Introduction

Infrastructor is an open-source server provisioning tool written in Groovy. It provides a powerful DSL to simplify configuration management of bare-metal servers and virtual machines. In this article, I'm going to show you how to set up a small Kubernetes cluster on top of virtual machines, which can be handy for its learning.

We are going to build a Kubernetes cluster on top of 3 VMs. We will create the VMs with Vagrant and VirtualBox: 1 master node (control plane), 2 worker nodes.

The number of the worker nodes can be easily increased and depends on your need and the capacity of your hardware. We will also use Calico 3.4 network plugin.

Prerequisites

Please make sure the following software is installed on your machine:

1. Java 8 - required to run Infrastructor

2. Infrastructor 0.2.2 (can be downloaded from GitHub releases page or installed with SDKMAN)

3. Vagrant 2.2.x

4. VirtualBox 5.2.x

The source code of the provisioning scripts can be found in the GitHub repository.

Setting Up VMs

Let's describe VM configurations in a Vagrantfile:

Vagrant.configure("2") do |config|
  config.vm.box = "ubuntu/bionic64"

  config.vm.define "master" do |node|
      node.vm.network "private_network", ip: "192.168.65.10"
      node.vm.provider :virtualbox do |vb|
        vb.memory = 4096
      end
  end

  (1..2).each do |i|
    config.vm.define "node-#{i}" do |node|
      ip = "192.168.65.1#{i}"
      node.vm.network "private_network", ip: ip
      node.vm.provider :virtualbox do |vb|
        vb.memory = 2048
      end
    end
  end
end

The master node has 4GB of RAM and the worker nodes have 2GB each. Feel free to adjust these settings if needed. Launch the nodes with the following command:

vagrant up

After the VMs are up and running we can run a provisioning script to configure a Kubernetes cluster.

Provision a Kubernetes Cluster

Let's describe a setup procedure of a Kkubernetes cluster with Infrastructor. We will define everything in a single file: provision.groovy. The provisioning script will have 5 main tasks:

1. Describe inventory: define all nodes and roles. Infrastructor will connect to the nodes using SSH.

def keypath(def machine) { ".vagrant/machines/$machine/virtualbox/private_key" }

def inventory = inlineInventory {
    node id: 'master', host: '192.168.65.10', username: 'vagrant', keyfile: keypath('master'), tags: [role: 'master']
    (1..2).each { 
        node id: "node-$it", host: "192.168.65.1$it", username: 'vagrant', keyfile: keypath("node-$it")
    }
}

2. Initialize hosts. Update hostnames to corresponding node IDs and turn off swap.

task name: "initializing host", parallel: ON_ALL_NODES, actions: {
    file {
        user    = 'root'
        target  = "/etc/hostname"
        content = node.id
    }

    insertBlock {
        user     = 'root'
        target   = '/etc/hosts'
        block    = "127.0.0.1\t${node.id}\n"
        position = START
    }

    shell "sudo hostname ${node.id}"
    shell "sudo swapoff -a"
}

3. Install kublet, kubeadm, kubectl and kubernetes-cni packages using apt and update the Kubernetes node IP to point to the correct network device (the local network instead of default NAT):

task name: "installing kubernetes", parallel: ON_ALL_NODES, actions: { node ->
    shell user: 'root', command: '''
        apt update
        apt install docker.io -y
        usermod -aG docker vagrant
    '''

    file {
        user = 'root'
        target = '/etc/apt/sources.list.d/kubernetes.list'
        content = 'deb http://apt.kubernetes.io/ kubernetes-xenial main'
    }

    shell user: 'root', command: '''
        curl -s https://packages.cloud.google.com/apt/doc/apt-key.gpg | apt-key add -
        apt update
        apt install kubelet kubeadm kubectl kubernetes-cni -y
    '''

    file  user: 'root', target: '/etc/default/kubelet', content: "KUBELET_EXTRA_ARGS=--node-ip=$node.host"
}

4. Initialize Kubernetes master node using  kubeadm init  and install Calico CNI plugin.

task name: "initializing kubernetes master", filter: {'role:master'}, actions: { node ->
    shell """
        sudo kubeadm init --token $TOKEN \
        --apiserver-advertise-address $node.host \
        --apiserver-bind-port $API_SERVER_BIND_PORT \
        --pod-network-cidr $POD_NETWORK_CIDR
    """

    shell '''
        mkdir -p $HOME/.kube
        sudo cp -i /etc/kubernetes/admin.conf $HOME/.kube/config
        sudo chown $(id -u):$(id -g) $HOME/.kube/config
    '''

    shell '''
        kubectl apply -f https://docs.projectcalico.org/v3.4/getting-started/kubernetes/installation/hosted/etcd.yaml
        kubectl apply -f https://docs.projectcalico.org/v3.4/getting-started/kubernetes/installation/hosted/calico.yaml
    '''
}

5. Join worker nodes to the cluster using  kubeadm join:

task name: "initializing worker nodes", filter: {!'role:master'}, parallel: ON_ALL_NODES, actions: {
    shell "sudo kubeadm join $MASTER_HOST:$API_SERVER_BIND_PORT --token $TOKEN --discovery-token-unsafe-skip-ca-verification"
}

