Setup Kubernetes Cluster with Kubeadm and Vagrant

Introduction

With reference to steps listed at Using kubeadm to Create a Cluster for setting up the Kubernetes cluster with kubeadm. I have been working on an automation to setup the cluster. The result of it is kubeadm-vagrant, a github project with simple steps to setup your kubernetes cluster with more control on vagrant based virtual machines.

Installation

• Download and install Vagrant specific to your OS

• Install few Vagrant plugins required for setting it up.

vagrant plugin install vagrant-hostmanager
vagrant plugin install vagrant-hosts
vagrant plugin install vagrant-env

• Clone the kubeadm-vagrant repo

git clone https://github.com/cloudnative-tech/kubeadm-vagrant

• For setting up up the Nodes configure the cluster parameters in .env file. Refer section below for details of configuration options.

vi .env

• Create a directory to store artifacts created during cluster creation.

mkdir .data

• Spin up the cluster

vagrant up

• This will spin up new Kubernetes cluster. You can check the status of cluster with following command,
  • For Master Node

    vagrant ssh k8s-master
    sudo su
    kubectl get pods --all-namespaces
    

  • For Worker Node

    vagrant ssh k8s-node-<NodeNumber>
    

Cluster Configuration Options

1. Generic Properties

• BOX_IMAGE is currently default with "generic/ubuntu1804" box which is custom box created which can be used for setting up the cluster with basic dependencies for kubernetes node.
• PROVIDER is currently default with "virtualbox" It may change in future with more providers.
• KUBE_VERSION is the version of Kubernetes cluster you want to set up with. The main branch will always have the latest available versions. You can find older versions in the releases.
• NW_PLUGIN is the network plugin you wish to enable with the Kubernetes cluster. Currently it supports calico an weave
• A secure dashboard is enabled by default. To enable INSECURE_DASHBOARD can be used. Set its value to yes
  • You can login to dashboard using https://<MASTER_IP>:31443 and the .kube/config file from your vagrant directory.
• A Load Balancer IP Pool can be provided with LB_IPPOOL. The format is the range of IP addresses in your host-only network.
• The sandbox by default will install HELM v 3.x. To disable it, update INSTALL_HELM to no
• The sandbox by default will enable operator-sdk and OLM. To disable it, update ENABLE_OLM to no

2. Master Node Properties

• Set SETUP_MASTER to true if you want to setup the node. This is true by default for spawning a new cluster. You can skip it for adding new worker nodes.
• Set MASTER_IP to a preferred value from the subnet of your virtualbox host-only network.
• Specify custom POD_NW_CIDR of your choice for particular NW_PLUGIN
• Specify Master node hostname with MASTER_HOSTNAME
• Provide your recommended values for MASTER_CPU AND MASTER_MEMORY. The minimum values for both are 2 and 2048 respectively.

3. Worker Noder Properties

• Set SETUP_NODES to true/false depending on whether you are setting up worker nodes in the cluster.
• Specify NODE_COUNT as the count of worker nodes in the cluster
• Specify NODE_IP_NW as the network IP which can be used for assigning dynamic IPs for cluster nodes from the same network as Master
• Specify Worker node hostname with NODE_HOSTNAME
• Provide your recommended values for NODE_CPU AND NODE_MEMORY. The minimum values for both are 1 and 1024 respectively.