How I keep digital keepsakes safe in Proxmox, Nextcloud, and restic without relying on one copy
I keep the working copy on Nextcloud and the real backup off the box.
That matters in a digital keepsakes backup setup. Photos, scans, old letters, PDFs, and notes need to stay easy to browse on the machine I use every day, but they also need a copy that survives a dead disk, a botched update, or a clumsy rm -rf at the worst possible time. One copy is not a strategy. It is optimism with a power cable.
Why local access stays on the machine while the real copy lives elsewhere
I use Nextcloud as the place where the files live day to day. That keeps browsing, tagging, sharing, and quick edits painless. If a scan needs renaming or a photo needs moving into a better folder, it happens there. No faff, no special mount, no hunting through backups to find the file I last touched.
The offsite copy is different. I treat it as the recovery copy, not the daily copy. If I have to restore from it, something has already gone wrong. That changes the job. It does not need to be fast for browsing. It needs to be complete, recent enough, and easy to trust.
I also keep the photos, scans, and notes together in one backup set. Splitting them across separate jobs turns restores into a scavenger hunt. When the file tree stays together, I can restore one folder, one album, or one notes directory without guessing where the rest went.
How Proxmox backups, restic, and snapshot retention fit around one Nextcloud box
Proxmox backups cover the VM or container state. That gives me a way back if the guest dies, the config gets mangled, or I want the whole thing back exactly as it was. For a Nextcloud box, that is useful, but I do not treat it as the only backup. A VM image is not a proper answer on its own when the files inside matter more than the machine.
restic handles the file-level copy of the keepsakes data. I point it at the Nextcloud data directory, plus any related paths I actually need, and back that up to storage that is not sitting on the same machine. If the VM explodes, restic still has the files. If restic fails, Proxmox still has the guest state. Two different layers. Less drama.
Snapshot retention needs a bit of discipline. Keep enough versions to catch accidental deletes and bad edits, but do not let old snapshots pile up forever. I set retention so I have recent points to fall back to, then I let older ones expire on purpose. That keeps storage under control and stops backup jobs from growing into a second hobby.
The 3-2-1 backup rule still makes sense here. Three copies, on two different types of storage, with one copy offsite. I do not care whether you do that with a NAS, external disk, object storage, or something else. The point is simple. If the machine, the local disk, and the room all go bad, you still have a way back.
Test one restore for Nextcloud, one for the backup store, and one for a single file
I test three restores.
First, I restore the Nextcloud VM or container from Proxmox backups. That checks the guest image and the ability to bring the service back.
Second, I restore the restic repository or the backup store itself. That checks the thing holding the file copies. If the backup target is broken, the rest of the plan is decorative.
Third, I restore one file, not a whole tree. One photo, one scan, one note. That tells me whether the path, permissions, and version history still work. A full restore that only works in theory is just expensive confidence.
I also check snapshot retention after a few cycles. Old versions should disappear when they are meant to. If they do not, storage fills up and the whole setup starts looking tidy right up until it does not fit on the disk any more.
That is the setup I trust. Nextcloud stays handy. Proxmox backs up the guest. restic keeps the files. Offsite saves the day. One copy never gets the final say.
