Used kops installed Kubernetes cluster on AWS.

Used alb-ingress-controller to do load balancing followed the official guide:

It works. Can run both dig and curl successfully.

Another sample: 2048-game got 504 Gateway Time-out error when access it from Record Set name in Route 53:

http://2048.mysite.com

But can been accessed from DNS name found from Load Balancers! So it seems that the Alias Target in Route 53 not work!

And, for Prometheus, used official chart to deploy and can dig successfully.

helm install prometheus

But can't run curl to get result:

curl server.mysite.com
curl: (7) Failed to connect to server.mysite.com port 80: Connection refused
curl alertmanager.mysite.com
curl: (7) Failed to connect to alertmanager.mysite.com port 80: Connection refused
curl pushgateway.mysite.com
curl: (7) Failed to connect to pushgateway.mysite.com port 80: Connection refused

I checked all the services in the Kubernetes cluster:

$ kubectl get svc --all-namespaces
NAMESPACE     NAME                                          TYPE        CLUSTER-IP       EXTERNAL-IP   PORT(S)         AGE
2048-game     service-2048                                  NodePort    100.68.230.211   <none>        80:31652/TCP    1h
default       kubernetes                                    ClusterIP   100.64.0.2       <none>        443/TCP         4d
default       steely-wombat-prometheus-alertmanager         ClusterIP   100.71.21.190    <none>        80/TCP          5m
default       steely-wombat-prometheus-kube-state-metrics   ClusterIP   None             <none>        80/TCP          5m
default       steely-wombat-prometheus-node-exporter        ClusterIP   None             <none>        9100/TCP        5m
default       steely-wombat-prometheus-pushgateway          ClusterIP   100.65.72.250    <none>        9091/TCP        5m
default       steely-wombat-prometheus-server               ClusterIP   100.65.239.188   <none>        80/TCP          5m
echoserver    echoserver                                    NodePort    100.64.176.267   <none>        80:31281/TCP    1h
kube-system   default-http-backend                          ClusterIP   100.71.27.31     <none>        80/TCP          3h
kube-system   kube-dns                                      ClusterIP   100.64.0.19      <none>        53/UDP,53/TCP   4d
kube-system   tiller-deploy                                 ClusterIP   100.70.101.11    <none>        44134/TCP       4d

I found that sample apps echoserver and 2048-game are NodePort type. The Pormetheus services are ClusterIP type. Is NodePort type necessary here?