Now the  provision.groovy  contains all the steps to set up a Kubernetes cluster:

def keypath(def machine) { ".vagrant/machines/$machine/virtualbox/private_key" }

def inventory = inlineInventory {
    node id: 'master', host: '192.168.65.10', username: 'vagrant', keyfile: keypath('master'), tags: [role: 'master']
    (1..2).each { 
        node id: "node-$it", host: "192.168.65.1$it", username: 'vagrant', keyfile: keypath("node-$it")
    }
}

inventory.provision { nodes ->

    def TOKEN = "change.me00000000000000"
    def POD_NETWORK_CIDR = "10.50.0.0/16"
    def API_SERVER_BIND_PORT = 8080
    def MASTER_HOST = nodes['master'].host
    def ON_ALL_NODES = nodes.size()

    task name: "initializing host", parallel: ON_ALL_NODES, actions: {
        file {
            user    = 'root'
            target  = "/etc/hostname"
            content = node.id
        }

        insertBlock {
            user     = 'root'
            target   = '/etc/hosts'
            block    = "127.0.0.1\t${node.id}\n"
            position = START
        }

        shell "sudo hostname ${node.id}"
        shell "sudo swapoff -a"
    }

    task name: "installing kubernetes", parallel: ON_ALL_NODES, actions: { node ->
        shell user: 'root', command: '''
            apt update
            apt install docker.io -y
            usermod -aG docker vagrant
        '''

        file {
            user = 'root'
            target = '/etc/apt/sources.list.d/kubernetes.list'
            content = 'deb http://apt.kubernetes.io/ kubernetes-xenial main'
        }

        shell user: 'root', command: '''
            curl -s https://packages.cloud.google.com/apt/doc/apt-key.gpg | apt-key add -
            apt update
            apt install kubelet kubeadm kubectl kubernetes-cni -y
        '''

        file  user: 'root', target: '/etc/default/kubelet', content: "KUBELET_EXTRA_ARGS=--node-ip=$node.host"
    }

    task name: "initializing kubernetes master", filter: {'role:master'}, actions: { node ->
        shell """
            sudo kubeadm init --token $TOKEN \
            --apiserver-advertise-address $node.host \
            --apiserver-bind-port $API_SERVER_BIND_PORT \
            --pod-network-cidr $POD_NETWORK_CIDR
        """

        shell '''
            mkdir -p $HOME/.kube
            sudo cp -i /etc/kubernetes/admin.conf $HOME/.kube/config
            sudo chown $(id -u):$(id -g) $HOME/.kube/config
        '''

        shell '''
            kubectl apply -f https://docs.projectcalico.org/v3.4/getting-started/kubernetes/installation/hosted/etcd.yaml
            kubectl apply -f https://docs.projectcalico.org/v3.4/getting-started/kubernetes/installation/hosted/calico.yaml
        '''
    }

    task name: "initializing worker nodes", filter: {!'role:master'}, parallel: ON_ALL_NODES, actions: {
        shell "sudo kubeadm join $MASTER_HOST:$API_SERVER_BIND_PORT --token $TOKEN --discovery-token-unsafe-skip-ca-verification"
    }
}

You can run it using the infrastructor CLI:

$ infrastructor run -f provision.groovy

The output should look like this:

$ infrastructor run -f provision.groovy
[INFO] running file: 'provision.groovy'
[INFO] task: 'initializing host'
[INFO] task: 'initializing host', done on 3 node|s
[INFO] task: 'installing kubernetes'
[INFO] task: 'installing kubernetes', done on 3 node|s
[INFO] task: 'initializing kubernetes master'
[INFO] task: 'initializing kubernetes master', done on 1 node|s
[INFO] task: 'initializing worker nodes'
[INFO] task: 'initializing worker nodes', done on 2 node|s
[INFO] file: 'provision.groovy' is done

EXECUTION COMPLETE in 2 minutes, 22.393 seconds

Then you can ssh to the master node and check the status of the cluster:

$ vagrant ssh master

$ kubectl get po --all-namespaces
NAMESPACE     NAME                                       READY   STATUS    RESTARTS   AGE
kube-system   calico-etcd-hs5wb                          1/1     Running   0          11m
kube-system   calico-kube-controllers-5d94b577bb-djj64   1/1     Running   0          11m
kube-system   calico-node-fc6gb                          1/1     Running   0          11m
kube-system   calico-node-rsxgs                          1/1     Running   0          11m
kube-system   calico-node-wrbfk                          1/1     Running   0          11m
kube-system   coredns-86c58d9df4-cmt9d                   1/1     Running   0          11m
kube-system   coredns-86c58d9df4-wlxjj                   1/1     Running   0          11m
kube-system   etcd-master                                1/1     Running   0          10m
kube-system   kube-apiserver-master                      1/1     Running   0          10m
kube-system   kube-controller-manager-master             1/1     Running   0          10m
kube-system   kube-proxy-6v2pm                           1/1     Running   0          11m
kube-system   kube-proxy-9t4vh                           1/1     Running   0          11m
kube-system   kube-proxy-ckh4j                           1/1     Running   0          11m
kube-system   kube-scheduler-master                      1/1     Running   0          10m

All services should be up and running. Your Kubernetes cluster is ready to host services. You can use the kubectl   command right from the master node to manage it. You can also modify Vagrantfile and provision.groovy files to add more worker nodes. Enjoy your Kubernetes learning!