Canary ¶

Ingress Nginx Has the ability to handle canary routing by setting specific annotations, the following is an example of how to configure a canary deployment with weighted canary routing.

Create your main deployment and service ¶

This is the main deployment of your application with the service that will be used to route to it

echo "
---
# Deployment
apiVersion: apps/v1
kind: Deployment
metadata:
  name: production
  labels:
    app: production
spec:
  replicas: 1
  selector:
    matchLabels:
      app: production
  template:
    metadata:
      labels:
        app: production
    spec:
      containers:
      - name: production
        image: registry.k8s.io/ingress-nginx/e2e-test-echo@sha256:6fc5aa2994c86575975bb20a5203651207029a0d28e3f491d8a127d08baadab4
        ports:
        - containerPort: 80
        env:
          - name: NODE_NAME
            valueFrom:
              fieldRef:
                fieldPath: spec.nodeName
          - name: POD_NAME
            valueFrom:
              fieldRef:
                fieldPath: metadata.name
          - name: POD_NAMESPACE
            valueFrom:
              fieldRef:
                fieldPath: metadata.namespace
          - name: POD_IP
            valueFrom:
              fieldRef:
                fieldPath: status.podIP
---
# Service
apiVersion: v1
kind: Service
metadata:
  name: production
  labels:
    app: production
spec:
  ports:
  - port: 80
    targetPort: 80
    protocol: TCP
    name: http
  selector:
    app: production
" | kubectl apply -f -

Create the canary deployment and service ¶

This is the canary deployment that will take a weighted amount of requests instead of the main deployment

echo "
---
# Deployment
apiVersion: apps/v1
kind: Deployment
metadata:
  name: canary
  labels:
    app: canary
spec:
  replicas: 1
  selector:
    matchLabels:
      app: canary
  template:
    metadata:
      labels:
        app: canary
    spec:
      containers:
      - name: canary
        image: registry.k8s.io/ingress-nginx/e2e-test-echo@sha256:6fc5aa2994c86575975bb20a5203651207029a0d28e3f491d8a127d08baadab4
        ports:
        - containerPort: 80
        env:
          - name: NODE_NAME
            valueFrom:
              fieldRef:
                fieldPath: spec.nodeName
          - name: POD_NAME
            valueFrom:
              fieldRef:
                fieldPath: metadata.name
          - name: POD_NAMESPACE
            valueFrom:
              fieldRef:
                fieldPath: metadata.namespace
          - name: POD_IP
            valueFrom:
              fieldRef:
                fieldPath: status.podIP
---
# Service
apiVersion: v1
kind: Service
metadata:
  name: canary
  labels:
    app: canary
spec:
  ports:
  - port: 80
    targetPort: 80
    protocol: TCP
    name: http
  selector:
    app: canary
" | kubectl apply -f -

Create Ingress Pointing To Your Main Deployment ¶

Next you will need to expose your main deployment with an ingress resource, note there are no canary specific annotations on this ingress

echo "
---
# Ingress
apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: production
  annotations:
spec:
  ingressClassName: nginx
  rules:
  - host: echo.prod.mydomain.com
    http:
      paths:
      - pathType: Prefix
        path: /
        backend:
          service:
            name: production
            port:
              number: 80
" | kubectl apply -f -

Create Ingress Pointing To Your Canary Deployment ¶

You will then create an Ingress that has the canary specific configuration, please pay special notice of the following:

The host name is identical to the main ingress host name
The nginx.ingress.kubernetes.io/canary: "true" annotation is required and defines this as a canary annotation (if you do not have this the Ingresses will clash)

The nginx.ingress.kubernetes.io/canary-weight: "50" annotation dictates the weight of the routing, in this case there is a "50%" chance a request will hit the canary deployment over the main deployment

echo "
---
# Ingress
apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: canary
  annotations:
    nginx.ingress.kubernetes.io/canary: \"true\"
    nginx.ingress.kubernetes.io/canary-weight: \"50\"
spec:
  ingressClassName: nginx
  rules:
  - host: echo.prod.mydomain.com
    http:
      paths:
      - pathType: Prefix
        path: /
        backend:
          service:
            name: canary
            port:
              number: 80
" | kubectl apply -f -

Testing your setup ¶

You can use the following command to test your setup (replacing INGRESS_CONTROLLER_IP with your ingresse controllers IP Address)

for i in $(seq 1 10); do curl -s --resolve echo.prod.mydomain.com:80:$INGRESS_CONTROLLER_IP echo.prod.mydomain.com  | grep "Hostname"; done

You will get the following output showing that your canary setup is working as expected:

Hostname: production-5c5f65d859-phqzc
Hostname: canary-6697778457-zkfjf
Hostname: canary-6697778457-zkfjf
Hostname: production-5c5f65d859-phqzc
Hostname: canary-6697778457-zkfjf
Hostname: production-5c5f65d859-phqzc
Hostname: production-5c5f65d859-phqzc
Hostname: production-5c5f65d859-phqzc
Hostname: canary-6697778457-zkfjf
Hostname: production-5c5f65d859-phqzc