Private only subnets topology cannot get the information of EC2 instances from EKS cluster

[tingchienc]

trafficstars

/kind bug

What steps did you take and what happened: [A clear and concise description of what the bug is.] I followed the instruction from https://cluster-api-aws.sigs.k8s.io/topics/bring-your-own-aws-infrastructure.html to use existing vpc to create EKS Clusters. I used kind cluster based on the get started guide. clusterctl generate cluster capi-eks-quickstart --flavor eks --kubernetes-version v1.26.0 --worker-machine-count=2 > capi-eks-quickstart.yaml I used VPC that has two private subnets and route tables and I specified ids into the yaml file.

At the end, the EKS cluster and EC2 instances run but I cannot get the EC2 information by EKS cluster. When typing clusterctl get kubeconfig capi-eks-quickstart > capi-eks-quickstart.kubeconfig kubectl get nodes --kubeconfig capi-eks-quickstart.kubeconfig It shows No resources found in default namespace.

kubectl logs capa-controller-manager-5d487d7d68-nj46j -n capa-system

What did you expect to happen: Shows something like here.

capi-quickstart-md-0-55x6t-5649968bd7-8tq9v Ready 6m5s v1.27.1 capi-quickstart-md-0-55x6t-5649968bd7-glnjd Ready 6m9s v1.27.1

Anything else you would like to add: [Miscellaneous information that will assist in solving the issue.] The cluster status looks weird. It always stucks here.

I tried to create ec2 endpoints to help connects but it seems no help.

Environment:

• Cluster-api-provider-aws version: v1.42
• Kubernetes version: (use kubectl version): v1.26.0
• OS (e.g. from /etc/os-release): Ubuntu 22.04