LAN Device Monitoring
The NAS, the printer, the camera and the machine in the server room have no public address, so nothing on the internet can check them. An agent inside your network can.
The check that runs on your side of the router
Every check we send from the outside starts on the internet and works inwards, which is exactly why none of them can see a device without a public address. A NAS on 192.168.1.20 is unreachable from outside by design, and so is every printer, camera, access point and virtual machine behind the same router. LAN monitoring turns the direction around. You install one agent on a machine that stays on — a server, a mini PC, a Raspberry Pi — it sweeps the private networks you enable and reports what answered. Everything it finds becomes an inventory you can look through; the devices you pick become ordinary Pulsitor monitors, with the same incidents, the same history and the same notification channels as your websites.
| 192.168.1.1 | gateway | monitored |
| 192.168.1.20 | nas | monitored |
| 192.168.1.31 | printer | monitored |
| 192.168.1.77 | camera-hall | inventory only |
The agent reports what it found. Which of those devices is worth an incident is your decision, taken one device at a time.
How LAN monitoring works
What makes Pulsitor powerful out of the box.
One binary, installed as a service
Run one command on a machine that is always on. The agent installs itself as a system service, starts with the machine and restarts itself if it ever stops. Nothing has to be installed beside it and there is no runtime to keep patched.
It sweeps, you pick
The agent reports which networks it is attached to and what answered on them. Whether any of those networks is swept at all is your decision, and so is which of the devices found is worth watching. The rest stays a list.
A picked device becomes a monitor
From then on the device is probed on its own interval, and one that stops answering opens an incident on the channels you already use. The agent measures; what counts as an outage is decided here, not on your network.
What people watch with it
Anything on the far side of the router that is only noticed once it is gone.
NAS and backups
The device every backup is written to is also the one nobody logs into for months. A NAS that dropped off the network takes every backup after it with it, silently.
Printers and label printers
A warehouse printer that lost its address at three in the morning stops the first shift, not the night shift. Hearing about it before anyone tries to print is the whole point.
Cameras and access points
A camera that goes dark records nothing, and an access point that reboots into a different address leaves a corner of the building without Wi-Fi. Neither reports itself.
Servers, switches and machines
Hypervisors, database servers, controllers on a production line, a virtual machine that did not come back after a host reboot — everything with an address and no way to call for help.
What the agent may and may not do
It runs inside your network, so what it is not allowed to do matters more than its features.
It never accepts a connection
Every exchange starts as an outgoing HTTPS request from your side. There is no port to forward, no VPN to build and no listening service on the machine for anyone to find. A firewall that allows ordinary outbound traffic is the whole requirement.
It sweeps only what you enable
The agent reports which networks it is attached to and waits. Nothing is swept until you turn that network on, only private address ranges are eligible, and a network too wide to sweep is refused outright rather than probed.
It decides nothing
Whether a device is missing, whether its address changed and whether either is worth waking somebody are judgements made on our side, from what the agent reported. It carries no thresholds, so a compromised agent cannot invent an outage.
Installed in one command
An enrollment token is the only thing you carry across — the agent already knows where to report.
Linux and macOS
curl -fsSL https://github.com/pulsitor/pulsitor-lan-agent/releases/latest/download/install.sh \ | sudo sh -s -- --enroll <token>
The script downloads the right build for the machine, checks it against the published checksums and hands over to the agent, which installs itself as a service.
Windows
iwr https://github.com/pulsitor/pulsitor-lan-agent/releases/latest/download/install.ps1 -OutFile install.ps1 .\install.ps1 -EnrollToken <token>
From an elevated PowerShell prompt. The agent registers itself with the service manager and starts with the machine.
A binary you already have
sudo ./pulsitor-lan-agent service install --enroll <token>
The installer is only a download-and-verify wrapper. One command copies the binary, redeems the token and writes the service definition.
Why watch a network this way
Sees what no probe can reach
Behind NAT, on a private range, on a device that will never have a public address. If the agent can reach it on the local network, it can be monitored — without exposing any of it to the internet.
Checks who answered, not just that something did
An address stops being an identity the moment DHCP hands the lease to somebody else. The agent matches the hardware address too, so a neighbour on the old IP is never read as your device being up.
Same incidents, same channels
A device that went missing is an incident like any other: the same alerting channels, the same history and the same status pages your team already reads.
LAN monitoring at a glance
What the agent actually does, in numbers.
| Property | Value |
|---|---|
| What is checked | That the device answers a probe sent now, and that it still holds the address it had |
| How a device is probed | ICMP echo from inside the network, with the hardware address verified against the answer |
| Inventory sweep | Every 5 minutes, across the private networks you enable |
| Check interval per device | From 15 seconds to 24 hours — the shortest interval depends on your plan |
| Before a device counts as offline | Unreachable for 3 minutes, and never on a single missed probe |
| Incidents it opens | Device offline, and device address changed — with the old and the new address in the alert |
| Runs on | Linux, macOS and Windows — one static binary, installed as a system service |
| What it needs from your firewall | Outbound HTTPS only. No inbound connection, no port forwarding, no VPN |
| Agents per account | Up to 10, so several locations are watched from one account |
| Notification channels | 14 |
| Available on | Paid plans, from 6 EUR per month |
LAN monitoring questions
What the agent sees, what it sends and what it never does.
Where does the agent have to run?
On any machine that stays on and sits on the network you want watched — a server, a mini PC, a Raspberry Pi, a virtual machine. It needs no fixed address of its own and nothing has to be able to reach it.
Do I have to open a port or set up a VPN?
No. The agent never accepts an inbound connection; it makes outgoing HTTPS requests to us, the same as any browser on the network. Nothing is exposed to the internet and there is no listening service to attack.
What does the agent send us?
The addresses, hardware addresses and names of the devices that answered on the networks you enabled, and later whether each monitored device is still answering. No traffic is captured, nothing is read from the devices themselves and no file ever leaves the machine.
Which networks does it scan?
Only the ones you switch on, and only from the private ranges. The agent reports what it is attached to and waits for that decision, so nobody — us included — can point it at a network you did not choose.
Does every device it finds count against my plan?
No. The inventory costs nothing and stays as long as you like. Only a device you pick becomes a monitor, and only then does it count against the monitor limit of your plan.
What happens when a device gets a new IP address?
The agent recognises it by its hardware address, so a device that came back on a different address is not reported as gone. It opens an address-change incident instead, carrying both the old and the new address.
Can I watch more than one location?
Yes. Each location runs its own agent and reports into the same account, so branches, a home lab and a data centre rack sit next to each other in one list. How many agents you may run at once depends on your plan.
What if a device does not answer ICMP?
A host that ignores an echo request reads as absent, so a device with a firewall that drops ICMP is not a good candidate. For services behind a public address, port monitoring is the check that fits.
Is LAN monitoring available on the free plan?
No. LAN agents are part of the paid plans, and the number of agents you may run at once grows with the plan.
Watching a network in 3 steps
Install the agent on a machine that stays on and enable the networks it may sweep.
Look through what it found and pick the devices worth an incident.
Choose the channels to be told on and let the agent report from the inside.