← back to projects

Proxmox Home Server

The foundation of my homelab — a Dell OptiPlex 7080 running Proxmox VE, hosting every VM and container in my setup.

Proxmox Dell OptiPlex 7080 ZFS Virtualisation Linux
Proxmox VE — Dell OptiPlex 7080 i5-10500T · 64GB DDR4 · 1TB NVMe (ZFS) · 256GB SATA SSD (ISOs) · 192.168.1.250 STORAGE NVMe ZFS Pool — VMs & Containers STORAGE SATA SSD — ISO Images & Templates ── LXC CONTAINERS ───────────────────────────────────────── jellyfinCT Ubuntu LXC 192.168.1.248 Jellyfin · Radarr · Sonarr · Prowlarr jellyseerrCT Debian LXC 192.168.1.247 Docker · Jellyseerr websiteCT Debian LXC 192.168.1.245 Caddy · Astro · Cloudflare Tunnel ── VIRTUAL MACHINES ──────────────────────────────────────── gameserver 192.168.1.249 Ubuntu · AMP Minecraft · Valheim · Zomboid displayserver 192.168.1.246 Ubuntu · Docker Grafana · Prometheus windowsdesktop 192.168.1.88 Windows 11 ProtonVPN · qBittorrent LEGEND LXC Container Virtual Machine (Linux) Virtual Machine (Windows) Running

The Hardware

I picked up the OptiPlex 7080 on eBay for around £230 — a solid deal for an i5-10500T, 64GB DDR4 and a 1TB NVMe. The main requirement was small form factor. Everything in my homelab sits inside a 10 inch DeskPi Rackmate T1, so the host machine and all peripherals had to fit within that.

Why Proxmox

I'd used Proxmox before so I was already familiar with the interface and how it handles VMs and containers. Beyond that it was an easy choice — it's open source, completely free, and has a huge community behind it. Whenever something breaks or behaves unexpectedly, the answer is usually already out there somewhere.

Storage Layout

The 1TB NVMe runs a ZFS pool which is where all VMs, containers and their data live. ZFS gives data integrity checksums and easy snapshot support which is useful when you're experimenting and want a quick rollback point. The 256GB SATA SSD is used purely for ISO storage — OS images, templates and anything needed for spinning up new machines.

The Road to Getting Here

Early on I was leaning heavily on AI tools to do config without fully understanding what was actually happening — which led to flattening VMs and starting over when things went wrong rather than properly diagnosing the issue. Not the most efficient approach but it taught me a lot about what not to do...

One issue that took a while to resolve was hardware transcoding. I originally ran Plex as my media server but 4K playback to my TV was problematic. I moved to Jellyfin as that has free hardware transcoding, — but even then the first LXC container I created for Jellyfin refused to see the host's iGPU. This time flattening did solve the issue.. I setup the container exactly the same as before and it just worked :)

These days the setup is stable and I have a much better understanding of what's running and why. The trial and error was a bit tiresome but I now know how everything fits together now, and can rebuild if needed

Next Plans

Add github push/pull requests for all servers so I can have all server configs saved into the cloud

Set a robust and regular backup schedule as since the proxmox host is on RAID0 there is currently no redundancy..

Add more VMs - Not sure what will be next exactly but I have a few ideas in mind :)

Current VMs

  • jellyfinCT — Ubuntu LXC running Jellyfin, Radarr, Sonarr and Prowlarr
  • jellyseerrCT — Debian LXC running Jellyseerr via Docker
  • gameserver — Ubuntu Server VM running CubeCoders AMP
  • displayserver — Ubuntu Server VM running Grafana and Prometheus
  • windowsdesktop — Windows 11 VM used as a download client
  • websiteCT — Debian LXC hosting this website