hadoop-yarn-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Thomas Graves (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (YARN-443) allow OS scheduling priority of NM to be different than the containers it launches
Date Wed, 06 Mar 2013 21:00:13 GMT

    [ https://issues.apache.org/jira/browse/YARN-443?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13595111#comment-13595111
] 

Thomas Graves commented on YARN-443:
------------------------------------

The container-executor generally takes in params on command line. One option though initialize_app
(which is basically localize) already takes a variable number of arguments though. So that
would require me to make the priority a mandatory parameter and then I would have to come
up with a special value for if its not set.  

Yeah I'm not a fan of env variables either so would rather not do that. 

It looks like nice should work on cygwin also so I think I can just use that and append it
to the command before calling, any objections?   For example DefaultContainerExecutor command
could become " nice -n <priority> bash WRAPPER_LAUNCH_SCRIPT..... 

The sched.priority was meant to stand for os scheduling priority. I can change it to whatever
people agree on would make the most sense. 
                
> allow OS scheduling priority of NM to be different than the containers it launches
> ----------------------------------------------------------------------------------
>
>                 Key: YARN-443
>                 URL: https://issues.apache.org/jira/browse/YARN-443
>             Project: Hadoop YARN
>          Issue Type: Improvement
>          Components: nodemanager
>    Affects Versions: 2.0.3-alpha, 0.23.6
>            Reporter: Thomas Graves
>            Assignee: Thomas Graves
>         Attachments: YARN-443.patch, YARN-443.patch
>
>
> It would be nice if we could have the nodemanager run at a different OS scheduling priority
than the containers so that you can still communicate with the nodemanager if the containers
out of control.  
> On linux we could launch the nodemanager at a higher priority, but then all the containers
it launches would also be at that higher priority, so we need a way for the container executor
to launch them at a lower priority.
> I'm not sure how this applies to windows if at all.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Mime
View raw message