Build a home server lab you can actually manage

You do not need a rack. Power draw matters more than raw speed, because the machine runs all day, every day.

Buy RAM before CPU. Containers and VMs are limited by memory long before a modern CPU becomes the bottleneck.

A common, sensible layout: Proxmox on the metal, one LXC container running Docker and a dashboard for the everyday apps, and VMs next to it for whatever you are learning this month. A broken experiment never touches the photo backup, and Proxmox can snapshot the app container before every upgrade.

This is how the Homeio team runs its own server. On the Proxmox host shell, fetch a Debian template and create an unprivileged container with nesting turned on, which Docker needs:

pveam update
pveam available --section system | grep debian-12
pveam download local debian-12-standard_12.7-1_amd64.tar.zst   # use the name the list shows

pct create 200 local:vztmpl/debian-12-standard_12.7-1_amd64.tar.zst \
  --hostname homeio --cores 2 --memory 4096 --swap 1024 \
  --rootfs local-lvm:32 --net0 name=eth0,bridge=vmbr0,ip=dhcp \
  --unprivileged 1 --features nesting=1,keyctl=1 --onboot 1
pct start 200

Open a shell inside the container and install Homeio. It brings Docker, PostgreSQL and Nginx with it:

pct enter 200
apt update && apt install -y curl
curl -fsSL https://get.homeio.app | bash

Give the container a DHCP reservation on your router, open its address in a browser, and create your account. Before each big change, take a snapshot from the Proxmox host so you can roll back in seconds:

pct snapshot 200 before-upgrade
# and if it goes wrong:
pct rollback 200 before-upgrade

Inside an LXC container some host features are limited: SMART data and disk partitioning belong to the Proxmox host, not to the container. Pass storage in as a mount point (pct set 200 -mp0 /tank/media,mp=/DATA/Media) rather than trying to manage disks from inside.

1. 1.Network-wide ad blocking with AdGuard Home or Pi-hole. Useful on day one, and it teaches DNS.
2. 2.Deploy a web server. Run one container, reach it on the LAN, then reach it from your phone over Tailscale (example below).
3. 3.Monitoring: Uptime Kuma for services and Beszel for machines, with notifications to your phone.
4. 4.A reverse proxy with HTTPS (Caddy or Nginx Proxy Manager) in front of your apps, with a local domain such as home.lan.
5. 5.Backups that you have restored: Proxmox Backup Server for guests, restic for data, one copy off-site.
6. 6.Infrastructure as code: move your compose files into a git repository and redeploy the whole stack from it.

Project 2 in practice. Start a web server on port 8080, check it from the server itself, then from your laptop:

mkdir -p ~/site && echo '<h1>Hello from my homelab</h1>' > ~/site/index.html
docker run -d --name site --restart unless-stopped -p 8080:80 -v ~/site:/usr/share/nginx/html:ro nginx:alpine
curl -s http://localhost:8080

Then open http://server-ip:8080 from another machine on the network. With Tailscale on the server and on your phone, the same page opens at http://server-name:8080 from anywhere, with no port forwarded on the router. In Homeio the container shows up on the desktop, where you can name it, give it an icon and a link.

• Write down what you set up and why, in a wiki such as BookStack or a plain git repository. Six months later you will not remember.
• Keep compose files and configuration in one folder (/DATA/AppData or a git repo), so a rebuild is a copy and a docker compose up.
• Update on a schedule, not when you remember: weekly image pulls, monthly OS updates, after a snapshot.
• Never forward ports for the dashboard or SSH. Use Tailscale or WireGuard for yourself, and a tunnel for the one app you share.