multi goroutine deal taskUnschedulable

[lishangyuzi]

In the scenario of scheduling large-scale jobs, I also encountered a problem. When the job fails to be scheduled, all the pods under this job will update the PodCondition. Since it is necessary to communicate with the apiserver, this will take a long time.Could we consider using the multi-goroutine approach to handle this part of the logic?

[volcano-sh-bot]

Welcome @lishangyuzi!

It looks like this is your first PR to volcano-sh/volcano.

Thank you, and welcome to Volcano. 😃

[volcano-sh-bot]

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by:
To complete the pull request process, please assign lowang-bh
You can assign the PR to them by writing /assign @lowang-bh in a comment when ready.

The full list of commands accepted by this bot can be found here.

Needs approval from an approver in each of these files:

• pkg/scheduler/OWNERS

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[lishangyuzi]

/assign @lowang-bh

[lowang-bh]

Have you increase the QPS of kubeclient in volcano scheduler?

[lishangyuzi]

Have you increase the QPS of kubeclient in volcano scheduler?

default qps of kubeclient has already met my expectations.It takes approximately 200 seconds for a job with 5000 pods to complete this stage.

volcano/cmd/scheduler/app/options/options.go

Lines 127 to 128 in b169623

	fs.Float32Var(&s.KubeClientOptions.QPS, "kube-api-qps", defaultQPS, "QPS to use while talking with kubernetes apiserver")
	fs.IntVar(&s.KubeClientOptions.Burst, "kube-api-burst", defaultBurst, "Burst to use while talking with kubernetes apiserver")

volcano/cmd/scheduler/app/options/options.go

Lines 40 to 41 in b169623

	defaultQPS = 2000.0
	defaultBurst = 2000

The parameters related to my API server QPS are as follows:

--max-mutating-requests-inflight=4000
--max-requests-inflight=2000
--watch-cache-sizes=node#2000,pod#10000