Karpenter task For OF

For Automate EKS cluster setup with Karpenter and Graviton on AWS task we will follow the least efforts path and re-use the terraform-aws-modules/eks/aws Terraform Community modules required to complete the setup, with some augmentations that allow us to satisfy acceptance criteria. We start with a simple deployment, so no advanced Cluster features, access policies, terraform tests, terraform remote state, etc will be configured. Because we don't have a VPN or AWS Direct Connect established yet, Cluster API server endpoint needs to be exposed to a public network. We expect the VPC and Subnets are properly configured, have ingress and egress traffic enabled and tagged according to the Karpenter expectations.

Requirements

You will need a terraform and kubectl installed and added to your shell $PATH. kubectl has to be configured to use correct Cluster context to perform queries against EKS cluster.

Name	Version
terraform	>= 1.3.2
aws	>= 5.74
helm	>= 2.7
kubectl	>= 2.0

Usage

1. Configure the main AWS provider in the eks-with-karpenter.tf file, by setting up the region and profile of your AWS account.

$ cat eks-with-karpenter.tf
provider "aws" {
  alias   = "main"
  region  = "your-region"
  profile = "aws-profile"
}
...

2. Initiate modules.

$ terraform init

3. Please define VPC and your Private Subnet identifiers in the fleet.tfvars file. Next generate and review the changeset for resources that will be created.

$ terraform plan -var-file="fleet.tfvars"

4. If the changeset satisfies your expectations - apply it.

$ terraform apply -var-file="fleet.tfvars"

5. Configure the EKS Cluster access for your kubectl and verify the result.

$ aws eks update-kubeconfig --region ${AWS_REGION} --name my-cluster --kubeconfig ${KUBE_CONFIG}
...

$ k get pods -A
NAMESPACE     NAME                                READY   STATUS    RESTARTS      AGE
kube-system   aws-node-4zkvh                      2/2     Running   0             68m
kube-system   coredns-5b7fb68d97-2qvhf            1/1     Running   0             67m
kube-system   coredns-5b7fb68d97-x55qv            1/1     Running   0             66m
...

6. Verify the Karpenter setup by deploying and scaling two resources from the InflateDeploymentExample.yaml.

One Deployment resource acts as an example of how to deploy amd-based workload and another as arm-based one.

$ kubectl apply -f InflateDeploymentExample.yaml
$ kubectl scale deployment inflate-amd --replicas 5
$ kubectl scale deployment inflate-arm --replicas 5
$ kubectl logs -n kube-system -l app.kubernetes.io/instance=karpenter -f

The result should be similar to