All features
Server monitoring

Server monitoring: spot trouble before it turns into downtime

A lightweight agent reports CPU, memory, disk, network and Docker containers every minute. You're alerted when a disk is filling up, CPU stays pinned or a container crashes, and the moment a server stops reporting.

A server's page with CPU, memory, disk and network charts over 24 hours

One command to install

Paste the command from the server's page. The agent runs as an unprivileged systemd service on x86_64 or ARM.

No ports to open

The agent only sends reports out over HTTPS. Nothing connects to your server, and it never runs anything it's sent.

Alerts that wait for a trend

CPU and memory alert when their average stays over your limit, so a short spike doesn't wake anyone. Disks alert as soon as one is over.

See it in Spot Downtime

The real app, with a demo team's monitors. Included on every plan, Free too.

A server's Docker containers: one killed for running out of memory, the rest healthy
Every container on the server, problems first: here, a worker killed for running out of memory.

What is server monitoring?

Uptime checks look at a server from outside: does it answer, is the website up, is the database port open. They tell you when something has already broken. Server monitoring looks from inside, at the resources that run out before things break: CPU, memory, disk space and network.

Spot Downtime's agent reads those numbers every minute and sends them to your workspace. Each server gets a page with live tiles, charts over 1 hour to 30 days, and every disk with its use and size.

Alerts that don't cry wolf

A busy server spikes to 100% CPU all the time, for a deploy, a cron job or a burst of traffic, and that's fine. What matters is CPU or memory that stays high. So those alerts use the average over a window you choose, 5 minutes by default, and only fire when the whole window is over your limit. They resolve once the value has been back under it for a full window, so they can't flap.

Disks are different: a full disk breaks things immediately, so a disk alerts as soon as it's over its limit, and the alert names the mount point, such as /var/log. And if the agent stops reporting, because the server crashed, lost its network or was switched off, the server is marked down and you get a normal down alert on every channel.

Built to be safe to install

The install script checks the download's checksum, creates a dedicated unprivileged user, and runs the agent as a hardened systemd service with a read-only view of the system and a 64 MB memory cap. It sends a test report before it finishes, so you know straight away whether the token and network are right.

The agent only makes outgoing HTTPS requests. It opens no ports, needs no firewall changes, and never runs anything it receives. Remove it any time with the same script and --uninstall.

How to set it up

  1. 1Click Add monitor and choose Server.
  2. 2Name it and pick the alert limits for CPU, memory and disks.
  3. 3Copy the install command from the server's page and run it on the server.
  4. 4The page fills in within a minute: live tiles, charts and every disk.

The details

Metrics
CPU, load, memory, swap, every disk, network in and out
Docker
Every container's state, health, restarts, CPU and memory, with alerts when one crashes or turns unhealthy (install with --docker)
Reported
Every minute; up to 3 hours buffered if the network drops
Down when
No report for 2 to 10 minutes (3 by default)
Resource alerts
CPU or memory averaging over a limit for 1 to 30 minutes; any disk over its limit
History
Charts over 1 hour, 24 hours, 7 and 30 days
Platforms
Linux with systemd, x86_64 and arm64
Plan
1 server on Free, 5 on Starter, 25 on Pro

Questions

What does the agent have access to?
It runs as its own unprivileged user in a locked-down systemd service. It reads /proc and filesystem sizes, and sends one HTTPS request a minute. It opens no ports and never runs commands.
Does it monitor Docker containers?
Yes. Install the agent with --docker and the server's page lists every container. You're alerted when one crashes, is killed for memory, keeps restarting or fails its health check, and you can make any container a monitor of its own.
Does it work on Windows or macOS?
Not yet: the agent runs on Linux. Watch Windows servers from outside with ping, TCP port and website checks in the meantime.
What happens if the server can't reach Spot Downtime?
The agent keeps up to 3 hours of samples and sends them once it's back, so the charts don't have a gap. If no report arrives in time, the server is marked down and you're alerted.
Are resource alerts sent by text?
No. Like slowness alerts, they go to your chat, email, PagerDuty and webhook channels. A server that stops reporting is a normal down alert and reaches every channel, SMS and calls included.
How do I remove it?
Run the install script with --uninstall. It stops the service and removes the agent, its user and its settings.

Set up server monitoring in under a minute

Included on the Free plan. No card needed.

Create a free account