Ollama Model Viewer

This web service provides a simple frontend to view loaded Ollama models and vRAM usage in the ollama namespace.

Features

• View currently loaded Ollama models
• Monitor vRAM usage on g5.2xlarge nodes
• Web-based interface with Bootstrap styling
• OpenShift OAuth integration for secure access

How to use

Prerequisites

• A running Kubernetes/OpenShift cluster
• kubectl or oc configured to connect to your cluster
• The ollama namespace must exist in your cluster
• An Ollama pod running with label app=ollama-serve

Running locally

1. Install Go dependencies:

   go mod tidy

2. Run the service:

   go run main.go

   The service will be available at http://localhost:8080. It will use your local kubeconfig for authentication.

Building and deploying with Docker

1. Build the Docker image:

   docker build -t quay.io/redhat-ai-dev/ollama-model-viewer:latest .

2. Run the Docker container:

   docker run -p 8080:8080 quay.io/redhat-ai-dev/ollama-model-viewer:latest

Deploying to OpenShift

To deploy this service to your OpenShift cluster with OAuth authentication:

1. Create secret resource

COOKIE_SECRET=$(openssl rand -base64 32) envsubst < deploy/secret.yaml | oc apply -f -

2. Deploy using the provided resources:

   oc apply -f deploy/

3. Get the external route:

   oc get route ollama-model-viewer -n ollama

See the deploy/README.md for detailed deployment instructions and security considerations.

Architecture

The application consists of:

• Main container: Go application serving the web interface
• OAuth proxy sidecar: Handles OpenShift OAuth authentication
• Service account: With necessary RBAC permissions to read pods and execute commands

Security

• Uses OpenShift OAuth for authentication
• Requires proper RBAC permissions to access the Kubernetes API
• Cookie-based session management for the OAuth proxy