[Documentation] Add HPA example to LWS site

[LuyuZhang00]

What type of PR is this?

/kind documentation

What this PR does / why we need it

add hpa docs

Which issue(s) this PR fixes

Fixes #652

Special notes for your reviewer

cc @Edwinhr716

Does this PR introduce a user-facing change?

No

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by: LuyuZhang00
Once this PR has been reviewed and has the lgtm label, please assign ahg-g for approval. For more information see the Code Review Process.

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

Needs approval from an approver in each of these files:

• site/OWNERS

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

[k8s-ci-robot]

Hi @LuyuZhang00. Thanks for your PR.

I'm waiting for a kubernetes-sigs 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-sigs/prow repository.

[netlify]

✅ Deploy Preview for kubernetes-sigs-lws ready!

Name	Link
🔨 Latest commit	c1e9ac6
🔍 Latest deploy log	https://app.netlify.com/projects/kubernetes-sigs-lws/deploys/68d1529cfe1aec000835a470
😎 Deploy Preview	https://deploy-preview-663--kubernetes-sigs-lws.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.

---

To edit notification comments on pull requests, go to your Netlify project configuration.

[Edwinhr716]

/ok-to-test

[Edwinhr716]

Is metrics server provided by most cloud providers? I've never had to do this step.

[LuyuZhang00]

This is a note, if user already have installed, they can safely ignore this step, I included it for my initial test

​​For open-source Kubernetes:​
By default, installations like kubeadm, kind, or minikube don't include the Metrics Server. The quickest way to get started with kubectl topor HPA is to install it manually.

For managed Kubernetes services (like GKE, EKS, AKS, etc.):​​
Most cloud providers include Metrics Server out of the box in their managed offerings—so it’s usually already there and good to go for things like HPA and kubectl top.

[Edwinhr716]

Makes sense, I'm fine with keeping this section. I think we should have a note saying: if using a managed Kubernetes service, skip this step

[Edwinhr716]

Would be good to add that model server metrics often take into account the whole group

[LuyuZhang00]

I tested adding CPU load to both the leader pod and the worker pods separately. In practice, the HPA only took effect when CPU load was increased on the leader pod. I noticed that in the updateStatus function within pkg/controllers/leaderworkerset_controller.go, it also monitors the leader pod. This seems like a potential area for improvement.

[Edwinhr716]

Yes, when using metrics exposed by the nodes themselves, it will only take the leader into account. My point was that we should add that model servers, such as vLLM, expose metrics that take into account the whole group.

This seems like a potential area for improvement.

I agree, but it is a complex area. We would need a way to aggregate the metrics of a group of pods, which should be covered by the revised Gang Scheduling KEP https://docs.google.com/document/d/1ulO5eUnAsBWzqJdk_o5L-qdq5DIVwGcE7gWzCQ80SCM/edit?pli=1&tab=t.0.

An idea I was thinking of earlier was to be able to change whether to target a leader pod or a worker pod for HPA, since there are deployment patterns where the leader only has a proxy server, so it isn't representative of when the accelerators being used have reached its limit.

[yankay]

HI @LuyuZhang00

Considering other examples, their YAML files are placed in the https://github.com/kubernetes-sigs/lws/tree/main/docs/examples/sample directory rather than within the markdown files.

Should we create an HPA directory under docs/examples and place the YAML files there? This would make maintenance easier.