We can confirm that the volume was mounted during the redeploy and PostgreSQL performed crash recovery on the existing data directory rather than running initdb. The recovery logs show WAL replay completed at a very low LSN position (around 67 MB into the WAL stream), which is consistent with what you're seeing now (11 MB, schema only) but would be unusually small for months of production ERP data. This suggests the volume may have already been in this state before the redeploy occurred. We do not retain automatic infrastructure-level volume snapshots or WAL archives, and the [Backups feature](https://docs.railway.com/volumes/backups) is opt-in and must be configured per volume. Please check your service's Backups tab to see if any scheduled or manual backups exist that could be restored.