b-behan/hello-kubernetes

Provides a demo image to deploy to a Kubernetes cluster. It displays a message, the name of the pod and details of the node it's deployed to.

Initial Setup

Install kubectl

The Kubernetes kubectl client is used to access the Kubernetes cluster. The following describes how to install the client: https://kubernetes.io/docs/tasks/tools/install-kubectl/

Add the kubeconfig file

The kubectl client uses the kubeconfig file to determine how to communicate with the Kubernetes API server in various clusters. By default it will look for a $HOME/.kube/config file. Take a backup of any existing config file, and save the provided config file to that location.

Check that kubectl is configured correctly

kubectl cluster-info

Access Monitoring Services

Foglight

An instance configured to monitor this cluster can be accessed here: https://dpfog.westus.cloudapp.azure.com:8443/

Kubernetes Dashboard

Start a proxy to the Kubernetes cluster:

kubectl proxy

Access the Kubernetes Dashboard: http://localhost:8001/api/v1/namespaces/kube-system/services/kubernetes-dashboard/proxy

Grafana

Start a second proxy to forward local traffic to the Grafana service:

kubectl port-forward svc/grafana 3000:80 

Access Grafana via the proxy: http://localhost:3000

Deploy a Service

1. Edit the yaml/hello-kubernetes.yaml file
2. Search and replace in the file to change all instances of “foo” to your username (e.g. “jsmith”)
3. Deploy the service with kubectl apply -f yaml/hello-kubernetes.yaml
4. Navigate to service specific endpoint http://13.88.137.222/jsmith

Change the Welcome Message

1. Edit the yaml/hello-kubernetes.yaml file
2. Change the value of the MESSAGE environment variable
3. Deploy the updated configuration with kubectl apply -f yaml/hello-kubernetes.yaml

Test Self Healing

1. Use kubectl to ssh into a pod: kubectl exec -it svc/hello-jsmith sh
2. List the processing running: ps -ef
3. Kill the node server (PID 17?): kill -9 17
4. Will be disconnected since the pod ends when the container exits
5. Should still be able to access the service specific endpoint http://13.88.137.222/jsmith

Force Deployment to Fail

1. Edit the yaml/hello-kubernetes.yaml file and modify the request and limit for memory to 1MiB:

          resources:
            requests:
              memory: 1Mi
              cpu: 0.05
            limits:
              memory: 1Mi

2. Deploy the updated configuration: kubectl apply -f yaml/hello-kubernetes.yaml
3. Pod enters crash loop
4. Service is still accessible because of rolling deploy leaves original version of pod running

Make Pod Unschedulable

1. Edit the yaml/hello-kubernetes.yaml file and modify the request and limit for memory to 1MiB:

          resources:
            requests:
              memory: 100Gi
              cpu: 0.05
            limits:
              memory: 100Gi

2. Deploy the updated configuration: kubectl apply -f yaml/hello-kubernetes.yaml
3. Kubernetes is not able to schedule the pod to run on any node in the cluster
4. Service is still accessible

Restore the Resource Configuration

1. Edit the yaml/hello-kubernetes.yaml file and restore the request and limit for memory to:

          resources:
            requests:
              memory: 32Mi
              cpu: 0.05
            limits:
              memory: 64Mi

2. Deploy the updated configuration: kubectl apply -f yaml/hello-kubernetes.yaml
3. Kubernetes is able to complete the deployment