Skip to content
zfs backup tool
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.

Your prayer, answered.

— Zeus


zeus is a tool for performing backups onto another zfs pool which is physically connected to the machine.

Quick Start

  1. Get zeus:
    go get

  2. Create pool named zbackup.

  3. Mark dataset which you want to backup:
    zfs zeus:backup=on <your-dataset>.

  4. Run zeus like this:
    sudo zeusd backup.


It is possible to use zeus with encrypted filesystems.

Currently, only pools with root filesystem encrypted are supported.

To use zeus with encryption:

  1. Create pool with encrypted root filesystem:
    sudo zpool create zbackup -O encryption=on -O keyformat=passphrase -m none <vdev-configuration>

  2. Provide executable named zfs-encryption-key in your path which will accept single argument with your backup pool name (e.g. zbackup) and will output encryption key to stdout.


    sudo -u operator sh -c "carcosa -p ~/.secrets/my -cG zfs/$1"

Note on holds

zeus will enforce additional protection for lately made snapshots by using zfs hold with zeus tag to avoid accidental deletion which will make further incremental backups impossible.

If you need to delete those snapshots, you always can release this lock by using:
zfs release zeus <snapshot>.

Note on backup filesystem names

By default zeus will receive source snapshots into paths like this:

  • <hostname> part can be configured via config file.
  • guid:<guid> is namespace based on source snapshot GUID which is unique and is required to avoid conflicts when you backup parts of the same hierarchy, like z/home/username/data and z/home/username.


zeus has two ways of configuration:

  • via config file
    • /etc/zeus/zeusd.conf if zeusd is run by root,
    • ~/.config/zeus/zeusd.conf if zeusd is run by user.
  • via zfs properties, namespaced with zeus:.

Config file is used mostly for static configuration and meant to be changed just once for initial configuration.

zfs properties are meant to be used for specifying which datasets you want to backup and fine-tune other backup process parameters.


Config file is written in TOML.

If you want to use different backup pool name you need to change it in config file.

Check out zeus.conf.example for structure of configuration file and options description.

zfs properties

All zeus-related properties are prefixed with zeus:.

Following zfs properties are supported:

  • zeus:backup:

    • on — enable backup on given filesystem,
    • off — disable backup on given filesystem.
  • zeus:housekeeping (default: by-count): specifying which housekeeping policy to apply after backup. Housekeeping is process of cleaning up old snapshots. zeus will attempt to clean up only snapshots managed by zeus.

    • none — do not apply any housekeeping.
    • by-count — clean up snapshots when their amount exceeds specified numbers. See next for more parameters for by-count housekeeping policy.
  • zeus:housekeeping:by-count:keep-on-source (default: 1): specifies how many snapshots keep on given filesystem. At least one snapshot required for incremental backups to work.

  • zeus:housekeeping:by-count:keep-on-target (default: 10): specifies how many snapshots keep on target backup pool. At least one snapshot required for incremental backups to work.


It is possible to test how zeus works without real backup pool.

  1. Create test file-based temporary pool by using:

    truncate --size 10G test.img
    sudo zpool create zbackup `pwd`/test.img -f -m none
  2. Set zeus:backup=on on any of yours filesystems.

  3. Run zeusd backup --no-export.

You can’t perform that action at this time.