From a bare Proxmox instance to a production-ready Kubernetes cluster — in under 5 minutes. One Bash script, zero dependencies.
K8s Setup · k8s-hetzner · 00:04:12 11:14:47 RUNNING 80% 16/20 ── Phases ────────────────────────────────────────────────────────────────── 1. ✓ Preflight Check 11. ✓ Control Plane init 2. ✓ Cloud-Init Template 12. ✓ CNI Plugin 3. ✓ Clone VMs 13. ✓ Join CPs (HA) 4. ✓ Port Forwarding 14. ✓ Join Workers 5. ✓ Start VMs + SSH 15. ✓ Labels + Taints 6. ✓ apt update + Reboot 16. ▶ Network Policies 7. ✓ SSH Keys + /etc/hosts 17. · Add-ons (Helm...) 8. ✓ Firewall (nftables) 18. · Fetch kubeconfig 9. ✓ Kubernetes packages 19. · Write inventory 10. ✓ HAProxy + keepalived 20. · Take snapshots ── Proxmox ───────────────────────────────────────────────────────────────── ● k8s-cp-1 running ● k8s-worker-1 running ● k8s-cp-2 running ● k8s-worker-2 running ● k8s-cp-3 running ── Log ───────────────────────────────────────────────────────────────────── [11:14:29] Worker-2 joined successfully. [11:14:31] Applying node labels and taints... [11:14:47] ✓ All 5 Nodes Ready. [q] Quit [+/-] Log lines
Why proxkube
A single Bash script. No Ansible, no Terraform, no external dependencies.
Cloud-Init template, network detection, storage detection — everything is automatically discovered and configured.
If setup aborts, resume continues exactly where it left off. No VM is recreated.
Built-in flicker-free TUI with 19 phases, progress bar and live log. No separate monitoring tool needed.
Secrets encryption at rest, audit logging, firewall source-IP restrictions, Pod Security Standards and proper kubelet TLS — enabled by default.
3 Control Planes with HAProxy and keepalived (Virtual IP). If one CP fails, another takes over automatically.
etcd backup with auto-rotation, VM snapshots, Velero for Persistent Volumes. Fully automated restore included.
Special mode for Proxmox on Hetzner root servers: private NAT network, port forwarding, iptables-persistent.
Prometheus + Grafana, Loki log aggregation and Falco runtime security — one flag to enable each.
Ecosystem
Just set to true — the rest happens automatically.
Traefik
cert-manager
Cloudflare DNS
ArgoCD
Vaultwarden
Longhorn
Velero
Prometheus
Grafana
metrics-server
Helm