Backups that actually restore

12 September 2026 · 6 min read

For years my backup strategy was a script that copied things somewhere and printed “done”. It never occurred to me to try getting the data back until the day I needed it. Half of it was there.

The setup now

Every night a single cron job runs restic against the folders that matter, then pushes the repository to object storage with rclone. Nothing clever:

restic backup /srv/data /etc --exclude-caches
restic forget --keep-daily 7 --keep-weekly 5 --keep-monthly 12 --prune

The part that matters

Once a month a second job restores a random snapshot into a scratch directory and compares checksums against the live copy. If the numbers do not match, I get an email. It has fired twice, both times because of a disk that was quietly dying.

A backup you have never restored is a hope, not a backup.