Pod CIDR in the generated spec is not used for cluster provision about eks-anywhere HOT 5 CLOSED

Did you get a cluster up?

from eks-anywhere.

No, the default CIDR conflict with the underlying host network.

from eks-anywhere.

The spec is the right place to configure the pods and service CIDR. You dont need to have to do anything else. Could you share the output of kubectl --kubeconfig dev-cluster-eks-a-cluster.kubeconfig get pods -A to check if all the pods are running? Could you also share the output of CIlium config on the cluster, you can get them with kubectl --kubeconfig dev-cluster-eks-a-cluster.kubeconfig get cm cilium-config -n kube-system -o yaml

from eks-anywhere.

Since the cluster-info dump proven wrong CIDR being used, there's no point keeping it as it conflicted with host network. The cluster didn't provision successfully (some sort of CA error) but this is expected since the CIDR is incorrect. That being said, since the configuration was correct, why the cluster still provision with default 192.168.x.x CIDR?

from eks-anywhere.

@gohmc thanks for the report! You are right, we are not currently respecting the pod/serviceCidr config for docker clusters. Notice the difference between the two templates, docker and vsphere

I'll add this to our current sprint to be fixed.

from eks-anywhere.