Where it is raised
GET /api/v1/outages/{slug}GET /api/v1/outages
What triggers it
The keyless outage JSON route allows 120 reads a minute per client address, counted before the slug is read. The throttle reply is never cacheable, so a shared cache cannot hand one caller's refusal to the next.
Example response
GET /api/v1/outages/github
HTTP 429
Retry-After: <seconds>
Cache-Control: no-store
{
"error": "Rate limit exceeded. Try again shortly.",
"code": "RU-6003"
}Common causes
- Polling one service's reading in a tight loop.
- Many callers behind one shared egress address.
How to fix it
- Honor the Retry-After header.
- Cache readings: responses already carry Cache-Control allowing two minutes of reuse.