When you create a backup, a snapshot of the resources (k8s object and PVC data) is taken that you can use to recover from data loss.
You can use a backup to fully restore the Bare Metal Orchestrator cluster. The backup contains all configuration data, state, and metadata for the cluster.
Backups are saved to the external S3 storage location.
You can use any Velero supported S3 storage provider as an external backup location. For more information, consult the official Velero website.
You can schedule backups to occur at regular intervals or perform a one-time backup at any time. For help with restoring a cluster from a backup, contact your Dell Support representative.
A full restoration of a cluster that has 75,000 servers can take from 5 to 6 hours to complete.
Restore behavior
When a backup of your Bare Metal Orchestrator cluster is taken, a snapshot of the current configuration of nodes and servers is captured.
In a disaster recovery scenario, all servers and the Bare Metal Orchestrator nodes are re-created. Restore now performs a disaster recovery scenario.
Limitations
Observe the following:
- Save backups to only one backup storage location.
- Partial backup is not supported.
- Backup files expire one month after creation and you cannot use expired files to restore the cluster.
- When using an external S3 storage location for the backup files, the server hosting the external backup store must support SSL. For more information, see Change the backup storage location.
- When restoring a cluster using a backup file:
- The number of nodes in the new cluster must be equal to or greater than the number of nodes that were in the original cluster.
- Objects that already exist in the cluster are not overwritten unless they have been modified.
- For restore to work on a new cluster, the IP address and host name should be the same as the one that was backed up.
- All backup storage locations should have common passwords. This is set as secret in the Velero namespace with name set to cloud-credentials.
- The container network interface (CNI) of the new cluster must be the same as the CNI of the original cluster. Note: Bare Metal Orchestrator reserves IP addresses in subnet ranges 10.42.0.0/16 and 10.43.0.0/16 by default for the Global Controller cluster communications and must not be used for any other purpose.
- The age status indicator for a restored server restarts and shows the time since the server was restored.