment that HS2 leaks sessions, which at
> least contributed to memory leaks in HS2. We further found that idle HS2
> sessions rarely get timed out and the number of live session keeps increasing
> as time goes on. Eventually, HS2 becomes irresponsive and demands a restart.
> Investigation shows that session idle timeout doesn't work appropriately.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
