needs updates and details
Backups should be made of all credentials (certificates), the etcd database and, if necessary, persistent volumes. Backups are useful when, e.g., a K8s upgrade fails or the user accidentally deleted an important resource. A cronjob dumps the etcd database every N minutes and ensures that the number of backup instances is contained. One candidate for managing the backups is borg.
Automated etcd backups can be configured in config/config.toml. To restore the cluster re-run kubeadm with an existing etcd database. The guide for restoration can be found in the operations handbook only accessible by C&H. The process is also available at the official Kubernetes documentation of etcd here.