Description
Hi team, is there any possibility of supporting NodeIP + NodePort as the TargetGroup backend?
As we know, EKS with VPC-CNI is a common setup in AWS, but there are also other clusters that either do not use EKS or run EKS with Calico or Cilium. In such cases, pod IPs that are not within the VPC CIDR cannot be reached from the Lattice service. By supporting NodeIP + NodePort, the Gateway API controller could be extended to more cluster types, enabling broader adoption of Lattice services.
Moreover, the AWS Load Balancer Controller supports NodeIP + NodePort by default. Therefore, I believe there should be no blockers for implementing this feature. What do you think about that, guys?
BTW, we are a company that heavily relies on AWS and are currently in the process of migrating from self-managed clusters to EKS. This migration involves enabling service-to-service communication across different clusters in multiple accounts