[ 
https://issues.apache.org/jira/browse/YARN-5892?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15956944#comment-15956944
 ] 

Eric Payne commented on YARN-5892:
----------------------------------

Build failure seems unrelated:
{noformat}
[ERROR] Failed to execute goal 
com.github.eirslett:frontend-maven-plugin:0.0.22:install-node-and-npm (install 
node and npm) on project hadoop-yarn-ui: Could not download Node.js: Got error 
code 522 from the server. -> [Help 1]
{noformat}
I re-kicked the YARN precommit build.
{quote}
bq. Which means that if a user's weight is < 1, the user's user limit will fall 
below the minimum user limit percent.
To me, YARN can ensure minimum weight of each user to >= 1.
{quote}
My concern about this approach is that if we really do have a user that should 
have a weight that is 0.5 compared to all of the other users in the queue, we 
would have to set all of the other users' weights to 2, and then remember to 
set new users' weights whenever they are added to that queue. We may have a use 
case for this, actually.

I discussed this with [~jlowe] offline. I think that if a user is really 
supposed to be less than the other users, it's acceptable for that user's user 
limit to be less than the MULP as long as that does not ever prevent that user 
from using the whole queue if they are the only user. I will investigate 
further.

> Capacity Scheduler: Support user-specific minimum user limit percent
> --------------------------------------------------------------------
>
>                 Key: YARN-5892
>                 URL: https://issues.apache.org/jira/browse/YARN-5892
>             Project: Hadoop YARN
>          Issue Type: Improvement
>          Components: capacityscheduler
>            Reporter: Eric Payne
>            Assignee: Eric Payne
>         Attachments: Active users highlighted.jpg, YARN-5892.001.patch, 
> YARN-5892.002.patch, YARN-5892.003.patch, YARN-5892.004.patch, 
> YARN-5892.005.patch, YARN-5892.006.patch, YARN-5892.007.patch
>
>
> Currently, in the capacity scheduler, the {{minimum-user-limit-percent}} 
> property is per queue. A cluster admin should be able to set the minimum user 
> limit percent on a per-user basis within the queue.
> This functionality is needed so that when intra-queue preemption is enabled 
> (YARN-4945 / YARN-2113), some users can be deemed as more important than 
> other users, and resources from VIP users won't be as likely to be preempted.
> For example, if the {{getstuffdone}} queue has a MULP of 25 percent, but user 
> {{jane}} is a power user of queue {{getstuffdone}} and needs to be guaranteed 
> 75 percent, the properties for {{getstuffdone}} and {{jane}} would look like 
> this:
> {code}
>   <property>
>     
> <name>yarn.scheduler.capacity.root.getstuffdone.minimum-user-limit-percent</name>
>     <value>25</value>
>   </property>
>   <property>
>     
> <name>yarn.scheduler.capacity.root.getstuffdone.jane.minimum-user-limit-percent</name>
>     <value>75</value>
>   </property>
> {code}



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: yarn-issues-h...@hadoop.apache.org

Reply via email to