Files
fm_be/docs/data-migration-wiki.md
2026-07-16 22:16:45 +07:00

95 lines
5.0 KiB
Markdown

# Data Migration Wiki
This page is the living technical document for migration scope, decisions, limitations, and open tasks.
Update it for every migration wave.
For CKEditor copy-paste, use the HTML version: `docs/data-migration-wiki-ckeditor.html`.
## Scope
- Database schema migrations in [`migrations/`](../migrations/).
- Data migration SQL (legacy to current domain model).
- Non-DB migration concerns: binary files (PDF/images), queue backlog, manual user actions.
## Source of Truth
- Versioned DB migrations: [`migrations/*.sql`](../migrations/).
- Legacy helicopter import flow:
- [`migrations/goose/20260414_helicopter_import_staging.sql`](../migrations/goose/20260414_helicopter_import_staging.sql)
- [`migrations/goose/20260415_helicopter_import_apply.sql`](../migrations/goose/20260415_helicopter_import_apply.sql)
- Deprecated helper note: [`migrations/scripts/migration_helicopter.sql`](../migrations/scripts/migration_helicopter.sql)
- Queue migration notes: [`docs/queue-sqs-migration.md`](./queue-sqs-migration.md)
- File manager APIs and attachment model:
- [`docs/file-manager-upload.md`](./file-manager-upload.md)
- [`docs/file-manager-attachment.md`](./file-manager-attachment.md)
## Implemented Data Migrations
1. `20260326_move_responsible_flight_rescuer_to_air_rescuer_profiles.sql`
- Copies `pilot_profiles.responsible_flight_rescuer` into `air_rescuer_profiles.responsible_flight_rescuer`.
- Drops old column from `pilot_profiles`.
2. `20260403_file_manager_file_lifecycle_phase1.sql`
- Normalizes legacy file lifecycle states (`pending/uploading`) into newer states (`validated/processing`) and carries `upload_error` into `failure_reason`.
3. `20260415_helicopter_import_apply.sql`
- Imports rows from `helicopter_old_import` into `helicopters`.
- Maintains old-to-new ID mapping via `helicopter_old_id_map`.
- Explicitly non-reversible (`Down` keeps data as audit trail).
## Known Migration Decisions and Gaps
### Confirmed "Will Not Migrate" (Current Decision)
1. Legacy helicopter columns intentionally ignored:
- `counter`
- `service`
- `disabled`
- `visible`
- Decision is documented in [`migrations/scripts/migration_helicopter.sql`](../migrations/scripts/migration_helicopter.sql).
### Confirmed "Not Migrated Yet" (Open)
1. Legacy SQS queue backlog
- Current worker does not consume old SQS backlog.
- Required action: drain/discard SQS backlog before/at cutover.
- Source: [`docs/queue-sqs-migration.md`](./queue-sqs-migration.md).
2. Binary files (PDF/images) into new file manager storage
- Current migrations create file manager tables and attachment relations, but no SQL migration imports binary objects or bulk-inserts legacy metadata into `file_files`/`attachments`.
- This means legacy documents may need manual re-upload unless a dedicated import job is created.
3. Legacy `photo_file_id` references in contact profile tables
- Contact profile schemas still include `photo_file_id` string columns.
- No migration currently maps these into `attachments` + `image_attachment_id`/`foto_attachment_id`.
## Manual Actions Required at Cutover
1. Validate which legacy PDFs/images must remain accessible on day 1.
2. For files without automated import, perform manual re-upload via file manager flow.
3. Re-link uploaded files to business references (attachments) and to entity-level attachment columns where relevant.
4. Obtain client sign-off for any files intentionally left unmigrated.
## Migration TODO Register
| ID | Item | Type | Status | Owner | Notes |
|---|---|---|---|---|---|
| MIG-001 | Decide final strategy for legacy PDFs/images: automated import vs manual re-upload | Decision | Open | TBD | Blocks final cutover checklist |
| MIG-002 | Produce inventory of all legacy file references and classify critical/non-critical | Analysis | Open | TBD | Include PDFs + photos |
| MIG-003 | If automation chosen, implement importer to populate `file_files` and `attachments` safely | Engineering | Open | TBD | Prefer idempotent batch import with audit log |
| MIG-004 | Define mapping from legacy contact `photo_file_id` to new attachment model | Engineering | Open | TBD | Covers pilot/doctor/air rescuer/technician/FA/staff |
| MIG-005 | Document and execute Redis backlog drain/discard plan before production cutover | Ops | Open | TBD | See queue migration doc |
| MIG-006 | Create client-facing exception list for data that cannot/will not be migrated | Governance | Open | TBD | Must be approved before go-live |
## Cutover Validation Checklist
1. Run DB migrations in target environment.
2. Run all required data migration scripts (including helicopter import flow if applicable).
3. Validate row counts and sample records for migrated domains.
4. Validate file accessibility for mandatory records (especially PDFs needed operationally).
5. Validate queue behavior post-cutover (`pending` to `published`, no legacy Redis dependency).
6. Record migration result, exceptions, and follow-up tickets in this page.
## Change Log
- 2026-04-17: Initial wiki created from current migration scripts and docs.