Canary Deployments

Canary deployments are a pattern for rolling out releases to a subset of users or servers. The idea is to first deploy the change to a small subset of servers, test it, and then roll the change out to the rest of the servers.

In our below example you will learn following

• using the kubernetes labels to define the canary deployments
• Running multiple version of the application parrallel ,leave it to kubernetes service to load balance between new and old deployments
• test functionality of the new deployment
• Gracefully bring-down the old version

Initial Status of the application

Intermediate Status of the application

Final status of the application

Below is an small spring boot application which respond with greeting message

with initial deployment -> Hello from web2.0

with Final deployment -> Hello from web3.0

Pre requisite installation

• Java 11
• maven
• kubernetes docker for desktop/minikube

Tech

• Java 11
• Springboot
• Lombok
• Maven
• kubernetes
• Docker

Clone

generate spring boot project command-line

curl https://start.spring.io/starter.tgz -d dependencies=web,lombok -d baseDir=canary-microservice-deployment -d bootVersion=2.6.3 -d javaVersion=11 | tar -xzvf -

Copy docker file to root directory with name Dockerfile

FROM openjdk:11.0-jdk-slim
ARG JAR_FILE=target/*.jar
COPY ${JAR_FILE} app.jar
ENTRYPOINT ["java","-jar","/app.jar"]

Build

mvn clean install

Build docker image 

docker build -t greeting . 

docker image ls  // list docker images

Deploy

Kubernetes initial Deployment object with 3 replicas , notice labels with app: greeting and type:canary.

create kubernetes deployment and service file under k8s directory

apiVersion: apps/v1
kind: Deployment
metadata:
  name: web2.0
  labels:
    app: greeting
    type: canary
spec:
  replicas: 3
  selector:
    matchLabels:
      app: greeting
  template:
    metadata:
      labels:
        app: greeting
        type: canary
    spec:
      containers:
        - name: greeting
          image: greeting:latest
          ports:
            - containerPort: 8080
          imagePullPolicy: IfNotPresent

Kubernetes service

apiVersion: v1
kind: Service
metadata:
  name: greetingservice
spec:
  selector:
    app: greeting
  ports:
    - port: 8080
      targetPort: 8080
  type: LoadBalancer
  

kubectl create -f k8s/deployment.yaml

kubectl get deployment,svc,po

Run

while true; do curl http://localhost:8080/; sleep 2; done

You will see the always spring-boot service will respond with Hello from web2.0 

Canary deployment

update kubernetes deployment which uses configmap and diff replicas with new deployment name

apiVersion: apps/v1
kind: Deployment
metadata:
  name: web3.0
  labels:
    app: greeting
    type: canary
spec:
  replicas: 1
  selector:
    matchLabels:
      app: greeting
  template:
    metadata:
      labels:
        app: greeting
        type: canary
    spec:
      containers:
        - name: greeting
          image: greeting:latest
          imagePullPolicy: IfNotPresent
          env:
            - name: greeting.message
              valueFrom:
                configMapKeyRef:
                  name: canary-config
                  key: greeting.message
          ports:
            - containerPort: 8080

Create the configmap required for the new deployment

kubectl create configmap canary-config --from-literal=greeting.message="Hello from web3.0"

check the deployment objects to verify config map deployment

kubectl get deployment,svc,po,cm


deploy updated kubernetes deployment with latest changes as mentioned above

kubectl create -f k8s/deployment-new.yaml

observe new deployment appeared with name web3.0 in the above image

Now 25% of the traffic will route to new service deployment, now verify the service response , which will have mixed response from old and new deployment

while true; do curl http://localhost:8080/; sleep 2; done

once you test and make sure the new changes in the deployment , now scale down the old deployment and make the new deployment 100%. currently we have an 3 replicas of old deployment and 1 replicas of new deployment

kubectl get deploy --show-labels

scale down the old deployment to zero instance and scale up the new deployment to 3 with below command

scale deploy web2.0 --replicas=0
kubectl scale deploy web3.0 --replicas=3

now test the application , which will respond always with new updated code.

while true; do curl http://localhost:8080/; sleep 2; done