Continuous Deployment in 3 hours

Transcript

1. Continuous Deployment in 3 hours Mike van Riel Mike van Riel · @mvriel
2. About me • Lead Developer of phpDocumentor • Software Composer at Ingewikkeld • Foster father of 2 happy dogs • 11 years of experience with PHP
3. What is Continuous Deployment, and what is Continuous Delivery?
5. The rules Mike van Riel
6. Feedback Feedback Feedback
7. Versioning is key
8. Build your artifacts once
10. one deployment process for all environments
11. Versioning Mike van Riel
12. Your Code, Configuration and Environment must be in Version Control
13. “To branch, or not to branch. That is the question.” - Bill Shakespeare*
14. Concepts Mike van Riel
15. Environments
16. Deployment Pipeline
17. Deployment Pipeline - optimized
18. Config flags • Toggles features on and off • Allows a team to continuously integrate • Can be used for Canary testing • Provides a rollback in case of flawed functionality
19. Aids Mike van Riel
20. Continuous Integration Servers Jenkins TeamCity Thoughtworks Go
21. Analysis tools • PHPUnit • Behat • PhpDocumentor • PHP_CodeSniffer • PHP_Depend • PHP_PMD • phploc • phpcpd
22. Build and deployment tools • Phing • Ant • Make • Capistrano • Webistrano • Deployinator • Maven
23. More feedback • Logstash • Graphite • Nagios
24. Exercises Mike van Riel
25. Environment • Ubuntu 12.04 • Apache • MySQL • PHP 5.3 • XDebug • One vhost – dev.www.blog.com • Packages – Git – Augeas-tools – Libaugeas-dev – Libaugeas-ruby – Jenkins
26. Exercise 1 Determine your workflow
27. Goal • Think before you start • Determine your process and workflow • Does your current flow match with Continuous Deployment
28. Think about the following • Versioning your application • Versioning your environment • Version numbering • Which tools to use • How to react on a build failure
29. Versioning • Master/trunk should always be deployable • Any change in master / trunk should result in a release candidate • Release branches, develop mainlines and other constructs do not play a role with Continuous Deployment – There are exceptions
30. Exercise 2 Planning your deployment pipeline
31. Goal • Determine how to start automating your build and deployment process • Review your deployment strategy • Automate your deployment process using phing
32. Stages • Commit stage – Pull changes – Clean workspace – Initialize workspace – Run unit tests • Deploy stage – Upload your application – Switch symlink
33. Deployment strategy • Have a symlink representing your current release • Upload your new release to a separate folder • Integrate configuration in the new folder • Create symlink of the new release • Move the symlink of the new release over the current using `mv -Tf`
34. Exercise 3 Apply Continuous Integration
35. Goal • Update jenkins and add phing plugin • Setup a job in Jenkins • Add deployment as a post-build action • Execute our first automated build and deployment
36. Update Jenkins $ wget ­q ­O ­ http://pkg.jenkins­ci.org/debian­stable/jenkins­ci.org.ke y | sudo apt­key add ­ Add to /etc/apt/sources/list: deb http://pkg.jenkins­ci.org/debian­stable binary/ And run: $ sudo apt­get update $ sudo apt­get install jenkins
37. Install phing plugin $ java ­jar jenkins­cli.jar ­s http://localhost:8080 install­plugin checkstyle cloverphp dry htmlpublisher jdepend plot pmd violations xunit git clone­workspace­scm build­pipeline­plugin $ java ­jar /vagrant/puppet/manifests/jenkins­cli.jar ­s http://localhost:8080 install­plugin phing credentials
38. Exercise 4 Adding support for environments using parameters
39. Goal • Extract all variable deployment options into parameters • Learn how to switch between environments
40. Exercise 5 Adding a staging environment
41. Goal • Add a new staging environment to puppet • Change our deployment action to deploy to staging instead of production – using the changes of last exercise
42. Exercise 6 Switching to Continuous Delivery
43. Goal • Add a new view for our project • Add a new deploy job • Persist workspace between jobs • Try out our new process
44. Adding a new Deploy job • Add a new job `blog-deploy` • Configure `blog` job to archive the build • Configure `blog-deploy` job to use the archived build
45. Exercise 7 Adding a UAT stage
46. Goal • Create a Behat test • Alter our build file and add a UAT stage • Add a new UAT job • Persist workspace between jobs • Automatically trigger UAT job after `blog` job • Move deployment to staging from `blog` to UAT job • Test our new process
47. UAT Stage • May run longer than 5 minutes • Can trigger a deployment to staging for exploratory testing • Automate!
48. Bonus exercise 1 Using the pipeline plugin
49. Goal • Use the pipeline plugin to create a new view with your process • Walk through your process
50. Questions? Please leave feedback on Joind.in, I'd appreciate it.
51. Credits • http://www.flickr.com/photos/joebenjamin/5009411920/ • http://en.wikipedia.org/wiki/Continuous_delivery • http://www.flickr.com/photos/robertlemke/4383863479 • http://continuousdelivery.com/2010/09/deployment-pipeline- anti-patterns/ •