Initial import NDC_1C

docs/historical_load_plan.md

@@ -0,0 +1,47 @@
+# Historical Load Plan
+
+Date: 2026-03-23  
+Status: executable in MVP mode
+
+## 1. Objective
+
+Build initial canonical baseline for analytics by loading a broad slice of entity sets.
+
+## 2. Preconditions
+
+1. OData access is healthy.
+2. `entity_sets.json` or metadata is available.
+3. Canonical DB is configured (`CANONICAL_DB_URL`).
+
+## 3. Run command
+
+Example:
+
+```powershell
+python scripts/run_refresh.py --mode historical --limit-per-set 500
+```
+
+Alternative explicit set list:
+
+```powershell
+python scripts/run_refresh.py --mode historical --entity-set DocumentSales --entity-set RegisterAccounting
+```
+
+## 4. Acceptance checks
+
+1. `logs/refresh_last_run.json` exists.
+2. Run status is `success` or `partial_success`.
+3. `store_stats.entities_total > 0`.
+4. `checkpoints_total > 0`.
+
+## 5. Recovery rules
+
+- If run is `failed`, fix connectivity/schema issue and rerun same command.
+- If run is `partial_success`, rerun failed entity sets explicitly via `--entity-set`.
+
+## 6. Post-load
+
+After successful baseline load:
+
+1. Freeze closed periods policy in ops notes.
+2. Switch daily operations to incremental mode.