Engineering
Observability & incidents
Lightweight logging, ownership, and incident habits
Observability & incidents
Keep this lightweight until volume demands a fuller SRE practice.
What to log
| Do | Don’t |
|---|---|
| Request ids / shopper ids (non-secret) | Access tokens, passwords, PAN, raw bind secrets |
| Auth failure reasons (safe enums) | Full Google tokens |
| Latency / error rates on critical routes | PII in marketing analytics dumps |
Ownership
| Symptom | First stop |
|---|---|
| Docs login broken | efacts-dev auth gate + Vercel protection |
| App cannot auth | Backend auth + iOS Keychain session |
| Missing receipts | Backend receipt services + POS ingest |
| Chrome / brand regressions | Docs or website frontend |
Incident basics
- Stabilize — rate limit, rollback, or disable flag
- Communicate — who is impacted + next update time
- Fix forward or roll back with a known-good artifact
- Write down root cause in Cognee / eng notes when meaningful