If you haven't installed Minikube then check out the following link
In this workshop you will create your own deployment. It contains several pods that will run an application. The application will look like this:
Now that we are into the cluster we need to create a namespace, but first let's see what namespaces are available.
kubectl get namespaceNAME STATUS AGE
default Active 1d
kube-system Active 1d
kube-public Active 1dYou probably see something similar to above. Those are the default namespaces that are always there. By default if you don't specify a namespace, all your deployments and changes will be done in the default namespace.
Now lets create your own namespace. Use the command below and change the name in the example to your own name or favorite alias.
kubectl create namespace insert-nameMost commands are easy to use in Kubernetes, switching between namespaces is a different story. Copy and execute the command below to switch to your own namespace. Don't forget to switch the namespace name to fit the one you used earlier. The actions which follow will now only execute within this namespace. Deployments in other namespaces will not be visible. Now lets validate the namespace change with the second command.
kubectl config set-context $(kubectl config current-context) --namespace=<insert-namespace-name>
# Validate it
kubectl config view | grep namespace:
# When using powershell |
kubectl config view | select-string namespace:
Now that we are in the right namespace we will deploy our replicaset. A replicaset is part of the manifest that tells kubernetes how many copies of your pods are desired.
kubectl apply -f https://raw.githubusercontent.com/Wesbest/KubernetesForEveryone/master/Training/k8s_deployment.yamlWe have just deployed the following deployment script to our namespace. It's a deployment with 3 pods.
apiVersion: apps/v1
kind: Deployment
metadata:
name: kubern8sdemo
labels:
app: kubern8sdemo
spec:
replicas: 3
selector:
matchLabels:
app: kubern8sdemo
template:
metadata:
labels:
app: kubern8sdemo
spec:
containers:
- name: kubern8sdemo
image: mvdmeij/k8sdemo:v1
ports:
- containerPort: 80To make sure the deployment went correctly we are going to check the status:
kubectl get deploymentsIf you want to delete the deployment you can use the following command:
kubectl delete deployment <deployment_name>If you didn't delete the deployment, the status should look similar to below. It shows the amounts of pods created during this deployment. As you can see you have a desired amount of 3 pods and the current amount of pods is also 3.
NAME DESIRED CURRENT UP-TO-DATE AVAILABLE AGE
kubern8sdemo 3 3 3 3 43s
Let's now take a closer look at the pods.
kubectl get podsNAME READY STATUS RESTARTS AGE
kubern8sdemo-68fcf74b6-55vjr 1/1 Running 0 1m
kubern8sdemo-68fcf74b6-h8ncl 1/1 Running 0 1m
kubern8sdemo-68fcf74b6-qddsp 1/1 Running 0 1mAs you can see there are three pods created each running with their own name. What would happen if we would delete one pod? Use the command below and change the name of the pod to yours.
kubectl delete pod kubern8sdemo-68fcf74b6-55vjrOkay we have now deleted the pod. Let's take a look at the status of the pods.
kubectl get podsNAME READY STATUS RESTARTS AGE
kubern8sdemo-68fcf74b6-h8ncl 1/1 Running 0 2m
kubern8sdemo-68fcf74b6-qddsp 1/1 Running 0 2m
kubern8sdemo-68fcf74b6-s4j2h 1/1 Running 0 24sIn this example you can see there are still 3 pods running, how is that possible? We have deleted a pod, but we have stated that the desired amount of the pods should be 3. Kubernetes has automatically created a new pod. Above you can see a new pod has been created 24s ago while the other 2 were created 2m ago. The kubernetes master node will always make sure to keep as many available as is desired.
So how do we reduce the amounts of pods? We will have to scale the replicaset. We are going to scale down the deployment to 2 pods. This will be a permanent change to the deployment, unless we later adjust the desired amount of pods again.
kubectl scale deployments kubern8sdemo --replicas=2 Now let's check the amount of pods again.
kubectl get podsNAME READY STATUS RESTARTS AGE
kubern8sdemo-68fcf74b6-h8ncl 1/1 Running 0 7m
kubern8sdemo-68fcf74b6-qddsp 1/1 Running 0 7mAs you can see the amount of pods has been reduced. Let's check the deployment to see the desired amount of pods.
kubectl get deploymentsNAME DESIRED CURRENT UP-TO-DATE AVAILABLE AGE
kubern8sdemo 2 2 2 2 10mThe desired amount is now 2.
