STATUS
Service status.
Measured availability for the Retrics application and API, the window it covers, and the limits of the measurement — read live from the checks themselves, not from a dashboard anyone edits.
LAST UPDATED · 2026-08-23 15:40 UTC
01Right now
The application and API are serving. The last completed check was 2026-08-23 15:39 UTC, 1 minute ago.
02Measured availability
99.43% over the trailing 30 days, from 6,278 of 6,426 expected one-minute checks.
A trailing window rather than month-to-date, deliberately: a month-to-date figure looks better every 1st and worst on the 31st, which flatters us at exactly the moment a merchant is most likely to be checking after an incident.
03Interruptions in the last 30 days
14 interruptions recorded. An interruption means at least two consecutive minutes with no answer. A single missed check is not counted: the checker runs on a scheduler that drifts across minute boundaries, and in practice that produced isolated one-minute gaps while the application was serving normally throughout.
- 2026-08-23 09:00 UTC — 2 minutes
- 2026-08-23 08:00 UTC — 3 minutes
- 2026-08-23 03:00 UTC — 3 minutes
- 2026-08-23 00:00 UTC — 3 minutes
- 2026-08-22 11:00 UTC — 3 minutes
- 2026-08-22 09:00 UTC — 2 minutes
- 2026-08-21 22:00 UTC — 2 minutes
- 2026-08-21 12:00 UTC — 2 minutes
- 2026-08-21 11:00 UTC — 3 minutes
- 2026-08-21 08:00 UTC — 3 minutes
- 2026-08-21 04:00 UTC — 3 minutes
- 2026-08-20 14:01 UTC — 2 minutes
- 2026-08-20 08:00 UTC — 3 minutes
- 2026-08-19 16:00 UTC — 2 minutes
04How this is measured, and what it cannot tell you
A check runs every minute. It passes only if the application is serving and its database answers — a page rendered by a process that cannot reach its database is not a working product, and scoring that as a good minute would make this figure flattering rather than true. Each passing minute records one row. Nothing records a failure.
Availability is therefore read from the gaps. A service that is down cannot report that it is down, but it also cannot report anything, and the absence is the evidence. That is why this page counts silence rather than trusting a self-reported status.
The limit worth stating plainly: the checker runs on the same platform as the application. If the scheduler stalls while the product is serving normally, this page records an interruption that no merchant experienced — so a reported interruption is not always one you felt. A fully independent verdict needs a checker hosted somewhere else, and Retrics does not have one yet. Until it does, this page says so rather than implying more.
If your experience disagrees with this page, your experience is the better evidence. Write to support@retrics.ai.