fix(recovery): restore source removed by production documentation merge

2026-11-10 03:42:10 +01:00
parent 585bdd4b44
commit 3d2fc6f913
957 changed files with 115754 additions and 0 deletions
@@ -0,0 +1,41 @@
+-- P014: fills a real, previously-undiscovered gap in the migration chain.
+-- sql/004_iceberg_write_boundary.sql's own comment already documented (but
+-- did not fix) that `opt_dataset_iceberg_mappings.dataset_id REFERENCES
+-- opt_datasets (dataset_id)` points at a table NO sql/ file ever defined --
+-- meaning `psql -d opt < sql/001... < sql/002... < sql/004...` (exactly the
+-- flow docs/runbook/backup-restore.md's own "Målbild" describes, and exactly
+-- what a real restore drill executes) has never actually been runnable
+-- end to end. This was found by running that exact sequence for real against
+-- a throwaway Postgres container during P014's restore drill, not by
+-- inspecting the code.
+--
+-- Numbered 003 (the previously-unused gap between 002_source_registry.sql
+-- and 004_iceberg_write_boundary.sql) so the existing ordering -- 001 -> 002
+-- -> 004 -> 005 -- becomes 001 -> 002 -> 003 -> 004 -> 005 without renumbering
+-- anything already shipped.
+--
+-- Columns mirror src/services/dataset-service.ts's `DatasetRecord` exactly
+-- (that file's own doc comment: "Storage shape: nullable columns, exactly as
+-- the eventual `opt_datasets` table would hold them") -- this is the real
+-- application contract, not a guess. `dataset_type`/`status` CHECK
+-- constraints mirror src/types/dataset.ts's `DatasetType`/`DatasetStatus`
+-- unions verbatim. DatasetService itself remains in-memory-only in this
+-- pass (wiring it to this table is a separate, larger change than closing
+-- the migration-chain gap); this table exists so the SCHEMA a fresh
+-- database gets by replaying sql/*.sql matches what 004/005 already assume.
+
+CREATE TABLE IF NOT EXISTS opt_datasets (
+  dataset_id TEXT PRIMARY KEY,
+  tenant_id TEXT NOT NULL,
+  project_id TEXT NOT NULL,
+  name TEXT NOT NULL,
+  dataset_type TEXT NOT NULL CHECK (dataset_type IN ('raw', 'structured', 'enriched', 'curated', 'feature', 'derived')),
+  status TEXT NOT NULL DEFAULT 'active' CHECK (status IN ('active', 'archived', 'deprecated')),
+  description TEXT,
+  correlation_id TEXT,
+  created_at TIMESTAMPTZ NOT NULL DEFAULT now(),
+  updated_at TIMESTAMPTZ,
+  created_by TEXT
+);
+
+CREATE INDEX IF NOT EXISTS opt_datasets_tenant_project_idx ON opt_datasets (tenant_id, project_id);