[ 
https://issues.apache.org/jira/browse/YARN-2325?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14068319#comment-14068319
 ] 

zhihai xu commented on YARN-2325:
---------------------------------

Hi Tsuyoshi OZAWA, thanks for your quick response to my patch.
I agree to your points above. If this transition occurs, it might be bug in the 
code. My patch is just to make sure we return early to avoid a 
NullPointerException for some unexpected code error which cause the node being 
removed.
I also find the current removeNode function did the same thing: check null 
pointer and return early.
    if (node == null) {
      return;
    }
If you think my patch is not needed for NPE prevention. I am ok to close this 
JIRA.

> need check whether node is null in nodeUpdate for FairScheduler 
> ----------------------------------------------------------------
>
>                 Key: YARN-2325
>                 URL: https://issues.apache.org/jira/browse/YARN-2325
>             Project: Hadoop YARN
>          Issue Type: Bug
>            Reporter: zhihai xu
>            Assignee: zhihai xu
>         Attachments: YARN-2325.000.patch
>
>
> need check whether node is null in nodeUpdate for FairScheduler.
> If nodeUpdate is called after removeNode, the getFSSchedulerNode will be 
> null. If the node is null, we should return with error message.



--
This message was sent by Atlassian JIRA
(v6.2#6252)

Reply via email to