In this blog post, I provide a basic guide on how to become a Kubernaut and set up your own Kubernetes Cluster.

Kubernetes (K8s)

… is Greek for helmsman or captain.
And this is what Kubernetes is about – Shipping your containerized applications.

Kubernetes is a system for automating

  • deployment
  • scaling
  • management

of containerized applications.

Kubernetes manages the

  • Computing Infrastructure
  • Networking Infrastructure
  • Storage Infrastructure

for your containers.

It manages everything around your application.

Read more.


Kubernetes Cluster Setup

The most simple Kubernetes Cluster consists of one Kubernetes-Master, which manages an arbitrary amount of Kubernetes-Nodes (Workers).

Reference: https://kubernetes.io/docs/tutorials/kubernetes-basics/deploy-app/deploy-intro/

1. Infrastructure Setup

  • Choose your Hardware and Operating System
    Since I’m a huge fan of Raspberry Pi’s, in my case I used three Raspberry Pi’s with Raspbian as an operating system (kube-master01, kube-node01 and kube-node02) for my Cluster Setup.

  • Connect the hosts to the network infrastructure and assign a static IP for easy access.
  • Enable SSH with Public-Key-Authentication to provide root SSH access from the Kubernetes Master to your Nodes.
    • Log into the future Kubernetes-Master Node and create a key-pair using ssh-keygen.
    • Copy the public key into the /root/.ssh/authorized_keys file of each Kubernetes nodes (master + workers)
    • Test the SSH connection from your Kubernetes-Master to each Worker-Node using
      ssh root@[IP] -i [private key]

2. Install Ansible

Install ansible on Kubernetes-Master executing
sudo apt-get install -y ansible
Check that the installed version is higher than Ansible 2.2 using
ansible --version.

3. Kubernetes Cluster Configuration and Setup

Once the Ansible is installed the required Ansible-Playbooks I provided can be downloaded into the Kubernetes-Master.
git clone https://gitlab.com/fabian.guschlbauer/k8cluster-blog.git

Hosts Configuration

  • The inventory package contains a hosts file that describes the components or to be precise the master and workers nodes of the Kubernetes Cluster.
    Modify inventory/hosts and add your future master and worker hostnames.
    My basic Kubernetes Cluster example consists of one Kubernetes-Master (kube-master01) and two Kubernetes-Workers (kube-node01, kube-node02).
root@kube-master01:~/k8cluster# cat inventory/hosts
[picluster:children]
master
workers

[master]
kube-master01

[workers]
kube-node01
kube-node02
  • The inventory/host_vars package contains the detailed configuration of the Kubernetes Cluster components. So a configuration file for each of the hostnames specified in inventory/hosts is needed, containing the following information.
root@kube-master01:~/k8cluster# cat inventory/host_vars/kube-master01
ansible_host: HOST_IP
ansible_port: 22
ansible_user: root
ansible_ssh_private_key_file: SSH_PRIVATE_KEY_FILEPATH
ansible_ssh_user: root

root@kube-master01:~/k8cluster# cat inventory/host_vars/kube-node01
ansible_host: HOST_IP
ansible_port: 22
ansible_user: root
ansible_ssh_private_key_file: SSH_PRIVATE_KEY_FILEPATH
ansible_ssh_user: root

...

Cluster Setup using Ansible

After the basic Kubernetes Cluster configuration, the provided Ansible Cluster Setup Playbook can be started from the root directory of the cloned setup repository using
ansible-playbook cluster.yml.

At the end of the setup there will be an output like:

Note: In case of any problems or errors, the playbook (cluster.yaml) or any tasks (listed below) can be modified.

root@kube-master01:~/k8cluster# find roles/ -name "main.yml"
roles/common/tasks/main.yml
roles/workers/tasks/main.yml
roles/master/tasks/main.yml
roles/cleanup/tasks/main.yml
roles/kubeadm/tasks/main.yml
roles/dashboard/tasks/main.yml

As the last step of the Kubernetes Cluster Setup, the available nodes can be checked using kubectl get nodes.

Congratulation Kubernaut, the Kubernetes Cluster setup is done.


Deploy, Deploy, Deploy

As a newly qualified Kubernaut, it is perfectly clear to deploy a first container to the new Kubernetes Cluster.
To do so, simply deploy any docker image you prefer using
kubectl create deployment example –image=<docker_image>

After a few moments, the docker container can be seen up and running.
kubectl get pods

For a detailed description of creating Deployments, I recommend visiting the Kubernetes documentation.

Categories: DevOpsInfrastructure

0 Comments

Leave a Reply

Avatar placeholder

Your email address will not be published. Required fields are marked *