[
https://issues.apache.org/jira/browse/HIVE-20560?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ashutosh Chauhan updated HIVE-20560:
------------------------------------
Summary: Set hive.llap.memory.oversubscription.max.executors.per.query to
be 1/3rd of num of executors per node (was: Set
hive.llap.memory.oversubscription.max.executors.per.query to be 1/3rd of num of
llap daemons)
> Set hive.llap.memory.oversubscription.max.executors.per.query to be 1/3rd of
> num of executors per node
> ------------------------------------------------------------------------------------------------------
>
> Key: HIVE-20560
> URL: https://issues.apache.org/jira/browse/HIVE-20560
> Project: Hive
> Issue Type: Improvement
> Components: Hive
> Reporter: Miklos Gergely
> Assignee: Miklos Gergely
> Priority: Major
> Fix For: 4.0.0
>
> Attachments: HIVE-20560.01.patch, HIVE-20560.02.patch,
> HIVE-20560.03.patch
>
>
> Unless hive.llap.memory.oversubscription.max.executors.per.query is not
> specified by the user,
> it should be calculated like this:
> {noformat}
> hive.llap.memory.oversubscription.max.executors.per.query = min(max(1,
> int(hive.llap.daemon.num.executors / 3)),8)
> {noformat}
> In container mode it remains 3 (unless specified otherwise).
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
