🏠 HOME LAB
INTEL® NUC

NUCleus

A Kubernetes cluster running on Proxmox VE, powered by Talos Linux. Immutable infrastructure, GitOps-driven, built to last.

Cluster operational
INTEL® NUC
Infrastructure
🖥️

Proxmox Cluster

2× Intel NUC (i7-8559U, 4C/8T each) · 96 GB total RAM · Proxmox VE 9.1. High-availability hypervisor cluster hosting all workloads.

8 cores · 16 threads · 96 GB
☸️

Kubernetes Cluster

5 nodes total — 1 control-plane + 4 workers, running Talos Linux. Each worker 4 vCPU, 8 GB RAM, with dedicated Longhorn storage disks.

5 nodes · 20 vCPU · 38 GB RAM
💾

Distributed Storage

Longhorn block storage with 2× replication across workers. ~50 GB usable capacity (100 GB raw across 4 nodes).

longhorn · zfs
🌐

Dual Ingress

Public traffic via Cloudflare Tunnel. Private access through Tailscale mesh VPN. No ports exposed.

cilium
Stack
⚙️ Proxmox VE Hypervisor
🐧 Talos Linux Immutable OS
☸️ Kubernetes v1.36
🌿 Cilium eBPF CNI
🔭 Hubble Network Observability
🔄 ArgoCD GitOps
📦 Longhorn Storage
🏗️ Terraform IaC
📋 Helmfile Deploy
Architecture
INTERNET INGRESS ☁️ Cloudflare Tunnel 🔒 Tailscale VPN CNI 🌿 Cilium — eBPF · kube-proxy replacement · L2 LB · 🔭 Hubble K8S control-plane 4 vCPU · 6 GB API · etcd · scheduler worker ×4 4 vCPU · 8 GB ea workloads ArgoCD Prom + Grafana Echo Server STORAGE 📦 Longhorn · ZFS/NFS · 2× replication COMPUTE 🖥️ nuc01 · i7-8559U 4C/8T · 64 GB · 1 TB NVMe + 250 GB SSD Proxmox VE 9.1 ⚡TB3 🖥️ nuc02 · i7-8559U 4C/8T · 32 GB · 240 GB SSD Proxmox VE 9.1
▸ Ingress Layer
Cloudflare Tunnel → public · Tailscale → private
▸ Service Mesh
Cilium — L2 load balancing, kube-proxy replacement, Gateway API · Hubble — network observability
▸ Workloads
ArgoCD · Prometheus + Grafana · Echo Server
▸ Storage
Longhorn — distributed block storage on ZFS, 2× replication (~50 GB usable)
▸ Compute
2× Intel NUC (i7-8559U, 96 GB total) running Proxmox VE → 5 K8s nodes (1CP + 4 workers)
▸ Provisioning
Terraform → VM provisioning · Helmfile → initial bootstrap · ArgoCD → day-2 reconciliation
Services
📊

Monitoring

Prometheus metrics collection with Grafana dashboards. Full cluster observability out of the box.

prometheus · grafana
🔊

Echo Server

HTTP echo service for testing ingress, load balancing, and connectivity. Available publicly and via Tailscale.

echo.hangas.net
🔭

Hubble

Network observability for Cilium — service map, flow logs, and DNS tracking. Real-time visibility into cluster networking.

cilium · hubble ui