This project is for demonstrating the different elements for setting up zero downtime deployments.
This demostrates three different levels of zero downtime deployments
- Barebones nginx deployment
- Added readiness and liveness probes
- Added preStop lifecycle hook with sleep statement
- A local kubernetes cluster, I'll provide instructions for minikube but kind would also work
- kubectl
- kustomize (should be included with kubectl)
- An http load tester, I recommend siege
- Start the minikube cluster with
minikube start - Install the traefic ingress controller by running the traefik-install-helm script
- Connect to the traefik load balancer from localhost by running
minikube tunnelin a new terminal window - Apply the base nginx deployment in the k8s folder with
kubectl apply -k k8s - Start running siege
siege -v 127.0.0.1 -c 2 - Modify the nginx deployment manifest
- Redeploy with
kubectl apply -k k8sand monitor the HTTP responses, you should see multiple 502 responses as the pods are terminated and started - Add the readiness / liveness probes to the deployment by running
kubectl apply -k kustom-probes - Modify the nginx deployment manifest again and redeploy, you should see less 502 responses
- Add the prestop lifecycle hook to the deploy by running
kubectl apply -k kustom-prestop-hook - Modify the nginx deployment manifest one more time and redeploy, you should see all 200 responses!