Yeah that sounds like it could be the problem. Try to switch it to cluster ip. You shouldn’t need to access the internal dispatch port externally, but if you did you could use kubectl port-forward.