WFGY/ProblemMap/GlobalFixMap/OpsDeploy/db_migration_guardrails.md

DB Migration Guardrails — OpsDeploy

Plan and execute schema or storage migrations with zero data loss and predictable retrieval quality. Applies to primary OLTP stores, vector stores, and search backends.

Open these first

• Boot order and deploy traps: Bootstrap Ordering, Deployment Deadlock, Pre-Deploy Collapse
• Vector index lifecycle: Vector Index Build & Swap
• Version locks: Version Pinning & Model Lock
• Backpressure and retries: Rate Limit Backpressure, Retry Backoff

Acceptance targets

• Data loss equals 0 verified by row counts and checksums
• Backfill completeness equals 100 percent with reconciliation pass
• ΔS drift between pre and post retrieval ≤ 0.02 on a gold set
• λ remains convergent across two seeds
• Cutover within the planned window with a reversible path

60-second checklist

1. Preflight
   Freeze schema in source, create target with explicit charset, collations, index types, vector metric, analyzer. Verify privileges and connection pools.
2. Dual-write flag
   Enable dual writes with idempotency keys. Record op_id, source_rev, target_rev, index_hash.
3. Backfill
   Copy in batches with capped concurrency. Validate counts and checksums per table or collection.
4. Shadow reads
   Use Shadow Traffic Mirroring to compare answers and latency.
5. Cutover
   Flip read source after a clean soak. Keep dual writes for a short tail.
6. Rollback plan
   Document exact steps and TTL for the rollback window with checkpoints.

Minimal playbook

• Contracts: protect schemas with Data Contracts so producers and consumers fail fast.
• Consistency: validate referential and vector invariants after each batch.
• Indexing: for vectors rebuild offline and swap atomically. See Vector Index Build & Swap.
• Traffic: rate limit heavy stages and prefer maintenance windows.
• Metrics: log migrated_rows, checksum_ok, dual_write_fail, ΔS_diff, λ_state, cutover_ms.

Common pitfalls → fix

• Metric mismatch after cutover produces wrong-meaning hits
  → confirm vector metric and normalization. See Embedding ≠ Semantic.
• Read timeouts during backfill
  → isolate reader replicas or use read-only mode during the heaviest phase. See Read-Only Mode.
• Dual-write divergence
  → enforce idempotency keys and reconcile with drift queries. See Idempotency & Dedupe.

Escalate

Abort and rollback if data checksums diverge or ΔS drift exceeds 0.05 on the gold set for longer than 15 minutes. Publish an incident update and run a focused postmortem with Postmortem & Regression Tests.

---

🔗 Quick-Start Downloads (60 sec)

Tool	Link	3-Step Setup
WFGY 1.0 PDF	Engine Paper	1️⃣ Download · 2️⃣ Upload to your LLM · 3️⃣ Ask “Answer using WFGY + <your question>”
TXT OS (plain-text OS)	TXTOS.txt	1️⃣ Download · 2️⃣ Paste into any LLM chat · 3️⃣ Type “hello world” — OS boots instantly

---

🧭 Explore More

Module	Description	Link
WFGY Core	WFGY 2.0 engine is live: full symbolic reasoning architecture and math stack	View →
Problem Map 1.0	Initial 16-mode diagnostic and symbolic fix framework	View →
Problem Map 2.0	RAG-focused failure tree, modular fixes, and pipelines	View →
Semantic Clinic Index	Expanded failure catalog: prompt injection, memory bugs, logic drift	View →
Semantic Blueprint	Layer-based symbolic reasoning & semantic modulations	View →
Benchmark vs GPT-5	Stress test GPT-5 with full WFGY reasoning suite	View →
🧙‍♂️ Starter Village 🏡	New here? Lost in symbols? Click here and let the wizard guide you through	Start →

---

👑 Early Stargazers: See the Hall of Fame —
⭐ WFGY Engine 2.0 is already unlocked. ⭐ Star the repo to help others discover it and unlock more on the Unlock Board.