Keeping Pipelines DRY

Conference : https://hasgeek.com/rootconf/2020-delhi/
Talk: https://www.youtube.com/watch?v=B-HR_2PrLKg
Rootconf Flash talk 2020

Transcript

1. Keeping Pipelines DRY Nancy Chauhan @nancy_nano_
2. Overview A typical enterprise runs hundreds or even thousands of CI/CD pipelines on Jenkins. • A lot of steps are common - build, test and deploy steps are usually same. • Such code is usually duplicated. • Making updates is difficult. You need to change shared code across all places.
3. Jenkins Shared Libraries To The Rescue • Jenkins pipeline is Groovy, has access to all JVM features. • You can define shared scripts, callable as functions from individual pipelines • Enables code reuse and reduce time to market for dev teams. • A code change to our shared library goes live immediately and is consumed the next time a team triggers their project's pipeline.
4. Example Use case: Deploying to Kubernetes Suppose you have a Kubernetes infrastructure with separate staging and production clusters. • Deployment steps : They are similar across different projects, typically involving. ◦ Language specific build ◦ Docker build ◦ Helm upgrade • All of this can be neatly packaged as a Jenkins shared library so that pipeline consumers can simply:
5. Example Use case: Deploying to Kubernetes stage (‘build’) { // language specific things here sh ‘./gradlew build shadowJar’ } stage (‘package’) { docker_build(‘java’) } stage (‘deploy staging’) { deploy(‘staging’) } And we are done!
6. Example Use case: Deploying to Kubernetes • The actual definitions of docker_build and deploy are stored in a central location as a Jenkins library which is backed by a git repository for the same. • Libraries can utilize third artifacts from Maven by using Groovy specific @Grab declarations giving you endless possibilities of extensibility such as importing Kube client for Java and making API calls from the pipeline code itself. • So let's say you wish to change your Kubernetes deployment tool from Helm to Tanka, you simply commit your changes to the git repo and it is available across all projects.
7. Thank You