No live probe yet — and we are not going to fake one
A status page earns its keep on the day it says something is broken. Publishing a hard-coded green dashboard before any probe exists would make this page decorative, so it says what is true instead.
Components
- In buildExchange API and routingLive probe not published yet — during the beta, incidents are announced in the app.
- In buildCloud workspacesLive probe not published yet.
- In buildSettlement and payoutsLive probe not published yet.
- PlannedPublic uptime history and incident logA real probe with real history, rather than a page that is green by construction.
What this page will report
- Availability of the completion endpoint, measured from outside our own network.
- Routing health — whether requests are finding a seat, separately from whether the API is up.
- Settlement lag, because money moving late is an incident even when nothing looks broken.
- An incident log with a written cause for anything that lasted long enough to notice.
How we will classify an incident
| Level | Means |
|---|---|
| Outage | Requests are failing. Announced immediately, cause published afterwards. |
| Degraded | Requests are served but slowly, or capacity is thin enough that routing is failing over often. |
| Accounting | Serving is fine but a figure is wrong — a mis-priced model, a settlement that did not post. Treated as an incident, because a wrong number is a wrong number. |