Writing / Article
Monitoring the Homelab with Beszel
A lightweight node-based dashboard across Proxmox, a Raspberry Pi 4 and Windows.
Once the lab grew past a single machine, I wanted one place to see how everything was doing. I didn’t want to run a full Grafana and Prometheus stack to get it, so I chose Beszel.
How it works
Two parts:
- The hub is a central web interface that collects and displays metrics.
- The agent is a small service on each monitored machine that sends data to the hub.
That’s the whole model, which is most of the appeal.
What I monitor
Three machines, because they’re the three that matter:
| Machine | What the dashboard showed (21 July 2026) |
|---|---|
| Proxmox host | About 3.5% CPU while running 74 services, with roughly 36% of its 32 GB of RAM free. |
| Raspberry Pi 4 | Pi-hole, Unbound and about a dozen other services (45 active services in total). Under 10% CPU and around 36.5 °C, with a dedicated fan. |
| Windows machine | Idles higher than the Linux systems, at about 11.5% CPU, and has a much larger disk footprint. |
These figures are a snapshot from the day I wrote the article. They are not live values and this site doesn’t track them.
Why Beszel
- It is light enough to leave running on small machines.
- It favours a clean at-a-glance view over detailed time-series graphs, which is what I actually look at.
- Checking CPU load, memory, temperatures and uptime in one place tells me quickly whether the lab is healthy.
At the time, both the Proxmox host and the Pi had plenty of headroom. If you run Proxmox LXCs or a Raspberry Pi and want quick observability without much overhead, it’s worth trying.
Where it fits now
Beszel is one of four monitoring tools in the lab, alongside Uptime Kuma for availability and Glances and Dozzle for investigating. See the homelab overview for how they sit with everything else.