Work needed for a functionally complete microshift v1.{}

• Before the kube-apiserver is ready, the view of all resources in the cluster is inconsistent.  When clients (and controllers) take action based on an incomplete view of all resources, the state in etcd can become corrupted.
• OCP handles this using a load balancer
• The solution must make the kube-apiserver inaccessible when the kube-apiserver is not ready.
• The remaining impact is not yet known