Docker image to backup your Docker container volumes
Why the name? Docker + Backup = Dockup
Instead of backing up volumes you can also run tasks that provide the files to be backed up. See the following projects as examples on building on Dockup for that purpose:
- wetransform-os/dockup-mongo - Uses
mongodump
andmongorestore
to backup and restore a MongoDB instance
You have a container running with one or more volumes:
$ docker run -d --name mysql tutum/mysql
From executing a $ docker inspect mysql
we see that this container has two volumes:
"Volumes": {
"/etc/mysql": {},
"/var/lib/mysql": {}
}
Launch dockup
container with the following flags:
$ docker run --rm \
--env-file env.txt \
--volumes-from mysql \
--name dockup wetransform/dockup:latest
The contents of env.txt
being something like:
AWS_ACCESS_KEY_ID=<key_here>
AWS_SECRET_ACCESS_KEY=<secret_here>
AWS_DEFAULT_REGION=us-east-1
BACKUP_NAME=mysql
PATHS_TO_BACKUP=/etc/mysql /var/lib/mysql
S3_BUCKET_NAME=docker-backups.example.com
S3_FOLDER=mybackups/
RESTORE=false
DOCKUP_COMPRESS=true
dockup
will use your AWS credentials to create a new bucket with name as per the environment variable S3_BUCKET_NAME
, or if not defined, using the default name docker-backups.example.com
. The paths in PATHS_TO_BACKUP
will be tarballed, gzipped, time-stamped and uploaded to the S3 bucket.
To place backups in a specific folder in the S3 bucket, provide it in the S3_FOLDER
variable.
It should either be empty or hold a path and end with a slash.
For more complex backup tasks as dumping a database, you can optionally define the environment variables BEFORE_BACKUP_CMD
and AFTER_BACKUP_CMD
.
Although dockup was initially designed to specifically support backups so AWS S3, you can also use different kinds of backup targets.
This is possible via the environment variable DOCKUP_MODE
.
It allows running custom logic to do the actual backup/restoration (see subfolders in scripts/
).
By default, the following modes are available:
aws
- store in AWS S3local
- store in a local folder (e.g. in a volume mounted into the dockup container, defaults to/dockup/target
)
When using local mode you may want to only keep a certain amount of backup archives in the target folder.
You can configure the environment variable LOCAL_CLEANUP_OLDER_HOURS
to a number of hours.
The backup script will delete any backup archives in the target folder that are older than this.
It will however not the delete the latest archive, even if it is older.
Please note that the archives are identified only by the file prefix of the BACKUP_NAME
, so you should make sure to not have any other files in there that could have the same prefix, to avoid any unintended file deletion.
Instead of providing paths manually you can set the PATHS_TO_BACKUP
to auto
.
Using this setting the backup script will try to the detect the volumes mounted into the running backup container and include these into the backup archive.
If you want dockup
to run the backup as a cron task, you can set the environment variable CRON_TIME
to the desired frequency, for example CRON_TIME=0 0 * * *
to backup every day at midnight.
Sometimes creating the TAR archive may fail, often due to modifications to the files while tar
is running.
If this happens very often, you should consider using a different option than creating TAR archives for backup.
The BEFORE_BACKUP_CMD
and AFTER_BACKUP_CMD
environment variables can help with that.
If this happens seldomly and you want to avoid a backup failing due to that, you can configure Dockup to retry creating the archive if it fails. For that, use the following environment variables:
- BACKUP_TAR_TRIES - maximum number of tries for the backup (defaults to
5
) - BACKUP_TAR_RETRY_SLEEP - number of seconds to wait between retries (defaults to
30
)
To restore your data simply set the RESTORE
environment variable to true
- this will restore the latest backup from S3 to your volume. If you want to restore a specific backup instead of the last one, you can also set the environment variable LAST_BACKUP
to the desired tarball name.
For more complex restore operations, you can define a command to be run once the tarball has been downloaded and extracted using the environment variable AFTER_RESTORE_CMD
.
You can use GnuPG to encrypt backup archives and decrpyt them again when you need to restore them. You need a GnuPG public key for encryption and the corresponding private key for decryption. Keep the private key safe (and secret), otherwise you will not be able to restore your backups.
For backup, the following environment variables need to be set:
- GPG_KEYRING - the location of the public keyring containing the public key you want to use for encryption
- GPG_KEYNAME - the user ID identifying the key
For restoring an encrypted file, the following environment variables need to be set:
- GPG_KEYRING - the location of the public keyring
- GPG_SECRING - the location of the secret keyring containing the private key you need for decryption
- GPG_PASSPHRASE - the passphrase needed to access the private key
To enable notifications for backups you can use the following environment variables:
- NOTIFY_BACKUP_SUCCESS - set to
true
to enable notifications on backup success - NOTIFY_BACKUP_FAILURE - set to
true
to enable notifications on backup failure
In addition, you need to configure a notification method.
Currently supported are the following notifications methods:
To configure Slack notifications you need to set at least the NOTIFY_SLACK_WEBHOOK_URL
environment variable.
Create an Incoming Webhook as a new integration in Slack and put the Webhook URL in here.
There is a handy script ./test-backup.sh
you can use for local testing.
All you need is Docker and configuring your S3 connection.
For that purpose, copy test-env.txt.sample
to test-env.txt
and adapt the variables accordingly.
Optionally generate a GPG key for testing encryption/decryption using ./gen-test-key.sh
.
It will be automatically used when you execute ./test-backup.sh
.
If you want to test w/o encryption after generating the key, rn ./test-backup.sh --no-encryption
.
Bucket naming guidelines: "Bucket names must be unique and should be DNS compliant. Bucket names can contain lowercase letters, numbers, hyphens and periods. Bucket names can only start and end with a letter or number, and cannot contain a period next to a hyphen or another period."
These rules are enforced in some regions.
Region name | Region |
---|---|
US Standard | us-east-1 |
US West (Oregon) | us-west-2 |
US West (N. California) | us-west-1 |
EU (Ireland) | eu-west-1 |
EU (Frankfurt) | eu-central-1 |
Asia Pacific (Singapore) | ap-southeast-1 |
Asia Pacific (Sydney) | ap-southeast-2 |
Asia Pacific (Tokyo) | ap-northeast-1 |
South America (Sao Paulo) | sa-east-1 |
To perform a restore launch the container with the RESTORE variable set to true