/~/blog/ My Homelab Setup
2026.11.02

The best advice I heard about setting up a homelab is to “use what you have”. I don’t remember where I heard that first, but it stuck with me, and I think it’s a great advice and starting point.

The second best advice I learned is to “be curious” and try to learn as much as you can, and while I had a infrastructure experience for the past 10+ years, I wanted to refresh my knowledge about ansible, and learn more about linux, setting up a fully secure, self-hosted, open-source, private homelab with some services that were helpful to me, and my partner.

I had a few old laptops and one raspberry pi that were sitting around from previous projects, so I decided to use them to set up my homelab. Below are the devices I used:

Device RAM Storage CPU GPU OS
Raspberry Pi 5 16GB 64GB MMC, 1TB SSD ARM N/A Raspberry OS Lite
Lenovo Thinkpad T450s 8GB 256GB SSD Intel i5 Intel HD 5500 Proxmox VE
MacBook Pro 2015 16GB 512GB SSD Intel i7 Intel, Nvidia GT750M Ubuntu, Archlinux
Desktop PC 32GB 512GB SSD Intel i5 Nvidia GTX1060 Archlinux

The following services are running on these devices:

On the Raspberry Pi 5:

On the PVE nodes1:

The implementation for these services is done in the following projects:

Ansible Distro Install

is a collection of ansible roles to install different Linux distributions on a variety of devices.

It supports:

For now, I’m trying to give the Macbook Pro a second life by installing linux on it, and it does work, but there are still some issues with the graphics drivers, so I will have to look into that in the future, or use it as a second PVE node.

The project consists of 3 roles:

While the bootstrap role could theoretically be built to run any distro installation from any Live ISO2, I opted-in to use more traditional method of using a live USB with the desired distro, and a small commands that needs to be ran manually to set up the network and ssh access, and then run the ansible playbook to install the distro.

Ansible Distro Configure

is a collection of self-contained, modular ansible roles to configure a bare linux installation to fully customized. It supports configuring Ubuntu and Archlinux setup with desktop environment, dot files, password manager, web-browsers, and applications.

It’s open-source, and doesn’t require the installation to be done via the ansible-distro-install project, but it integrates easily with it.

The roles are split into 3 categories:

There is also the upgrade.yml playbook, which is used to upgrade the system and the services.

Architecture

Except for the git, caldav, and headscale services, that are publicly accessible -the CalDAV service doesn’t need to be publicly accessible, but it is added for convenience-, all the other services are only accessible from the local network, or via VPN (Wireguard, Tailscale) on .internal domain names, that are resolved by the Pi-hole DNS resolution, while still retaining the same VPN functionality mentioned before in My Home (VPN) Network Setup.

Here is the almost correct full digram of the homelab setup:

  ---
config:
    theme: redux
    look: neo
---
graph TD
    Internet[🌍 Internet] <--> ISP[ISP: 2.213.76.?]
    Internet <--> VPN[VPN: 149.40.8.?]
    ISP <--> SUBNET[Router Subnet: 192.168.0.1]
    VPN <--> VPN_SUBNET[VPN Subnet: 10.2.0.2]
    subgraph "Local Area Network (LAN)"
        SUBNET <--direct mode--> RPI[Raspberry Pi: 192.168.0.254]
        VPN_SUBNET <--VPN mode-->  RPI
        RPI <--> D1[TV: 192.168.0.100]
        RPI <--> D3[Desktop: 192.168.0.210]
        RPI <--> PROXY[proxy.intenral]
        subgraph "Proxmox VE Cluster"
          PROXY <--> TS[tailscale]
          PROXY <--> MONITOR[monitor.internal]
          PROXY <--> SUBS[subscriptions.internal]
          PROXY <--> ...
        end
    end
TS <--> D4[Laptop: 163.116.173.?]
TS <--> D5[Phone: 2a02:3032::1::?]

Closing thoughts

I have been managing my whole homelab setup with this ansible playbook approach for a while now, and it has been working great, I can easily add new services, or change the configuration of existing ones, the data of all these services is backed up to the garage S3 object store, and I can easily restore the backups when needed.

The next steps are to add more nodes to the PVE cluster, investigate the approach of LXC containers vs KVM virtual machines, add more services (ex. jellyfin), add offsite-backup replication, and maybe build a server case for these devices 😅.

Feel free to reach out if you are interested in this project, have questions, recommendations for a follow write-ups, ideas to add, or just want to say “hi”!

  1. Right now PVE is deployed on a single node (Lenovo Thinkpad T450s), but I plan to add more nodes for redundancy or reuse the MacBook as a second node in the future. ↩︎

  2. The Proxmox VE installation is using a Debian live ISO, because the Proxmox VE ISO which comes already with its own installer, but the installer can’t be automated via a bash script or ansible playbook. ↩︎