buttersink VS btrbk

for data I rely on my filesystem. Both ZFS and BRTFS support atomic snapshots which need no additional storage space (i.e. copy-on-write), and you can sync those to wherever you need using rsync or whatever. As long as you sync the snapshot you are guaranteed to have a consistent state. You can however also efficiently sync the snapshots themselves to other volumes of the same type. There is e.g. buttersink (https://github.com/AmesCornish/buttersink) which allows you to incrementally sync snapshots over the network, as long as both source and target are a btrfs volume. For restore copy the most recent snapshot back to the new drive and off you go.

For backup I use Buttersink (https://github.com/AmesCornish/buttersink seems abandoned, but still works well on Python2) to sync my snapshots to a separate disk every once in a while.

I've thought about saving snapshots to S3, but I like the idea of being able to randomly access any single file from S3, so some sort of sync (e.g. s3cmd sync) is more attractive.

That's still easier using snapshots and something like btrbk. Snapshot the directory at start, prune if there are too many snapshots (or snapshots get too old).

I'm on Arch, but you might still find it useful: Btrfs snapshots Arch Wiki - Incremental backup to external drive GitHub - btrbk

the smoothest backup tool i have seen for Linux is btrbk works real nice and is customizable for almost all use-cases BTRFS rocks :)

I recommend redundancy and regular verification is you want to insure your data against corruption. If you do that, you can forget about things like ECC. My setup is a NUC server running Ubuntu with a USB3-connected storage drive running BTRFS. I use btrbk to auto-snapshot and auto-replicate via incremental sends to my BTRFS backup drive, and RotKraken to track integrity of the data with a monthly verification run so that I notice corruption in time to correct it.

In pondering my backup strategy, I was wondering if I could use BTRFS snapshots and a backup tool like btrbk, which is a nice integrated snapshot/backup solution I've used happily on desktop Linux. BTRFS needs subvolumes for snapshots, so I couldn't backup the host itself (which wasn't installed with a / subvolume like other distributions I've used), but it could snapshot the VMs and containers, which have their own individual subvolumes. Then btrbk can send that snapshot in an incremental fashion to external storage.

I then installed btrbk and tried to follow the instructions to create snapshots of root and home on the SSD and then send/receive those to the HDD. I mainly used https://github.com/digint/btrbk and https://mutschler.dev/linux/fedora-btrfs-35/, but I don't use luks.

For anyone using btrfs on their system, I heartily recommend btrbk, which has served me very well for making incremental backups with a customizable retention period: https://github.com/digint/btrbk

btrbk is the best solution I know.

I use BTRBK to make and copy the BTRFS snapshots to my HDD. I schedule it to run every 3 hours using a Sytemd unit file through my own script to avoid running the backup at inconvenient moments: