Exploring kubernetes

Installations

kind

Kind is used to create kubernetes cluster

curl -Lo ./kind https://kind.sigs.k8s.io/dl/v0.15.0/kind-linux-amd64
chmod +x ./kind
sudo mv ./kind /usr/local/bin/kind

Dummy server

1. Build docker

   cd dummy_server/
   docker build --tag dummy-server:v1 .
   

2. Docker run command

   docker run --rm --name server -p 8080:8080 dummy-server:v1
   

Getting started

Setup the kubernetes cluster

1. Create a kubernetes cluster

   kind create cluster
   kubectl cluster-info --context kind-kind  # Verify the cluster info

2. Load the docker images into the cluster

   kind load docker-image dummy-server:v1
   kind load docker-image dummy-server:v2

3. Get the images loaded in the node

   kubectl get nodes
   docker exec -ti <node-name> crictl images

Deploy the app

kubectl create deployment dummy-server --image=dummy-server:v1

Other useful commands

kubectl get deployments  # Get deployment information
kubectl get pods  # Get pods information
export POD_NAME=$(kubectl get pods -o go-template --template '{{range .items}}{{.metadata.name}}{{"\n"}}{{end}}')  # Fetch pod name
kubectl describe pods  # Describes the pods
kubectl describe deployments  # Describes the deployments
kubectl logs $POD_NAME  # Get logs from the pod
kubectl exec $POD_NAME -- env  # Get environments of a pod
kubectl exec -ti $POD_NAME -- /bin/sh  # Enter into the pod container

Expose the port of the app

kubectl expose deployment/kubernetes-bootcamp --type="NodePort" --port 8080
kubectl get services  # List out the services, when we expose a deployment or pod it creates a service

Verify the app

export NODE_PORT=$(kubectl get services/dummy-server -o go-template='{{(index .spec.ports 0).nodePort}}')  # Fetch the node port exposed for this app
kubectl get nodes -o wide  # Get the node ip from the output of this command
curl $NODE_IP:$NODE_PORT

Scale the app

kubectl get deployments
kubectl get rs  # Get replica statuses
kubectl describe deployments/dummy-server
kubectl scale deployments/dummy-server --replicas=2  # Scale up to 2 replicas
kubectl describe deployments/dummy-server

Rolling update

If we keep running a simple script which keeps fetching from the server, initially the response would be This is v1 and it would change to This is v2 as the update completes. This shows us two things, first, the update has taken place, and second, there is no downtime.

kubectl set image deployments/dummy-server dummy-server=dummy-server:v2  # Update the image and k8s will take care of rolling out the update without downtime
kubectl get pods -o wide
kubectl describe deployments/dummy-server
kubectl scale deployments/dummy-server --replicas=1  # Scale down to 1 replica

If the update had a bug/issue, we can rollback to previous version

kubectl rollout undo deployments/dummy-server
kubectl describe services
kubectl describe pods
kubectl describe deployment

Delete the deployments

kubectl delete deployments/dummy-server
kubectl get deployments
kubectl get pods -o wide

Delete the kind cluster

kind delete cluster
kind get clusters

References

1. Kubernetes basics
2. Kubernetes cheatsheet