Deflake scheduler PostBind integration test

[Huang-Wei]

What type of PR is this?

/kind flake
/sig scheduling

What this PR does / why we need it:

For a successful scheduling, PostBind() is for sure called after Bind(). However, when the test detects the pod is assigned ".spec.nodeName" right after Bind() (before PostBind()), and process to check the postFilter counter, the test may flake.

This PR tries to fix this flake by waiting for PostBind() to complete. Additionally, it also wraps the scheduler initialization logic into the sub-test, to make each sub-test independent.

Which issue(s) this PR fixes:

Fixes #93659

Special notes for your reviewer:

The flake doesn't seem to block 1.19 release, so I think it can be merged to 1.20.

Does this PR introduce a user-facing change?:

NONE

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: Huang-Wei

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

The pull request process is described here

Details

Needs approval from an approver in each of these files:

• test/integration/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

[Huang-Wei]

diff -w would get a better diff for review.

/assign @ahg-g
/priority important-longterm

[Huang-Wei]

/retest

[soulxu]

Emm...this check should be useless anymore? since if postBindPlugin is called, the pluginInvokeEventChan will return, if the postBindPlugin isn't called, the timeout will return.

[Huang-Wei]

if the postBindPlugin isn't called, the timeout will return.

t.Errorf won't return - the logic proceeds to L1344.

[soulxu]

if the postBindPlugin isn't called, the timeout will return.

t.Errorf won't return - the logic proceeds to L1344.

Yea, I mean only the t.Errorf at line 1342 is enough, line 1342 already ensure the plugin isn't called. so the line 1344 to 1346 can be removed.

[Huang-Wei]

I see your point. Although the root cause of L1342 and L1344 ~ L1346 is the same, but their symptoms are different, so IMO keeping L1344 ~ L1346 as is can make the semantics more explicit.

[soulxu]

got it, thanks

[alculquicondor]

/lgtm

[ahg-g]

/milestone v1.19

I think we should still include it in 1.19.