Status
What is happening, and what is about to
Current incidents, planned work, and the figures behind the availability claim. If something is wrong you should find out here rather than by opening a ticket.
Everything is working normally.
How availability is measured
Three things have to be stated for an availability figure to mean anything. Most published figures state none of them.
- Window
Calendar month, not a rolling year - Measured from
From three locations outside our own network, because a check run on our own machines would only ever agree with us. - Interval
every thirty seconds - An outage starts
At the first failed check, not when a ticket is opened
- An outage ends
At the first successful check after it - Planned work
Counted, not excluded - Target
99.95% a month, covering the API, the build system and every running service - If missed
Credited automatically, without being asked
Being told
subscribe
Subscribe in the console and you are notified about the regions and services you actually use rather than about everything.
api
$ curl https://status.example.com/v1/summary
{"state":"operational","incidents":[]}