README.md

Kube 101

Requirements

• Docker Engine 19.03+ (tested successfully with Docker4Mac or Docker Linux)
• k3d binary installed on your PATH (https://github.com/rancher/k3d)
• kubectl binary installed on your PATH (https://kubernetes.io/docs/tasks/tools/install-kubectl/)
• A bash prompt

Create Kubernetes Cluster

• Using the k3d command, create a Kubernetes cluster with the following properties:
  • 1 master node and 3 worker nodes,
  • ports 80 and 443 published on the localhost interface of your host machine,

k3d create --publish="80:80" --publish="443:443" --workers="3"

• 4 docker containers are now running: 1 for the master node and 3 for the worker nodes:

docker ps

Configure kubectl

• Use the configuration generated by k3d for kubectl, by following the instructions from k3d output:

export KUBECONFIG="$(k3d get-kubeconfig --name='k3s-default')"
kubectl cluster-info

You should see the following output:

Kubernetes master is running at https://localhost:6443
CoreDNS is running at https://localhost:6443/api/v1/namespaces/kube-system/services/kube-dns:dns/proxy

To further debug and diagnose cluster problems, use 'kubectl cluster-info dump'.

Interact with Kubernetes API

• Check the cluster topology:

kubectl get nodes
# You should see 1 master and 3 workers

• Check the pods running, cluster wide:

$ kubectl get pod --all-namespaces

NAMESPACE     NAME                         READY   STATUS      RESTARTS   AGE
kube-system   coredns-b7464766c-mtj2k      1/1     Running     0          3m45s
kube-system   helm-install-traefik-bw2tg   0/1     Completed   0          3m46s
kube-system   svclb-traefik-9fknp          2/2     Running     0          3m27s
kube-system   svclb-traefik-ppffn          2/2     Running     0          3m27s
kube-system   svclb-traefik-vxdlk          2/2     Running     0          3m27s
kube-system   svclb-traefik-z7wfh          2/2     Running     0          3m27s
kube-system   traefik-56688c4464-r8ctk     1/1     Running     0          3m27s

• Open the URL http://localhost: you should see a 404 page not found page. It means an HTTP server is answering in the cluster on port 80.

Deploy the Example Web Application

• Create the namespace named whoami:

$ kubectl get namespaces
NAME              STATUS   AGE
default           Active   5m56s
kube-node-lease   Active   5m56s
kube-public       Active   5m56s
kube-system       Active   5m56s
# No "whoami" namespace

$ kubectl apply -f ./whoami-namespace.yml
namespace/whoami created

$ kubectl get namespaces
NAME              STATUS   AGE
default           Active   6m46s
kube-node-lease   Active   6m46s
kube-public       Active   6m46s
kube-system       Active   6m46s
whoami            Active   14s
# New "whoami" namespace

• Create the "deployment", which should create 2 pods (because 2 replicas). The "whoami" container is a web application listening on its port 80 which respond to HTTP GET requests by responding (body) the headers of the request:

$ kubectl get pod --namespace=whoami
No resources found in whoami namespace.

$ kubectl apply -f ./whoami-deployment.yml
deployment.extensions/whoami created

kubectl get pod --namespace=whoami
NAME                      READY   STATUS    RESTARTS   AGE
whoami-756586b9ff-h4vlq   1/1     Running   0          8s
whoami-756586b9ff-j6sqg   1/1     Running   0          8s

• Expose + Load Balance the 2 replicated pod of the deployment by creating a new "service":

$ kubectl get service --namespace=whoami
No resources found in whoami namespace.

$ kubectl apply -f ./whoami-service.yml
service/whoami created

$ kubectl get service --namespace=whoami
NAME     TYPE        CLUSTER-IP      EXTERNAL-IP   PORT(S)   AGE
whoami   ClusterIP   <IP>            <none>        80/TCP    13s

• Spawn a new pod in the same namespace, with an interactive shell to request the "service" and see loadbalancing happening (look at IP: field on the response body):

kubectl run --tty -i --rm --image=alpine --namespace=whoami -- sh
(...)
/ # apk add --no-cache curl # Install curl
(...)
/ # curl http://<IP>:80 # Wher <IP> is the "cluster IP" of the service
(...)
/ # curl http://<IP>:80 # Wher <IP> is the "cluster IP" of the service
(...)
/ # exit
(...)
deployment.apps "sh" deleted

• Publish the "service" to outside the cluster by creating a new "Ingress", that will be implemented by Traefik, the Ingress Controller of the cluster:

$ kubectl get ingress --namespace=whoami
No resources found in whoami namespace.

$ kubectl apply -f whoami-ingress.yml
ingress.extensions/whoami created

$ kubectl get ingress --namespace=whoami
NAME     HOSTS       ADDRESS      PORTS   AGE
whoami   localhost   172.30.0.2   80      7s

• You can now open the URL http://localhost:80 and access the web service. If you reload the page, you'll see the same "load balancing" behavior as earlier with curl.

Cleanup

Destroy the cluster with k3d delete (check with docker ps).

Going Further

• https://medium.com/@geraldcroes/kubernetes-traefik-101-when-simplicity-matters-957eeede2cf8
• https://kubernetes.io/docs/tutorials/kubernetes-basics/

whoami-deployment.yml

---
kind: Deployment
apiVersion: apps/v1
metadata:
  name: whoami
  namespace: whoami
spec:
  replicas: 2
  selector:
    matchLabels:
      app: whoami
  template:
    metadata:
      labels:
        app: whoami
    spec:
      containers:
      - name: whoami
        image: containous/whoami
        imagePullPolicy: Always

whoami-ingress.yml

---
apiVersion: extensions/v1beta1
kind: Ingress
metadata:
  name: whoami
  namespace: whoami
  labels:
    app: whoami
  annotations:
    kubernetes.io/ingress.class: 'traefik'
    traefik.frontend.passHostHeader: 'true'
spec:
  rules:
  - host: localhost
    http:
      paths:
      - backend:
          serviceName: whoami
          servicePort: 80

whoami-namespace.yml

---
apiVersion: v1
kind: Namespace
metadata:
  name: whoami

whoami-service.yml

---
apiVersion: v1
kind: Service
metadata:
  name: whoami
  namespace: whoami
  labels:
    app: whoami
spec:
  type: ClusterIP
  ports:
  - port: 80
    name: whoami
  selector:
    app: whoami

Good job on this!!