You are viewing a plain text version of this content. The canonical link for it is here.
Posted to yarn-issues@hadoop.apache.org by "balloons (JIRA)" <ji...@apache.org> on 2017/09/25 12:15:00 UTC
[jira] [Comment Edited] (YARN-7247) Assign multiple will lead to
hot point problems of physical resource consumption
[ https://issues.apache.org/jira/browse/YARN-7247?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16178926#comment-16178926 ]
balloons edited comment on YARN-7247 at 9/25/17 12:14 PM:
----------------------------------------------------------
[~templedf] Thank templedf. But it was a very difficult choice for us:
* Turning on the dynamic assignment will cause the same resource consumption type of container placed on one node.
* Turning off the dynamic assignment will lead to very bad scheduling performance.
So, I want the fair scheduler to support turn on multi assignments, but each time the resource is allocated to a different application.
What do you think?
was (Author: balloons):
[~templedf] Thank templedf. But it was a very difficult choice for us:
* Turning on the dynamic assignment will cause the same resource consumption type of container placed on one node.
* Turning off the dynamic assignment will lead to very bad scheduling performance.
So, I want the fair scheduler to support turn on multi assignments, but each time the resource is allocated to a different node.
What do you think?
> Assign multiple will lead to hot point problems of physical resource consumption
> --------------------------------------------------------------------------------
>
> Key: YARN-7247
> URL: https://issues.apache.org/jira/browse/YARN-7247
> Project: Hadoop YARN
> Issue Type: Bug
> Components: fairscheduler
> Reporter: balloons
> Assignee: Daniel Templeton
>
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: yarn-issues-help@hadoop.apache.org