To support FastAndFurious upgrades, it would be good to be able to have a full snapshot of the cluster previous the start of the upgrade procedure, in case a rollback is needed. As FastAndFurious is going to upgrade the version of the RHEL system on the computes along with everything else, only restoring the control plane would leave clusters in a version not compatible with the underlying RHEL version (i.e. 16.2 compute nodes with RHEL 9). In order to have a really quick way to rollback, keeping backups of the compute nodes helps a lot.
Superseded by #2069624