[jira] [Commented] (SPARK-7831) Mesos dispatcher doesn't deregister as a framework from Mesos when stopped

["Timothy Chen (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/SPARK-7831?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14556382#comment-14556382 ] 

Timothy Chen commented on SPARK-7831:
-------------------------------------

Hi Luc, actually that's the default behavior that I put into the dispatcher framework, since I expect the dispatcher to be a long running framework and even when it goes away it's expected to be resumed and all the tasks should be running.
To really shut down the framework a user can ask Mesos to terminate the framework via the shutdown REST api call.
I think what we should do here is to make a configuration flag to trigger this behavior or not, and default I think should be what its currently like.
What you think?

> Mesos dispatcher doesn't deregister as a framework from Mesos when stopped
> --------------------------------------------------------------------------
>
>                 Key: SPARK-7831
>                 URL: https://issues.apache.org/jira/browse/SPARK-7831
>             Project: Spark
>          Issue Type: Bug
>          Components: Mesos
>    Affects Versions: 1.4.0
>         Environment: Spark 1.4.0-rc1, Mesos 0.2.2 (compiled from source)
>            Reporter: Luc Bourlier
>
> To run Spark on Mesos in cluster mode, a Spark Mesos dispatcher has to be running.
> It is launched using {{sbin/start-mesos-dispatcher.sh}}. The Mesos dispatcher registers as a framework in the Mesos cluster.
> After using {{sbin/stop-mesos-dispatcher.sh}} to stop the dispatcher, the application is correctly terminated locally, but the framework is still listed as {{active}} in the Mesos dashboard.
> I would expect the framework to be de-registered when the dispatcher is stopped.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org