HTTP status
Confirms the page returns the code it should. A server that responds with a 500 is down as far as your customers are concerned, and a plain ping would call it healthy.
Three parts that work together: checks that catch real failures, incident management that keeps everyone informed, and a status page that stays honest without anyone maintaining it.
A ping only proves a server answered. These check whether it answered correctly.
Confirms the page returns the code it should. A server that responds with a 500 is down as far as your customers are concerned, and a plain ping would call it healthy.
An error page that answers 200 is not up either, and a check that only reads the code will not see it. We fetch the page and look for a word or sentence you chose, and open an incident when it is gone.*
* We search what your server sends back, as text. We do not run JavaScript, so a text that a page only builds in the browser is not found, and we do not read individual fields of a JSON answer.
A certificate that quietly lapses is one of the most avoidable outages there is. You hear 30, 14, 7 and 1 days ahead instead of from a browser security screen, and you hear it when a certificate is in date but issued for the wrong name or by an authority browsers do not trust.
Sites rarely fail all at once, they get slow first. Tracking response time turns that slide into a warning instead of a surprise, and tells you when a service is slower than is normal for it, well before it crosses a limit.
When a domain name stops resolving, everything behind it is gone at once, and a record that changes without anyone knowing is how a site ends up somewhere it should not be. We look your name up through two independent DNS services, open an incident when it stops resolving, and tell you when it starts pointing somewhere new.
A name that is not renewed takes the website, the mail and every service on it along. Where the registry publishes the date, you hear 30, 7 and 1 days before the registration runs out, and once more if the date passes.
Every outage moves through a clear lifecycle, so your team and your customers always know exactly what is happening. A slow response opens and closes its incident by itself; the two steps in the middle are for an outage somebody is working on.
Opened automatically at the first check that fails, with what the check ran into. No manual step, no missed outage.
Your team sets the status while working the problem, and the status page shows it.
Customers can see you are past diagnosis and actively shipping a repair.
Closed automatically once the check passes again. An incident whose recovery never arrived can be closed by hand, and its timeline says that it was.
Checks are only useful if the right people see them. These are the three places that happens.
Email reaches you, and each person on the account chooses whether they get it. Slack, Microsoft Teams and generic webhooks land alerts in the tools your team already has open, all at once if you like. A service that keeps going down is reported as unstable, instead of at every short interruption.
A page with every service you monitor and its current state, updating itself from the checks. Open to your customers, or private so that only the people you give access can open it. Live incident history, and maintenance banners your visitors see before the work starts rather than after.
Uptime percentages and response-time trends you can read at a glance, plus a PDF when someone needs the numbers on paper. On the first of the month the report of each status page arrives by mail, with a short summary of how the month went.
Up to 250 client status pages under a single account, each carrying the client's branding instead of ours. A shared pool of 250 services you divide however you like, logins for 20 people of your own team, and read-only access for clients who just want to see their own page.
We monitor Lunolyte with Lunolyte. Our own status page is public, including the bad days.