Server monitoring
Server monitoring from the inside, with a fix when something stops.
Uptime checks from outside tell you that a site is down. The server agent tells you why: a full disk, a stopped database, memory running out. One command installs it on Debian, Ubuntu and the RHEL family. It connects outbound only and opens no ports. And when a service stops, a quick fix can restart it.
Website monitoring is free. The server agent is included from the Pro plan; the number of servers depends on the plan.
Installed with one command
Add a server in the app, copy the command, run it as root. The agent is online in about a minute and starts reporting.
curl -fsSL https://app.centoculus.com/install/<your-token> | sh- Debian, Ubuntu and the RHEL family (AlmaLinux, Rocky Linux and others), as .deb or .rpm packages for amd64 and arm64
- The installer checks the package checksum before anything is installed
- Outbound connections only: the agent sends its data out, nothing connects in, no port has to be opened in your firewall
- Each server has its own key; remove a server in the app and its key stops working
- Open source agent, small footprint, no containers and no runtime to install
Not yet available: an agent for Windows servers. Websites on Windows servers can still be monitored from outside.
What the agent reports
The numbers that explain an outage, from inside the server.
-
CPU and load
Load average and CPU usage, so you see a runaway process before the site slows to a crawl.
-
Memory and swap
Free memory and swap usage. A server that starts swapping is a server that will go down soon.
-
Disks
Fill level of every file system. A full disk is the most common reason a database or mail server stops.
-
Services
The services you choose, for example nginx, Apache, PHP-FPM, MySQL or Postfix. You are alerted when one of them stops.
-
Database and web server
The agent recognizes MySQL and MariaDB, Redis, nginx and Apache and reports their key figures.
-
Network and uptime
Traffic per interface and the time since the last reboot, so an unexpected restart does not go unnoticed.
Alerts that tell you what to do
An alert names the server, the service or disk and the value, in plain words: "nginx stopped", "disk /var 96 % full". You get it by email and app push, on Pro and up also by Telegram, on Business and up by webhook.
- Thresholds for every disk, for memory and for load, with sensible defaults
- Alert when a chosen service stops, and again when it is running
- Alert when the agent itself stops reporting, so a server that went offline is never silent
- Server and website checks in one account: you see that the site is down and that the database stopped, side by side
Fix automatically when this server fails
The agent reports; the fix comes in through the Responder (a separate small program, outbound HTTPS only) or SSH as root. Switch automatic fixes on per server once one of them is set up; they draw from your prepaid balance or the included minutes of your "Automatic fixes" plan.
-
A check fails
A service stopped or a disk is full. A waiting time you set filters out short glitches.
-
The quick fix runs
It restarts the service after its configuration test passes, frees disk space or renews a certificate. Fixed price, never installs packages, never reboots, never deletes website files or databases.
-
A deep fix if needed
If the quick fix cannot solve it, it reports what it found. You, or the automatic mode, order a deep fix billed in 15-minute blocks; the estimate is the upper limit.
Automatic fixes are off until you switch them on, per server. They are paid from prepaid cover: businesses from their balance, everyone else from the included minutes of a 12-month plan "Automatic fixes" (unused minutes lapse at the end of the term). Safeguards: a waiting time, a monthly spending limit, enough balance or included minutes, no second fix while one runs. Details under fixes when something breaks.
Who it is for
-
A VPS with a few sites
You know when the disk fills up and when MySQL stops, before your visitors see error 500. Website checks and the agent in one view.
-
Agencies
Client servers next to client sites, one status page per client, and fixes you can order for them. Checks as often as every 15 seconds on the Agency plan.
-
Hosting providers
Monitoring and fixes your customers buy from you, under your brand. See white label.
Questions about server monitoring
Which Linux distributions are supported?
Debian, Ubuntu and the RHEL family (AlmaLinux, Rocky Linux and compatible systems), on amd64 and arm64. The installer picks the .deb or .rpm package for your system.
Does the agent open a port on my server?
No. The agent only sends data out over an encrypted connection. Nothing connects in, so your firewall stays as it is.
How much load does the agent add?
Very little. It reads values the kernel already keeps and sends a few kilobytes at a time. You will not notice it on a small VPS.
Can I monitor a Windows server?
Not from the inside yet. Websites, ports and certificates on a Windows server can be monitored from outside on every plan.
Is the agent the way in for fixes?
No. The agent only monitors and reports metrics. Fixes come in through one of three ways: the Responder, a separate small program installed with one command as root that fetches its work over outbound HTTPS and can be limited to an allowlist of commands; SSH as root on your own server; or restricted SSH access to a webspace.
How do I remove it?
Remove the server in the app, which disables its key, and uninstall the package with apt or dnf. Nothing else is left on the server.
Is server monitoring in the free plan?
No. The free plan covers website monitoring: 50 monitors, checked every 5 minutes. The server agent is included from Pro; see the plans and prices.
See inside your servers, and get them fixed.
One command to install, alerts that name the cause, and a fix when a service stops. Website checks included, free to start.
Related: Fixes when something breaks · Website monitoring · Status pages · Plans and pricing · Website down checker