EKS with 2 nodes groups stacks in private subnets only

VPC: 2 available zones, 2 private and 8 public subnets

Load Balancer is in public subnets.

EKS nodes are in private subnets.

Environment name

test12

Diagram

VPC Subnets:

SUBNET_ID="12"

VpcBlock 10.${SUBNET_ID}.0.0/16

SubnetPublicA1 10.${SUBNET_ID}.0.0/22
SubnetPublicA2 10.${SUBNET_ID}.4.0/22
SubnetPublicA3 10.${SUBNET_ID}.8.0/22
SubnetPublicA4 10.${SUBNET_ID}.12.0/22

SubnetPublicB1 10.${SUBNET_ID}.16.0/22
SubnetPublicB2 10.${SUBNET_ID}.20.0/22
SubnetPublicB3 10.${SUBNET_ID}.24.0/22
SubnetPublicB4 10.${SUBNET_ID}.28.0/22

SubnetPrivateA 10.${SUBNET_ID}.192.0/18
SubnetPrivateB 10.${SUBNET_ID}.128.0/18

EKS nodes groups

EKS nodes groups:

main
websites

main is for Kubernetes core service pods.

websites is for our websites pods.

We could scale main and websites groups individually.

Variables, Configuration

Configuration is in init.sh file.

Usage

It is as simple as:

. ~/venv-aws-cli/bin/activate
source ./all.sh 
eksCreateCluster

After ~ 28 minutes you will see all Workers as Ready:

$ kubectl get nodes
NAME                                          STATUS     ROLES    AGE   VERSION
ip-10-12-162-237.eu-west-1.compute.internal   Ready      <none>   14m   v1.12.7
ip-10-12-228-122.eu-west-1.compute.internal   Ready      <none>   14m   v1.12.7
ip-10-12-248-78.eu-west-1.compute.internal    NotReady   <none>   19s   v1.12.7

Cleanup

To delete every resources (VPC, Workers, EKS cluster)

. ./all.sh
eksCleanup

I would take ~ 20 minutes to clean up.

Note: AWS keyPair will be kept.