Backfill MongoDB data into PostgreSQL and verify ongoing parity
- Dominant language
- JavaScript
- Stars
- 30
- Forks
- 9
- PR merge metrics
- No merged PRs in 30d
Description
Part of #199.
## Goal
Turn PostgreSQL dual writes into a trustworthy migration target by copying historical MongoDB data and continuously checking parity.
For the competition-room foundation, this backfill must target the new Room/RaceSession model rather than preserving the current assumption that a room is itself one racing session.
## RaceSession mapping
- Convert each existing MongoDB room into one PostgreSQL Room plus one RaceSession for its current event.
- Attach every embedded attempt to that RaceSession while preserving deterministic attempt and solve IDs.
- Populate session participant state according to the approved domain contract.
- Preserve room identity, access code, ownership, privacy, timestamps, and deletion/expiration state.
- Make the mapping explicit and repeatable; do not attempt to infer event sessions whose attempts were already deleted by the legacy event-change behavior.
- Verify concurrent live writes are neither lost nor duplicated during the backfill.
## Acceptance criteria
- [ ] Implement an idempotent, resumable backfill for users/preferences, rooms, race sessions, participant state, attempts, durable solve results, and sanitized metrics.
- [ ] Support dry-run, bounded batches, progress checkpoints, and safe retries.
- [ ] Do not copy OAuth access tokens or prohibited metric fields.
- [ ] Produce reconciliation output for record counts, missing identifiers, duplicate identifiers, room/session relationships, penalty fields, and representative aggregates.
- [ ] Verify every attempt belongs to the expected RaceSession and every RaceSession belongs to the expected Room.
- [ ] Verify writes arriving during the backfill are not lost or duplicated.
- [ ] Document operational invocation, expected runtime/load, alerts, failure recovery, and how to rerun only failed ranges.
- [ ] Add focused tests for deterministic IDs, cursor boundaries, retry behavior, session mapping, and reconciliation failures.
## Dependencies
- Stable production dual writes from PR #171, deployed through #176.
- The Room/RaceSession domain contract and expansion schema from #200 and #203.
- The schema and migration history must remain compatible with rollback requirements.
## Unlocks
- #202
- #174
- #160
Contributor guide
Assessment
This issue has not been assessed yet.