Continuous Delivery

Tools - Workflows - Culture

Sebastian Helzle

https://github.com/Sebobo@sebobo @punkt_de

Michael Lihs

lihsmi.ch@kaktusmimi @boschglobal

Agenda

• Introduction
• Tools
• Workflow(s)
• Culture

Continuous Delivery - Why?

Tools - Workflow(s) - Culture

Part 1

Tools

Development Environment

• VirtualBox
• Vagrant
• Proper IDE

Version Control

• Git
• Gitlab
• Sourcetree

Continuous Integration

• Jenkins
• Pipeline Plugin
• Docker for ephemeral build containers

Pipeline Example

    node {

        stage 'Clone from Git'
        git url: 'GIT_URL'

        stage 'Build'
        sh 'composer install'
        sh 'grunt'

        stage 'Unit Tests'
        sh 'phpunit --bootstrap src/autoload.php tests/unit'

        stage 'Deploy to Test'
        sh './flow surf:deploy project'

    }

Pipeline Example

Artifact Storage

• Artifactory
  • JS / CSS builds
  • PHP packages (composer)
  • Docker Images

Testing

• PHPUnit
• Selenium
• Behat

Delivery (a.k.a. Deployment)

• Surf
• Fabric

Environments

• Integration, Staging, Production
• Use immutable infrastructure
• If you really need it, use either
  • Ansible
  • Chef
  • ...
• But really: use immutable infrastructure

Logging, Monitoring

• ELK Stack
• Kibana
• Dashing

Tools - Overview

Part 2

Workflow(s)

Development Workflow

• Develop locally
  • Dev Servers are dead
• Pair Programming
• Code Review

Version Control

• Short living feature branches
  • Develop branches are evil
  • Regularly integrate into master
• Merge Request (Gitlab)
• Problem: Deliver tasks independently

Merge Request Workflow

• Create a feature branch (PRJ-123)
• Implement your feature
• Push branch to Gitlab server
• Create merge request
• Assign it to your colleague
• Merge or decline

Feedback Loops

Feedback Loops

• Establish a Continuous Delivery pipeline
• Separate stages from fast to slow(er)
  • Run tests locally (before push)
  • Build on CI server
  • Run Unit Tests
  • Run Integration Tests
  • ...
  • Deploy to Staging
  • Deploy to Production

Do NOT program on the CI Server

• Use pipeline plugin
• Make build pipeline part of project
  • Write a (simple) script
  • Version it

Do not run jobs on Jenkins Master

• Use agents
• Run agents in short living (ephemeral) containers
• Separate build environments
• Provide greenfield to each build

Build once, use everywhere

• Make builds traceable
• Do not build things more than once
• Build an re-use

Part 3

Culture

Feedback

• Personal Feedback
  • Pairing - close session with feedback
  • Retrospectives
  • Customer feedback
  • User feedback
• Automated Feedback
  • Tests
  • Monitoring

KISS

• Keep It Simple Stupid
• Strive for understanding
• Make whole team understand the toolchain
• Motivate by simplicity not by complexity

Prodcut Thinking

over Project Thinking

• Cross-functional teams
• Create value for customer, not just finish projects
• Support and re-think supporting business process
• Take responsibility for success of product

Strive for Agility

• Individuals and interactions >> processes and tools
  • Stable teams
  • Iteratively develop and track processes
• Working software >> comprehensive documentation
  • Convention over documentation
  • Frameworks, well-known software and tools

Strive for Agility

• Customer collaboration >> contract negotiation
  • Include customer in planning and review (and retrospectives?)
  • Provide access to JIRA, test system, monitoring, test results
• Respond to change >> follow a plan
  • Small batches, iterations
  • Get changes on the road quickly
  • Incorporate customer and user feedback

Thank You!

Slides: http://michaellihs.github.io/presentation-cicd-t3cs2016/

by Dani

References

• Jenkins Pipeline Plugin
• Gitlab CI
• Artifactory (Open Source Edition)
• TYPO3 Surf (standalone)
• TYPO3 Console by Helmut Hummel
• ELK Stack
• Jenkins on Docker by Riot Games (part 1)
• Jenkins on Docker by Riot Games (part 2)
• Jenkins on Docker by Riot Games (part 3)
• Jenkins on Docker by Riot Games (part 4)
• Jenkins on Docker by Riot Games (part 5)
• Agile Manifesto

TODOs

• Add goals for Dev Environment
  • Be as close to production as possible
• Add example for Behat tests
• Add slide about traceability
• Provide images for collaboration models
  • cross-functional teams
  • devops teams
• Add some information about (Jenkins) notifications
• Add blue-green deployment
• Add further ideas for imm. infrastrct
• Refine image for CD
• Refine tools overview image
  • Monitoring
  • Dashboards
• RFT idea: show workflow based on "real world example"
  • how to add a feature