We have our application ready, let's make a start to expose it to the outside world. In order to do that we need to create a service. This service will be used to link to the deployment with the use of the selectors and labels. Create the service:
Let's apply the service:
kubectl create -f https://raw.githubusercontent.com/smii/KubernetesForEveryone/master/Training/k8s_service.yaml
To check the status of the service, use the command below.
kubectl get serviceNAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
kubern8sservice NodePort 10.108.57.90 <none> 80:32227/TCP 18mSomething similar should appear. The service has created and can be reachable within the kubernetes space by using the 'kubern8sservice' as name but it's not reachable from the outside world!
Let's add a hostname to the service by using an ingress controller (nginx) to make it reachable from the outside world!
kubectl create -fhttps://raw.githubusercontent.com/smii/KubernetesForEveryone/dev/Training/k8s_ingress.yamlNow let's check which hostname it created for our service!
kubectl get ingressNAME CLASS HOSTS ADDRESS PORTS AGE
example-ingress <none> hello-world.test 172.31.232.29 80 8m54sWe have now connected the ingress to our 'kubern8sservice' service which is connected to the pods with the help of labels and selectors. We can now access our application from the browser as it is exposed through the service ingress. Open a new tab and copy the hostname that returned in a browser. A webapplication should appear. Enable the autorefresh button. We will need that later.
Our developers have worked hard to create a new version of our application. Let's update the deployment! Remember the deployment?
apiVersion: apps/v1
kind: Deployment
metadata:
name: kubern8sdemo
labels:
app: kubern8sdemo
spec:
replicas: 3
selector:
matchLabels:
app: kubern8sdemo
template:
metadata:
labels:
app: kubern8sdemo
spec:
containers:
- name: kubern8sdemo
image: mvdmeij/k8sdemo:v1
ports:
- containerPort: 80As you can see our old deployment uses a docker image with tag v1. The developers recently finished version 5. With a single command we can update the version of our deployed pods to that version. If you want to see how quick the application is updated, make sure to keep an eye on your browser. Before proceeding to the next step, wait till the application has been updated in your browser.
kubectl set image deployment/kubern8sdemo kubern8sdemo=mvdmeij/k8sdemo:v5
It seems that version 5 was a bit too much for our users. Lets roll back a few versions and see what kubernetes does under the hood. If you want to watch live changes to the pods include -w as shown below. Keep an eye on the terminal while executing the version change.
kubectl get pods -wkubectl set image deployment/kubern8sdemo kubern8sdemo=mvdmeij/k8sdemo:v3While watching the live changes, you should see something similar to this:
NAME READY STATUS RESTARTS AGE
kubern8sdemo-7b4db456b-j59sd 1/1 Running 0 2m
kubern8sdemo-7b4db456b-npk75 1/1 Running 0 2m
kubern8sdemo-84ff7fd64-w4ksc 0/1 ContainerCreating 0 2s
kubern8sdemo-84ff7fd64-w4ksc 1/1 Running 0 3s
kubern8sdemo-7b4db456b-j59sd 1/1 Terminating 0 2m
kubern8sdemo-84ff7fd64-m4rzh 0/1 Pending 0 0s
kubern8sdemo-84ff7fd64-m4rzh 0/1 Pending 0 0s
kubern8sdemo-84ff7fd64-m4rzh 0/1 ContainerCreating 0 0s
kubern8sdemo-7b4db456b-j59sd 0/1 Terminating 0 2m
kubern8sdemo-7b4db456b-j59sd 0/1 Terminating 0 2m
kubern8sdemo-7b4db456b-j59sd 0/1 Terminating 0 2m
kubern8sdemo-84ff7fd64-m4rzh 1/1 Running 0 4s
kubern8sdemo-7b4db456b-npk75 1/1 Terminating 0 2m
kubern8sdemo-7b4db456b-npk75 0/1 Terminating 0 2m
kubern8sdemo-7b4db456b-npk75 0/1 Terminating 0 2m
kubern8sdemo-7b4db456b-npk75 0/1 Terminating 0 2mKubernetes is creating a new container before terminating the old one until our desired amount is reached. Use control/cmd c to stop it the pod watch.
Below you can see how your application should now look in the browser:
Clean up your deployment and your service before you go to the next task.
Try to get Super Mario running in your browser.
*tips
Image source: https://hub.docker.com/r/pengbai/docker-supermario/ Don't reinvent the wheel. Use our templates. https://github.com/smii/KubernetesForEveryone/tree/master/Templates Use your favorite editor but the real masters may also use vim ofcourse!
Requirements:
Create a deployment with 2 pods.
Create a services with type NodePort and re-use your public ip adress.
Create an ingress that points to the label previous created service
Well that's about it. Raise your hand if you have questions!