Technical notes and manifests for running Ollama, Open WebUI, and FastMCP (with Azure integration) on Kubernetes, with a focus on Windows / WSL2 environments.
k8s/— Kustomize bases for FastMCP, storage, and NVIDIA-related configuration.helm/— Helm values and references for ingress, Ollama, Open WebUI, and SearXNG.notes/— Step-by-step install and setup notes (Kubernetes, NVIDIA on WSL2, Ollama).
- FastMCP + Azure — MCP server deployment and secrets (see
k8s/fastmcp/). - Inference & UI — Ollama and Open WebUI (Helm charts / values).
- Ingress — NGINX Ingress controller configuration.
- Search — SearXNG (optional companion service).
- A working Kubernetes cluster (local or remote).
- For GPU workloads: NVIDIA drivers and device plugin setup as described in
notes/. - Azure app registration details where FastMCP uses Azure APIs (see
k8s/fastmcp/secrets.yamlpatterns).
Apply manifests with Kustomize from the repo root, or install Helm releases using the files under helm/ once your cluster and namespaces are ready. See notes/ for environment-specific setup (especially WSL2).
| Topic | Notes |
|---|---|
| Kubernetes install | notes/Kubernetes/Install_k8s.md |
| NVIDIA + Kubernetes on WSL2 | notes/Kubernetes/Install_nvidia_k8s_wsl2.md |
| Ollama | notes/AI/Install_Ollama.md |