Use case
Linux server monitoring
Most outages on a single server are visible an hour before they happen: a disk filling, swap climbing, load creeping up after a deploy. Server monitoring is about seeing that hour.
NoDisrupt's agent is a small Go binary that reports host metrics every 60 seconds. The same account checks the services on that server from outside, so one incident shows both the cause and the effect.
Needs the Pro plan: £24 or $29 a month.
Start freeHow it works
- Step 1
Create a server metrics monitor
The app generates an install command with your credentials filled in.
- Step 2
Run one command on the server
The installer downloads the agent for amd64 or arm64, verifies its SHA-256 checksum, writes a root-only config and starts a hardened systemd service. No inbound ports are opened: the agent only makes outbound HTTPS requests.
- Step 3
Add the outside view
Add HTTP, API, port or ping checks for what the server serves, and heartbeat monitors for its cron jobs, so you see whether customers are affected, not only whether the host is busy.
- Step 4
Route alerts
Thresholds feed escalation policies: email first, then SMS or a phone call if nobody acknowledges.
NODISRUPT_AUTH_KEY='<your-auth-key>' NODISRUPT_METRIC_ID='<your-monitor-id>' \ sudo -E bash -c "$(curl -fsSL https://downloads.nodisrupt.com/agent/install.sh)"
What the agent reports
| Watch | Metric or check | Example alert |
|---|---|---|
| CPU | cpu_percent | Above 95% |
| Memory | mem_used_percent | Above 90% |
| Swap | swap_used_percent | Above 25%: the host is short of RAM |
| Load average | load1, load5, load15 | load5 above the core count |
| Disk space and inodes | disk_<device>_used_percentage, _inodes_remaining | Disk above 85%, or inodes nearly gone |
| Network | net_bytes_sent, net_bytes_recv | Traffic stops: the box is up but not serving |
- Linux only. There is no Windows agent.
- The installer needs systemd; hosts without it need a manual process supervisor.
- Host metrics, not application tracing: there is no APM, profiling or log search.
- The agent is a Pro plan feature (250 monitors, 5 users). Uptime, port and ping checks are on every plan; cron heartbeats start on Core.
Common questions
What does the server agent collect?
CPU utilisation, memory and swap use, the 1, 5 and 15-minute load averages, used space and free inodes per physical disk, and network bytes and packets. The full list with units is in the docs.
Does it work on Ubuntu, Debian and RHEL?
Yes. Any Linux distribution with systemd on amd64 or arm64, which covers current Ubuntu, Debian, RHEL, Rocky, Alma, Amazon Linux and 64-bit Raspberry Pi OS.
Is the agent safe to run?
It runs as a hardened systemd unit with NoNewPrivileges, PrivateTmp and a read-only system, its config is readable only by root, the download is verified against a published checksum, and it only makes outbound HTTPS calls.
How often does it report?
Every 60 seconds by default. The interval is set in /etc/nodisrupt/agent.yaml.
Can I monitor Windows servers?
Not with the agent. You can still check a Windows server from outside with HTTP, port and ping monitors, and push any metric you can script to an asset monitor.