pipeline-cli

The pipeline-cli is an opinionated utility wrapper around oc CLI. It enables developer to run stages of their continuous delivery pipeline from their own workstation or form within a CI/CD server.

The utitility support and promote the following Continuous Delivery patterns:

• Single Command Environment: "run a single command to build and deploy the application to any accessible environment, including the local development."
• Single Path to Production: "Any change can be tied back to a single revision in a version-control system."
• Short-Lived Branches: "Branches must be short lived – ideally less than a few days and never more than an iteration."

As well as support a Github Flow.

How we got here

Version 1: OpenShift Pipeline

This is OpenShift built-in pipeline build strategy.

Pro:

• Easy to setup
• Small footprint

Con:

• It does not managed itself as it requires that the BuildConfig already exists and it is manually managed/maintained
• Issues related to managed Jenkins memory/CPU
• Not very reusable (aside from copy-and-paste)
• High Dependency to Jenkins and Jenkinsfile
• Long Jenkisfile.
• Can't build/deploy from workstation (Without Jenkins)
• No support for PR-based build/deployment.

Version 2: Jenkins Shared Library

Jenkins Shared Library addresses the reusablity aspect, and make it easier to share common pipeline tasks/stages.

Pro:

• Reusable
• Brings consistency accross multiple instances of Jenkins
• Smaller Jenkinsfile
• Opinionated/Configurable

Con:

• Hard to develop/evold the pipeline: Debugging/Troubleshooting requires adding a binch of echo to try to identify issues.
• High Dependency to Jenkins and Jekninsfile
• Can't build/deploy from workstation (Without Jenkins)

Version 3: Groovy Script Library

Pro:

• Reusable
• Single command line build/deployment from Workstation
• Low dependency to Jenkins

Con:

• Java with Gradle + Groovy has a considerably large footprint and long statup. It takes about 5min to actually start build/deployment.
• Complex configuration file

Version 4: JavaScript Node/NPM Module

Pro:

• Reusable
• Single command line build/deployment from Workstation
• Low dependency to Jenkins
• Easy to extend
• Less opnionated

Con:

• Bigger footprint/boilerplate

Why Not ...?

Helm

Helm helps you manage Kubernetes applications — Helm Charts help you define, install, and upgrade even the most complex Kubernetes application.

Operator Framework

The Operator Framework is an open source toolkit to manage Kubernetes native applications, called Operators, in an effective, automated, and scalable way.

OpenShift Do (Odo)

Odo utilizes Source-to-Image to create reproducible Docker images from source code. Odo handles the complex task of building, pushing and deploying your source code.

Coverage of fast tests:

DEBUG=info:* npm run coverage

Reset dependencies

rm -rf node_modules; npm install

Debug

npm link