Follow a request from admission through backend selection, retries, and the final response.
- Validation and TTL checks happen before queue admission.
- A worker selects an eligible backend and retries within the request budget.
- The final response includes headers that expose queueing, processing, and backend selection.
The request TTL is the end-to-end budget across queue time and every backend attempt.
client → validate → priority queue → worker → select backend
→ call backend → retry/requeue if eligible → respond
- The listener receives and validates the request.
- User profile and request headers determine priority and overrides.
- The request enters the priority queue.
- A worker chooses backends by path, health, load-balancing order, and circuit state.
- The worker retries or requeues eligible failures while TTL remains.
- The proxy returns the final response with diagnostic headers and telemetry.
Note
Exact settings and response fields are defined in Configuration and Headers and Status Codes.
Warning
If the TTL expires in the queue or during retries, the proxy returns its TTL-expired response instead of starting another attempt.