AWS deployment using kubeadm

Introduction

This project presents the creation of a small cluster on AWS. The cluster will be composed of a master node and three worker nodes.

The files to achieve this deployment are available here

We mount a NFS server on the master node too, from which workers will upload .dll.

AWS credentials

You must create and provide your AWS programmatic access keys in your dev/test environment:

mkdir -p ~/.aws
cat <<EOF | tee ~/.aws/credentials
[default]
aws_access_key_id = <ACCESS_KEY_ID>
aws_secret_access_key = <SECRET_ACCESS_KEY>
EOF

Generate a SSH key pair

Use the following procedure to generate a SSH key pair and save it in ~/.ssh:

ssh-keygen -b 4096 -t rsa -f ~/.ssh/cluster-key

The generated SSH key pair cluster-key will be used to ssh the instances of the cluster.

Deploy a cluster

We will create a cluster on AWS composed of four ec2 instances:

• a master node
• three worker nodes

In parameters.tfvars:

• set the value of the parameter ssh_key with the content of the public SSH key ~/.ssh/cluster-key.pub and the path to the private SSH key, for example:
  shell

  ssh_key = {
    private_key_path = "~/.ssh/cluster-key"
    public_key       = "ssh-rsa AAAAB3NzaC1yc2EAAAADAQABAAABAQD3F6tyPEFEzV0LX3X8BsXdMsQz1x2cEikKDEY0aIj41qgxMCP/iteneqXSIFZBp5vizPvaoIR3Um9xK7PGoW8giupGn+EPuxIA4cDM4vzOqOkiMPhz5XK0whEjkVzTo4+S0puvDZuwIsdiW9mxhJc7tgBNL0cYlWSYVkz4G/fslNfRPW5mYAM49f4fhtxPb5ok4Q2Lg9dPKVHO/Bgeu5woMc7RY0p1ej6D4CKFE6lymSDJpW0YHX/wqE9+cfEauh7xZcG0q9t2ta6F6fmX0agvpFyZo8aFbXeUBr7osSCJNgvavWbM/06niWrOvYX2xwWdhXmXSrbX8ZbabVohBK41 email@example.com"
  }
  

  Copy to clipboard
• set the ID of an existing VPC and its subnet:
  shell

  vpc_id    = "<VPC_ID>"
  subnet_id = "<SUBNET_ID>"
  

  Copy to clipboard

To deploy the cluster execute the command:

make all

The outputs display the public IP of each instance, like:

master_public_ip = {
  "ip" = "54.185.23.147"
  "name" = "i-0168c936872babdf2"
}
worker_public_ip = [
  {
    "ip" = "54.184.45.26"
    "name" = "i-06b8aeab6cb62750a"
  },
  {
    "ip" = "35.87.249.26"
    "name" = "i-0e4c32d39bfcf8aac"
  },
  {
    "ip" = "54.244.169.65"
    "name" = "i-0c691f1d971e62150"
  }
]

Prerequisites

You must open the following inbound ports:

Accessing the cluster from outside

Copy /etc/kubernetes/admin.conf from the master on your machine located outside the cluster as ~/.kube/config. Then replace localhost or the private address IP with the public IP of the Kubeadm server (master node). kubectl can now manage your Kubeadm cluster from your local machine.

Destroy the cluster

To delete all resources of the cluster created on AWS, execute the command:

make destroy