Extender bind should respect IsInterested

[yqwang-ms]

What type of PR is this?
/kind bug

What this PR does / why we need it:
If one extender want to manage its own extended resource exclusively, such as it want to always serve the binding request for its own extended resource xxx.com/gpu exclusively (because it also needs to set additional Pod annotations to help track resource placement when binding atomically).

However, currently, during initailization, the kube scheduler will just pick any extender which has bind verb, it may picks wrong extender, such as the one interests on xxx.com/fpga.
And then when executing a Pod binding, it find that the bind extender does not interest on xxx.com/gpu, so fall back to default binder, then even though the pod is bound, but the annotations are missed to be bound together.

This PR delays the bind extender selection when executing a Pod binding, so it will find the interested extender for each pod, instead of pre-find a global one to serve all cluster binding.

Which issue(s) this PR fixes:

Fixes # NO See issue above.

Special notes for your reviewer:

Does this PR introduce a user-facing change?:

Extender bind should respect IsInterested

[k8s-ci-robot]

Hi @yqwang-ms. Thanks for your PR.

I'm waiting for a kubernetes member to verify that this patch is reasonable to test. If it is, they should reply with /ok-to-test on its own line. Until that is done, I will not automatically test new commits in this PR, but the usual testing commands by org members will still work. Regular contributors should join the org to skip this step.

Once the patch is verified, the new status will be reflected by the ok-to-test label.

I understand the commands that are listed here.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

[yqwang-ms]

/assign @wojtek-t

[yqwang-ms]

/assign @k82cn
/assign @Huang-Wei

Could you please take a look at this small fix when you free? :)
Or could you please tell me what should I do next?
Thanks again!

[yqwang-ms]

@Huang-Wei Could you please also take a look at this? Thanks!

[Huang-Wei]

/ok-to-test

[yqwang-ms]

Hi @Huang-Wei All tests are passed, ready to approve? :)

[Huang-Wei]

/lgtm
/approve

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: Huang-Wei, yqwang-ms

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

The pull request process is described here

Needs approval from an approver in each of these files:

• pkg/scheduler/OWNERS [Huang-Wei]

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

[tedyu]

I proposed similar change on May 23rd:

#78273

[tedyu]

My PR maintains extenderBinders so that non-bind extender isn't involved in the closure.

@Huang-Wei
Do you think I should submit a PR for this optimization ?

Cheers

[yqwang-ms]

@tedyu Sorry for not found your PR previously.
From my view, at least the relaxed validation is still worth to be in master.

[tedyu]

@yqwang-ms
What do you think of the extenderBinders which narrows the extenders evaluated inside the closure ?

[yqwang-ms]

@tedyu I think it is a minor optimization given in most clusters we only have very few extenders. Anyway you can benchmark it.
Overall, I think the extenderBinders optimization is better to have, and the relaxed validation is must have. :)