Uptime

Uptime is the share of time a service stays reachable over a measurement window. In hosting, what a monitor reports is not the state of a machine but the HTTP response it receives for a specific URL from a specific location. That gap explains why the same month can show 100% on the provider page and a handful of incidents in your own dashboard.

Server uptime versus site availability

A server can stay online while the site serves nothing useful. A machine that answers ping but returns Error 503 after running out of memory counts as alive to a check that only looks at the TCP connection, while visitors see a broken service. Availability asks whether users actually got the page they requested, so a useful monitor validates the HTTP status and the content, not just an open port.

What uptime monitoring checks

A monitor requests your URL every few minutes from several locations and marks the site down when there is no response or a status outside its healthy list. That produces false negatives. A monitor blocked by a Linux firewall reports an outage while every real visitor is fine. A 404 page can read as alive because the connection succeeded. A server returning 503 can still be counted up when the check only verifies TCP. A single location can also be wrong in the other direction, because a routing or CDN fault makes one probe fail while the rest of the world is unaffected.

SLA, guarantees, and credits

An uptime promise lives inside a hosting SLA as a credit, not cash and not compensation for lost sales. Provider math usually excludes scheduled maintenance, upstream interruptions, and faults on your side, so the official number will almost always look better than an independent tracker. Claiming a credit needs your own timestamped evidence, and the SLA normally sets a short deadline to file.

The single-location trap

Monitoring from one city reports a local routing problem as a global outage. If a Singapore probe fails while three other regions stay green, the likely culprit is the network path rather than your server. Rule out domain propagation and stale cache before you blame the host. Watch from several locations at a one to five minute interval, turn on alerts, and keep the timestamps so you have evidence. Monitor the URL visitors actually use, not just the homepage, because a maintenance page that returns a 404 still reads as up.

FAQ