Running Sunbird the new way

Old Approach

Pros:

  • Sunbird was installed using a single script.

  • Very minimal set of variables to be defined prior to install (about 40 odd variables)

  • 1 hour installation

  • Low technical knowledge required to setup

Cons:

  • The configuration file was not version controlled. It was only a local file which had inventory and variables defined
  • Customization was tough for docker images

  • No control on published contents

  • Cannot analyze data patterns

  • Only core components are available for use but not Knowledge Platform and Data Pipeline


New Approach

Pros:

  • All components can be installed - LP, DP and Core service's 

  • Easy customization of all components by forking the desired repository

  • Spin up a Jenkins instance and start building on your own

  • Everything is a Jenkinsfile

  • Version control of build, provision and deploy process using Jenkinsfile

  • Visibility on how Sunbird does build and deploy by just looking at the Jenkinsfiles

  • Docker images can be stored in any registry and artifacts in azure blobs or locally on Jenkins

  • Seamless and quick rollbacks

  • All features, updates and process of Sunbird available to end users

  • 85% of Ansible variables refactored and simplified

  • Only about 100 odd variables needs to initialized for entire setup

  • 600 variables refactored to about 100 odd variables

  • Every configuration can be overridden based on requirements

  • Easy upgrades by just using the latest GA certified tag. One click build and deploy.

  • Total control on publishing content and frameworks 

  • Minimal hardware - Just 10 servers for entire setup

  • Easy scaling of infrastructure - All through Ansible inventory and Jenkinsfiles

  • Production ready setup and easy promotion of code from dev to production (Securing Jenkins is out of scope)

  • Provided scripts to create environment specific jenkins jobs
  • Anyone can run a full stack Sunbird with home brewed customization's

Cons:

  • Takes a day's effort

  • Medium level technical skills required to setup

  • Only azure blobs supported to store artifacts at this point of time (can be skipped if stored locally on Jenkins)

Variables:

Old Inventory and Variables Structure

Private                                                                               Public

                                             


New Inventory and Variables Structure

Private                                                                               Public

                                      


           

How was this done?

  • Converted every freestyle Jenkins job to a Jenkinsfile

  • Published every Jenkinsfile along with ready to use job configurations

  • Constant and non mandatory variables were set to its default which is as per standards 

  • Flexibility to override the defaults and add their own variables
  • Setting up Jenkins and all required jobs using 3 scripts.

  • Mandatory variables provided in a standard template along with detailed descriptions which needs to be filled

  • Documentation on how to setup

Pictorial representation


How will an adopter run this process?

  • Clone sunbird-devops

  • Run jenkins scripts and configure Jenkins (3 scripts)

  • Fill mandatory ansible variables

  • Run provision jobs to setup your environment which enables required softwares

  • Start building the jobs and deploy to servers

  • Create a framework and start using Sunbird

How will an adopter upgrade?

  • Clone sunbird-devops

  • Run the same jenkins scripts which will upgrade existing jobs and create new jobs as per release

  • Add / remove variables as per changelog provided with every release

  • Run build and deploy from new GA tag