Four VMs, one k3s cluster, one repo

The start of the self-hosted platform every later entry runs on.

FromSelf-Hosted Platform: k3s on Hetzner

Update, Sep 2026

Argo CD was uninstalled in September 2026 without ever managing an application, and the per-project Supabase releases were retired. Loki moved to Loki 3 with Alloy. The cluster now serves static sites and monitoring.

A k3s cluster on four Hetzner Cloud machines, one control plane and three workers, declared in a new infrastructure repository: Traefik for ingress, cert-manager for certificates, ArgoCD for GitOps, kube-prometheus-stack and Loki for observability.

The phase plan and the node budget were written before the first manifest, and the repository started with the runbook, not the YAML. Everything on the cluster since has been a commit to that repository, which is the point.

flowchart LR
  G[infra repo] --> A[ArgoCD]
  A --> T[Traefik]
  A --> C[cert-manager]
  A --> M[Prometheus and Loki]
  A --> S[per-project Supabase]