Features · Server monitoring

See inside your servers, live.

Uptime checks tell you a site is down. Server monitoring shows why: a full disk, memory running out, a process eating the CPU. You paste one command on a Linux server and a small agent starts reporting. It only reads and reports, and nothing can be sent back to it.

SERVER LIVE · web-1
  • CPU34% · Normal
  • Memory61% · Normal
  • Disk /92% · Critical
Alert sent: Disk / is 92% full (your limit is 90%)

Inside

What it does

CPU, memory, disks, Docker containers and processes, live and read-only.

  1. 01

    Live numbers

    While a server's page is open, CPU, memory and the process list refresh every couple of seconds, like watching top or htop in your browser.

  2. 02

    The whole machine

    CPU per core, memory split into programs, cache and free, every disk, network traffic, NVIDIA GPUs, uptime and logged-in users.

  3. 03

    Processes

    The busiest processes with user, CPU and memory. Passwords and tokens in command lines are masked on your server before anything is sent.

  4. 04

    Docker containers

    Included automatically when Docker is installed. Each container's state, restarts, CPU and memory, with alerts when one you choose stops.

  5. 05

    Attack signals

    Signs of a flood read from the server's own network counters: connection requests it cannot queue, connections being dropped, far more traffic than is normal for that server. We tell you; nothing is blocked.

  6. 06

    Security signs

    A database port open to the internet, a new open port or login account, SSH passwords being guessed, a server sending spam, a program hiding under a kernel name. Signs to check, found without giving the agent any extra permission.

  7. 07

    Alerts

    CPU or memory above your limit, a disk filling up, a server that stops reporting, a Docker container you chose that stops, a process that looks suspicious, or signs of an attack or a break-in (when you switch those on). One message when it starts, one when it is over, and problems that start together arrive as a single message.

Setup

How to start

  1. 1

    Add a server and give it a name.

  2. 2

    Copy the install command and run it on the server with sudo.

  3. 3

    The server appears online within a minute.

The fine print, up front

Good to know

  • Read-only by design: the agent opens no port and accepts no commands. We can't restart, change or run anything on your server. The one thing it listens for is that you removed the server from your account; then it stops itself.
  • It is one readable Python file. The installer checks its SHA-256 before installing, and you can print exactly what it would send with --check.
  • It runs as its own locked-down user with no shell and no sudo. Every report is signed with a secret unique to that server.
  • Linux with systemd and Python 3.8 or newer. Windows and macOS are not supported.
  • When Docker is installed, the agent joins the server's docker group so it can list containers. It only ever reads from Docker. Add --no-docker to the install command if you prefer to leave Docker out.
  • Suspicious-process detection and security signs use simple, cheap checks (miner names, programs in /tmp or a hidden folder, open ports, new accounts). They are hints, not an antivirus or a firewall.

Plans

What each plan includes

Free$0Pro$9/moBusiness$24/mo
1 server5 servers25 servers
History saved every 5mHistory saved every 1mHistory saved every 30s
History kept for 24 hoursHistory kept for 30 daysHistory kept for 90 days
Failure alerts for 5 Docker containers per serverFailure alerts for 15 Docker containers per serverFailure alerts for 15 Docker containers per server
Live view and alertsLive view and alertsLive view and alerts
See full pricing

Questions

Frequently asked questions

The honest answers, before you sign up.

1

What happens if someone steals a server's secret?

The worst they can do is send fake numbers for that one server. The secret gives no access to your account or the server, and you can replace it from the server's page in one click.

2

How much does the agent use?

Very little. It is capped at 20% of one CPU core and 128 MB of memory by its service settings, and normally uses far less.