This repo is an prototype on the deployment architecture of a microservice and currently dev in progress.
This should be visualized as a baseline to create a microservice deployment leveraging the following capabilities
| SNo | Artifact | Capabilities |
|---|---|---|
| 1 | Core Kubernetes | Replication Set,HPA,Service Discovery |
| 2 | Helm Chart | Deployment template,Umbrella chart,subcharts |
| 4 | Resource Monitoring & Logging | k8s dashboard,Grafana,ServiceGraph,Jaeger,Kibana |
Note This repo doesn't cover the OAuth2 spring implementation code itself.That is part of a different repo and will be published soon.This repo talks only on the deployment part of the infrastructure.
Minikube installed,kubectl is configured to talk to some K8S cluster, also helm and Istio installed.
os - Windows 10 with headless Vbox
kubernetes Cluster -Minikube
As this local deployment focus only on CD part of the flow, Docker images are build locally and injected to Minikube
To inject docker image to a cluster simply login to the cluster using minikube ssh command and then execute docker build command
The Docker images under /docker have 2 basic spring boot app build as myapp as a frontend app which interns calls a backend microservice backapp.
Also Ideally in an actual cluster there should be an LB that will relay traffic to Istio gateway but as this is a local Minikube instance we are going to expose the Istio ingress service as NodePort so that we can connect to it using Kube IP. Use Kubectl apply command to inject /local/ingress_add_port.yamlconfig changes to the cluster
The Umbrella chart installs its' subcharts, which in turn depend on the Common chart Use the following command
helm-dep-up-umbrella.bat <<rc1>>
where rc1 is the release name.
This will first compile charts and then deploy the entire set of microservice frontend,backend and network to K8s cluster.To modify the image tags and other configurations,tweak below to config files as per need
--/global/values holds pod deployment related configs modify image tag version,canary type,service port, hpa etc here
--/network/route/values lays down the config for intelligent routing
to deploy a canary deployment update /global/values for istio.carnary to true and probably use a different image tag.Then run the umberella command as usal to deploy a new set
helm-dep-up-umbrella.bat <<rc2>>
where rc2 is the canary release name.
Once deployed check for all deployments using
kubectl get pods,svc,virtualservice,deployment,destinationrule,hpa
Open various mangement portals using the link set displayed in console post executing the umbrella script.
K8s Dashboard: run minikube dashboard
grafana : http://localhost:3000
tracing : http://localhost:16686
ServiceGraph : http://localhost:8088/force/forcegraph.html
Kibana : http://192.168.99.119:30621 (Disabled in Local for memory usage)
Check loadbalancing, autoscaling, tracing features firing up the pods with random requests as below
$while true; do curl -s http://192.168.99.119:31391/hello; echo ""; sleep 1; done
use liveness and readiness probes for a graceful shutdown.
Note: Current readiness probe is not functioning properly and it is in dev in progress.
Use the umbrella chart values to deploy a set of microservice as needed by modifying /global/values
Internally the frontend microservice calls the backend microservice using the K8s Service registry DNS.The virtual service on top of this creates the additional intelligent routing based on the weight-age or other header parameters passed.
public String index() throws UnknownHostException {
LOG.info("Printing Hello World message");
RestTemplate restTemplate = new RestTemplate();
String resourceUrl = "http://"+"backapp";
ResponseEntity<String> response = restTemplate.getForEntity(resourceUrl, String.class);
return " Message from frontend modified is "+version+ InetAddress.getLocalHost().getHostName()+"\n\n"+"Message from backend is: " + response.getBody();
}```