OpenShift didn't check the hostPort usage before creating a nodePort for a same port.
Neither our document mentioned user should avoid overlapping with OCP default ports[1]

This would allow user creating a nodePort overlapping with an existing hostPort by accident. 


[1] - https://github.com/openshift/enhancements/blob/master/dev-guide/host-port-registry.md

1. Expand the node port range to 9000-32767 [1]
2. Create any example pod and a NodePort using 9101 which should be already used by openshift-sdn pod as a hostPort.
3. Curl the nodeIp:9101. It's connect to our example pod, instead of openshift-sdn 

User's nodePort will conflict with OpenShift default ports. Causing networking trouble.

OCP should not allow user to expand the node port overlapping with default ports.
Or, scheduler or controller-manager should detect the hostPort usage before creating a nodePort