Installing to EKS via Helm but using a pre-existing ELB

I’ve successfully managed to install Kong onto my EKS cluster with helm and it’s working quite well. We have one small issue and I cannot tell if it’s something that we have to live with or not.

We use Terraform to deploy all of our cloud environment, but the helm installation of Kong creates its own Amazon ELB (unmanaged by Terraform.) This leads to us being unable to destroy the cluster via Terraform since some of Terraform’s managed components are unable to be deleted/detached because the Kong-generated ELB is still active.

Is there a means of telling the Kong helm installation to not create an ELB, but instead use a provided one (via an ARN or something)? We can easily have Terraform create the ELB and hand that over to Kong, but the documentation for Kong’s Helm values.yaml doesn’t have anything that keys me into addressing this.

Any ideas?


you can change the kong proxy service to NodePort to a specific target port and separately map the ELB target group to that port.

I’ve been remiss in replying, but this absolutely worked.