This repository provides a Docker image for borgmatic, a simple and efficient backup tool based on Borgbackup. The image is designed to make it easy to set up and run borgmatic (with Borg and optionally Cron daemon) within a Docker container, enabling you to streamline your backup process and ensure the safety of your data.
Warning As of 2023-06-23 msmtp and ntfy flavors have been discontinued. This image has now switched to apprise.
Warning Secrets will be implemented differently from October 2024. From
*_FILE
toFILE__*
Before proceeding, ensure that you have Docker installed and properly configured on your system. Refer to the Docker documentation for installation instructions specific to your operating system. If you want to use docker-compose, you may also need to install it seperately. Alternatively, you can also use podman to run this image.
Run this command to create data directories required by this image under your prefered directory.
mkdir data/{borgmatic.d,repository,.config,.ssh,.cache}
Configure a copy of borgmatic's config.yaml in data/borgmatic.d
and run the container. You can modify any of the host mount point to fit your backup configuration.
docker run \
--detach --name borgmatic \
-v /home:/mnt/source:ro \
-v ./data/repository:/mnt/borg-repository \
-v ./data/borgmatic.d:/etc/borgmatic.d/ \
-v ./data/.config/borg:/root/.config/borg \
-v ./data/.ssh:/root/.ssh \
-v ./data/.cache/borg:/root/.cache/borg \
-v ./data/.state/borgmatic:/root/.local/state/borgmatic \
-e TZ=Europe/Berlin \
ghcr.io/borgmatic-collective/borgmatic
See Other usage methods below for more options.
When you run the container for the first time, you'll need to execute into the container and run a command to initialize the repository in the directory you've specified in your docker configuration:
docker exec -it borgmatic /bin/sh
borgmatic init --encryption repokey
In addition, it may be a good idea to manually perform a backup to ensure everything performs as expected:
docker exec -it borgmatic /bin/sh
borgmatic --stats -v 1 --files
Both these commands will use the borgmatic.d/config.yaml
file you provided, along with the BORG_PASSPHRASE
and other environment variables in your docker configuration.
Note/Gotcha for archive names: By default borgmatic uses
{hostname}
for naming (and then pruning, compacting archives). However the docker containers hostname changes every time it's rebuilt. To ensure consistent naming across archives and a properly working prune/compact you should specifically set the archive name in the config.yaml e.g.archive_name_format: 'my-pc-backup-{now:%Y-%m-%d-%H%M%S}
.
The following volumes are available for mounting:
Volume | Description |
---|---|
/mnt/source |
Your data you wish to backup. For some safety you may want to mount read-only. borgmatic is running as root so all files can be backed up. |
/mnt/borg-repository |
Mount your borg backup repository here. |
/etc/borgmatic.d |
Where you need to create crontab.txt and your borgmatic config.yml |
/root/.borgmatic |
Note this is now redundant and has been deprecated, please remove this from your configs |
/root/.local/state/borgmatic |
Here are the state files for periodic checks. |
/root/.config/borg |
Here the borg config and keys for keyfile encryption modes are stored. Make sure to backup your keyfiles! Also needed when encryption is set to none. |
/root/.ssh |
Mount either your own .ssh here or create a new one with ssh keys in for your remote repo locations. |
/root/.cache/borg |
A non-volatile place to store the borg chunk cache. |
To generate an example borgmatic configuration, run:
docker exec borgmatic \
bash -c "cd && borgmatic config generate -d /etc/borgmatic.d/config.yaml"
You can set the following environment variables:
Variable | Description |
---|---|
TZ |
Time zone, e.g. TZ="Europe/Berlin"' . |
BORG_RSH |
SSH parameters, e.g. BORG_RSH="ssh -i /root/.ssh/id_ed25519 -p 50221" |
BORG_PASSPHRASE |
Repository passphrase, e.g. BORG_PASSPHRASE=DonNotMissToChangeYourPassphrase |
BACKUP_CRON |
Cron schedule to run borgmatic. Default:0 1 * * * |
RUN_ON_STARTUP |
Run borgmatic on startup. e.g.: RUN_ON_STARTUP=true |
DOCKERCLI |
Install docker client executable to manipulate (start/stop) containers before, or after backup. See here for a detailed explanation. |
You can also provide your own crontab file. If data/borgmatic.d/crontab.txt
exists, BACKUP_CRON
will be ignored in preference to it. In here you can add any other tasks you want ran
0 1 * * * PATH=$PATH:/usr/local/bin /usr/local/bin/borgmatic --stats -v 0 2>&1
Beside that, you can also pass any environment variable that is supported by borgmatic. See documentation for borgmatic and Borg and for a list of supported variables.
You can set any environment variable from a file by using a special prepend FILE__
.
As an example:
-e FILE__BORG_PASSPHRASE=/run/secrets/mysecretvariable
Will set the environment variable BORG_PASSPHRASE
based on the contents of the /run/secrets/mysecretvariable
file.
It is important to know that this environment variable is not simply available via docker (compose) exec borgmatic sh
but only for the automatic call via the defined cron.
