How to restore a controller from a backup
The restore tool is useful for rolling back to a previous backup, for example, if a
juju upgrade-model failed for some reason. It must be run on the primary controller. At a high level, it performs the following actions:
- Run various pre-checks, for example, ensuring the backup version matches the controller version
- Confirm with the user they definitely want to proceed
- Stop the controller agent
- Reinstate the model data in MongoDB using
mongorestore(already installed on the controller)
- Restart the controller agent
The tool works on high-availability (HA) controllers, as long as there’s SSH connectivity from the primary controller (where you run
juju-restore) to the secondary controller machines. We check for connectivity before beginning the restore, just in case.
Restoring a backup used to be done using the
juju restore-backup command. This command is now deprecated (and no longer works on controllers running Focal); use
juju-restore is written in Go and can be downloaded from the Releases tab of the GitHub project. To “install”
juju-restore, download the executable and copy it to the controller machine.
# Download the latest release binary (Linux, AMD64) wget https://github.com/juju/juju-restore/releases/latest/download/juju-restore chmod +x juju-restore # Switch to the controller model juju switch -m controller # Copy juju-restore to the machine hosting the MongoDB primary juju scp juju-restore 0:
juju-restore should be run on the primary controller machine in the MongoDB replica set. This is typically machine 0 in the controller model. All replica set nodes need to be healthy and in
juju-restore requires a backup file created with
juju create-backup. If you saved a local copy, use
juju scp to copy this to the controller machine.
To perform the restore:
# SSH into the controller machine juju ssh 0 # Start the restore! (it will ask for confirmation) juju-restore /path/to/backup
By default, you can only restore from a backup created with exactly the same Juju version as the controller. If you run
juju-restore with the
--allow-downgrade option, you can also restore from a backup created with an earlier Juju version. This means that if you take a backup before upgrading the controller, and then hit a bug in the new Juju version, you can restore the backup to get back to the original version.
High availability (HA) operation
juju-restore supports restoring backups to high availability controllers, and can automatically stop and restart Juju agents and Mongo daemons on the secondary controller machines. If
--manual-agent-control is specified or the user responds “no” when prompted,
juju-restore expects the user to do this manually.
Additional options include:
--yes: answer “yes” to confirmation prompts (for non-interactive mode)
--include-status-history: restore the status history collection for machines and units (which can be large, and usually isn’t needed)
--password, and related options, to override the defaults for connecting to MongoDB
For the full list of options, type:
Last updated a month ago.