That typically indicates a hardware fault in the underlying physical host that your database Machine was running on. Usually, you can restore from the most recent volume snapshot:
https://community.fly.io/t/no-databases-in-postgres-after-restoring-from-a-snapshot/21432
https://fly.io/docs/postgres/managing/backup-and-restore/
I think the default retention period for snapshots in such cases is now a generous 60 days, 
, giving you ample time to catch up.
Hope this helps!