[ 
https://issues.apache.org/jira/browse/YARN-8509?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16559043#comment-16559043
 ] 

Chandni Singh commented on YARN-8509:
-------------------------------------

Couple of nits and questions:

1. This is a javadoc block which should be above the method. I understand that 
this test is moved out of another test class but this gives a good opportunity 
to fix this.
{code:java}
/**
 * Test case: Submit three applications (app1/app2/app3) to different
 * queues, queue structure:
 *
 * <pre>
 *               Root
 *            /  |  \  \
 *           a   b   c  d
 *          30  30  30  10
 * </pre>
 *
 */
{code}

2. Why explicitly setting the log level to debug in the code?
{code}
    Logger.getRootLogger().setLevel(Level.DEBUG);
{code}

3. Can you explain the comment? 
{code}
   // We should release pending resource be capped at user limit, think about
    // a user ask for 10000maps. but cluster can run a max of 1000. In this
    // case, as soon as each map finish, other one pending will get scheduled
    // When not deduct reserved, total-pending = 3G (u1) + 20G (u2) = 23G
    //          deduct reserved, total-pending = 0G (u1) + 20G (u2) = 20G
{code}


> Fix UserLimit calculation for preemption to balance scenario after queue 
> satisfied  
> ------------------------------------------------------------------------------------
>
>                 Key: YARN-8509
>                 URL: https://issues.apache.org/jira/browse/YARN-8509
>             Project: Hadoop YARN
>          Issue Type: Bug
>          Components: yarn
>            Reporter: Zian Chen
>            Assignee: Zian Chen
>            Priority: Major
>         Attachments: YARN-8509.001.patch, YARN-8509.002.patch
>
>
> In LeafQueue#getTotalPendingResourcesConsideringUserLimit, we calculate total 
> pending resource based on user-limit percent and user-limit factor which will 
> cap pending resource for each user to the minimum of user-limit pending and 
> actual pending. This will prevent queue from taking more pending resource to 
> achieve queue balance after all queue satisfied with its ideal allocation.
>   
>  We need to change the logic to let queue pending can go beyond userlimit.



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

---------------------------------------------------------------------
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