Homelab monitoring
Homelab monitoring needs two layers, detailed checks inside the lab, usually Uptime Kuma, and one watcher outside it, because a monitor that shares your power, router and internet line goes silent exactly when those fail.
Most homelabs end up with Uptime Kuma in a container, and it is a good choice. It checks every VM, container and device on the LAN by HTTP, ping, TCP or DNS, sends to around 90 notification providers and costs nothing. The gap is where it lives. When the power goes, the router reboots into a bad config or the ISP drops you for an hour, Kuma goes down with everything else, or it sees the failure and has no route out to tell you. You find out when you get home, or when someone in the house asks why Jellyfin is broken.
The fix is not replacing Kuma. It is one thing outside the lab that expects to hear from it. Services behind NAT cannot be checked from the internet without opening ports, so the lab pushes instead: a small script checks each service locally and sends a heartbeat out only when the service answered. No port forwarding, no tunnel, no public IP.
Heartbeats out of the lab
#!/bin/sh
# Usage: lab-heartbeat <local url> <logdash monitor id>
# Pings Logdash only while the local service answers. Six pings a minute,
# because a Pro push monitor expects one in every 15-second window.
for i in 1 2 3 4 5 6; do
curl -fsS -m 2 -o /dev/null "$1" &&
curl -fsS -m 2 -o /dev/null -X POST "https://api.logdash.io/ping/$2"
sleep 10
done
# chmod +x it, then crontab -e on an always-on box, one line per service:
# * * * * * /usr/local/bin/lab-heartbeat http://192.168.1.20:8096/ JELLYFIN_MONITOR_ID
# * * * * * /usr/local/bin/lab-heartbeat http://192.168.1.30:8123/ HASS_MONITOR_IDPush monitors are a Pro feature, $15 a month for up to 50 services, and on Pro Logdash expects a heartbeat in every 15-second window. Cron runs at most once a minute, so the script loops six times with 2-second timeouts, which keeps every gap under 15 seconds. One missed window is an alert, so run it on a box with a wired connection. A flaky Wi-Fi link turns into pairs of down and up messages.
A homelab monitoring stack that covers both failures
- Inside: Uptime Kuma for every container, disk and device. Fine-grained, free, and on the LAN where it can reach everything.
- Outside: one heartbeat per service that matters, sent by the script above. When all of them go quiet at once, the problem is power, the router or the line, not a service.
- Public services: anything you expose through a reverse proxy or a tunnel can take a plain HTTP monitor instead. Five of those run on the free plan, checked every 5 minutes. Logdash refuses private addresses, so a 192.168 URL has to use a heartbeat.
- A status page: the free plan includes one public page, so the household can check it before they message you.
- Create the push monitors One service per thing you care about, monitor mode set to push. Copy each id into its crontab line.
- Install the script Save it, make it executable, add the crontab lines. Within a minute every monitor is green and Telegram gets an up message for each, so you know the path works.
- Unplug the WAN cable Every heartbeat stops at once. Within 30 seconds Logdash marks each monitor down, and the Telegram alerts land on your phone over mobile data, one per service, while the whole lab is still offline.
Logdash vs Uptime Kuma in the lab
| Feature | Logdash | Uptime Kuma |
|---|---|---|
| Sees LAN-only services | Through heartbeats your script sends | Directly, by HTTP, ping, TCP or DNS |
| Alerts on a power cut or ISP outage | Yes, the silence is the alert | No, it is down or offline too |
| Adding a service | One crontab line per service | A few clicks in the UI |
| Notification channels | Telegram and webhook | Around 90 providers |
| Maintenance | Nothing to update | One container to update and back up |
| Cost | Heartbeats need Pro, $15 a month | Free on hardware you already run |
When Uptime Kuma alone is the better pick
- You only care about single services failing, and you would notice a power cut or a dead line anyway because you are at home.
- You want the outside watcher free. Kuma plus a free Healthchecks.io account, 20 jobs with a Telegram integration, covers both layers for nothing.
- You want ping, TCP and DNS checks on every device. Logdash does HTTP and heartbeats, nothing else.