[ 
https://issues.apache.org/jira/browse/YARN-7682?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16305698#comment-16305698
 ] 

Arun Suresh edited comment on YARN-7682 at 12/28/17 7:51 PM:
-------------------------------------------------------------

bq. affinity with more than 2 in cmin is actually a cardinality constraint. I 
don’t think we still call it affinity (but I might not remember well). If 
that’s the case we can have both min and max cardinalitie
So - this is why I think exposing cMin externally is a bit dicey - Suppose cMin 
= 5, and we say target expression IN some node - but numContainers/Allocations 
in the SchedulingRequest = 2. Does that mean - we have to wait till the AM 
sends 3 more SchedulingRequests ? - which seems to fall more into the realm of 
gang scheduling. or should we fail the scheduling request ?

I feel, for the time-being, We let the user specify either: anti-affinity 
(where we set cMin = 0), affinity (cMin = 1, cMax = infinity) and 
max-cardinality (cMin = 1, and cMax = some x)




was (Author: asuresh):
bq. affinity with more than 2 in cmin is actually a cardinality constraint. I 
don’t think we still call it affinity (but I might not remember well). If 
that’s the case we can have both min and max cardinalitie
So - this is why I think exposing cMin externally is a bit dicey - Suppose cMin 
= 5, and we say target expression IN some node - but numContainers/Allocations 
in the SchedulingRequest = 2. Does that mean - we have to wait the AM sends 2 
more SchedulingRequests ? - which seems to fall more into the realm of gang 
scheduling. or should we fail the scheduling request ?

I feel, for the time-being, We let the user specify either: anti-affinity 
(where we set cMin = 0), affinity (cMin = 1, cMax = infinity) and 
max-cardinality (cMin = 1, and cMax = some x)



> 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: yarn-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: yarn-issues-h...@hadoop.apache.org

Reply via email to