[
https://issues.apache.org/jira/browse/YARN-7682?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16306416#comment-16306416
]
Konstantinos Karanasos commented on YARN-7682:
----------------------------------------------
What about the following: given that the external API is there in the
PlacementConstraints, what if we let users specify random cmin and then fail
(ie, canAssign=false) if there are not already that many containers in the
target. That would not be gang scheduling and it allows us to not differentiate
between source!=target or source==target cases, which means we keep it simpler
for the first version.
Then instead of restricting the API, we can restrict such use cases when we add
the constraint validation later on (for now we just assume non gang semantics).
I agree that extending canAssign for multiple containers should not be done for
the first version.
> Expose canAssign method in the PlacementConstraintManager
> ---------------------------------------------------------
>
> Key: YARN-7682
> URL: https://issues.apache.org/jira/browse/YARN-7682
> Project: Hadoop YARN
> Issue Type: Sub-task
> Reporter: Arun Suresh
> Assignee: Panagiotis Garefalakis
> Attachments: YARN-7682-YARN-6592.001.patch, YARN-7682.wip.patch
>
>
> As per discussion in YARN-7613. Lets expose {{canAssign}} method in the
> PlacementConstraintManager that takes a sourceTags, applicationId,
> SchedulerNode and AllocationTagsManager and returns true if constraints are
> not violated by placing the container on the node.
> I prefer not passing in the SchedulingRequest, since it can have > 1
> numAllocations. We want this api to be called for single allocations.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