If you want to initialize a repository manually or start a backup outside of the cron job, proceed as follows:
- Initialize repository
docker exec borgmatic /bin/sh -c 'export BORG_PASSPHRASE=$(cat /run/s6/container_environment/BORG_PASSPHRASE) && borgmatic init --encryption repokey'
- Trigger manual backup
docker exec borgmatic /bin/sh -c 'export BORG_PASSPHRASE=$(cat /run/s6/container_environment/BORG_PASSPHRASE) && borgmatic create --stats -v 0'
To enhance your experience with Borgmatic, we'll show you a quick example of how to use Apprise for notifications. Apprise is a versatile tool that integrates with a variety of services and is built into Borgmatic. With the upcoming version 1.8.4 also natively. Here's a quick example of how you can use Apprise.
In an unmodified Borgmatic installation, your cronjob.txt
might look something like this:
0 1 * * * /usr/local/bin/borgmatic --stats -v 0 2>&1
To incorporate Apprise notifications, you can modify it like this:
*/5 * * * * PATH=$PATH:/usr/local/bin /usr/local/bin/borgmatic --stats -v 0 > /tmp/backup_run.log
Add the following lines to your Borgmatic configuration file (config.yaml
):
before_backup:
- echo "Starting a backup job."
after_backup:
- echo "Backup created."
- apprise -vv -t "✅ SUCCESS" -b "$(cat /tmp/backup_run.log)" "mailtos://smtp.example.com:[email protected]&pass=YourSecurePassword&[email protected]&[email protected]"
on_error:
- echo "Error while creating a backup."
- apprise -vv -t "❌ FAILED" -b "$(cat /tmp/backup_run.log)" "mailtos://smtp.example.com:[email protected]&pass=YourSecurePassword&[email protected]&[email protected]"
If you don't want to send the log file, you can replace -b "$(cat /tmp/backup_run.log)"
with a custom message like -b "My message"
.
Apprise offers a variety of services to send notifications to, such as Telegram, Slack, Discord, and many more. For a complete list, visit the Apprise GitHub page.
Apprise allows you to notify multiple services at the same time:
after_backup:
- echo "Backup created."
- apprise -vv -t "✅ SUCCESS" -b "$(cat /tmp/backup_run.log)" "mailtos://smtp.example.com:[email protected]&pass=YourSecurePassword&[email protected]&[email protected],slack://token@Txxxx/Bxxxx/Cxxxx"
Starting from version 1.8.4, Borgmatic has native support for Apprise within its configuration. This makes it even easier to set up notifications. Below is how you can add Apprise directly to your Borgmatic config.yaml
.
apprise:
states:
- start
- finish
- fail
services:
- url: mailtos://smtp.example.com:[email protected]&pass=YourSecurePassword&[email protected]&[email protected]
label: mail
- url: slack://token@Txxxx/Bxxxx/Cxxxx
label: slack
start:
title: ⚙️ Started
body: Starting backup process.
finish:
title: ✅ SUCCESS
body: Backups successfully made.
fail:
title: ❌ FAILED
body: Your backups have failed.
And as of borgmatic 1.8.9+, borgmatic's logs are automatically appended to the body
for each notification.
Apprise provides a flexible and powerful way to handle notifications in Borgmatic. Be sure to check out the official Apprise documentation for a full range of options and capabilities.
This image can be used to run borgmatic like a binary by passing the borgmatic command while running the container. It allows you to isolate your system and execute borgmatic commands without directly installing borgmatic on your host system and only keeping persistent data.
To execute borgmatic commands, you can run your container by passing borgmatic subcommands:
docker run --rm -it \
MOUNT_FLAGS_HERE \
ghcr.io/borgmatic-collective/borgmatic \
list
NOTE Replace MOUNT_FLAGS_HERE
placeholder with appropriate mount flags and optionally environment flags. See above for more clues.
This will execute borgmatic list
in your container. The idea is to create symlink to a script which executes this. Now create a new file borgmatic-docker.sh
somewhere like your workspace or home directory.
#!/bin/sh
docker run --rm -it \
MOUNT_FLAGS_HERE \
ghcr.io/borgmatic-collective/borgmatic \
"$@"
Modify the above script as per your needs and copy it's path. Now you can either create a symbolic link to this script or add it as alias.
- Create a symlink to a directory that exists in your PATH variable e.g.:
chmod +x /path/to/script/borgmatic-docker.sh
sudo ln [-s] /path/to/script/borgmatic-docker.sh /usr/local/bin/borgmatic
- Or, to create an alias add this to your
~/.bashrc
or similar file for other shells.
alias borgmatic="sh /path/to/script/borgmatic-docker.sh"
Tip You can view list of available command line options in borgmatic's docs
To keep the container always running for continous backup, you can run it in detached mode. If you do not pass the command, by default it'll start the cron daemon which will run borgmatic at interval set in crontab.txt file.
docker run -d --restart=always \
MOUNT_FLAGS_HERE \
ghcr.io/borgmatic-collective/borgmatic \
If you ever need to run borgmatic manually, for instance to view or recover files, run:
docker exec -it container_id_or_name bash
Then you can run borgmatic
directly within that shell.
