Skip to content

feat: prove retained real feature restart recovery - #47

Open
zhouning wants to merge 2 commits into
feat/ar1-metadata-fabric-retained-real-feature-terminal-successfrom
feat/ar1-metadata-fabric-retained-real-feature-restart-recovery
Open

feat: prove retained real feature restart recovery#47
zhouning wants to merge 2 commits into
feat/ar1-metadata-fabric-retained-real-feature-terminal-successfrom
feat/ar1-metadata-fabric-retained-real-feature-restart-recovery

Conversation

@zhouning

Copy link
Copy Markdown
Owner

Summary

  • bind recovery to the checked M3-24 Chongqing 20-feature evidence and retained authority, without re-ingestion or new ResourceVersion/Run creation
  • restart retained PostgreSQL, MinIO, Gravitino and dedicated GDA Control PostgreSQL in order while requiring stable infrastructure identity and rotated process identity
  • independently re-read Iceberg/Parquet, Gravitino and GDA Control; require unchanged content, quality, projection, ledger and exact succeeded@3 replay with zero new facts
  • register the local rehearsal in platform truth, add ADR-071, advance roadmap/system-of-record, and wire evidence/unit/PostgreSQL checks into CI

Live evidence

  • contract SHA: 83ed15ae4eed85e0c261c2b3a04ea2ad559f3deb7b86c7c2f2dedd0cf28d23d0
  • evidence SHA: 1b5a5ceeadee88868bab6237b3f3280c8b13793cc54193592fec7dbbfdd4e8a6
  • evidence file SHA: 6880ff81dcde37f824ab3c7d04f62863375d5a6f1ada2a2dbfa832e77da7cfb1
  • retention ID: m3-24-229740ac50ebb53b
  • expiry: 2026-08-07T04:15:23.082316Z
  • Iceberg snapshot: 8034081021802585202
  • Parquet body SHA: 6cc0fc9eaf48f8106f9afe192704c44407c86c9ea119ae20894bf369a8e74779
  • row-set SHA: c26ff708f4b6be082327dff63a6a8659420dbc4cab37dea1cac7b40f147512df
  • GDA Control facts SHA: 5c0b8a58729e551c250b0410bcebdfe3f019f215a50b20d503f094fe1562d8b5
  • final Run remains succeeded@3; exact terminal replay returns promotion_created=false

All three Kubernetes Pod UIDs and the control PostgreSQL PID rotated. Namespace, StatefulSet, Service, PVC, image, control container and volume identities remained stable. The retained resources remain available for audit and must not be cleaned early.

Production boundary

This proves local process restart continuity for one retained real-data authority on a single development host. It does not prove backup/PITR, independent failure domains, protected production identity/storage/tenant attestation, persistent production scheduler/executor, production restart recovery or production_ready; all remain false.

Verification

  • M3-25 evidence validator: valid
  • platform truth validator: valid
  • required-platform suite: 903 passed, 9 warnings
  • M3-25 plus M3-24 focused suite: 14 passed
  • M3-25 plus platform truth suite: 18 passed
  • PostgreSQL reconciliation/promotion/terminal suites: 7 passed
  • Ruff and git diff --check
  • frozen implementation/script/evidence SHA checks

Stacked on #45.

Closes #46

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant