I’ve only recently discovered Borg (with Vorta), and have started using it as my main backup system in place of rsync.
The benefits are clear, in terms of automation, restoration, and disk size.
But I’m holding off taking the final step, which is to have my external backup drive contain nothing but copies of the Borg files (from my internal backup drive).
What’s stopping me is a feeling that the files as copied over by rsync are somehow more ‘real’ than those made by Borg.
Am I being daft? It’s all just ones and zeroes in the end, and everything needs some software to interpret it.
Or is there something to 'real’ copies that makes a clone-type backup something of added value to a Borg-type backup?
I am using Vorta/Borg backup for my private weekly backup on external USB-Drive. It is fast and deduplicating safes a lot of disk space.
As I am a Gnome-fan I also tried deja-dup. From the descroption I thought it would be the same in Gtk - but it wasn’t. So, I went back to Vorta.
I have already tested restoring data from the disk - worked. Another nice feature: you can also mount a “snapshot” into your filesystem and read files from that directly.
Those are definitely among the reasons I’ve been trying it. Does it worry you that you need borg to open those archives again? One advantage of rsync that I keep coming back to is that I could theoretically take the external drive to any computer at all (doesn’t even have to be one of mine) and access my files.
It’s a weakness of mine to spend ages trying to find the ‘solution’ in situations where there is no obvious right answer. Oh well…
Maybe borg on the internal backup and rsync on the external one makes most sense after all.
How often do you expect you will be restoring a backup? I personaly use borg with borgmatic. This means that during a total system restore I just have to install a borgmatic package and copy its config to /etc/… and then run borgmatic mount if I want to hand pick some files. Vorta probably allows you to do something like that as well.
But my use case is not a full-system-snapshot restore. I just backup my “home” files because the system files can be easily reinstalled from arch repos.
Failure rates on external drives are very high unless they are based on laptop drives due to all the heat from the enclosures.
But maybe the bigger reason is I want my backups to “hands off” except for my annual tests. With external drives you have to rotate them out by hand and manually have at least one at a different location. I used to use hard drives as part of my backup strategy and it was so much more work.
My backups are fully automated. At all times I have at least 3 copies of my data with 1 copy being offsite.
Makes sense. I use SSDs or NVMe in icy box enclosures, which seems to work, but I haven’t really battle tested them yet.
But I have a second hard drive in the machine for regular backups.
One system I find unexpectedly great for backups is Syncthing. More than a few times I’ve panicked about having overwritten some important document only to be relieved to remember there’s a copy on my laptop and phone because of Syncthing.
Syncthing is no backup, but a real great tool to keep data in sync (music, movies, settings, …) This of course could be thought of a “backup”, because it also works remotely with minimal overhead.
It has flawlessly synced literally several million files over 12+ years between 10+ machines here and remote without a single glitch, even when the Syncthing versions weren’t the same on all machines. A fantastic piece of software! “Unexpectedly great” indeed.
Of course you need to check the sync conflict files (*.sync-conflict-*) once in a while, but at least it keeps the “before” and “after” so you can select which is the valid one.
Oh, Borg backups. Yes, I trust them. Still using other means most of the time (TrueNAS, rsync, Grsync, Syncthing).
I personally have been using Pika backup which is based on Borg and have found it to be really easy and straightforward to use. Backups can be scheduled, encrypted with a history of archives
Just out of curiosity how long does it take for borg to create one archive of this size. I have experience only with small-ish ~10GB archives. Does the time scale linearly or exponentially?
I found this thread after searching for Pika Backup after watching a video on YT extolling its virtues. I have syncthing running OK, which keeps my documents/photos synced with an Unraid share.
I’ve been using URBackup hosted on my Unraid server for over a year backing up our Windows (yuk) clients without problems. But I can not get URbackup client working on my EnOS laptop so I’m looking at other options. Is Pika a good choice for both file/image backups? Or are there other easy options out there.
So here’s an alternate viewpoint - I used to use Vorta and Borg, but needed my backups to be platform independent. So I switched a 3-2-1 backup strategy, but with rsync instead, and version control via NAS, alongside a cloud backup (encrypted) for resilience.
Vorta & Borg has its use cases, and I can see where you’re coming from. If I didn’t have a Mac in the mix, I’d absolutely be standarding on borg, - but, I would keep a separate airgapped rsync-based backup on a luks drive. Having all my eggs in one backup basket, it doesn’t sit right with me.
Without dragging things off topic, I’ve used Vorta/Borg for many years for my home directory, not system files, and it has saved me repeatedly. Kept on external USB drive sitting nearby, which is admittedly a bit dangerous. (Should probably further refine my storage tactics…)
But I use timeshift for my system files only. Also on that same external USB drive. Mmmm, yeah, probably need some refinement to being reliant in that external hard drive for backups…