Difference between the Rancher backup operator and cluster-level etcd snapshots

A Rancher upstream cluster

In order to be prepared for certain scenarios, such as a disaster recovery, or a possible migration of your Rancher upstream cluster to another environment, it is crucial to understand the different backing-up tools SUSE Rancher offers.

We have at your disposal two different tools, with two differentiated purposes, when backing up an upstream (local) cluster:

• Rancher Backup Operator: It will help you back up the specific Rancher resources, getting a copy of the CRDs pertinent to Rancher as YAML so that it can be imported into another Rancher instance. Then, it will only backup Rancher-related resources, but not the whole cluster information. The Rancher Backup Operator is useful, for example, when you want to migrate your Rancher instance to another cluster. The Rancher Backup Operator will help you to back up the Rancher state via the Rancher-related resources, but not all the information of your cluster.
• etcd snapshots: They will back up the entire cluster (which limits you to restoring to IP addresses that match what exists in etcd), not only the Rancher-related resources. Therefore, if a disaster scenario appears, the etcd snapshots will be the way to go in most cases, as they back up the entire etcd datastore, which is the "brain" of a Kubernetes cluster. Therefore, etcd snapshots are helpful in Disaster Recovery scenarios, or when rolling back the Rancher or Kubernetes version after a failed upgrade.
  • RKE2 etcd snapshots
  • k3s etcd snapshots
  • RKE etcd snapshots

Please note that you can configure an S3 location to save both the snapshots and the Rancher operator backups. This would be a recommended approach, as they will remain even if something catastrophic happens to the cluster, and it is no longer accessible.