Multi-tenant Elastic Resource Quota #48

Question

Multi-tenant Elastic Resource Quota #48

kaiohenricunha opened this issue 8 months ago · comments

Kaio Cunha commented 8 months ago

Area

Scheduler
Controller
Helm Chart
Documents

Other components

No response

What happened?

The current design of the capacity scheduler for elastic quotas doesn't fit a multi-tenant scenario, if I understand it correctly.

If I have the following namespaces:

acme-dev-us-west-2-webhook
acme-dev-us-west-2-kube-janitor
apple-dev-us-west-2-webhook

It would be great if we could configure the "acme" tenant resources to be shared only among namespaces starting with "acme".

What did you expect to happen?

Tenant's resources not shared with other tenants

How can we reproduce it (as minimally and precisely as possible)?

No response

Anything else we need to know?

No response

Kubernetes version

1.28

Scheduler Plugins version

Latest

Wei Huang · Answer 1 · Tue Dec 05 2023 06:16:22 GMT+0800 (China Standard Time)

My 2 cents is that there is no one-size-fit-all quota design. If an ElasticQuota needs to support cross-namespace sharing, there are more areas need to be considered:

semantics of selector (label selector or regex or mix, etc.)
how to deal with overlapping selectors
how it impacts preemption
...

If you have a through design and can make it backwards-compatible, I'm happy to review to discuss the next steps.

Kaio Cunha · Answer 2 · Tue Jan 09 2024 02:33:03 GMT+0800 (China Standard Time)

My 2 cents is that there is no one-size-fit-all quota design. If an ElasticQuota needs to support cross-namespace sharing, there are more areas need to be considered:

semantics of selector (label selector or regex or mix, etc.)

how to deal with overlapping selectors

how it impacts preemption

...

If you have a through design and can make it backwards-compatible, I'm happy to review to discuss the next steps.

Thanks for the ideas. I'll see if I can work on that and submit a PR. Could be an interesting use case.

刘达 · Answer 3 · Mon Feb 26 2024 11:10:07 GMT+0800 (China Standard Time)

My 2 cents is that there is no one-size-fit-all quota design. If an ElasticQuota needs to support cross-namespace sharing, there are more areas need to be considered:

semantics of selector (label selector or regex or mix, etc.)

how to deal with overlapping selectors

how it impacts preemption

...

If you have a through design and can make it backwards-compatible, I'm happy to review to discuss the next steps.

Thanks for the ideas. I'll see if I can work on that and submit a PR. Could be an interesting use case.

Hi, may I ask if you are working on it? Maybe a new queue based plugin can be introduced.

Kubernetes Triage Robot · Answer 4 · Sun May 26 2024 11:33:17 GMT+0800 (China Standard Time)

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

Kubernetes Triage Robot · Answer 5 · Tue Jun 25 2024 11:48:35 GMT+0800 (China Standard Time)

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

Kubernetes Triage Robot · Answer 6 · Thu Jul 25 2024 12:17:33 GMT+0800 (China Standard Time)

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

Kubernetes Prow Robot · Answer 7 · Thu Jul 25 2024 12:17:37 GMT+0800 (China Standard Time)

@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-sigs/prow repository.