You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hive.apache.org by "Bharathkrishna Guruvayoor Murali (JIRA)" <ji...@apache.org> on 2018/04/09 20:50:00 UTC
[jira] [Commented] (HIVE-19093) some parts of the Driver runs from
the "Background-Pool" in HS2
[ https://issues.apache.org/jira/browse/HIVE-19093?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16431204#comment-16431204 ]
Bharathkrishna Guruvayoor Murali commented on HIVE-19093:
---------------------------------------------------------
I also noticed the same while working on : HIVE-19133
We are explicitly passing PerfLogger as well to the background thread (as I have commented on HIVE-19133) which is leading to wrong results and race conditions.
> some parts of the Driver runs from the "Background-Pool" in HS2
> ---------------------------------------------------------------
>
> Key: HIVE-19093
> URL: https://issues.apache.org/jira/browse/HIVE-19093
> Project: Hive
> Issue Type: Bug
> Reporter: Zoltan Haindrich
> Priority: Major
>
> I was looking into perflog results; and the fact that Driver.run open / close happens on a different thread caught my eye - this might cause real problems since {{Session.get()}} will return an entirely different session in the aftermath...most notably there are some lock related calls like: releaseLocks
> {code}
> 2018-04-03T08:36:53,488 DEBUG [2c81c6c1-aa6f-4609-8250-5b1a5360a8ba HiveServer2-Handler-Pool: Thread-16242]: log.PerfLogger (PerfLogger.java:PerfLogBegin(132)) - <PERFLOG method=Driver.run from=org.apache.hadoop.hive.ql.Driver>
> 2018-04-03T08:37:21,791 DEBUG [HiveServer2-Background-Pool: Thread-16247]: log.PerfLogger (PerfLogger.java:PerfLogEnd(172)) - </PERFLOG method=Driver.run start=1522744613488 end=1522744641791 duration=28303 from=org.apache.hadoop.hive.ql.Driver>
> {code}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)