HOW WE MEASURE

Evidence you can inspect.

Blossom Status probes the public services you use to deploy and run Apps on Blossom, every minute, from more than one location. Every result is stored and published per day as raw CSV.

What we check

Component	Request	Passes when
Dashboard	`https://app.blossom88.com/users/sign_in`	HTTP 200 and body containing “Log in”
API	`https://api.blossom88.com/api/v1`	HTTP 200 and body containing “api up”
Live updates	`https://ws.blossom88.com/cable/up`	HTTP 200
MCP server	`https://mcp.blossom88.com/up`	HTTP 200 and body containing “healthy”
Runtime images	`https://registry-system.blossom88.com/v2/`	HTTP 401 and header docker-distribution-api-version: registry/2.0
App registry	`https://registry.blossom88.com/v2/`	HTTP 401 and header docker-distribution-api-version: registry/2.0
Docs	`https://docs.blossom88.com/`	HTTP 200 and body containing “Blossom”
Blossom Radar	`https://radar.blossom88.com/up`	HTTP 200

Each probe opens a fresh HTTPS connection and sends one GET without following redirects, so response times include DNS, TCP and TLS. A probe fails on a timeout (5 seconds to connect, 10 to respond, 20 overall), a connection or certificate error, or a response that does not match. A failed attempt is retried once after three seconds; only a second failure is recorded as failed.

A check never changes meaning. If we change what “up” means for a component, the old check is retired and a new one starts its own history.

Two kinds of location

Blossom Cloud (Falkenstein, Germany): runs inside this status app, which is itself hosted on Blossom.
Independent server (Germany): runs outside Blossom’s hosting. It buffers results on disk and delivers them later if this site is unreachable.

If Blossom’s hosting fails, the probe running on Blossom stops with it. The independent location keeps measuring, so that outage is still recorded rather than silently missing.

Minutes, uptime and coverage

Operational: every location that reported in that minute passed.
Degraded: some locations passed and some failed. Usually a network problem near one probe; counted as available and shown separately.
Down: every location that reported failed.
Unmonitored: no location reported. Unmonitored time is never counted as uptime.

Uptime is available minutes divided by monitored minutes. Coverage, shown next to every percentage, is the share of the window with results; always read the two together. The 24-hour window is rolling; longer windows and daily bars use UTC calendar days. Percentages are truncated, so 99.996% shows as 99.99%, not 100%.

An outage is a run of consecutive down minutes. In the daily bars, a day with fewer than 30 down minutes is a short outage, and 30 or more is an outage.

What this does not measure

These checks cover Blossom’s shared services and one App hosted on Blossom. They do not measure each customer’s App, build and deploy success, Blossom BYOC (bring your own cloud) providers, or your network path. This page reports what we observed; it is not a service level agreement.

Use the data

Open a component for daily evidence and CSV downloads. /api/v1/status returns the current status and uptime windows as JSON. Local times use your browser’s time zone; without JavaScript they stay in UTC.