Most Linux users find out how good their snapshot tool is at the worst possible moment: after a package upgrade that leaves the shell in a broken state, or after a rogue rm -rf that runs a level too high. The good news is that the tools have quietly gotten very good. openSUSE’s default install now survives a full-system wipe if Snapper’s boot-menu entry survives, and Timeshift on Mint or Ubuntu will restore a system from a live USB in about the time it takes to make coffee.
We tested seven rollback and snapshot apps across Ubuntu, Fedora, openSUSE Tumbleweed, and Arch. The list covers Btrfs-native tools, filesystem-agnostic options, and the alternative approach of rebuilding the system from a declarative configuration.
What matters in a rollback tool
- Boot-menu integration. A snapshot you cannot reach when the kernel refuses to boot is not a snapshot.
- Pre and post transaction hooks. Grabbing a snapshot around every package operation catches the update-that-broke-the-system case.
- Space efficiency. Rolling snapshots on a 250 GB SSD should not eat the whole disk after a week.
- Excludes done right. Never snapshotting
/homeprotects data, and never snapshotting/var/cachesaves gigabytes. - Restore without reboot. Some rollbacks can be applied live, which saves the trip to a rescue USB.
Quick comparison
| App | Best for | Filesystem | Free plan | Notes |
|---|---|---|---|---|
| Snapper | openSUSE, Fedora Silverblue | Btrfs (native) | Free, open-source | Pre/post hooks, boot-menu integration |
| Timeshift | Ubuntu, Mint, Debian | Btrfs or rsync | Free, open-source | GUI-first, restore from live USB |
| btrbk | Btrfs power users | Btrfs (native) | Free, open-source | CLI, cron-driven send/receive |
| Restic | Any filesystem | Any | Free, open-source | Deduplicating, encrypted |
| BorgBackup | Long-term archives | Any | Free, open-source | Deduplicating, compressed |
| Relax-and-Recover | Bare-metal recovery | Any | Free, open-source | Rebuilds bootable ISO from live system |
| NixOS generations | NixOS specifically | ZFS or ext4 | Free, open-source | Every rebuild is a rollback target |
The apps
1. Snapper, the default on openSUSE for a reason
Snapper is the tool that made the “roll the system back with two keys at the boot menu” trick famous. On openSUSE it hooks into zypper, so every package operation gets a paired pre and post snapshot on the Btrfs root subvolume. When something goes wrong, boot into the bootloader, pick the snapshot from the list, and the system reads-only reverts to that point. Confirm the rollback and Snapper promotes it to the new default subvolume.
Where it falls short: Btrfs only. Setup on non-openSUSE distros is fiddly because the subvolume layout matters and most other installers do not lay it out the way Snapper expects. Snapshots do not travel between machines out of the box.
Pricing:
- Free and open-source.
- Ships in openSUSE by default.
Platforms: Linux (Btrfs).
Download: en.opensuse.org/openSUSE:Snapper_Tutorial or github.com/openSUSE/snapper
Bottom line: The gold standard for “the package manager broke my system” recovery on openSUSE and any Btrfs distro willing to adopt the same subvolume layout.
2. Timeshift, the pick for Ubuntu and Mint
Timeshift is the graphical tool the Linux Mint team ships pre-installed. It takes system snapshots on either Btrfs or rsync backends, keeps a rotating set of hourly, daily, weekly, and monthly copies, and its Restore workflow is the least intimidating on this list. From a live Mint or Ubuntu USB you can install Timeshift, mount the target disk, and roll back a broken system in one flow.
Where it falls short: The rsync backend excludes /home by default. You have to remember that a Timeshift restore does not touch personal files, which is the source of most surprise. Very large system trees on rsync take a while for the first snapshot.
Pricing:
- Free and open-source.
- Pre-installed on Linux Mint.
Platforms: Linux (Btrfs and rsync).
Download: github.com/linuxmint/timeshift or your distro’s repos
Bottom line: The right pick on a Mint or Ubuntu box, especially for users who want a rollback tool with a proper graphical restore.
3. btrbk, the CLI tool serious Btrfs users end up on
btrbk is a scripted wrapper around btrfs send and btrfs receive that automates snapshot rotation, retention, and send-to-remote. A cron-driven config file replaces the ad-hoc scripts most Btrfs users end up writing. It scales down to a laptop with a single external disk and up to a small server farm shipping snapshots between hosts.
Where it falls short: No GUI. The config syntax rewards patience. Not the tool to reach for if you have never touched btrfs subvolume.
Pricing:
- Free and open-source.
- No paid tier.
Platforms: Linux (Btrfs).
Download: github.com/digint/btrbk or your distro’s repos
Bottom line: The right choice when Snapper’s opinions clash with your subvolume layout and you already know Btrfs well.
4. Restic, filesystem-agnostic and fast
Restic takes a completely different approach: it encrypts and deduplicates snapshots into an object-storage-friendly repository that lives on a local disk, a NAS, or S3. Nothing depends on Btrfs. Restore is a mount command that surfaces any snapshot as a directory, which is faster to browse than an rsync tree. On modern hardware the initial backup of a 250 GB laptop finishes in about half an hour.
