I have a program that deployed on the cluster and this program create other pods by rest request that auto assign new NodePorts for each new pod created, this software auto scale instances of the multiplayer app. My problem is that I need to connect using any ip to assigned for any node created from cluster, ok it's works, but that ip from nodes its not static so when ip is renewed I to lost connection from service.
I tried use load balance by ingress, but I will used many ports and ingress not permite this use and NatConfig but not success too.
I searched any solution in options from gcp but I don't find.
How long do your multiplayer pods run for? And do you have some way of preserving state such that any pod can serve a specific multiplayer instance?
Yes I can, I use the code developed in golang using library of kubernets, and using the the repeated state check from app to server I controller the state of pod. But my problem is keep ip to node instances from cluster.
I thinking to used only one url by load balance for used with the functions of server, how to create, delete and control states and when created new pod instance to app I will to send the ip from any nodes running in cluster and port of pod.