So i created a cluster containing 4 machines using this command

gcloud container clusters create "[cluster-name]" \
  --machine-type "n1-standard-1" \
  --image-type "COS" \
  --disk-size "100" \
  --num-nodes "4"

and i can see that it's creating 4 VM instances inside my compute engine. I then setup deployments pointing to one or more entry(ies) in my container registry and services with a single service exposing a public ip

all of this is working well, but it bothers me that all 4 VM instances it has created is having public ip(s), please do correct me if i am wrong, but to my understanding here's what happen behind the scene

1. A container is created
2. VM instances is created based on #1
3. An instance group is created, with VM instances on #2 as members
4. (Since i have one of the service exposing a public ip) a network load balancer is created pointing to the instance group on #3 or the VM instances on #2

Looking at this, i don't think i need a public ip on each of the VM instances created for the cluster.

I have been reading the documentation(s), although i think i might have missed something, but i can't seem to find the configuration arguments that will allow me to achieve this