[
https://issues.apache.org/jira/browse/YUNIKORN-3243?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18067516#comment-18067516
]
Wilfred Spiegelenburg commented on YUNIKORN-3243:
-------------------------------------------------
For the queue sorting Paul and Craig did YUNIKORN-2678 which was added to the
YuniKorn 1.6 release. The sorting does partially take into account guaranteed
share on purpose. It was discussed.
I gave an example of not setting the guaranteed on the queue which results in a
similar state as having a large ratio between the two guarantees. We should
have another look at not starving a queue with the large ratio but need to be
careful to keep sorting as consistent as possible.
We do still have YUNIKORN-2840 open for some smaller updates to be worked on to
improve performance and sibling sorting. I do not think that they cover this
case.
> Fair-share queue sorting causes starvation of sibling queues with asymmetric
> guaranteed resources
> -------------------------------------------------------------------------------------------------
>
> Key: YUNIKORN-3243
> URL: https://issues.apache.org/jira/browse/YUNIKORN-3243
> Project: Apache YuniKorn
> Issue Type: Bug
> Components: core - scheduler
> Reporter: Shubham Mishra
> Assignee: Shubham Mishra
> Priority: Major
> Fix For: 1.6.1
>
>
> When two sibling queues under the same parent have vastly different
> guaranteed resources (e.g., 3600:1 ratio), the fair-share queue sorting in
> {{TryAllocate}} causes the smaller queue to be completely starved — its
> {{app.tryAllocate()}} is never called. This has two consequences:
> # {*}Scheduling starvation{*}: The smaller queue's asks are never evaluated
> for allocation, even when nodes have capacity.
> # {*}Autoscaler blindness{*}: Because
> {{[SetSchedulingAttempted|https://github.com/apache/yunikorn-core/blob/cb7f2381b6098f8936fe57dd7f13f205939a0021/pkg/scheduler/objects/application.go#L1065](true)}}
> is only set inside {{app.tryAllocate()}} (line 1065 of
> {{{}application.go{}}}), the starved queue's asks never get this flag.
> {{inspectOutstandingRequests}} skips them, so the cluster autoscaler (e.g.,
> Karpenter) is never notified that capacity is needed.
> The second issue is the more critical one — even if scheduling is delayed,
> the autoscaler should be able to provision nodes in parallel. But with the
> current design, the autoscaler signal is gated on queue visitation.
> Here are the unit-tests to reproduce this -
> [https://github.com/apache/yunikorn-core/pull/1077]
> h3.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
