Skip to content

Conversation

@bhoomi1024
Copy link

Add Detailed KServe Integration Documentation

This PR introduces comprehensive documentation for integrating KServe with kgateway, addressing Issue #327.


📘 Added Documentation

1. Serverless Installation.md

A step-by-step guide for installing and configuring KServe in a serverless environment.

Key highlights:

  • Autoscaling features (scale to/from zero)
  • Revision-based canary rollout
  • Version compatibility

Includes:

  • Version matrix for Kubernetes, Knative, and kgateway
  • Knative Serving setup
  • Cert-Manager installation
  • KServe installation using Helm and YAML
  • Notes on feature flags, compatibility, and TLS requirements

2. KServe Ingress with kgateway.md

A comprehensive guide to using kgateway as an ingress for KServe’s ML services.

Covers:

  • Advanced ingress routing via VirtualService and RouteTable
  • TLS termination setup
  • CORS configuration
  • Secure endpoint handling using kgateway features
  • Troubleshooting and diagnostic commands
  • Cleanup instructions for all resources

✅ 3. _index.md Updates

  • Adds the new KServe section to the documentation index
  • Ensures proper navigation and discovery within the site

🔗 Related Issue

Closes #327


🧪 Tested On

  • Kubernetes v1.30+
  • Knative v1.16
  • kgateway using gateway.networking.k8s.io/v1beta1

Verification:

  • Setup instructions validated
  • TLS configuration tested
  • Inference requests confirmed using curl

@artberger
Copy link
Contributor

Hi @bhoomi1024 thanks for opening this PR to add KServe integration docs to kgateway. As I was reviewing it locally, I was wondering whether you had successfully tested this as written? For example, the ingress guide in the before you begin has you verifying the KServe installation, but there are no steps about installing it. Also, it mentions several kgateway CRs like VirtualService and RouteTable that I do not have.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants