Getting Started

Installing

Many Linux distributions include Borg in their package managers. In Fedora it is referred to as borgbackup. In this case you would install borg by running the following:

$ sudo dnf install borgbackup

Alternately, you can download a precompiled version from Borg Github Releases, which allows you to install Borg as an unprivileged user. You can do so with following commands (they may need to be adjusted to get the latest version):

$ cd ~/bin
$ wget https://github.com/borgbackup/borg/releases/download/1.2.6/borg-linux64
$ wget https://github.com/borgbackup/borg/releases/download/1.2.6/borg-linux64.asc
$ gpg --recv-keys 6D5BEF9ADD2075805747B70F9F88FB52FAF7B393
$ gpg --verify borg-linux64.asc
$ rm borg-linux64.asc
$ chmod 755 borg-linux64

Finally, you can install it using pip:

$ pip install --user borgbackup

Download and install Emborg as follows (requires Python3.6 or better):

$ pip install --user emborg

Or, if you want the development version, use:

$ git clone https://github.com/KenKundert/emborg.git
$ pip install --user ./emborg

You may also need to install and configure either a notification daemon or a mail daemon. This allows errors to be reported when you are not running Emborg in a terminal. More information can be found by reading about the notifier and notify Emborg settings.

Configuring Emborg to Backup A Home Directory

The basic idea behind Emborg is that you place all information relevant to your backups in two configuration files, which allows you to use Emborg to perform tasks without re-specifying that information. Emborg allows you to have any number of setups, which you might want if you wanted to backup to multiple repositories for redundancy or if you want to use different rules for different sets of files. Regardless, you use a separate configuration for each set up, plus there is a common configuration file shared by all setups. You are free to place most settings in either file, whichever is most convenient. All the configuration files are placed in ~/.config/emborg. If you run Emborg without creating your configuration files, Emborg will create some starter files for you. A configuration is specified using Python, thus the content of these files is formatted as Python code and is read by a Python interpreter.

As a demonstration on how to configure Emborg, imagine wanting to back up your home directory in two ways. First, you want to backup the files to an off-site server. Here the expectation is that you would backup once a day on average and you would do so interactively so that you can choose an appropriate time. Second, you have some free space on your machine that you would like to dedicate to recent snapshots of your files. The idea is that you find that you occasionally overwrite or delete files that you just spent time creating, and you want to run local backups every 10-15 minutes so that you can easily recover these files. To accomplish these two things, you need three configuration files.

Shared Settings

The first file is the shared configuration file:

configurations = 'backups snapshots'
default_configuration = 'backups'

This is basically the minimum you can give. Your two configurations are listed in configurations. It could be a list of strings, but you can also give a single string, in which case the string is split on white space. Then you specify your default configuration. In this example backups is to be run interactively and snapshots is to be run on a schedule by cron, so the default is set to backups to make it easier to run interactively.

Configuration for a Remote Repository: backups

The second file is the configuration file for backups:

repository = 'backups:archives'
prefix = '{host_name}-'
encryption = 'keyfile'
passphrase = 'crone excess mandate bedpost'

src_dirs = '~'
excludes = '''
    ~/.cache
    **/*~
    **/.git
    **/__pycache__
    **/.*.swp
'''
exclude_if_present = '.nobackup'

check_after_create = 'latest'
prune_after_create = True
compact_after_delete = True
keep_daily = 7
keep_weekly = 4
keep_monthly = 12
keep_yearly = 2

This configuration assumes that you have a backups entry in your SSH config file that contains the appropriate user name, host name, port number, and such for the server that contains your remote repository. It also assumes that you have shared an SSH key with this server so you do not need to specify a password each time you back up, and that that key is pre-loaded into your SSH agent. The repository is actually in the archives directory on that server, and each back-up archive will be prefixed with your local host name, allowing you to share this repository with other machines.

You specify what to backup using src_dirs and what not to backup using excludes. Nominally both src_dirs and excludes take lists of strings, but you can also specify them using a single string, in which case the strings are broken into individual lines, any blank lines or lines that begin with # are ignored, and then the white space is removed from the front and back of each line.

This configuration file ends with settings that tell Emborg to run check and prune operations after creating a backup, and it gives the desired prune schedule.

This is just an example, and a rather minimal one at that. You should not use it without understanding each of the settings. The encryption setting is a particularly important one for you to understand and set properly. More comprehensive information about configuring Emborg can be found in the section on Configuring.

With this configuration, you can now initialize your repository and use it to perform backups. If the repository does not yet exist, initialize it using:

$ emborg init

Then perform a back up using:

$ emborg create

or simply:

$ emborg

This works because create is the default action and backups is the default configuration.

Then, you can convince yourself it is working as expected by moving a directory out of the way and using Emborg to restore it:

$ mv bin bin-saved
$ emborg restore bin

Configuration for a Local Repository: snapshots

The third file is the configuration file for snapshots:

repository = '/mnt/snapshots/{user_name}'
prefix = '{config_name}-'
encryption = 'none'

src_dirs = '~'
excludes = '''
    ~/.cache
    **/*~
    **/.git
    **/__pycache__
    **/.*.swp
'''
prune_after_create = True
compact_after_delete = True
keep_within = '1d'

In this case the repository is on the local machine and it is not encrypted. It again backs up your home directory, but for this configuration the archives are only kept for a day.

The repository must be initialized before it can be used:

$ emborg -c snapshots init

Here the desired configuration was specified because it is not the default. Now, a cron entry can be created using crontab -e that creates a snapshot every 10 minutes:

*/10 * * * *  emborg --config snapshots --mute create

Once it has run, you can pull a file from the latest snapshot using:

$ emborg -c snapshots restore passwords.gpg

Overdue Backups

Emborg allows you to easily determine when your files were last backed up using:

$ emborg due

However, you must remember to run this command. Emborg also provides emborg-overdue to provide automated reminders. You configure emborg-overdue using a configuration file: ~/.config/emborg/overdue.conf. For example:

default_maintainer = 'me@mydomain.com'
dumper = 'me@mydomain.com'
default_max_age = 36 # hours
root = '~/.local/share/emborg'
repositories = [
    dict(host='laptop (snapshots)', path='snapshots.lastbackup', max_age=0.2),
    dict(host='laptop (backups)', path='backups.lastbackup'),
]

Then you would configure cron to run emborg-overdue using something like:

00 * * * * ~/.local/bin/emborg-overdue --quiet --mail

This runs emborg-overdue every hour on the hour, and it reports any delinquent backups by sending mail to the appropriate maintainer (the message is sent from the dumper). You can specify any number of repositories to check, and for each repository you can specify host (a descriptive name), path (the path to the repository from the root directory, a max_age in hours, and a maintainer. You can also specify defaults for the maintainer and max_age. When run, it checks the age of each repository and sends email to the appropriate maintainer if it exceeds the maximum allowed age.

In this example the actual repository is not checked directly, rather the lastbackup file is checked. This is a file that is updated by Emborg after every back up. This file is found in the Emborg output directory. Every time Emborg runs it creates a log file that can also be found in this directory. That logfile can be viewed directly, or you can view it using the log command:

$ emborg log

Configuring Emborg to Backup an Entire Machine

The primary difference between this example and the previous is that Emborg needs to be configured and run by root. This allows all the files on the machine to be backed up regardless of who owns them. Other than being root, the mechanics are very much the same.

To start, run emborg as root to create the initial configuration files:

# emborg

This creates the /root/.config/emborg directory in the root account and populates it with three files: settings, root, home. You can delete home and remove the reference to it in settings, leaving only:

configurations = 'root'
default_configuration = 'root'

This assumes that most of the settings will be placed in root:

repository = 'backups:backups/{host_name}'
prefix = '{config_name}-'
passphrase = 'western teaser landfall spearhead'
encryption = 'repokey'

src_dirs = '/'
excludes = '''
    /dev
    /home/*/.cache
    /proc
    /root/.cache
    /run
    /sys
    /tmp
    /var
'''

check_after_create = 'latest'
compact_after_delete = True
prune_after_create = True
keep_daily = 7
keep_weekly = 4
keep_monthly = 12

Again, this is a rather minimal example. In this case, repokey is used as the encryption method, which is only suitable if the repository is on a server you control.

When backing up the root file system it is important to exclude directories that cannot or should not be backed up. Those include: /dev, /proc, /run, /sys, and /tmp.

As before you need to initialize the repository before it can be used:

# emborg init

To assure that the backups are run daily, the following is added to /etc/cron.daily/emborg:

#/bin/sh
# Run root backups

emborg --mute --config root create

This is preferred for laptops because cron.daily is guaranteed to run each day as long as machine is turned on for any reasonable length of time.