Before being able to run `kubeadm init` successfully and set up a simpe k8s cluster additional steps are required - at least some of these could be made part of the kubeadm package installation (or be fixed elsewhere: boo#1218687).

1. kubeadm checks for the presence of `conntrack` and `iptables`, therefore, the
   packages 'conntrack-tools' and 'iptables' should be 'Requires:' of 
   kubernetes-kubeadm.
2. The `containerd` service should be running which may require a 
   `systemctl enable --now containerd`
   Regarding `containerd` also see boo#1218687!
3. The `kubelet` service should be enabled which may require a
   `systemctl enable kubelet.service`
4. A hostname should be set and it should be part of /etc/hosts
   While this cannot be done as part of the installation, a message could be 
   shown after the installation of kubernetes-kubeadm.

An 'kubernetes-kubeadm' post-install script could output a message asking the user to ensure this.

After running `kubeadm init` successfully, the 'coredns' container remains in a 'Pending' state since no network plugin is installed. The installation does provide a manifest for Flannel in package 'flannel-k8s-yaml' which is, however, outdated (boo#1218694). However, even using an updated one from the upstream sources doesn't render a working 'flannel' pod.

It seems that the k8s packages have been receiving regular refreshes to new k8s versions, however, boo#1218687 and boo#1218694 appear to indicate that the results have never been tested :(