Redacteu

Status

Checking the API…
How we measure

Every minute, euRedact checks that the API answers and that a GPU node is serving redaction. A minute counts as available only when both hold; a minute that was not measured counts as unavailable.

The maintenance window, the hour from 03:00 to 04:00 UTC each day, is not counted, whether or not maintenance runs in it. The General Terms leave maintenance windows out of availability (Article 1), Article 8.3 sets the target, and Article 8.11 says maintenance windows are not counted as unavailability. Maintenance runs under cover of a second node, so the service normally stays up in that hour; a day’s tooltip still shows how many of its window minutes were unavailable. Work outside the window counts like any other minute. Article 1 also leaves force majeure out; we do not, because a measurement cannot tell it from an outage, so those minutes count against us.

A day’s bar shows the share of that day’s minutes outside the maintenance window that were available, counted from when measuring began; the figure under the bars covers the days of the last 90 since then. When a day lost minutes, its tooltip says why: the API not answering, no GPU node serving, or not measured; some days also carry a short note from us on what happened. This page asks the API when you open it and every minute after; it never shows a status saved in advance.

When something goes wrong that you would notice, we post an incident here with updates as it develops, in UTC; it reaches this page within a minute or two. While an incident is open, its impact sets the status at the top of the page if it is worse than what we measure. It never makes a measured problem look better. Incidents do not change the availability figures, which come only from the minute-by-minute measurement.

euRedact Core, the open-source SDK, runs on your own machine and does not depend on this service.