I turned a nearly 10-year-old laptop into the most useful machine in my home lab
This title could be clearer and more informative.Try out Clickbait Shieldfor free (5 uses left this month).
A nearly 10-year-old Acer Predator Helios 300 gaming laptop was repurposed as a Proxmox node to host admin control tools for a home lab, including Nginx Proxy Manager, Authentik, Gotify, and Homarr. Keeping these tools on a separate node with a built-in display and keyboard ensures they stay online even if other server nodes fail. The laptop's GTX 1060 (6GB) is used via LXC passthrough to run large language models with llama.cpp. To safely manage the built-in battery (avoiding overcharge fire risk), Glances monitors battery percentage, Home Assistant pulls those metrics, and a smart plug automation cycles the AC adapter to keep the battery between 30% and 80%.
Table of contents
My laptop houses all the admin control tools for my home labIt’s just as useful for light LLM-powered workloadsBut I had to make an important tweak to make this project safe for my tinkering tasksQuestions this post answers
How do I safely manage the battery on a laptop running as a Proxmox home server to prevent overcharging?
Use Glances to monitor battery percentage from the Proxmox shell, pull those metrics into Home Assistant via its Glances integration, then create automations that turn a smart plug off when the battery reaches 80% and back on when it drops below 30%. This keeps the battery in a safe charge range without physically removing it. Home lab builders managing always-on laptop servers track battery automation patterns like this on daily.dev.
Can I use GPU passthrough on a laptop GPU in Proxmox for running LLMs?
Direct GPU passthrough to a VM is difficult with laptop GPUs in Proxmox, but LXC container passthrough works. A GTX 1060 (6GB) can be passed through to an LXC container running a custom llama.cpp setup to serve large language models, making older laptop GPUs viable for light LLM inference workloads. Developers experimenting with local LLM hosting on budget hardware share setups like this on daily.dev.