hive-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Eugene Koifman (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HIVE-12403) Too many delta files during Compaction - OOM Part Deux
Date Thu, 07 Jul 2016 00:28:11 GMT

    [ https://issues.apache.org/jira/browse/HIVE-12403?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15365387#comment-15365387
] 

Eugene Koifman commented on HIVE-12403:
---------------------------------------

Another thing worth considering here is run the Minor compaction jobs in this case in parallel.
 This would help if the issue is memory but not cluster capacity.

> Too many delta files during Compaction - OOM Part Deux
> ------------------------------------------------------
>
>                 Key: HIVE-12403
>                 URL: https://issues.apache.org/jira/browse/HIVE-12403
>             Project: Hive
>          Issue Type: Bug
>          Components: Transactions
>    Affects Versions: 1.3.0
>            Reporter: Eugene Koifman
>
> HIVE-11540 solution is somewhat incomplete.
> Suppose hive.compactor.max.num.delta=10 and there are 1000 delta files to compact.
> CompactorMR will currently run 100 jobs on 10 files each to create 100 wider delta files
but then it will run 1 job which will include 100 of the wider deltas.
> it's a simple fix to make sure the max.num.delta is always respected



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message