Homelab Setup Overview
Hardware Specs:
- 3x Intel NUC 13th Gen (32GB RAM, 1TB NVMe each)
- Synology DS1821+ NAS (40TB raw storage)
- Ubiquiti Dream Machine Pro
- Raspberry Pi 4 cluster (8 nodes)
Core Stack:
- Proxmox VE (Virtualization)
- TrueNAS Scale (Storage)
- K3s Kubernetes (Container Orchestration)
- Docker Swarm (Legacy container management)
Core Infrastructure Services
1. Proxmox VE
- Purpose: Bare-metal hypervisor for VM management
- Key Features: LXC containers, ZFS support, cluster management
- Why I Use It: Perfect balance between enterprise features and homelab usability
2. TrueNAS Scale
- Storage Setup: 40TB ZFS pool (RAID-Z2)
- Services Hosted:
- SMB/NFS shares for media storage
- iSCSI targets for VM storage
- S3-compatible object storage via MinIO
3. K3s Kubernetes Cluster
- Nodes: 3 master nodes (NUCs), 8 workers (RPi cluster)
- Key Components:
- Longhorn distributed storage
- Traefik ingress controller
- Cert-Manager for SSL certificates
- ArgoCD for GitOps deployments
Networking Services
1. OPNSense Firewall
- Features: VLAN segmentation, IDS/IPS, WireGuard VPN
- Bandwidth: 1Gbps symmetric fiber handling
- Custom Rules: Isolated IoT network, guest WiFi restrictions
2. Pi-hole + Unbound
- Stats: Blocks ~30% of DNS queries (50k+ domains)
- Setup: Dockerized with gravity-sync for redundancy
- Custom: Whitelist for streaming services
3. Tailscale Mesh VPN
- Use Case: Secure remote access to homelab services
- Integration: Exit node for secure browsing anywhere
- ACLs: Role-based access to specific services
Media & Entertainment Stack
1. Plex Media Server
- Library Stats: 4K HDR movies, automated TV show downloads
- Hardware Acceleration: Quadro P2000 passthrough
- Users: 15 family members with shared access
2. *Arr Suite
- Components: Sonarr, Radarr, Lidarr, Readarr
- Automation: Triggered by Overseerr requests
- Storage: 20TB media library on TrueNAS
3. Audiobookshelf
- Content: 500+ audiobooks, 300 podcasts
- Features: Mobile sync, progressive web app
- Backup: Nightly sync to Backblaze B2
Development & Automation Tools
1. GitLab CE
- Features: CI/CD pipelines, container registry
- Integration: Auto-deploys to K3s via ArgoCD
- Stats: 150+ private repositories
2. Home Assistant
- Devices: 80+ connected IoT devices
- Automations: 45 Node-RED flows
- Dashboards: 6 custom UIs for different family members
3. Vaultwarden
- Usage: Family password manager
- Security: Behind Authelia with 2FA
- Backup: Encrypted daily snapshots
Monitoring & Observability
1. Grafana + Prometheus
- Dashboards: 15+ custom boards
- Metrics: 200k+ time series collected
- Alerts: Slack/Pushover notifications
2. Uptime Kuma
- Monitored Services: 45 endpoints
- Status Page: Public-facing with SLA metrics
- Incidents: Automatic Discord notifications
3. NetData
- Real-time Monitoring: Per-second metrics
- Anomaly Detection: ML-powered alerts
- Integration: Kubernetes cluster insights
Backup Strategy
-
3-2-1 Rule Enforcement:
- Local ZFS snapshots (hourly/daily/weekly)
- Offsite to Backblaze B2 (encrypted)
- Cold storage on LTO-6 tapes (quarterly)
-
Critical Data:
# Database backup example mysqldump -u root -p"$DB_PASSWORD" --all-databases | \ age -r "age1qy..." > /backups/db-$(date +%s).sql.age
-
Disaster Recovery:
- Automated Proxmox VM backups
- etcd snapshots for Kubernetes
- Offline SSH recovery keys
FAQs
Q: Why not use cloud services instead?
A: Complete data ownership, cost efficiency for large storage needs, and learning experience.
Q: How much does this cost to run?
A: ~$150/mo for power + $50/mo for cloud backups. Hardware amortized over 5 years.
Q: Security concerns with self-hosting?
A: All services sit behind VPN, use 2FA, and receive automatic security updates.
Q: Time investment required?
A: ~5-10hrs/week for maintenance, but most services now run autonomously.
Q: Recommended starting point for beginners?
A: Begin with Docker on a single NUC, expand to Proxmox as needs grow.
Q: How to handle power outages?
A: APC UPS with NUT shutdown triggers, ~45min runtime.
Q: Most valuable homelab skill learned?
A: Troubleshooting distributed systems and failure scenario planning.
This homelab setup continues to evolve, with plans to implement Ceph storage and experiment with GPU passthrough for AI workloads. The key lesson? Start small, automate everything, and always have a rollback plan! 🚀