Find the first failing boundary here, then follow its owner: the operational runbook for exact procedures, the quickstart for the smallest local path, the deployment walkthrough for environment choices, the full API reference for request and authentication contracts, or the contributor guide for change validation.
| Symptom | Narrow first | Procedure owner |
|---|---|---|
| The first local run fails | Setup, preparation, or API boot | Quickstart |
| An optional service stack does not settle | Prerequisite or service health | Deployment walkthrough, then runbook |
| The API is unavailable or its address is occupied | Missing process, unhealthy process, or address conflict | API incident procedure |
| Event flow is slow or stopped | Source, transport, processing, or store | Incident response |
| Local data is missing or unexpected | Selected profile, preparation step, or concurrent writer | Local pipeline operations |
| A request is rejected | Authentication, authorization, or request contract | Full API reference |
- Reproduce the smallest applicable quickstart path.
- Classify the failure as environment, request path, or event path.
- Record the first failed boundary, then follow its owner.
- Do not reset local state unless the procedure requires it.
| Change or incident | Verification owner |
|---|---|
| Documentation, runtime, backend, or client change | Contributor guide |
| Incident response or recurring maintenance | Operational runbook |
| Current acceptance or external gate | Engineering status |
When escalating, include the selected profile, first error, expected and actual results, and the latest relevant change. Exclude credentials and private data.