k8s // app setup with ingress

assuming three nodes k8s or minikube cluster

warning / lessons learned

svc: we need to expose the service anyways (ClusterIP is fine enough)
ingress: take good care of ingress class name
ingress: take good care of destination service port

architecture

we use a full-blown setup to have ingress listen on node’s hostNetwork.

app setup

sample hello world app on 8080/tcp

cat > test-lbs.yaml <<EOF
apiVersion: apps/v1
kind: Deployment
metadata:
  name: test-lbs
  labels:
    app: test-lbs
spec:
  replicas: 2
  selector:
    matchLabels:
      app: test-lbs
  template:
    metadata:
      labels:
        app: test-lbs
    spec:

      containers:
        - name: test-lbs
          image: gcr.io/google-samples/hello-app:1.0
          # listens on 8080 anyhow
          #ports:
          #- containerPort: 80
EOF

cluster ip service on 80/tcp

cat > test-lbs-svc.yaml <<EOF
apiVersion: v1
kind: Service
metadata:
  name: test-lbs
  labels:
    app: test-lbs
spec:

  #type: NodePort

  ports:
    - protocol: TCP
      targetPort: 8080
      port: 80
      # no need to force it (30000-32767)
      #nodePort: 30000

  selector:
    app: test-lbs
EOF

ingress for vhost hello.world pointing to the service – beware of the ingress class you are using.

domain=hello.local
class=nginx

cat > test-lbs-ingress.yaml <<EOF
apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: test-lbs
  annotations:
    nginx.ingress.kubernetes.io/rewrite-target: /$1
spec:

  ingressClassName: $class

  rules:
    - host: $domain
      http:
        paths:
          - path: /
            pathType: Prefix
            backend:
              service:
                name: test-lbs
                port:
                  number: 80
EOF

ready to go

kubectl apply -f test-lbs.yaml
kubectl apply -f test-lbs-svc.yaml
kubectl apply -f test-lbs-ingress.yaml

kubectl get deploy test-lbs
kubectl get pods | grep ^test-lbs
kubectl get svc test-lbs
kubectl get ingress test-lbs
kubectl get pods -n ingress-nginx
kubectl get pods -n ingress-nginx -o wide

check ingress through HTTP

assuming full-blown ingress setup

node2=192.168.49.3
node3=192.168.49.4

nmap -p 80,443 $node2
nmap -p 80,443 $node3

curl -i --resolve $domain:80:$node1 $domain
curl -i --resolve $domain:80:$node2 $domain

# ingress-nodeport alternative
    curl -i --resolve hello.local:30080:$node2 http://hello.local:30080/
    curl -i --resolve hello.local:30080:$node3 http://hello.local:30080/

troubles

ingress controller replica pending

Warning  FailedScheduling  2m24s  default-scheduler  0/2 nodes are available: 1 node(s) didn't have free ports for the requested pod ports, 1 node(s) didn't match Pod's node affinity/selector. preemption: 0/2 nodes are available: 1 No preemption victims found for incoming pod, 1 Preemption is not helpful for scheduling..

==> missing required label on nodes (primary=true on minikube vs. ingress-ready=true on kind?)

resources

k8s // app setup with ingress

warning / lessons learned

architecture

app setup

ready to go

troubles

resources

tutorials

rewrite

port vs. targetport

troubles