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

Posted to mapreduce-issues@hadoop.apache.org by "Vinod Kumar Vavilapalli (Commented) (JIRA)" <ji...@apache.org> on 2011/10/31 07:00:35 UTC

[jira] [Commented] (MAPREDUCE-3314) Support AM specific blacklisting of nodes on the RM

    [ https://issues.apache.org/jira/browse/MAPREDUCE-3314?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13139942#comment-13139942 ] 

Vinod Kumar Vavilapalli commented on MAPREDUCE-3314:
----------------------------------------------------

Thanks for filing this. I ran into the same issue but was lazy enough not to create it. Essentially what happens for me is that on a small enough cluster (single node actually), if all nodes are blacklisted by the AM, it will just keep getting containers and wasting them.
                
> Support AM specific blacklisting of nodes on the RM
> ---------------------------------------------------
>
>                 Key: MAPREDUCE-3314
>                 URL: https://issues.apache.org/jira/browse/MAPREDUCE-3314
>             Project: Hadoop Map/Reduce
>          Issue Type: Bug
>          Components: mrv2
>    Affects Versions: 0.23.0
>            Reporter: Hitesh Shah
>             Fix For: 0.23.1
>
>
> The AM blacklists certain nodes due to task failures but the RM does not know of this and can allocate blacklisted nodes to the AM again and again. 
> RM should support a protocol that allows AM to inform the RM of nodes that it should not allocate to it. 

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira