We are provisioning the node groups as EKS managed node groups.
Once the node group is up we need to set some proxy on each of the node group instances, but the problem is that the outputs of the EKS managed group doesn't give any details about the instance details.
Once we get the EC2 instance id we can use data source and can use remote_exec to update the proxy.
This approach doesn't really help you because you'd need Terraform to run every single time the managed node group increases in size or replaces an unhealthy instance.
The normal thing with instances in autoscaling groups is to either bake things directly into the AMI and/or use user data to configure the instance with anything extra that isn't baked into the AMI. Unfortunately managed node groups don't allow either the AMI to be changed or custom user data to be passed in.
The AWS containers roadmap on Github is tracking a feature request for adding custom user data but until then your only reliable option here is to forgo managed node groups and use normal autoscaling groups with your own AMI or user data to provide the configuration you need.
