Kubectl has support for auto-completion allowing you to discover the available options. This is applied to the current terminal session with source <(kubectl completion bash)

Kubernetes is managed by creating, editing, or deleting objects and resources. Each object has a structure and properties that instructs Kubernetes about the desired state, allowing Kubernetes to make the changes required. To deploy an applicaton, a Deployment needs to be created using kubectl create deployment examplehttpapp --image=katacoda/docker-http-server

View all deployments with kubectl get deployments

A deployment will launch a set of Pods. A pod is a group of one or more containers deployed across the cluster. View all the pods with kubectl get pods

What happens when you run kubectl get pods -o wide?

More details on a Pod can be found asking for the description using kubectl describe pod <pod-name>

Auto complete will help you auto-complete the names of the pod.

Unless specified, Pods are started within the default namespace. List all namespaces with in the cluster with kubectl get namespaces. The namespaces can be used to filter queries to the available objects. kubectl get pods -n kube-system

Pro Tip: Use the shorthand kubectl get ns

As with Deployments, namespaces can be created withkubectl create ns testns

Namespaces can be specified with -n <namespace> such as kubectl create deployment namespacedeg -n testns --image=katacoda/docker-http-server

kubectl get pods -n testns

Kubectl can help scale the number of Pods running for a deployment, referred to as replicas.

kubectl scale deployment examplehttpapp --replicas=5

View the state of the scale with kubectl get deployments as you scale up/down, the number of Pods will update to reflect the desired configuration kubectl get pods

Under the covers, everything within Kubernetes is controllable as YAML. This provides increased control and potentially easier interactive for making changes.

The Kubernetes Object Definitions can be edited using kubectl edit deployment examplehttpapp.

On saving and exiting, the changes requested will be applied. Click Show Solution for Vim tips!

Image can be changed using the set image command kubectl --record=true set image deployment examplehttpapp docker-http-server=katacoda/docker-http-server:v2

kubectl rollout status deployment examplehttpapp

Mistakes happen! Likewise with Rollout status, it can be rolled back to the previous version using kubectl rollout undo deployment examplehttpapp

We've learned how to successfully deploy and manage applications, but we need to make them accessible. This is managed via Services.

The expose command will create a new service for a deployment. The port specifies the port of the application we want to available kubectl expose deployment examplehttpapp --port 80

Like with Pods, we can get and describe Services kubectl get svc kubectl describe svc examplehttpapp

But how does Kubernetes know where to send traffic? That is managed by Labels. Each Object within Kubernetes can have a label attached, allowing Kubernetes to discover and use the configuration. With the previous example, the service knows which Pods to send traffic to based on the label called app. We can use this label to select all the pods for that deployment, within that namespace.

kubectl get services -l app=examplehttpapp -o go-template='{{(index .items 0).spec.clusterIP}}'

curl $(kubectl get services -l app=examplehttpapp -o go-template='{{(index .items 0).spec.clusterIP}}')

Now the application is receiving data, if something goes wrong we need to be able to investigate. Using kubectl it's possible to view the logs for Pods.

The command combines label selectors and querying the Go Template to get the first Pod name.

kubectl logs $(kubectl get pods -l app=examplehttpapp -o go-template='{{(index .items 0).metadata.name}}')

To continue debugging, it's something required to view the CPU or Memory usage of a node or Pod.

View the Node CPU/Memory status kubectl top node

View the Pod CPU/Memory status kubectl top pod