README.md

Sample Python Application to demonstrate loadbalancing & application update strategies in Openshift

• Rolling Updates
• Blue/Green Deployments
• A/B Deployments This will display the pod hostname which is servicing the web request

This is a fork of the below github repository https://github.com/openshift/django-ex.git

Create a New Project for the sample application deployment (Optional)

$ oc new-project demoproject --display-name="Demo Project" --description="Project for demo purpose"
$ oc project demoproject

Deploy the sample python application and create route to expose it

$ oc new-app python:3.6~https://github.com/dineshsadasivam/demoapp --name=demoapp
$ oc expose svc demoapp --name=demoapp

Display pods servicing demoapp

$ oc get pods -o wide -l app=demoapp

Change the route loadbalacing algorith from the default Source IP to roundrobin and disable cookies for testing.

This will enabled testing of application loadbalacing from single workstation.

$ oc annotate route/demoapp haproxy.router.openshift.io/balance=roundrobin 
$ oc annotate route/demoapp haproxy.router.openshift.io/disable_cookies=true 

Command to validate load-balancing and deployment changes from CLI

$ while true ; do curl http://demoapp-demoproject.router.default.svc.cluster.local 2>/dev/null |grep Hello ; done

Rolling Update/Rollback:

Modify the application (https://github.com/dineshsadasivam/demoapp/tree/master/templates/hello.html) and trigger a new build for rolling update

$ oc start-build demoapp
$ oc rollback demoapp 

Blue-Green Deployments:

The blue-green deployment strategy minimizes the time it takes to perform a deployment cutover by ensuring you have two versions of your application stacks available during the deployment. We can make use of the service and routing tiers to easily switch between the two running application stacks, hence it is very simple and fast to perform a rollback.

A/B Deployments:

A/B deployments get their name from the ability to test the new application features as part of the deployment. This way you can create a hypothesis, perform an A/B deployment, test whether your hypothesis is true or false, and either roll back to your initial application state (A) or proceed with your new application state (B).