My solution creates a new namespace for Nginx to run in, and uses a FQDN in the Nginx config to proxy the Pi web app - and specifies the correct port:
front-end namespacehttp://pi-web-np.pi.svc.cluster.local:8030 as the proxy_pass settingDeploy the namespace first:
kubectl apply -f labs/namespaces/solution/01-namespace.yaml
Then the original proxy setup:
kubectl apply -n front-end -f labs/namespaces/specs/reverse-proxy
Browse to http://localhost:30040 - you'll get an error from your browser
Check the logs and you'll see the proxy won't run if the "upstream" server can't be found:
kubectl logs -n front-end -l app=pi-proxy
You'll see nginx: [emerg] host not found in upstream "pi-web-internal" in /etc/nginx/nginx.conf:28
Update the ConfigMap with the correct FQDN, using the app in the existing pi namespace, and then you'll need to rollout new Pods to pick up the config change:
kubectl apply -f labs/namespaces/solution/nginx-configMap.yaml
kubectl rollout restart -n front-end deploy/pi-proxy
Browse to http://localhost:30040/pi?dp=40000 - now the proxy loads the content from the Pi app; the response will take a couple of seconds
Confirm the cache is being used:
kubectl exec -n front-end deploy/pi-proxy -- ls /tmp
Refresh the web app and your response will be instant
Back to the exercises.