If I upgrade a kubeware using a rolling-upgrade with service replace strategy, the status of the load balancer associated to the service dissapears in K8s.

This happens when we try to preserve the current nodePort for the service port exposed. If we don't specify the current nodePort then K8s will change it for a new random one and the load balancer will stop working since it'll be using the old one, so either we preserve it or somehow update the LB config.