Yikun commented on PR #37802:
URL: https://github.com/apache/spark/pull/37802#issuecomment-1247701070

   > Is there a way to specify a queue dynamically when using volcano to submit 
spark applicaiton?
   
   @zheniantoushipashi Currnently, as you know we can only specify volcano 
queue by [using 
PodGroupTemplate](https://github.com/apache/spark/blob/master/resource-managers/kubernetes/integration-tests/src/test/scala/org/apache/spark/deploy/k8s/integrationtest/VolcanoTestsSuite.scala#L255-L260).
 @dongjoon-hyun meant Spark + Yunikorn can set queue by setting label but 
volcano couldn't. 
   
   Actually, there are mainly two style to specify scheduler hint in common 
scheduler:
   - Using annotation/label
   - Using a separate CRD
   
   Because in K8S scheduling, the queue is only one of the capabilities, 
cooporate with other PodGroup information. If it is also placed in the label or 
annotation, will be very complicated, it is not the same way as [K8S annotation 
suggests](https://kubernetes.io/docs/concepts/overview/working-with-objects/annotations/#attaching-metadata-to-objects).
   
   The queue and other parts of the PodGroup are usually considered at the same 
time, so the queue is used as a field of the PodGroup in Volcano. Users can 
configure all batch scheduling through the pod group CRD.
   
   Therefore, you might want to `use PodGroupTemplate to cover the whole scene, 
and then increase flexibility through configuration items` in your own 
downstream version at current.
   
   Also cc @william-wang
   
   
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to