[
https://issues.apache.org/jira/browse/TAJO-818?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13994805#comment-13994805
]
Jaehwa Jung edited comment on TAJO-818 at 5/13/14 2:33 AM:
-----------------------------------------------------------
Hi [~dilve].
Thank you for your report.
I also found identical problem at
[TAJO-819|https://issues.apache.org/jira/browse/TAJO-819]. I have took more
tests and it happened regardless of catalog type.
Fortunately, I found that TajoWorkerResourceManager can't return allocated disk
slots occasionally, and I'm going to upload the patch for this bug in a few
days.
Cheers.
Jaehwa
was (Author: blrunner):
Hi [~dilve].
Thank you for your report.
I also found identical problem at
[TAJO-819|https://issues.apache.org/jira/browse/TAJO-818]. I have took more
tests and it happened regardless of catalog type.
Fortunately, I found that TajoWorkerResourceManager can't return allocated disk
slots occasionally, and I'm going to upload the patch for this bug in a few
days.
Cheers.
Jaehwa
> TajoWorkerResourceManager can't return allocated disk slots occasionally.
> -------------------------------------------------------------------------
>
> Key: TAJO-818
> URL: https://issues.apache.org/jira/browse/TAJO-818
> Project: Tajo
> Issue Type: Bug
> Components: resource manager
> Affects Versions: 0.8.0, 0.9.0
> Environment: 32 machine cluster, HCatalog
> Reporter: Ilhyun Suh
> Assignee: Jaehwa Jung
>
> I made a table via Hive and tried some queries with Tajo.
> But I can see resource allocation status in web interface is still very high
> even though the query finished successfully. Disk resource usage does not
> come back to 0 after query finished.
> And because of this the next queries could not be executed without restarting
> Tajo.
> Here's a brief summary of how i've met this issue
> {noformat}
> Tajo version : 0.8.0 and latest version on master branch, both HCatalog
> enabled while building
> Table : Table is partially materialized cube
> No Column name Type
> 1 time_hour TEXT
> 2 client TEXT
> 3 cache TEXT
> 4 httpcode TEXT
> 5 method TEXT
> 6 requestaddr TEXT
> 7 fqdn TEXT
> 8 server TEXT
> 9 direct TEXT
> 10 user TEXT
> 11 contenttype TEXT
> 12 referer TEXT
> 13 smartfilter TEXT
> 14 device TEXT
> 15 os TEXT
> 16 browser TEXT
> 17 sumts INT8
> 18 sumtt INT8
> 19 counts INT8
> 20 gid TEXT
> Table was made with Hive using GROUP BY and GROUPING SETS,and it is about 32GB
> Query : Executed a simple selection query
> (e.g. SELECT sumTS,sumTT,countS, client FROM htable WHERE
> client='58.186.210.28' AND gid ='2')
> {noformat}
> Unfortunately, I could not get any information from the log files, no
> exception has occured.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
