Server host metrics (agents)

Outside-in monitors tell you whether a service is reachable. To see what's happening inside a server — CPU, memory, disk, temperature — install the NetTools metrics agent. It's available on AI Premium and up (up to 10 hosts).

How it works — and why it's safe

The agent is outbound-only. It pushes a small metrics sample to us over HTTPS on a schedule; nothing connects to it. That means:

• No inbound ports to open on your firewall.
• No credentials stored on our side — the agent authenticates with a token that only permits metric pushes, nothing else.
• Uninstalling is removing one file and its scheduled task.

It runs on Linux, macOS, and Windows.

Install

Open the HOSTS tab in the web app and add a host — you get a one-line command with your token baked in. On Linux:

curl -fsSL https://nth02.abr.cloud/cloud/agent/script -o abr-agent.sh \
  && sudo ABR_AGENT_TOKEN=<your-token> sh abr-agent.sh install

macOS is the same with ?os=macos; Windows gets a PowerShell one-liner. The tab always shows the correct command per OS. The agent installs a scheduled task and begins reporting within a minute.

What you get

• Live vitals — CPU, memory, disk, plus load, swap, and (on Linux) temperature and per-core detail, with OS and CPU model auto-detected.
• Sustained-only alerts — per-host thresholds that fire only when a level stays high across several checks. A brief 99% spike never emails you; a genuinely overloaded server does.
• Daily coverage bars and history over your plan's retention window.
• Token rotation and a clickable update badge when a newer agent ships — one copy-paste updates a host without losing its history.

Alerts

Agent alerts cover: host went offline / came back, and sustained high CPU, memory, load, or disk. Thresholds default to sensible values (CPU/RAM 90%, disk 95%) and are adjustable per host from the card's EMAIL ALERTS row.

See Monitoring for the outside-in side, and Plans & billing for limits.