Where it falls short: Not really a boot-menu recovery tool. Restore assumes you can boot a working system to invoke the CLI. Encryption is on by default, which is right, but losing the repository password is fatal.
Pricing:
- Free and open-source.
- No paid tier.
Platforms: Linux, plus Windows, macOS, and BSD builds.
Download: restic.net or your distro’s repos
Bottom line: The best choice when the root filesystem is ext4 or XFS and you want efficient, encrypted snapshots that also cover /home.
5. BorgBackup, the long-term archive companion
BorgBackup is the older cousin of Restic and still popular for the same reason it always was: deduplication and compression that hold up over years of retained snapshots. On a system with slow rotational storage for the backup target Borg tends to end up smaller than Restic for the same corpus. Restore is done with borg extract or by mounting the archive.
Where it falls short: Single-writer repositories, so two machines cannot back up to the same Borg repository at once without an intermediary like Borgmatic. No native cloud target, though rclone bridges close the gap.
Pricing:
- Free and open-source.
- No paid tier.
Platforms: Linux, macOS, BSD.
Download: borgbackup.org or your distro’s repos
Bottom line: The choice for years-long archives where deduplication matters more than fresh installs.
6. Relax-and-Recover (ReaR), the bare-metal option
Relax-and-Recover builds a bootable ISO or PXE image directly from a running Linux system, snapshotting the layout, bootloader, and package manifest so a completely dead machine can be reimaged onto new hardware. It is the tool used to move an entire server from a failing disk to a fresh one without reinstalling. On desktops it is overkill for a package-update rollback, but it is the only option on this list that recovers from a physically dead drive.
Where it falls short: The initial setup is a config-file exercise, not a click. Recovery lands you in a rescue shell before rebooting the fresh system, which is not for beginners.
Pricing:
- Free and open-source.
- No paid tier.
Platforms: Linux (most distros).
Download: relax-and-recover.org or your distro’s repos
Bottom line: The right pick when the risk you are hedging is dead hardware, not a bad update.
7. NixOS generations, rollback as a side effect of the whole design
NixOS treats every system change as a new generation, so the whole “roll back to before that upgrade” workflow is a single bootloader entry pick. Every nixos-rebuild switch produces a new generation, and old ones stay around until you garbage-collect them. There is no separate snapshot tool because the entire system state is derived from a text file.
Where it falls short: Not applicable to any distro that is not NixOS. Migrating to NixOS is a project in its own right. Home directory data still needs a real backup on top.
Pricing:
- Free and open-source.
- No paid tier.
Platforms: NixOS only.
Download: nixos.org
Bottom line: The most bulletproof option, if you are willing to reinstall onto NixOS to get it.
How to pick the right one
- On openSUSE or a Btrfs-first setup → Snapper. It is what the distro is designed around.
- On Ubuntu or Mint with a graphical restore path → Timeshift. Pre-installed on Mint, one apt away on Ubuntu.
- Running Btrfs and comfortable in the terminal → btrbk. More control than Snapper, less opinion.
- On ext4 or XFS and you want
/homecovered too → Restic. Cross-platform, encrypted, deduplicating. - Keeping years of archives → BorgBackup. Deduplication holds up over long retention.
- Preparing for dead-hardware recovery → Relax-and-Recover. Bootable image out of a live system.
- Willing to switch distros for rock-solid rollback → NixOS. Every rebuild is already a rollback point.
For a normal desktop Linux user hedging against a bad package upgrade, the answer is either Snapper on a Btrfs distro or Timeshift on a Btrfs or ext4 distro. Set retention to keep the last 10 hourly and the last four daily snapshots and move on.
FAQ
Does Timeshift back up my home directory?
Not by default. Timeshift’s default include list is /, /usr, and /etc with /home explicitly excluded. The reasoning is that a system rollback should not destroy personal files. Enable /home inclusion in the settings only if you understand that a restore then reverts documents alongside the OS.
Can Snapper roll back a system from the boot menu on any distro?
It can, but only if the distro sets up the Btrfs subvolume layout Snapper expects and installs the snapper-plugins package that adds boot-menu entries. openSUSE and Garuda Linux do this for you. On Fedora or Arch you can do it manually, but it is a deliberate reinstall exercise.
What is the fastest way to recover from a fatal Linux update?
If Snapper or Timeshift with a Btrfs backend is set up, boot into the bootloader, pick the last known-good snapshot, and confirm. Total time is under two minutes. Without either, boot a live USB, mount the target disk, and restore from the most recent Restic or Borg snapshot. Budget half an hour.
Do snapshot tools protect against ransomware?
Only if the snapshots live somewhere the compromised system cannot write to. Snapper and Timeshift on the same disk are vulnerable to malware that deletes snapshots. Send snapshots to an external disk with btrbk, or push to a Restic or Borg repository on a NAS with read-only append-only credentials for the client, and the ransomware cannot delete the history.
Are there GUI options besides Timeshift?
Snapper has snapper-gui, though it is minimal. KDE users can look at Kalpa Backup and the Discover snapshot integration. Everything else on this list is terminal-only, which is why Timeshift dominates the graphical category.