[jira] [Commented] (YARN-4861) Define ContainerExitStatus for OPPORTUNISTIC containers that get killed

["Konstantinos Karanasos (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/YARN-4861?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15209596#comment-15209596 ] 

Konstantinos Karanasos commented on YARN-4861:
----------------------------------------------

An OPPORTUNISTIC container might be killed in one of the following cases:
* by the AM while running;
* by the AM while queued;
* by the NM, while running, in order to free up resources for a GUARANTEED container to start its execution;
* by the NM, while queued, in order to reduce the length of the queue.

In all these cases, we need to define the proper Exit Status for the container.
Then, we need to make sure that the AM reacts properly to the defined Exit Statuses (e.g., by rescheduling killed OPPORTUNISTIC containers).

Currently, in YARN-2883, OPPORTUNISTIC containers that got killed by the NM while running get a KILLED_BY_APPMASTER ExitStatus.
In YARN-4738, OPPORTUNISTIC containers that got killed while queued are get an ABORTED ExitStatus.

> Define ContainerExitStatus for OPPORTUNISTIC containers that get killed
> -----------------------------------------------------------------------
>
>                 Key: YARN-4861
>                 URL: https://issues.apache.org/jira/browse/YARN-4861
>             Project: Hadoop YARN
>          Issue Type: Task
>            Reporter: Konstantinos Karanasos
>
> When we kill an OPPORTUNISTIC container, which is either running or queued, we need to define its Exit Status.



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