Platform status
Measured from the outside, the same way our customers' own uptime figures are: a site counts as up when we fetch its address and it answers.
All systems operational
This page reports the platform. A single site that is not answering is not published here — its owner is told directly and sees it on their own page in the console.
Day by day
Green is a day everything served. Amber is a day on which something was down without it being platform-wide. Red is a day with an interruption of the kind listed below. A blank day is one we have no measurements for.
Interruptions
| Started | Ended | Lasted | State |
|---|---|---|---|
| 11 Aug 12:50 | 11 Aug 12:52 | 2 minutes | Resolved |
| 11 Aug 10:49 | 11 Aug 12:43 | 1 hour | Resolved |
How this is measured
Every site we host is fetched from outside the platform on a short interval. A site counts as up when the request comes back with a page — not when its container is running, which is a different and much easier question to pass.
Once a minute we count how many of those sites are not answering, and write the answer down. The availability above is the average of those counts, so it is not a figure anybody types in, and it does not change afterwards — including when a site is deleted, which used to erase its outages from this page along with it.
An interruption is listed here when enough of the platform is affected at once for it to be a fault of ours rather than of one site: a failed deployment, a single broken application or a site somebody stopped is not a platform outage, and saying so on this page would tell every other customer something untrue about the service. Whoever is affected is always told directly, whether or not it appears here.
Maintenance we scheduled, restarts a customer asked for and sites deliberately stopped are logged but excluded — they are not availability a customer lost. Customers see the numbers for their own sites, including everything that happened only to them, on the Uptime tab in the console.