Trace answers a different question from Activity, database audit, and logs: it follows the technical path of one operation across service boundaries. Use it when you need to know where a request spent time, where it failed, or whether asynchronous work was accepted versus completed.
Start with the right identifier
For HTTP endpoint support, capture the UTC timestamp, method/path, response status, and x-revo-oid response header. Do not collect the bearer credential or any signed URL.
Investigation sequence
An accepted asynchronous request only establishes that work was queued or started. Confirm a terminal Job/run state, delivery result, or finalized Storage artifact before concluding that work succeeded.
What to look for
- Ingress and authorization: did the request reach the intended active Endpoint under an authorized principal?
- Route and validation: was a definition selected and did its guard reject the structured input before execution?
- Resolution: which Component version, custom runtime revision, schema Component, library set, and execution policy were selected?
- Execution: which segment is slow or failed; did the configured timeout or cancellation boundary apply?
- Async handoff: is a queue acceptance followed by a durable processor/Agent/automation result?
- External work: did application code record a safe provider correlation ID that lets you verify an ambiguous timeout without replaying it?
Trace, logs, and history together
Safe support bundle
Provide a narrow time range, IDs, sanitized status/error text, resource name/ID, selected version, and reproducible non-secret input shape. Redact headers, cookies, API keys, Secret values, private prompts, Storage download URLs, and customer payloads unless an approved secure support channel specifically requires them.
Use the Observability guide for the complete operator workflow, then return here when the investigation needs a trace-first view. Last modified on September 5, 2026
