New and cached work can run
At least one serving strategy is healthy, the relay answers, and the proxy budget remains available.
Service health
Checking live status
No health claim is shown until the public endpoint answers.
Refreshes every 60 seconds.
Cold work is impaired
Cached transcript reads may still serve. Translation shedding is reported separately and does not automatically stop plain transcript work.
Components
Every row is read from the public endpoint. Translation shedding has its own row because it affects only requests that ask for a translated track.
| Component | State | Detail |
|---|---|---|
| Relay | — | Waiting for the endpoint. |
| Translation shedding | — | Waiting for the endpoint. |
| Proxy capacity | — | Waiting for the endpoint. |
Serving strategies
| Strategy | State | Success rate | Last success | Last failure | Error |
|---|---|---|---|---|---|
| Waiting for the endpoint. | |||||
Canaries
Canaries probe every client ScriptHaul knows, including ones that YouTube has retired; a failing canary that is not a serving strategy is expected, not an incident.
| Client | State | Last OK | Last probe | Error |
|---|---|---|---|---|
| Waiting for the endpoint. | ||||
Machine-readable status
The public response includes service, relay, translation, capacity, serving-strategy, canary, and measured-latency states without internal addresses, credentials, byte totals, or customer data.
curl https://api.scripthaul.com/v1/statusMeasured API latency
Waiting for the rolling authenticated-request sample window.
These are measured Worker handling times across authenticated API-key requests in the rolling 24-hour window, aggregated into privacy-safe five-minute histogram buckets. Cache-hit and cold-transcript cohorts remain separately labelled in the machine response.