Hi everyone, I need urgent guidance for a production data recovery situation on Railway.

Incident:

- We run a Python service using SQLite on a mounted Railway Volume.

- A disconnect callback path mistakenly deleted rows from credit_ledger (should have been anonymized, not deleted).

What we already did:

- Patched app logic to anonymize ledger rows instead of deleting.

- Added a DB safeguard trigger to prevent physical deletion of credit_ledger rows.

- Took in-container file backups before/after recovery attempts.

- Ran best-effort recovery by rebuilding from iap_orders references.

Current state:

- credit_ledger had many expected historical rows, but now only a very small subset is available.

- Best-effort app-level reconstruction restored only one row.

- We need older historical rows that are not reconstructable from current tables.

Question:

- Is there any way on Railway to restore an older snapshot/version of a mounted volume (or a single file) to a separate location for forensic merge?

- If full snapshot restore is not available, is there any platform-assisted path to recover prior file versions for SQLite-backed volume data?

- Recommended safest workflow for point-in-time recovery without overwriting current production data?