Repair live canonical migrations and scanner cache bootstrap

docs/implplan/SPRINT_20260309_014_Platform_live_runtime_fault_repair.md

@@ -0,0 +1,74 @@
+# Sprint 20260309_014 - Live Runtime Fault Repair
+
+## Topic & Scope
+- Repair live backend/runtime faults uncovered after the full 60-image rebuild and fresh `stella-ops.local` redeploy.
+- Keep the rebuilt stack client-ready underneath the clean UI shell by fixing background workers, runtime contracts, and hardened-container assumptions instead of hiding errors behind empty states.
+- Working directory: `src/Platform/**`.
+- Cross-module edits allowed for this sprint: `src/JobEngine/**`, `src/Concelier/**`, `src/Scanner/**`, `devops/compose/**`, and linked docs in `docs/**`.
+- Expected evidence: targeted `.csproj` test runs, live API verification, live Playwright rechecks on impacted routes, and runtime log validation after redeploy.
+
+## Dependencies & Concurrency
+- Depends on the scratch rebuild baseline and live search runtime repair from `SPRINT_20260309_013_AdvisoryAI_live_unified_search_corpus_runtime_repair.md`.
+- Safe parallelism: avoid unrelated web/search feature edits already in flight from other agents; stage only the runtime-fault hunks touched here.
+
+## Documentation Prerequisites
+- `docs/modules/platform/architecture-overview.md`
+- `docs/modules/jobengine/architecture.md`
+- `docs/modules/concelier/architecture.md`
+- `docs/modules/scanner/architecture.md`
+- `docs/code-of-conduct/CODE_OF_CONDUCT.md`
+- `docs/code-of-conduct/TESTING_PRACTICES.md`
+
+## Delivery Tracker
+
+### TASK-014-001 - Diagnose live runtime failures from rebuilt stack
+Status: DONE
+Dependency: none
+Owners: QA, Developer
+Task description:
+- Rebuild all services, redeploy the compose stack, then inspect live route behavior and backend logs to identify runtime faults that survive basic page rendering.
+
+Completion criteria:
+- [x] Full image matrix rebuild completed.
+- [x] Fresh compose recreate completed.
+- [x] Live evidence captured for runtime faults and impacted routes.
+
+### TASK-014-002 - Repair scheduler and analytics runtime contract faults
+Status: DONE
+Dependency: TASK-014-001
+Owners: Developer
+Task description:
+- Fix PostgreSQL type/function mismatches causing scheduler planner loops and platform analytics maintenance to fail after startup.
+
+Completion criteria:
+- [x] Scheduler planner queries no longer emit `run_state = text` errors.
+- [x] Platform analytics maintenance invokes `analytics.compute_daily_rollups` with the correct PostgreSQL parameter type.
+- [x] Focused tests prove the repaired contracts.
+
+### TASK-014-003 - Repair canonical advisory DI and scanner cache runtime assumptions
+Status: DONE
+Dependency: TASK-014-001
+Owners: Developer
+Task description:
+- Restore Concelier canonical advisory service registration under the live WebService and align scanner cache paths with writable hardened-container storage so maintenance jobs stop failing after deploy.
+
+Completion criteria:
+- [x] `/api/v1/canonical` resolves through registered services without runtime DI failure.
+- [x] Scanner cache maintenance no longer writes into read-only `/app` paths in live containers.
+- [x] Focused tests and live verification cover the repaired contracts.
+
+## Execution Log
+| Date (UTC) | Update | Owner |
+| --- | --- | --- |
+| 2026-03-09 | Sprint created after full rebuild/redeploy exposed live runtime faults in scheduler planning, platform analytics maintenance, Concelier canonical DI, and scanner cache maintenance. | Codex |
+| 2026-03-09 | Repaired scheduler enum/query typing and platform analytics date binding; focused `.csproj` verification passed and post-redeploy logs stopped emitting the runtime faults. | Codex |
+| 2026-03-09 | Added Concelier startup migration registration, fixed Scanner worker env-prefix bootstrap, and introduced compose cache ownership bootstrap; focused tests passed, `/api/v1/canonical` returned `200`, cache paths resolved to `/var/lib/stellaops/cache/scanner`, and live Playwright rechecks passed (`111/111` routes, changed-surfaces pass). | Codex |
+
+## Decisions & Risks
+- This sprint intentionally treats background worker failures as product defects even when the frontdoor UI still renders. A clean route sweep is insufficient if the live services are erroring underneath.
+- Cross-module edits are permitted because the faults span runtime contracts across Platform, JobEngine, Concelier, Scanner, and compose deployment wiring.
+- Microsoft Testing Platform projects in this sprint require `dotnet test <project>.csproj -- --filter-class ...`; `--filter` against the project silently ran whole suites and was rejected as verification evidence.
+- Hardened Scanner containers need both a writable cache root and ownership bootstrap. The compose stack now uses `scanner-cache-init` to prepare the named volume for the non-root runtime user.
+
+## Next Checkpoints
+- Targeted repair commit once runtime faults are fixed, revalidated live, and staged without unrelated agent changes.