Docker & Kubernetes
Putting it together - multi-tier
Apply all resources to deploy the app.
Putting it together: the multi-tier deployment
All the files are ready. The deployment order: namespace, secret (by command), settings, database, API, web. Then it's verified that everything is ready:
kubectl apply -f k8s/00-namespace.yaml
kubectl create secret generic db-gizli -n magaza --from-literal=POSTGRES_PASSWORD="$DB_PAROLA"
kubectl apply -f k8s/
kubectl rollout status deploy/db deploy/api deploy/web -n magaza
kubectl get all -n magaza
End-to-end testing
The last step is seeing a request follow the web → API → database path end to end. A temporary container calls the API from inside the cluster; from the browser, the web service is opened:
kubectl run -n magaza deneme --rm -it --image=curlimages/curl:8.10.1 --restart=Never -- curl -s http://api:8000/saglik
{"durum":"ok","veritabani":"bagli"}
minikube service web -n magaza --url
http://192.168.49.2:30080
Exercise Deploy and verify
Write four commands: apply the k8s/ folder; wait for the api deployment's rollout to finish in the magaza namespace; list all basic objects in magaza; print the web service's address with minikube service without opening a browser.
Show example solution
kubectl apply -f k8s/
kubectl rollout status deployment/api -n magaza
kubectl get all -n magaza
minikube service web -n magaza --url
Test yourself
-
What happens if kubectl apply -f k8s/ is run a second time?
- Everything is recreated
- It errors
- The database is deleted
- Only changed objects are updated; the rest stay "unchanged"
Show answer
Only changed objects are updated; the rest stay "unchanged" — apply is idempotent.
-
Delete the temporary pod when done.
kubectl run -n magaza deneme ____ -it --image=curlimages/curl:8.10.1 --restart=Never -- curl -s http://api:8000/saglik
Show answer
kubectl run -n magaza deneme --rm -it --image=curlimages/curl:8.10.1 --restart=Never -- curl -s http://api:8000/saglik
-
Match.
- 00-namespace.yaml
- Applied first
- create secret
- Secret outside the repo
- rollout status
- Wait for ready
- curl http://api:8000
- Internal end-to-end test
Next lesson: kubectl logs
