README.md

Control plane

• Create a VPC (intra, private and public subnets)
• Create a management EKS cluster in a single zone
• Use SPOT instances
• Use bottlerocket AMI
• Install and configure Karpenter
• Install and configure Flux
• Write a secret that contains the cluster's specific variables that will be used with Flux. (please refer to variables substitutions)
• Deploy Cilium

How to apply this?

1. Edit the file backend.tf and put your own S3 bucket name.
2. Create a file that contains your own variables. Here's an example:

variables.tfvars

env          = "dev"
cluster_name = "mycluster-0" # Generated with petname

github_owner    = "Smana"
github_token    = <REDACTED>
repository_name = "cilium-gateway-api"

tags = {
  GithubRepo = "cilium-gateway-api"
  GithubOrg  = "Smana"
}

3. Apply with terraform apply -var-file variables.tfvars

Cleaning things up

In order to really clean everything you should follow these steps:

1. Suspend Flux reconciliations

   flux suspend kustomization --all

2. Delete Gateways (These create AWS loadbalancers)

   kubectl delete gateways --all-namespaces --all

3. Wait 3/4 minutest and delete all IRSA

   kubectl delete irsa --all-namespaces --all

4. terraform destroy --var-file variables.tfvars

One step:

flux suspend kustomization --all && \
kubectl delete gateways --all-namespaces --all && sleep 60 && \
kubectl delete irsa --all-namespaces --all && sleep 30 && \
terraform destroy --var-file variables.tfvars