Cluster autoscaler to support custom schedulers' Filter logic

[jdheyburn]

Which component are you using?:
Cluster autoscaler (CA)

Is your feature request designed to solve a problem? If so describe the problem this feature should solve.:

When a Pod is being scheduled with a custom scheduler (i.e. non default-scheduler) that uses the Filter API, it can be marked as Unschedulable. When CA polls for unschedulable pods, it uses a different Filter API (I believe the non default-scheduler) to determine whether it can be scheduled on existing nodes (from this block). This difference in logic can cause a pod to be stuck in Pending given that differing logic is used to determine its placement.

Describe the solution you'd like.:

To allow CA to be extended to use the same Filter API as the scheduler used for the Pod that is currently in Unscheduled state, so that it can scale up the node that is required.

Describe any alternative solutions you've considered.:
N/A

Additional context.:

I am doing testing with the placement-policy scheduler plugin for this.

• https://github.com/Azure/placement-policy-scheduler-plugins/

I am also open to any other workaround to allow CA to play nicely with other schedulers, of which I am probably unaware of.

Thanks for taking the time to read this!

[k8s-triage-robot]

The Kubernetes project currently lacks enough contributors to adequately respond to all issues and PRs.

This bot triages issues and PRs according to the following rules:

• After 90d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, lifecycle/rotten is applied
• After 30d of inactivity since lifecycle/rotten was applied, the issue is closed

You can:

• Mark this issue or PR as fresh with /remove-lifecycle stale
• Mark this issue or PR as rotten with /lifecycle rotten
• Close this issue or PR with /close
• Offer to help out with Issue Triage

Please send feedback to sig-contributor-experience at kubernetes/community.

/lifecycle stale

[jdheyburn]

/remove-lifecycle stale

[k8s-triage-robot]

The Kubernetes project currently lacks enough contributors to adequately respond to all issues.

This bot triages un-triaged issues according to the following rules:

• After 90d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, lifecycle/rotten is applied
• After 30d of inactivity since lifecycle/rotten was applied, the issue is closed

You can:

• Mark this issue as fresh with /remove-lifecycle stale
• Close this issue with /close
• Offer to help out with Issue Triage

Please send feedback to sig-contributor-experience at kubernetes/community.

/lifecycle stale

[k8s-triage-robot]

The Kubernetes project currently lacks enough active contributors to adequately respond to all issues.

This bot triages un-triaged issues according to the following rules:

• After 90d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, lifecycle/rotten is applied
• After 30d of inactivity since lifecycle/rotten was applied, the issue is closed

You can:

• Mark this issue as fresh with /remove-lifecycle rotten
• Close this issue with /close
• Offer to help out with Issue Triage

Please send feedback to sig-contributor-experience at kubernetes/community.

/lifecycle rotten

[k8s-triage-robot]

The Kubernetes project currently lacks enough active contributors to adequately respond to all issues and PRs.

This bot triages issues according to the following rules:

• After 90d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, lifecycle/rotten is applied
• After 30d of inactivity since lifecycle/rotten was applied, the issue is closed

You can:

• Reopen this issue with /reopen
• Mark this issue as fresh with /remove-lifecycle rotten
• Offer to help out with Issue Triage

Please send feedback to sig-contributor-experience at kubernetes/community.

/close not-planned

[k8s-ci-robot]

@k8s-triage-robot: Closing this issue, marking it as "Not Planned".

In response to this:

The Kubernetes project currently lacks enough active contributors to adequately respond to all issues and PRs.

This bot triages issues according to the following rules:

• After 90d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, lifecycle/rotten is applied
• After 30d of inactivity since lifecycle/rotten was applied, the issue is closed

You can:

• Reopen this issue with /reopen
• Mark this issue as fresh with /remove-lifecycle rotten
• Offer to help out with Issue Triage

Please send feedback to sig-contributor-experience at kubernetes/community.

/close not-planned

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.

[vadasambar]

There are 2 options here:

1. Import your own scheduler code into CA, re-compile and use the new CA image
2. Convert your filter plugin into an extender (design proposal, official docs reference, discussion around extenders in #sig-scheduling)

[vadasambar]

/reopen

[k8s-ci-robot]

@vadasambar: You can't reopen an issue/PR unless you authored it or you are a collaborator.

In response to this:

/reopen

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.

[vadasambar]

/remove-lifecycle rotten

[vadasambar]

Looks like I can't re-open the issue. @jdheyburn you might have to re-open the issue.

[jdheyburn]

/remove-lifecycle rotten

[jdheyburn]

/reopen

[k8s-ci-robot]

@jdheyburn: Reopened this issue.

In response to this:

/reopen

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.

[jdheyburn]

@vadasambar Thanks for picking this up, although I'm not in a position to test it out, I'll look to have a play around with it once it is released