fatieelamri / aksdotnet

Sample for AKS and .NET Core

aksdotnet

Kubernetes deployment samples for AKS and .NET Core

Images

Repo is here

• docker.io/dgkanatsios:0.1 - normal working image tagged 0.1
• docker.io/dgkanatsios:0.2 - normal working image tagged 0.2
• docker.io/dgkanatsios:0.2-startupfails - app fails to start
• docker.io/dgkanatsios:0.2-healthfails - health probe fails after 10 seconds

Script

Run watch -n 0.5 kubectl get pods -o wide --sort-by=.metadata.creationTimestamp

Make sure cluster is there and working fine

• kubectl cluster-info: get info about AKS cluster
• kubectl get nodes -o wide: get info about nodes

Imperative creation of deployment

• kubectl create deploy nginx --image=nginx: create an nginx deployment
• kubectl expose deploy nginx --port=80 --type=LoadBalancer: expose it to the outside world
• kubectl get deploy nginx -o yaml: see some YAML for the deployment
• kubectl delete svc nginx && kubectl delete deploy nginx: cleanup

Declarative creation of deployment

Deployment creation

• kubectl apply -f deploy.yaml: this will create version 0.1
• kubectl get po --show-labels: observe that all Pods have the same label as the Deployment template
• kubectl get rs --show-labels: same for the replica set
• kubectl describe rs ....: check the details for the replica set
• kubectl describe po | grep -C 3 'Controlled By': see that the Pods "belong" to the Replica Set

Verify that the service is working

• kubectl get svc: get the public IP and check IP/hello from a browser to verify that the deployment works
• kubectl run -it busybox --rm --image=busybox sh and wget -q0- aksdotnet-deployment/hello: see that version 0.1 is returned
• Get the Pod name in the browser, do kubectl exec -it PodName -- bash to connect to the Pod. Run apt update && apt install net-tools tcpdump procps, then...
  • ps -aux to see the running processes (observer that the .NET process has a PID of 1). Try also running top for real-time information
  • netstat -tulpn to see the open ports from processes
  • tcpdump port 80 and refresh the browser to see the packets coming in
• Do kubectl logs for one of the pods to see logs of incoming requests

ReplicaSet reconciliation loop

• kubectl get po: get the pod names
• kubectl delete po ....: delete a random Pod
• kubectl get po: verify that a new Pod was created
• kubectl describe rs: check the ReplicaSet events

Scaling

• kubectl scale deployment aksdotnet-deployment --replicas=10: this will scale the deployment to 10 replicas
• kubectl describe deploy: check the events about replica set being scaled
• kubectl describe rs: check the events about the Pods being created

maxUnavailable and updating with a wrong image

• kubectl edit deploy aksdotnet-deployment: update tag to a non-existent one (alternatively you can do kubectl set image deployment aksdotnet-deployment aksdotnet docker.io/dgkanatsios/aksdotnet:wrong-tag --record or edit the yaml file and then kubectl apply -f deploy.yaml --record)
• kubectl get rs and kubectl rollout status deploy aksdotnet-deployment: see what happens with maxUnavailable
• kubectl rollout history deploy aksdotnet-deployment: see the deployment history
• kubectl rollout history deploy aksdotnet-deployment --revision=2: see details about a specific revision
• kubectl rollout undo deploy aksdotnet-deployment: undo the last deployment
• kubectl get deploy aksdotnet-deployment: see that everything is good again

Updating to a correct version

• kubectl edit deploy aksdotnet-deployment: update tag to 0.2 - Pod template changes so an upgrade is triggered
• kubectl rollout status deployment aksdotnet-deployment: see the status of the deployment
• echo $?: get the last exit code (of kubectl rollout status deploy) and see that it is not zero
• kubectl describe deploy aksdotnet-deployment: see that the version was upgraded to 0.2
• kubectl rollout history deployment aksdotnet-deployment: see all deployments
• kubectl get rs: see that there is one ReplicaSet per deployment update

Updating to a version with an entrypoint that will error

• kubectl edit deploy aksdotnet-deployment: change tag to 0.2-startupfails and observe what happens with the pods (kubectl get po -w)

Updating to a version with a liveness probe that will error

• kubectl edit deploy aksdotnet-deployment: change tag to 0.2-healthfails and observe what happens with the pods (kubectl get po -w)

I WANNA DO MOARRRR KUBECTL

For more kubectl fun, check out my CKAD-exercises repo.