kubernetes-cluster-local

Automatically provision multi node Kubernetes cluster locally using Vagrant, Ubuntu, Ansible and Kubeadm

Features

A Kubernetes cluster with a single master and two worker nodes will be provisioned
Weave Net will be deployed as the pod network
kubectl will be installed and configured in the master node
helm will be installed and configured in the master node
tiller will be deployed and initialized in the master node
BIND9 DNS server will be installed and configured in the master node. All nodes will be using it as the nameserver
NFS server will be configured in the master node
NFS client with StorageClass of nfs-client will be deployed in the master node for dynamic provisioning of PersistentVolumes
Nginx ingress will be deployed as the ingress controller in the master node
Metallb load balancer will be deployed with controller in the master and speaker in all nodes

Prerequisite

Install the following components into local machine

Provision cluster

Automatically provision multi node Kubernetes cluster from scratch using base Vagrant box (Use this when you want the latest Docker, Kubelet, Kubeadm, Kubectl and Helm to be installed)

$ ./cluster.sh start

Automatically provision multi node Kubernetes cluster using a prebuilt Vagrant box with Docker, Kubelet, Kubeadm, Kubectl and Helm preinstalled (Use this if you want subsequent cluster provisioning to be faster after a cluster teardown)

$ ./cluster.sh start prebuilt

Prebuilt Vagrant box from https://app.vagrantup.com/zulhilmizainuddin/boxes/kubernetes-ubuntu18.04 will be used

Prebuilt Vagrant box was built using https://github.com/zulhilmizainuddin/kubernetes-vagrant-box

SSH into master node

SSH into master node after Kubernetes cluster provisioning has completed

$ ./cluster.sh ssh master

Deployment example

kubectl example

Nginx deployment

Deploy nginx

$ kubectl run nginx --image=nginx --port=80 --replicas=2

Expose nginx deployment with ClusterIP service

$ kubectl expose deploy nginx --port=80

Deploy nginx ingress resource

$ cat <<EOF | kubectl apply -f -
apiVersion: extensions/v1beta1
kind: Ingress
metadata:
  annotations:
    kubernetes.io/ingress.class: nginx
  name: nginx-ingress
  namespace: default
spec:
  rules:
    - host: nginx.k8s.zone
      http:
        paths:
          - backend:
              serviceName: nginx
              servicePort: 80
            path: /
EOF

Get load balancer ip address

$ kubectl get svc -n kube-system | grep LoadBalancer | tr -s ' ' | cut -d' ' -f4
192.168.33.240

Set nginx.k8s.zone resolution in host machine /etc/hosts file

$ echo "192.168.33.240    nginx.k8s.zone" | sudo tee -a /etc/hosts

Access nginx at http://nginx.k8s.zone on host machine

helm example

Prometheus deployment

Deploy prometheus

$ helm upgrade \
  --install \
  --set server.ingress.enabled=true \
  --set server.ingress.annotations."kubernetes\.io/ingress\.class"=nginx \
  --set server.ingress.hosts={"prometheus.k8s.zone"} \
  --set server.persistentVolume.storageClass=nfs-client \
  --set alertmanager.persistentVolume.storageClass=nfs-client \
  --set nodeExporter.tolerations[0].key=node-role.kubernetes.io/master \
  --set nodeExporter.tolerations[0].operator=Exists \
  --set nodeExporter.tolerations[0].effect=NoSchedule \
  prometheus \
  stable/prometheus