Use docker compose for easily management of your borgmatic container. You can also use this image with your existing docker-compose configuration to immediate setup backups for your deployed containers and/or the host.
- Copy
.env.template
to.env
and edit it to your needs.
cp .env.template .env
You will need to configure environment variables for volumes. You can also directly configure docker-compose.yml
file.
Beside these, you can also set other configuration variables in your .env
file. See Environment section for more details.
- Start the container
docker-compose up -d
- To view logs
docker-compose logs -f
If you want to run borgmatic commands using this configuration instead of starting the container as daemon, you can run:
docker-compose run --rm borgmatic borgmatic list
If a container is already running, you can execute borgmatic commands in it by running:
docker-compose exec borgmatic ls
# or to run a shell
docker-compose exec borgmatic bash
- Stop the backup container:
docker-compose down
- Modify volume
/host/mount/location
indocker-compose.restore.yml
file to point to the location where you want to restore your backup. - Run an interactive shell:
docker-compose -f docker-compose.yml -f docker-compose.restore.yml run borgmatic
- Fuse-mount the backup:
borg mount /mnt/borg-repository <mount_point>
- Restore your files
- Finally unmount and exit:
borg umount <mount_point> && exit
.
Tip In case Borg fails to create/acquire a lock: borg break-lock /mnt/repository
In case you are using the container to backup docker volumes used by other containers, you might want to make sure that the data is consistent and doesn't change while the backup is running. The easiest way to ensure this is to stop the affected containers before the backup and restart them afterwards.
There are two ways to achieve the start and stops. The first use Docker CLI. The second Use Docker HTTP-POST API.
You can use the appropriate borgmatic hooks and control the docker engine through the API, or via the docker client labrary (see options) using the hosts docker socket.
Please note that you might want to prefer the *_everything
hooks to the *_backup
hooks, as
after_backup
will not run if the backup fails for any reason (missing disk space, etc.) and
therefore the containers stay stopped.
First mount the docker socket from the host by adding -v /var/run/docker.sock:/var/run/docker.sock
to your run
command or in the volume list of your docker-compose.yml
.
Now, pick one of these two options.
Add the following environment to your docker run command line -e DOCKERCLI='true'
to your run
command or in the enviroment section of your docker-compose.yml
. This is in addition to the above mentioned socket to add.
Now the docker command is available in your container.
Then add the following in your config.yaml:
...
constants:
...
containernames: "container-a container-b container c"
...
before_backup:
- echo {containernames} | xargs -n 1 echo | tac | xargs docker stop
after_backup:
- echo {containernames} | xargs docker start
...
This way all the containers are stopped in reverse order before the backup, and restarted in order after the backup. This way, for instance, you can ensure the back-end gets stopped last and started first.
Note: Make sure you put the names of the containers in a single, quoted string, separated by spaces, as the containernames constant shows.
Then use the following example to create the start/stop hooks in the config.yml
for the containers
that you want to control.
hooks:
before_everything:
- echo "Stopping containers..."
- 'echo -ne "POST /v1.41/containers/<container1-name>/stop HTTP/1.1\r\nHost: localhost\r\n\r\n" | nc local:/var/run/docker.sock 80 > /dev/null && echo "Stopped Container 1" || echo "Failed to stop Container 1"'
- 'echo -ne "POST /v1.41/containers/<container2-name>/stop HTTP/1.1\r\nHost: localhost\r\n\r\n" | nc local:/var/run/docker.sock 80 > /dev/null && echo "Stopped Container 2" || echo "Failed to stop Container 2"'
- echo "Containers stopped."
- echo "Starting a backup."
after_everything:
- echo "Finished a backup."
- echo "Restarting containers..."
- 'echo -ne "POST /v1.41/containers/<container1-name>/start HTTP/1.1\r\nHost: localhost\r\n\r\n" | nc local:/var/run/docker.sock 80 > /dev/null && echo "Started Container 1" || echo "Failed to start Container 1"'
- 'echo -ne "POST /v1.41/containers/<container2-name>/start HTTP/1.1\r\nHost: localhost\r\n\r\n" | nc local:/var/run/docker.sock 80 > /dev/null && echo "Started Container 2" || echo "Failed to start Container 2"'
- echo "Containers restarted."
While the parameters defined in above examples are sufficient for regular backups, following additional privileges will be needed to mount an archive as FUSE filesystem:
--cap-add SYS_ADMIN \
--device /dev/fuse \
--security-opt label:disable \
--security-opt apparmor:unconfined
Depending on your security system, --security-opt
parameters may not be necessary. label:disable
is needed for SELinux, while apparmor:unconfined
is needed for AppArmor.
To init the repo with encryption, run:
docker exec borgmatic \
bash -c "borgmatic --init --encryption repokey-blake2"
Backup Docker using borgmatic - Thank you @modem7