Server monitoring
One agent per host, reporting the numbers that predict an outage — usually well before response time moves at all.
Installed with one line
Copy the command from the console with your key already in it. The host registers, reports within the minute, and updates itself in place.
curl -fsSL https://portal.serverpal.net/install.sh \ | SERVERPAL_AGENT_KEY=<your-key> bash
- Outbound only — no inbound firewall rule and no listening port added
- Uninstalls cleanly when you decommission the host
- Reports its version, and the console tells you when it lags
What the agent reports
CPU
Utilisation on every sample, plus a flag when the host is being throttled — the reason a box can look busy and idle at the same time.
Memory
Used against total, so a percentage always comes with the numbers behind it.
Disk
Every mounted filesystem tracked separately. A full /var never hides behind a mostly-empty root.
Load average
Alerted against a per-core threshold, so one setting means the same thing on a 2-core and a 32-core machine.
OS health
Pending package updates and reboot-required, surfaced as information and never as an alarm.
systemd
A count of failed units, so a service that died quietly between deploys still gets noticed.
Thresholds you set, per host
A database server and a cache do not agree on what busy means, so thresholds are set per host for CPU, memory, disk and load-per-core. Nothing turns amber unless a threshold you configured has actually been crossed.
- Group hosts the way you already think about them, and roll a group up to its worst member
- Routine OS facts stay informational, so amber keeps its meaning
- History kept for the length of your plan, so you can see what changed before it broke
Put the agent on a server.
The free plan covers one host and twenty checks, with no card.