This is a playground to play with kubernetes (k8s) in its one-node local cluster mode with the help of minikube. Minikube will run a virtualized, ready to play k8s cluster.

To install minikube for your OS, use the provision.sh script from this repo. It will detect if you have Linux or OSX. The script will install kubectl and minikube.

Start minikube by executing: minikube start. Some very helpful minikube commands are:

• status - check the status of the minikube,
• ip - check IP address of the minikube cluster,
• logs - show logs,
• service - show URL for a service running in the k8s cluster; we will use this command later.

We can manage the k8s cluster with kubectl command. Let's see nodes of our cluster.

➜  kubectl get nodes
NAME       STATUS    AGE
minikube   Ready     76d

A very detailed information for a given node can be achived via kubectl describe node minikube. You can also start a fancy Web dashboard ! minikube dashboard will automatically open in your default Web browser, to get only the URL, add --url.

Let's run a deployment using Docker image darek/goweb:1.0 and exposes port 8080.

➜ kubectl run goweb --image=darek/goweb:1.0 --port=8080

You can add labels to pods, for a faster search and discoverability. Let's add labels to all pods we have (for now we have only one). Labels are great to select pods you want if you have many of them.

➜ kubectl label pods --all app=goweb
➜ kubectl label pods --all environement=production
➜ kubectl label pods --all release=stable

The same effect can be achieved by using a YAML file: kubectl create -f goweb_deployment.yml.

apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  name: 
spec:
  replicas: 1
  template:
    metadata:
      labels:
        app: goweb
        release: stable
        environment: production
    spec:
      containers:
      - name: goweb
        image: darek/goweb:1.0
        ports:
        - containerPort: 8080

Remember that kuberneter needs to pull the image, so the container may not beed provisioned right away. Let's see pods:

➜ kubectl get pods
NAME                    READY     STATUS              RESTARTS   AGE
goweb-907329357-dnycx   0/1       ContainerCreating   0          2m

And now by labels:

➜ kubectl get pods -l app=goweb --show-labels
NAME                     READY     STATUS              RESTARTS   AGE       LABELS
goweb-907329357-dnycx    0/1       ContainerCreating   0          13m       app=goweb,pod-template-hash=2461254821,stage=testing,what=test

ContainerCreating says that the container is in progress. Use a describe command to see detailed information.

➜ kubectl describe pod goweb-907329357-dnycx 
Name:		goweb-907329357-dnycx 
Namespace:	default
Node:		minikube/192.168.99.100
Start Time:	Mon, 12 Dec 2016 07:26:08 +0100
Labels:		pod-template-hash=907329357
		    run=goweb
Status:		Pending
IP:
Controllers:	ReplicaSet/goweb-907329357
Containers:
  goweb:
    Container ID:
    Image:			darek/goweb:1.0
    Image ID:
    Port:			8080/TCP
    State:			Waiting
      Reason:			ContainerCreating
    Ready:			False
    Restart Count:		0
    Environment Variables:	<none>
Conditions:
  Type		Status
  Initialized 	True
  Ready 	False
  PodScheduled 	True
Volumes:
  default-token-f3ech:
    Type:	Secret (a volume populated by a Secret)
    SecretName:	default-token-f3ech
QoS Tier:	BestEffort
Events:
  FirstSeen	LastSeen	Count	From			SubobjectPath		Type		Reason		Message
  ---------	--------	-----	----			-------------		--------	------		-------
  2m		2m		1	{default-scheduler }				Normal		Scheduled	Successfully assigned goweb-907329357-jiau5 to minikube
  2m		2m		1	{kubelet minikube}	spec.containers{goweb}	Normal		Pulling		pulling image "darek/goweb:1.0"

You can that the pod is pulling the image, Events section is very helpful. After some time the pod, and the deployment should be ready.

➜ kubectl get deployments
NAME      DESIRED   CURRENT   UP-TO-DATE   AVAILABLE   AGE
goweb     1         1         1            1           3m

Remember that in order to get a detailed insight into pods and deployments with a describe command. You can also see logs from a pod with a kubectl logs PODNAME command.

➜ kubectl logs goweb-907329357-dnycx 
2016/12/12 06:33:26 Starting goweb 1.0

Let's define a service, which will group our pods, expose a port and provide discoverability. Type NodePort says that the port will be exposed.

➜ kubectl expose deployment goweb --type=NodePort --port=8080
➜ kubectl get services
NAME         CLUSTER-IP   EXTERNAL-IP   PORT(S)    AGE
goweb        10.0.0.191   <nodes>       8080/TCP   9s
kubernetes   10.0.0.1     <none>        443/TCP    76d

In order to access the service to its exposed port, you can use minikube's helped command: minikube service goweb. It will open your browser right to the address:port of the service. If you have the httpie tool installed (pip install httpie), you use this oneliner:

http $(minikube service  goweb --url)
HTTP/1.1 200 OK
Content-Length: 130
Content-Type: text/plain; charset=utf-8
Date: Sun, 11 Dec 2016 17:35:09 GMT

hi from d75e0a562f0f ! version 1.0
Accept-Encoding:[gzip, deflate]
Accept:[*/*]
User-Agent:[HTTPie/0.9.2]
Connection:[keep-alive]

Let's now scale the service to 4 pods:

➜ kubectl scale --replicas=4 deployment/goweb

Now, when you check the deployment, you should see 4 pods running:

➜ kubectl get deployment
NAME      DESIRED   CURRENT   UP-TO-DATE   AVAILABLE   AGE
goweb     4         4         4            4           33m

Let's check if service is keeping our pods safe, let's kill one:

➜ kubectl get pods
NAME                    READY     STATUS    RESTARTS   AGE
goweb-988856142-0dc3a   1/1       Running   0          2m
goweb-988856142-dxqvb   1/1       Running   0          1m
goweb-988856142-jbdyl   1/1       Running   0          1m
goweb-988856142-xhm5y   1/1       Running   0          6m
➜ kubectl delete pod goweb-988856142-xhm5y
pod "goweb-988856142-xhm5y" deleted
➜ kubectl get pods
NAME                    READY     STATUS              RESTARTS   AGE
goweb-988856142-0dc3a   1/1       Running             0          3m
goweb-988856142-dxqvb   1/1       Running             0          2m
goweb-988856142-xhm5y   1/1       Running             0          7m
goweb-988856142-z1p7b   0/1       ContainerCreating   0          1s

The above listing shows that immediately k8s healed our service, taking it back to intended 4 pods running. Let's now do an update:

➜ kubectl set image deployment/goweb goweb=darek/goweb:2.0

Go to the Web brower and check if a container is anouncing a 2.0 version:

http $(minikube service  goweb --url)
HTTP/1.1 200 OK
Content-Length: 130
Content-Type: text/plain; charset=utf-8
Date: Sun, 11 Dec 2016 17:35:09 GMT

hi from e95e0d562e0a ! version 2.0
Accept-Encoding:[gzip, deflate]
Accept:[*/*]
User-Agent:[HTTPie/0.9.2]
Connection:[keep-alive]

You can also verify if a pod is indeed updated to 2.0 by executing kubectl describe pod PODID or accessing the logs, where you should see version 2.0: 2016/12/11 17:51:38 Starting goweb 2.0.

We can now delete the cluster and stop the minikube.

➜ kubectl delete service,deployment goweb
service "goweb" deleted
deployment "goweb" deleted
➜ minikube stop

Dariusz Dwornikowski @tdi