We are building a new Kubernetes cluster in GCP, using a Shared VPC. The cluster was created with VPC-native (i.e., alias IP) enabled. Everything about the Shared VPC seems to be in order. The node, pod, and service IP addresses are all as expected from the Shared VPC. I even have one service on an internal load balancer on the Shared VPC subnet which works just fine.
But when I run the simplest of tests for an external service
kubectl create deployment hello-web --image=nginxdemos/hello
kubectl expose deployment hello-web --type=LoadBalancer --port 80 --target-port 80
kubectl get service hello-web
the public IP indicated by the last command, gets "This site can't be reached, x.x.x.x took too long to respond" from Chrome.
I ran the above test in our existing clusters which are not VPC-native nor Shared VPC and it works. So what might I be missing?