README

Service Design

Dockerized and used Java and Go applications with the names/tags of

• Java: mert/wr:java
• Go: mert/wr:go

Since we're focused on Canary Deployment by specifying weights for applications, I decided to use one service and two deployments. Components are as listed

• 1 Service: wr
• 2 Deployments(Subsets): wr-go and wr-java
• 1 Virtual Service: wr
• 1 Destination Rule: wr
• 1 Ingress Gateway: wr-gateway

I'd like to point out that Istio also supports weighted routing between two different services. But this is not the case here.

Setup

This step includes the environment to implement the solution.

Environment

• Windows 10 Host
• Ubuntu 18.04 LTS Guest running on Oracle VMBox with
  • 4 Processors
  • 12 GiB Memory
  • 20 GiB VDI Storage

Installed Tools on Ubuntu

• Minikube (v1.5.2)
• kubectl (Client: v1.16.3, Server: v1.15.2)
• Istio (v1.4.0)

Operations

1-) Initiliazing Minikube and Single Node Kubernetes Cluster

$ sudo minikube start --memory=8192 --cpus=4 --kubernetes-version=v1.15.2 --vm-driver none
$ sudo mv /home/\$USER/.kube /home/\$USER/.minikube \$HOME
$ sudo chown -R \$USER /home/\$USER/.kube /home/\$USER/.minikube

2-) Installing Istio

$ curl -L https://git.io/getLatestIstio | ISTIO_VERSION=1.4.0 sh -
$ cd istio-1.4.0
$ for i in install/kubernetes/helm/istio-init/files/crd*yaml; do kubectl apply -f $i; done
$ kubectl apply -f install/kubernetes/istio-demo.yaml
$ watch -n1 kubectl get pods -n istio-system #Wait pods to be Running
$ kubectl label namespace default istio-injection=enabled #Enable sidecar injection

3-) Deploying the Service and two Deployments

• After ensuring that you've the Java and Go Docker images, extract the Kubernetes manifests from .zip, deploy service.yml and deployment_go.yml, deployment_java.yml by

   $ kubectl apply -f service.yml
   $ kubectl apply -f deployment_go.yml
   $ kubectl apply -f deployment_java.yml
  

• (Optional) Added two HPA files for both deployments. If you'd like to provision pod autoscaling, you can execute

   $ kubectl apply -f hpa_go.yml
   $ kubectl apply -f hpa_java.yml
  

• To test whether the deployments are working or not, port-forward the deployments by

  $ kubectl port-forward deployment/wr-go 8080:8080 (curl http://localhost:8080)
  $ kubectl port-forward deployment/wr-java 8081:8080 (curl http://localhost:8081)
  $ # OR
  $ curl $(kubectl get service -n default |grep wr | awk '{print $3}'):8080
  

• After ensuring that the deployments are working as expected, create an Ingress Gateway, Virtual Service and Destination Rule by

  $ kubectl apply -f ingress_gateway.yml 
  $ kubectl apply -f virtual_service.yml 
  $ kubectl apply -f destination_rule.yml
  

• Modify /etc/hosts file to map Ingress Gateway IP with wr.com by

  $ echo "$(kubectl get service -n istio-system | grep ingress | awk '{print $3}') wr.com" | sudo tee -a /etc/hosts
  

• To test the connection for Ingress Gateway and make requests for Kiali monitoring later, execute

  - for ((i=1;i<=100;i++)); do curl -s -o /dev/null -v "wr.com"; sleep 0.2; done
  

• After making 100 requests, we can now see that the Go subset gets approximately two out of three requests and Java subset gets one of three.

  - xdg-open "https://$(minikube ip):$(kubectl -n istio-system get service kiali -o jsonpath='{.spec.ports[?(@.port==20001)].nodePort}')/kiali/console"