[ 
https://issues.apache.org/jira/browse/YARN-5829?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15923135#comment-15923135
 ] 

ASF GitHub Bot commented on YARN-5829:
--------------------------------------

Github user kambatla commented on a diff in the pull request:

    https://github.com/apache/hadoop/pull/201#discussion_r105785747
  
    --- Diff: 
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/scheduler/fair/FSAppAttempt.java
 ---
    @@ -831,7 +831,11 @@ private Resource assignContainer(
         }
     
         // The desired container won't fit here, so reserve
    -    if (isReservable(capability) &&
    +    // Important: Do not enable this kind of reservation in case of 
preemption,
    +    // since the preempted application will re-reserve the preempted 
resources
    +    // and prevent the starving application to get them
    +    if (!scheduler.getConf().getPreemptionEnabled() &&
    --- End diff --
    
    This seems very bad. What happens if preemption is enabled, but no 
containers are preempted. However, if there are apps requesting large 
containers that don't fit in the free resources created by smaller containers 
finishing, these apps end up being starved forever. 


> FS preemption should reserve a node before considering containers on it for 
> preemption
> --------------------------------------------------------------------------------------
>
>                 Key: YARN-5829
>                 URL: https://issues.apache.org/jira/browse/YARN-5829
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>          Components: fairscheduler
>            Reporter: Karthik Kambatla
>            Assignee: Miklos Szegedi
>
> FS preemption evaluates nodes for preemption, and subsequently preempts 
> identified containers. If this node is not reserved for a specific 
> application, any other application could be allocated resources on this node. 
> Reserving the node for the starved application before preempting containers 
> would help avoid this.



--
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