fix production issues

docs/ops/phase4-measurement-template.md

@@ -0,0 +1,54 @@
+# Phase 4 — Extraction measurement template
+
+Fill one copy per candidate (`notifications` | `payments-webhooks` | `media`). Store dated copies under `docs/ops/evidence/` (create as needed). Do not extract without G1–G7.
+
+## Candidate
+
+- Name:
+- Date:
+- Author:
+- Environment measured: staging / production
+
+## G1 — Soak baseline
+
+- Command / duration:
+- Error rate / p95:
+- Link or paste `/metrics` summary:
+
+## G2 — Bottleneck
+
+- Symptom (SLO miss):
+- Attribution (route, job, DB, Redis, storage):
+- Supporting graph or log sample:
+
+## G3 — Blast radius / compliance
+
+- Why a separate deployable (not just more workers):
+
+## G4 — Ownership
+
+- Eng owner:
+- On-call:
+- Budget for dual-run:
+
+## G5 — Contract
+
+- Sync API or async events:
+- Contract test location (planned):
+
+## G6 — Dual-run plan
+
+- Shadow period:
+- Diff/alert strategy:
+
+## G7 — Rollback
+
+- Cutback steps:
+- Drill date (must be practiced once):
+
+## Decision
+
+- [ ] **Do not extract** — pursue in-monolith item: _______________
+- [ ] **Extract** — write service ADR and implementation plan (separate change)
+
+Sign-off: _______________ date: _______________