I'm a junior sysadmin and I built the monitoring cockpit I wanted at work (open source)

Bezo — Thu, 05 Nov 2026 10:21:14 +0000

Hi DEV! I'm Enzo, a junior system and network administrator. I'm just starting out in IT, and this is my first real open-source project: NetCockpit.

👉 Live demo, nothing to install: https://b2zo0.github.io/netcockpit/ 👉 Code: https://github.com/B2ZO0/netcockpit

The problem I had

At work, the network is managed by an outside provider. They have the keys to the firewall and the switches. I don't.

So when something broke, I had to jump between a ping, the vSphere client, Grafana, Zabbix, and a network map in a Visio file nobody had updated in years. I kept asking myself the same questions: is it the server, the switch, or the Internet? Which port is this machine even plugged into?

I wanted one screen that answers that in one sentence.

What NetCockpit does

It doesn't replace Prometheus, Grafana or Zabbix. It installs them, configures them, and adds the missing piece: a cockpit that understands your sites, your VMware and your network, without anyone drawing anything.

🗺️ A live topology, built automatically: Internet → ISP → firewall → buildings → networks → switches → servers → VMs.
🔌 Which port is everything plugged into? It reads CDP/LLDP through the ESXi hosts, so it works even without access to the switches. Add read-only SNMP and you get the full port-by-port map (UniFi, Cisco…).
🚦 No false alarms: a laptop that went home shows as absent, not down. Red is reserved for what really matters: servers, hypervisors, firewalls, switches, domain controllers.
🖥️ VMware at a glance: hosts, VMs, datastores and alarms, from vCenter or a single ESXi.
🔔 One alert pipeline to Telegram, Discord or email. If the firewall dies, you get one message, not 80.
🤖 Optional AI diagnosis, with a cloud API or a local model (Ollama) so nothing leaves your network.
📦 One command: ./setup.sh detects your network, asks a few questions and starts everything. Upgrades never touch your configuration.
🌍 English and French, day and night themes.

I run it every day, in my homelab and in production at a small company with two buildings.

Try it in one minute

The easiest way is the live demo: a simulated two-site network with a few failures, so you can click around.

To run it locally with Docker:

bash
git clone https://github.com/B2ZO0/netcockpit.git
cd netcockpit
docker compose -f docker-compose.demo.yml up --build

Then open http://localhost:8088. Nothing is scanned and no credentials are needed.

For a real install, there's a step-by-step guide for newcomers, and scenario guides for different setups: one network or several buildings, provider-managed network, no VMware, re-addressing…

What I learned
Never lose the user's config. My first update script rewrote the .env file and wiped settings. Now every run makes a backup and only touches the lines it manages. Lesson learned the hard way, on my own production.
Documentation is half the product. The first person I showed it to didn't know what to type at the first question. Now each question in the setup wizard explains where to find the answer.
Different isn't wrong. I call them buildings, other people say sites or branches. The interface now lets you choose the word.
Real networks are messy. Undeclared subnets, a Windows agent that stopped reporting RAM after an update, an IP plan about to change… Each one became a fix or a guide.
I'd love your feedback

I have a lot to learn, so any feedback, advice or criticism is welcome, whether you're a beginner or a seasoned pro. Issues and pull requests are open, and I'll keep shipping releases based on what you tell me: new features and fixes for the problems you run into.

If you find it useful, a ⭐ on GitHub or a follow really helps. I plan to build more tools for the community, for beginners, intermediates and pros alike.

Thanks for reading! 🙏

DEV Community: Bezo

I'm a junior sysadmin and I built the monitoring cockpit I wanted at work (open source)