[
https://issues.apache.org/jira/browse/YUNIKORN-1378?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17628440#comment-17628440
]
Craig Condit commented on YUNIKORN-1378:
----------------------------------------
How do we ensure cache coherency? The filters are called multiple times because
the state of the pods or nodes can change over time. Getting the result of a
Filter call wrong is very dangerous; this could result in a pod being scheduled
where it is invalid, or a node being considered unsuitable for a pod when it
really isn't. That could result in unexpected autoscaler scale-ups, or worse.
What level of performance will be gained? I'd like to see some benchmarking
around this to prove that it's necessary.
> Add predicate with equivalence cache to improve the scheduler performance
> -------------------------------------------------------------------------
>
> Key: YUNIKORN-1378
> URL: https://issues.apache.org/jira/browse/YUNIKORN-1378
> Project: Apache YuniKorn
> Issue Type: Task
> Components: shim - kubernetes
> Reporter: Qi Zhu
> Assignee: Qi Zhu
> Priority: Major
> Labels: pull-request-available
>
> Now we call Filter every time when we predicate, we can cache some operations
> for predicate so that we can improve the scheduler performance.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
