Installation on AWS

We can use https://eksctl.io which is the AWS supported way of setting up a K8's cluster (EKS).

1. Get your credentials

export AWS_ACCESS_KEY_ID=...
export AWS_SECRET_ACCESS_KEY=...

2. Setup your EKS install config

Copy the below (changing it as necessary) into a file called cluster.yaml. You'll need to set the {ACCOUNT_ID} to your account.

apiVersion: eksctl.io/v1alpha5
kind: ClusterConfig
metadata:
  name: bionic-gpt
  region: us-east-2
managedNodeGroups:
- name: bionic-gpt
  instanceType: t2.large
  minSize: 2
  maxSize: 4
iam:
  withOIDC: true
  serviceAccounts:
  - metadata:
      name: ebs-csi-controller-sa
      namespace: kube-system
    attachPolicyARNs:
    - "arn:aws:iam::aws:policy/service-role/AmazonEBSCSIDriverPolicy"
    wellKnownPolicies:
      ebsCSIController: true
    roleName: eksctl-cluster-ebs-role
    roleOnly: true
addons:
- name: aws-ebs-csi-driver
  serviceAccountRoleARN: "arn:aws:iam::{ACCOUNT_ID}:role/eksctl-cluster-ebs-role"

Dry Run `eksctl`

First a dry run

eksctl create cluster --dry-run -f cluster.yaml

You should see something like

accessConfig:
  authenticationMode: API_AND_CONFIG_MAP
apiVersion: eksctl.io/v1alpha5
availabilityZones:
- us-east-2a
- us-east-2b
- us-east-2c
 ....
 ....

Now Create the Cluster

Run the command without the --dry-run.

eksctl create cluster -f cluster.yaml

And wait....

Finally after 20 minutes or more you'll hopefully have an EKS cluster.

Accessing the Kubeconfig file

The below will export the kubeconfig to your home directory.

eksctl utils write-kubeconfig --cluster bionic-gpt --region us-east-2

You can then

kubectl get nodes