Keith and Mario's Guide to Continuous Deployment

This is a presentation that @mariovisic (http://twitter.com/mariovisic) and I (http://twitter.com/keithpitt) performed at RubyConf Australia, 2014.

Recently it has become common practise for development teams to deploy their code several times a day, as well as encouraging new developers to deploy on their first day at work.

In our talk Mario and I will discussed how we used continuous deployment to push these practises to the extreme. Automatically deploying the master branch on new changes is an awesome way to improve your development process.

Automatically deploying master will fundamentally change how you work. Gone are the days of the epic pull request. You'll quickly find yourself writing smaller more manageable chunks of code, that overall have a great impact on the quality of the software you produce.

By the end of the talk you'll know how to change the GitHub merge pull request button into a deploy button - and have the confidence to do so.

Some things we'll go over in the talk:

- How to setup your CI environment for deployments
- Why having fast tests are important
- How to use your Staging environment for testing deployments
- How to use feature flags to hide deployed features from some users
- Zero downtime deploys, even when there are database migrations
- Your new deploy button, AKA The GitHub merge pull request button
- What to do when deployment goes wrong

Proposal: https://github.com/keithpitt/rubyconfau-2014-cfp/tree/master/keith-and-marios-guide-to-continuous-deployment

Transcript

1. CONTINUOUS DEPLOYMENT Keith & Mario’s Guide To @keithpitt @mariovisic OR Why automatically deploying your master branch to production is a good idea.
2. Keith and Mario’s Guide to FAST WEBSITES @keithpitt @mariovisic Previously…
18. microlancer
19. Desktoppr
21. ALRIGHT, LISTEN UP
24. WHY CONTINUOUS DEPLOYMENT?
25. FAST FEEDBACK CONFIDENCE PREDICTABILITY
26. INGREDIENTS
27. INGREDIENTS Feature Planning! Code Reviews! Feature Toggles! Automated Tests! Smoke Tests! Zero Downtime Deploys! Rollback Strategy! Handling Database Migrations! Application Monitoring
28. ! FEATURE PLANNING
29. SMALLEST THING THAT YOU CAN SHIP TODAY Feature Planning
30. HOW DO YOU DEPLOY IT WITH ZERO DOWNTIME? Feature Planning
31. HOW WILL WE HANDLE DATABASE MIGRATIONS? Feature Planning
32. CODE REVIEWS
33. GITHUB PULL REQUESTS Reviewing Code
35. GITHUB.COM/SQUARE/CANE Reviewing Code
36. GITHUB.COM/BBATSOV/RUBOCOP Reviewing Code
37. FEATURE TOGGLES
38. KEEP IT GHETTO Feature Toggles
39. Feature Toggles GITHUB.COM/FETLIFE/ROLLOUT
40. Feature Toggles GITHUB.COM/PDA/FLIP
41. AUTOMATED TESTS
42. WRITE JUST ENOUGH TESTS TO TRUST THE CODE Automated Tests
43. SO THEY’RE FAST Automated Tests
44. COREY HAINES FAST RAILS TESTS Automated Tests
45. SMOKE TESTS
46. RUNS POST DEPLOY AGAINST REMOTE SERVER Smoke Tests
47. BONUS POINTS: THE SAME TESTS RUN AGAINST REMOTE SERVER & LOCAL Smoke Tests
48. ZERO DOWNTIME DEPLOYS
49. DEPLOYING NEW CODE WITHOUT MAINTENANCE PAGE Zero Downtime Deploys
50. HEROKU PRO TIP: HEROKU LABS:ENABLE PREBOOT Zero Downtime Deploys
51. RAILSCASTS TUTORIAL Zero Downtime Deploys
52. DATABASE MIGRATIONS
53. HARDEST PART Database Migrations
55. PGError: ERROR: column "notes" does not exist
56. EASY WAY: MAINTENANCE PAGE Database Migrations
57. HARD WAY: DEPLOY MIGRATIONS FIRST Database Migrations
58. Database Migrations PEDRO BELO ARTICLE
59. BEWARE OF DATABASE LOCKING Database Migrations
60. Database Migrations GITHUB.COM/SOUNDCLOUD/LHM
61. ROLLING BACK DEPLOYS
62. JUST ROLL FORWARD Rolling Back Deploys
63. JUST TURN THE FEATURE OFF Rolling Back Deploys
64. IF YOU HAVE TO: GIT REVERT AND DEPLOY Rolling Back Deploys
65. BEAR GRYLLS WORST CASE SCENARIO
66. TURN ON MAINTENANCE PAGE Rolling Back Deploys
67. APPLICATION MONITORING
68. EXCEPTION TRACKING Application Monitoring
69. PERFORMANCE MONITORING Application Monitoring
70. DAVID GOODLAD BUSINESS METRIC MONITORING Application Monitoring
71. GETTING NOTIFIED Application Monitoring
72. INGREDIENTS Feature Planning! Code Reviews! Feature Toggles! Automated Tests! Smoke Tests! Zero Downtime Deploys! Rollback Strategy! Handling Database Migrations! Application Monitoring
73. TIME TO MIX
77. THE FINAL PRODUCT
92. FAST FEEDBACK CONFIDENCE PREDICTABILITY
93. FAST FEEDBACK CONFIDENCE PREDICTABILITY ✔ ✔ ✔
94. AVERAGE NUMBER OF DEPLOYS PER DAY Manual! Deploys Continuous! Deploys
95. AVERAGE NUMBER OF DEPLOYS PER DAY 1.9 4.7 130 Additions! Per Deploy 245 Additions! Per Deploy Manual! Deploys Continuous! Deploys
97. CONTINUOUS DEPLOYMENT Keith & Mario’s Guide To @keithpitt @mariovisic FIN.