[
https://issues.apache.org/jira/browse/HIVE-12162?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14956138#comment-14956138
]
Sergey Shelukhin edited comment on HIVE-12162 at 10/14/15 2:07 AM:
-------------------------------------------------------------------
That might be caused by the way the thing shuts down. In fact I'm starting to
think that session shutdown is hurting us more than session startup w/o session
reuse.
If you look for {noformat}
2015-10-13T18:52:31,532 INFO [main]: tez.TezSessionState
(TezSessionState.java:close(412)) - Closing Tez Session
{noformat}
line and go from there, there are tons of exceptions from various components
seemingly because they shut down in the wrong order. I think HBase metastore
stuff makes it worse.
ZK appears to shut down one of the first, then there are tons of errors from
HDFS related to that (leases), HBase errors related to HDFS, all kinds of mess
that takes forever. I was actually looking at something else, just an
observation.
was (Author: sershe):
That might be caused by the way the thing shuts down. In fact I'm starting to
think that session shutdown is hurting us more than session startup w/o session
reuse.
If you look for {noformat}
2015-10-13T18:52:31,532 INFO [main]: tez.TezSessionState
(TezSessionState.java:close(412)) - Closing Tez Session
{noformat}
line and go from there, there are tons of exceptions from various components
seemingly because they shut down in the wrong order. I think HBase metastore
stuff makes it worse.
ZK appears to shut down one of the first, then there are tons of errors from
everywhere related to that, HBase errors related to HDFS, all kinds of mess
that takes forever. I was actually looking at something else, just an
observation.
> MiniTez tests take forever to shut down
> ---------------------------------------
>
> Key: HIVE-12162
> URL: https://issues.apache.org/jira/browse/HIVE-12162
> Project: Hive
> Issue Type: Bug
> Reporter: Sergey Shelukhin
> Assignee: Vikram Dixit K
>
> Even before LLAP branch merge
> (http://ec2-174-129-184-35.compute-1.amazonaws.com/jenkins/job/PreCommit-HIVE-TRUNK-Build/5618/testReport/org.apache.hadoop.hive.cli/TestMiniTezCliDriver/)
> and with AM reuse
> (http://ec2-174-129-184-35.compute-1.amazonaws.com/jenkins/job/PreCommit-HIVE-TRUNK-Build/5628/testReport/org.apache.hadoop.hive.cli/TestMiniTezCliDriver/),
> there's this:
> {noformat}
> estCliDriver_shutdown 1 min 8 sec Passed
> testCliDriver_shutdown 1 min 7 sec Passed
> testCliDriver_shutdown 1 min 7 sec Passed
> testCliDriver_shutdown 1 min 6 sec Passed
> testCliDriver_shutdown 1 min 6 sec Passed
> testCliDriver_shutdown 1 min 5 sec Passed
> testCliDriver_shutdown 1 min 5 sec Passed
> testCliDriver_shutdown 1 min 5 sec Passed
> testCliDriver_shutdown 1 min 4 sec Passed
> testCliDriver_shutdown 1 min 4 sec Passed
> testCliDriver_shutdown 1 min 4 sec Passed
> testCliDriver_shutdown 1 min 4 sec Passed
> testCliDriver_shutdown 1 min 4 sec Passed
> testCliDriver_shutdown 1 min 4 sec Passed
> testCliDriver_shutdown 1 min 3 sec Passed
> testCliDriver_shutdown 1 min 3 sec Passed
> testCliDriver_shutdown 1 min 3 sec Passed
> testCliDriver_shutdown 1 min 3 sec Passed
> testCliDriver_shutdown 1 min 3 sec Passed
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
