I created a script that I dropped into /etc/cron.hourly which does the following:

1. Use rsync to mirror my root partition to a btrfs partition on another hard drive (which only updates modified files).
2. Use btrfs subvolume snapshot to create a snapshot of that mirror (which only uses additional storage for modified files).
3. Moves “old” snapshots into a trash directory so I can delete them later if I want to save space.

It is as follows:

#!/usr/bin/env python
from datetime import datetime, timedelta
import os
import pathlib
import shutil
import subprocess
import sys

import portalocker

DATETIME_FORMAT = '%Y-%m-%d-%H%M'
BACKUP_DIRECTORY = pathlib.Path('/backups/internal')
MIRROR_DIRECTORY = BACKUP_DIRECTORY / 'mirror'
SNAPSHOT_DIRECTORY = BACKUP_DIRECTORY / 'snapshots'
TRASH_DIRECTORY = BACKUP_DIRECTORY / 'trash'

EXCLUDED = [
    '/backups',
    '/dev',
    '/media',
    '/lost+found',
    '/mnt',
    '/nix',
    '/proc',
    '/run',
    '/sys',
    '/tmp',
    '/var',

    '/home/*/.cache',
    '/home/*/.local/share/flatpak',
    '/home/*/.local/share/Trash',
    '/home/*/.steam',
    '/home/*/Downloads',
    '/home/*/Trash',
]

OPTIONS = [
    '-avAXH',
    '--delete',
    '--delete-excluded',
    '--numeric-ids',
    '--relative',
    '--progress',
]

def execute(command, *options):
    print('>', command, *options)
    subprocess.run((command,) + options).check_returncode()

execute(
    '/usr/bin/mount',
    '-o', 'rw,remount',
    BACKUP_DIRECTORY,
)

try:
    with portalocker.Lock(os.path.join(BACKUP_DIRECTORY,'lock')):
        execute(
            '/usr/bin/rsync',
            '/',
            MIRROR_DIRECTORY,
            *(
                OPTIONS
                +
                [f'--exclude={excluded_path}' for excluded_path in EXCLUDED]
            )
        )

        execute(
            '/usr/bin/btrfs',
            'subvolume',
            'snapshot',
            '-r',
            MIRROR_DIRECTORY,
            SNAPSHOT_DIRECTORY / datetime.now().strftime(DATETIME_FORMAT),
        )

        snapshot_datetimes = sorted(
            (
                datetime.strptime(filename, DATETIME_FORMAT)
                for filename in os.listdir(SNAPSHOT_DIRECTORY)
            ),
        )

        # Keep the last 24 hours of snapshot_datetimes
        one_day_ago = datetime.now() - timedelta(days=1)
        while snapshot_datetimes and snapshot_datetimes[-1] >= one_day_ago:
            snapshot_datetimes.pop()

        # Helper function for selecting all of the snapshot_datetimes for a given day/month
        def prune_all_with(get_metric):
            this = get_metric(snapshot_datetimes[-1])
            snapshot_datetimes.pop()
            while snapshot_datetimes and get_metric(snapshot_datetimes[-1]) == this:
                snapshot = SNAPSHOT_DIRECTORY / snapshot_datetimes[-1].strftime(DATETIME_FORMAT)
                snapshot_datetimes.pop()
                execute('/usr/bin/btrfs', 'property', 'set', '-ts', snapshot, 'ro', 'false')
                shutil.move(snapshot, TRASH_DIRECTORY)

        # Keep daily snapshot_datetimes for the last month
        last_daily_to_keep = datetime.now().date() - timedelta(days=30)
        while snapshot_datetimes and snapshot_datetimes[-1].date() >= last_daily_to_keep:
            prune_all_with(lambda x: x.date())

        # Keep weekly snapshot_datetimes for the last three month
        last_weekly_to_keep = datetime.now().date() - timedelta(days=90)
        while snapshot_datetimes and snapshot_datetimes[-1].date() >= last_weekly_to_keep:
            prune_all_with(lambda x: x.date().isocalendar().week)

        # Keep monthly snapshot_datetimes forever
        while snapshot_datetimes:
            prune_all_with(lambda x: x.date().month)
except portalocker.AlreadyLocked:
    sys.exit('Backup already in progress.')
finally:
    execute(
        '/usr/bin/mount',
        '-o', 'ro,remount',
        BACKUP_DIRECTORY,
    )

Borg daily to the local drive then copied across to a USB drive, then weekly to cloud storage. Script is triggered by daily runs of topgrade before I do any updates

• Offline Backup on 2 separate HDD/SSD
• Backup on HDD within my desktop pc
• Backup offsite with restic to Hetzner Storage Box

Borg Backup, whenever I feel like it - usually monthly.

I use external drive for my important data and if my system is borked (which never happen to me) I just reinstall the OS

I use Duplicacy to backup to my TrueNAS server. Crucial data like documents are backed up a second time to my GDrive, also using Duplicacy. Sadly it’s a paid solution, but it works great for me.

Backup? What?

I have a server with a RAID-1 array, that makes daily, weekly, and monthly read only btrfs snapshots. The whole thing (sans snapshots) is sync’d with syncthing to two rPi’s in two different geographic locations.

I know neither raid nor syncthing are “real” backup solutions, but with so many copies of the files living in so many locations (in addition to my phone, laptop, etc.) I’m reasonably confident its a decent solution.

i do backups of my home folder with Vorta, tha uses borg in the backend. I never tried restic, but borg is the first incremental backup utility i tried that doesnt increase the backup size when i move or rename a file. I was using backintime before to backup 500gb on a 750gb drive and if I moved 300gb to a different folder, it would try to copy those 300gb again onto the backup drive and fail for lack of storage, while borg handles it beautifully.

as an offsite solution, i use syncthing to mirror my files to a pc at my fathers house that is turned on just once in a while to save power and disc longevity.

My systems are all on btrfs, so I make use of subvolumes and use brkbk to backup snapshots to other locations.

My KVM hosts use “virsh backup begin” to make full backups nightly.

All machines, including the KVM hosts and laptops, use rsync with --link-dest to create daily incremental versioned backups on my main backup server.

The main backup server pushes client-side encrypted backups which include the latest daily snapshot for every system to rsync.net via Borg.

I also have 2 DASs with 2 22TB encrypted drives in each. One of these is plugged into the backup server while the other one sits powered off in a drawer in my desk at work. The main backup server pushes all backups to this DAS weekly and I swap the two DASs ~monthly so the one in my desk at work is never more than a month or so out of date.

My conclusion after researching this a while ago is that the good options are Borg and Restic. Both give you incremental backups with cheap timewise snapshots. They are quite similar to each other, and I don’t know of a compelling reason to pick one over the other.

Using timeshift. Very, very easy, works great.

Borg to a NAS.

500GB of that NAS is “special” so I then rsync that to a 500GB old laptop hdd, of which is is duplicated again to another 500GB old laptop hdd.

Same 500GB rsync’d to Cloud Server.

All my configs are in gitlab or a self hosted forgejo server and all files are in seafile or a self hosted service running on proxmox. Then I use proxmox backup server on a storage VPS for off-site backup