In this https://docs.docker.com/ucp/high-availability/backups-and-disaster-recovery/ document, it appears that we only need a backup from one controller node to restore the cluster. However, from https://docs.docker.com/ucp/reference/backup/, it appears that we have to backup each node. Which one is it?
There is a difference here between backup and restore.
When a node fails, you can use the backup of that node in order to restore it and make the cluster healthy. Thus you only need to restore the problem node.
However, restore only works when using the backup of that specific node. That means you will want to take backups of all the controller nodes, so that you can restore no matter which controller node fails in the cluster.
We’re currently running Docker Enterprise 2.0 with Kubernetes. Would you have some information on backing up Kubernetes or will this be taken cared of when UCP is backed up? Thanks.