You are viewing a plain text version of this content. The canonical link for it is here.

Posted to yarn-issues@hadoop.apache.org by "Beckham007 (JIRA)" <ji...@apache.org> on 2014/06/17 13:51:01 UTC

[jira] [Created] (YARN-2169) NMSimulator of sls should catch more Exception

Beckham007 created YARN-2169:
--------------------------------

             Summary: NMSimulator of sls should catch more Exception
                 Key: YARN-2169
                 URL: https://issues.apache.org/jira/browse/YARN-2169
             Project: Hadoop YARN
          Issue Type: Bug
    Affects Versions: 2.4.0
            Reporter: Beckham007


In the method middleStep() of NMSimulator , sending heart beat may cause InterruptedException or other Exception if the load is heavily. If not handler these exceptions, the task of NMSimulator cloud not add to the executor queue again. So the NM will lost.
In my situation, the pool size is 4000, nm size is 2000, and am is 1500. Some NMs will lost.



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