Backups & snapshots¶

What is backed up: the app configs and databases of the media stack (/volume1/docker: Plex, Sonarr, Radarr, Prowlarr, Bazarr, Seerr, Tautulli, qBittorrent, gluetun, Portainer).

Layer	Protects against	Where	Schedule	Retention
Btrfs snapshots (Snapshot Replication)	Deletion, bad edits, ransomware	NAS	`docker` hourly · `data` daily	more for `docker`, short for `data`
restic → B2	NAS loss, fire, theft	Backblaze B2, encrypted	backup nightly · prune and check weekly	daily / weekly / monthly tiers
App-native	A bad app update	`*/Backups/scheduled`	weekly (*arrs), every 3 days (Plex DB)	the app's defaults

The restic job runs outside the container update window. Exact times and retention counts: see private runbook.

Not backed up off-site: the media (data, about 9 TB). It can be re-acquired, and the *arr databases (backed up) know what to fetch.

What restic includes¶

/volume1/docker, excluding Plex thumbnails/cache/codecs/logs, *arr MediaCover, logs, #snapshot dirs and restic.env. Before each run pre-backup.sh makes consistent SQLite copies (online backup API) of the Sonarr, Radarr, Prowlarr, Bazarr, Seerr and Plex DBs into restic/dbdump/. When restoring a DB, use the dbdump copy; the live files are only crash-consistent. Setup: restic → B2.

Snapshots setup¶

DSM → Snapshot Replication → Snapshots → select the share → Settings: set the schedule and retention (see private runbook). Keep #snapshot hidden on data, so Plex and the *arrs don't scan it.

Restore: one file or one app¶

# from a snapshot (docker share)
ls "/volume1/docker/#snapshot/"            # GMT<offset>-YYYY.MM.DD-HH.MM.SS folders
# from restic
sudo docker exec restic-backup /scripts/r.sh snapshots
sudo docker exec restic-backup /scripts/r.sh restore latest --target /tmp/r --include /source/docker/sonarr
sudo docker cp restic-backup:/tmp/r/source/docker/sonarr /volume1/docker/_restore-sonarr   # inspect, then stop the app and copy into place

For a DB: stop the app, replace <app>.db with the dbdump copy, and delete any stale -wal/-shm files.

Restore after total loss¶

Rebuild DSM, the shares and the users (NAS setup), keeping the same PUIDs (recorded in the private runbook).
On any machine with restic, export the four variables from restic.env (they are in the password manager) and run:
```
restic -r b2:<your-bucket>:<repo-path> restore latest --target ./restore
```
Copy restore/source/docker/* → /volume1/docker/, then put the dbdump DBs in place.
Start Portainer on the restored portainer-ce/ data. Every stack and its env comes back. Redeploy them in the order given.

The restic password is the single point of failure

Without RESTIC_PASSWORD, the B2 data can't be read. Keep it and the B2 key in a password manager and on paper somewhere away from home.

Test restores¶

Quarterly: restore one config file and one dbdump DB, and compare their sha256 with the source. Snapshots: browse one file in #snapshot.