You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@hive.apache.org by "Sergey Shelukhin (JIRA)" <ji...@apache.org> on 2018/10/01 17:11:00 UTC

[jira] [Commented] (HIVE-20648) LLAP: Vector group by operator should use memory per executor

    [ https://issues.apache.org/jira/browse/HIVE-20648?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16634338#comment-16634338 ] 

Sergey Shelukhin commented on HIVE-20648:
-----------------------------------------

LlapProxy.isDaemon() should be used for LLAP detection

> LLAP: Vector group by operator should use memory per executor
> -------------------------------------------------------------
>
>                 Key: HIVE-20648
>                 URL: https://issues.apache.org/jira/browse/HIVE-20648
>             Project: Hive
>          Issue Type: Bug
>    Affects Versions: 4.0.0, 3.2.0
>            Reporter: Prasanth Jayachandran
>            Assignee: Prasanth Jayachandran
>            Priority: Major
>         Attachments: HIVE-20648.1.patch
>
>
> HIVE-15503 treatment has to be applied for vector group by operator as well. Vector group by currently uses MemoryMX bean to get heap usage and heap max memory which will not work for LLAP. Instead it should use memory per executor as upper bound to make flush decision.  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)