W3M20035.1-cloud-infrastructures

Use Case: Skalierbares Pizza backen mit UI für Metriken Beispiel: Knopf für Pizzaorder wird gedrückt -> Pizzaofen erscheint im UI und backt Pizza (max. Kapazität x Pizzen und Dauer von den Pizzen y min zu Demonstrationszwecken)

Technologie:

• UI (Next.js/React)
• Backend (Typescript/Node.js)
• Oven (C#)
• Minikube/K8s etc.
• Prometheus (Dashboard)
• Ansible (Provisioning und Setup)
• Grafana (Dashboard für Metriken)
• Helm (Kubernetes Deployment)

Getting started

• Minikube
• kubectl

minikube start
minikube addons enable ingress
kubectl get pods -A (to verify that everything is running)
kubectl get svc -n ingress-nginx

Build in minikube context:

macOS / linux: eval $(minikube docker-env)
windows: @FOR /f "tokens=*" %i IN ('minikube -p minikube docker-env --shell cmd') DO @%i or minikube -p minikube docker-env --shell powershell | Invoke-Expression

Docker commands

Frontend: docker build -f containers/frontend/Dockerfile --tag frontend:latest --build-arg NEXT_PUBLIC_WS_URL=ws://chart-example.local/ws .
Backend: docker build -f containers/backend/Dockerfile --tag backend:latest .
Oven: docker build -f containers/furnace/Dockerfile --tag oven:latest .

Helm commands

helm dependency build deployment/charts/application/
helm dependency build <path-to-charts>

helm install test deployment/charts/application/
helm install <releasename> <path-to-charts>

helm dependency update deployment/charts/application/
helm upgrade test deployment/charts/application/

helm uninstall test
helm uninstall <releasename>

kubectl get pods (um laufende Pods zu bekommen)
kubectl logs <podname> (Logs des Pods/Container))

To access the frontend (not needed if ingress is working correctly):
kubectl port-forward svc/test-frontend 3000:3000

Troubleshooting:

Q: I have a ImagePullBackOff error when reading kubectl get pods
A:

• Build all docker images that can not be found (docker build -t <name-of-image-as-in-helm-values.yaml>:<tag-as-in-helm-values.yaml> in the correct folder) -> see Docker commands
• minikube image load <name-of-built-image>:<tag-of-built-image> for each image
• Verify that the error is gone with kubectl get pods

Q: How do I update my pods to the latest code change? TODO: test again and if working remove this todo
A:

• Build the docker image
• Remove the old docker image from minikube (minikube ssh - docker rmi <image-name> - exit)
• minikube image load <name-of-built-image>:<tag-of-built-image> for the updated image
• helm uninstall <releasename>
• helm install <releasename> <path-to-charts>
• Verify that the error is gone with kubectl get pods and kubectl logs <podname>

Q: I can't access pods from hosts although ingress is running. A:
The command kubectl get svc -n ingress-nginx should show LoadBalancer for ingress and an external ip.

NAME                                 TYPE           CLUSTER-IP       EXTERNAL-IP   PORT(S)                      AGE
ingress-nginx-controller             LoadBalancer   10.102.26.215    127.0.0.1     80:30225/TCP,443:30293/TCP   4h31m
ingress-nginx-controller-admission   ClusterIP      10.100.180.199   <none>        443/TCP                      4h31m

Otherwise run: kubectl patch svc ingress-nginx-controller -n ingress-nginx -p '{"spec": {"type": "LoadBalancer"}}' (linux/macos)
on windows: kubectl patch svc ingress-nginx-controller -n ingress-nginx -p '{\"spec\": {\"type\": \"LoadBalancer\"}}'

Please run update-hosts.sh (macos/linux) to update your hosts file based on the ingress ip.

Now, you need to run minikube tunnel to reach the application on 'chart-example.local'.

Q: Restart after updating the image A:
Restart Pods if needed

Sometimes Kubernetes won’t notice :latest changed (because the tag is the same). To force reload:

kubectl rollout restart deployment test-frontend

This deletes old pods and pulls your updated frontend:latest.

Application erreichen:

Anpassen der etc/host datei:
Ort Windows: %windir%\system32\drivers\etc
Hinzufügen des eintrages chart-example.local zur minikube ip: 192.168.49.2 chart-example.local

Anpassen der eigenen etc/hostdatei mit 127.0.0.1 chart-example.local
Aufmachen eines SSH Tunnels für chart-example.local zur vm z.B mit VSCode SSH extension