Host Uptime Kuma without installing anything
HTTP, TCP, DNS or certificate checks every minute, and the alert going out where you will actually read it.
Starting at €4.99/month
Introduction
Uptime Kuma installed, updated and backed up by us, on a machine in France. You install nothing and configure no server: the application is online a few minutes after you order, under your own name.
The plan that suits it is App Essentiel at €4.99/month incl. VAT — 2 GB of memory. Uptime Kuma is a Node.js application that keeps every heartbeat of every monitor in a local database and pushes changes live to any dashboard left open. Thirty monitors checked every sixty seconds means a reading to write, aggregate and display every two seconds: that continuous flow, plus the history kept behind it, is what sets the tier.
It is the direct replacement for UptimeRobot or StatusCake when you want to decide the interval, the number of monitors and what becomes public. There is no automatic import from those services — the models do not match — but recreating twenty monitors takes half an hour and lets you drop the ones that no longer matter. Around a hundred notification targets are supported, from plain email to your own webhook, by way of Telegram, Discord, Slack, Matrix, ntfy or Gotify.
What managed hosting changes
HTTP, TCP, ping, DNS, certificate
A page that must contain a given word, a port that must answer, a DNS record that must point to the right place, a certificate expiring in thirty days: every monitor has its own type, interval and number of retries before it declares an outage.
The alert where you will read it
Email, Telegram, Discord, Slack, Matrix, ntfy, Gotify, Teams or your own webhook: around a hundred destinations are built in, and a single monitor can notify several of them at once.
Public status pages
One public page per service or per client, with its monitors grouped, an incident history and a banner to display during an outage. It can carry your own domain name.
Push monitors for silent jobs
A backup or a scheduled task calls a URL every time it finishes successfully. If the call does not arrive within the expected window, the alert goes out: it is the only reliable way to learn that a nightly job has stopped running.
Why hand it over rather than install it
An online shop
Home page, checkout and certificate monitored separately: you know straight away whether the whole site is down or only the payment step.
A contractor looking after a fleet of sites
One monitor per client, one group per site, and a public status page that answers half of Monday morning's calls before they are made.
A backup that has to run every night
The script calls Uptime Kuma when it finishes. Two nights without a call and the alert goes out: you find out before you need the backup, not on the day you go looking for it.
A charity with a website and a cloud
Three monitors, one email alert and one on Telegram. Nobody has to be a systems administrator to read the board and understand what is wrong.
A webmaster who will not miss a renewal again
The HTTPS monitor warns thirty days before the certificate expires, and the same screen shows how long the service has been up without interruption.
Hosted in France
In our own datacentre in Nouvelle-Aquitaine, which we operate ourselves. Your readings and your status pages stay with you, and the person on the phone speaks your language.
Frequently asked questions
No, and that is the costly mistake. A monitor sitting next to what it watches will not see a network outage in the datacentre: everything looks fine on its screen while your visitors cannot reach you. Uptime Kuma with us for a site hosted elsewhere — or the other way round — gives a measurement that means something, because the network path is part of what gets tested.
There is no automatic import; the data models are too different. In practice, recreating twenty monitors takes half an hour, and it is a good moment to delete the ones watching a service that was shut down two years ago. After that, the configuration and the history live in a database you take with you if you leave.
It tells you that one observation point got no answer, which is not quite the same thing. A brief hiccup on the network path can raise a false alarm. That is why every monitor has a retry count before declaring an outage and a delay before repeating the alert: two settings that separate useful monitoring from a flooded mailbox.
Nothing. You choose the application, fill in the fields it asks for — often an email address and a name — and it goes online. Access arrives by email with the address and the credentials.
Yes, they are part of the service. On a monitoring tool the real risk is not a vulnerability: it is the tool quietly stopping and watching nothing, without anyone noticing, precisely because it no longer sends alerts.