[
https://issues.apache.org/jira/browse/YARN-5216?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15365262#comment-15365262
]
Hitesh Sharma commented on YARN-5216:
-------------------------------------
Thank you for the insights, [~kkaranasos]!
Sorry, rebalancing wasn't the right terminology to use. I was referring to
killing of queued containers that happens during
{{shedQueuedOpportunisticContainers}} to enforce the queue limits, which in
turns follows the paths you mention above.
It might be a good idea to use start container to imply resume when the
container is paused, but at the same time it also overloads the meaning of
start container and given how different they are it can impose some challenges.
Anyways, we can discuss this more in [YARN-5292].
{quote}
As far as I can see, all you need from the NM to support preemption is (let me
know if there are more things that I am missing):
# Determine the way a container stops (option 1: kill, option 2: preempt).
# Determine the way it start (that is, resume it if it's paused, instead of
starting it from the beginning).
# Decide which container to start (you might want to start first containers
that are paused instead of new ones).
{quote}
How do you propose to do 3 without having an extension point to pick a
container to start? The moment we have an extension point to pick a container
to start we also need an extension point to pick up a container to kill for
enforcing queue limits or something else.
Appreciate the feedback and help. Thanks a lot!
> Expose configurable preemption policy for OPPORTUNISTIC containers running on
> the NM
> ------------------------------------------------------------------------------------
>
> Key: YARN-5216
> URL: https://issues.apache.org/jira/browse/YARN-5216
> Project: Hadoop YARN
> Issue Type: Bug
> Reporter: Arun Suresh
> Assignee: Hitesh Sharma
> Attachments: YARN5216.001.patch, yarn5216.002.patch
>
>
> Currently, the default action taken by the QueuingContainerManager,
> introduced in YARN-2883, when a GUARANTEED Container is scheduled on an NM
> with OPPORTUNISTIC containers using up resources, is to KILL the running
> OPPORTUNISTIC containers.
> This JIRA proposes to expose a configurable hook to allow the NM to take a
> different action.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
