[ 
https://issues.apache.org/jira/browse/YARN-6197?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15870013#comment-15870013
 ] 

Bibin A Chundatt commented on YARN-6197:
----------------------------------------

[~sunilg]/[~varun_saxena]
For activating and checking {{amIfStarted}} should be consider 
{{rmContext.getScheduler().getMinimumResourceCapability}}.
As per the current code 
{code}
 public FiCaSchedulerApp(ApplicationAttemptId applicationAttemptId,
      String user, Queue queue, AbstractUsersManager abstractUsersManager,
      RMContext rmContext, Priority appPriority, boolean isAttemptRecovering,
      ActivitiesManager activitiesManager) {
    if (rmApp == null || rmApp.getAMResourceRequest() == null) {
      // the rmApp may be undefined (the resource manager checks for this too)
      // and unmanaged applications do not provide an amResource request
      // in these cases, provide a default using the scheduler
      amResource = rmContext.getScheduler().getMinimumResourceCapability();
      partition = CommonNodeLabelsManager.NO_LABEL;
    } else {
      amResource = rmApp.getAMResourceRequest().getCapability();
      partition =
          (rmApp.getAMResourceRequest().getNodeLabelExpression() == null)
          ? CommonNodeLabelsManager.NO_LABEL
          : rmApp.getAMResourceRequest().getNodeLabelExpression();
    }
{code}
Any info on why the minimum was considered instead of {{Resource.NONE}} ?? We 
could have activated AM directly with out AM Limit check but following ordering 
policy.

> CS Leaf queue am usage gets updated for unmanaged AM
> ----------------------------------------------------
>
>                 Key: YARN-6197
>                 URL: https://issues.apache.org/jira/browse/YARN-6197
>             Project: Hadoop YARN
>          Issue Type: Bug
>            Reporter: Bibin A Chundatt
>            Assignee: Bibin A Chundatt
>
> {{LeafQueue#activateApplication()}} for unmanaged AM  the am_usage is updated 
> with scheduler minimum allocation size. Cluster resource/AM limit headroom 
> for other apps in queue will get reduced .
> Solution: FicaScheduler unManagedAM flag can be used to check AM type.
> Based on flag the queueusage need to be updated during activation and removal
> Thoughts??



--
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