Continuous Deployment at Kitchensurfing
10 likes6,393 views
Continuous deployment is the automated process of shipping code to production with every push to the master branch. Some benefits include earlier feedback, faster development, and lower risk of deployments. Kitchensurfing uses continuous deployment with Ruby on Rails, MongoDB, and Heroku. Their process includes code reviews, testing, and automated deployment to Heroku after successful builds. MongoDB helps avoid schema migrations during product development. They emphasize trusting test suites, releasing in small pieces, using feature flags, and observing production after deployments.
Continuous Deployment at Kitchensurfing
- 1. CONTINUOUS DEPLOYMENT Lars Kluge, CTO and Co-founder of Kitchensurfing
- 6. WHO IS FAMILIAR WITH …?
- 7. DEFINITION
- 8. Continuous Deployment is the automated process of shipping your product to production, with every push to master.
- 10. MOTIVATION? Earlier feedback for your business: get features in front of users as early as possible Faster development: develop, push, next feature Lower the risk of deployments Motivation for everyone involved: changes can be done immediately--no wait for the next scheduled release
- 11. WHO IS USING IT? Facebook, Etsy, Quora, Linkedin, …
- 12. IS IT PRACTICAL FOR SMALLER STARTUPS?
- 13. YES. We use it and love it at Kitchensurfing.
- 14. OUR STACK Ruby on Rails, MongoDB, Heroku, …
- 15. OUR WORKFLOW
- 16. 1. Pick up a ticket in Pivotal Tracker 2. Code 3. Commit with reference to ticket id 4. Pull Request on Github 5. Code Review 6. Multiple Staging Environments if manual check necessary 7. CI: Codeship runs test suite for pull request 8. Merge into master 9. Github notifies Pivotal Tracker that ticket is merged 10. CI runs again 11. On successful build, Codeship deploys to Heroku 12. Release Notes Email sent by Heroku
- 17. OUR LEARNINGS
- 18. MONGODB HELPS.
- 19. NO SCHEMA.
- 20. ETSY source
- 21. WITH MONGODB: While (re)inventing your product, no* schema migration necessary.
- 22. YOUNG PRODUCT = A LOT SCHEMA CHANGES
- 23. TRUST YOUR TEST SUITE
- 24. RELEASE BIG PRODUCTS IN SMALL PIECES
- 25. USE FEATURE FLAGS Show new features only to your beta user group Avoid the 'big bang' release
- 26. OBSERVE PRODUCTION AFTER DEPLOY Not only exception tracking How are business #s changing? Cloud behavior MMS (MongoDB Monitoring Service) New Relic
- 27. BEHAVIOR CHANGE Is your team ready to make the behavior change? The whole team needs to support it. Introduce Continuous Deployment as early as possible; it's getting harder down the road.
- 28. PRODUCT TEAM How to break down features into small, easy to release pieces? What is the order of operation?
- 29. COMMUNICATION Keep your team in the loop Release Notes Email What is online, what's not? Ticket finished, does it mean it's online?
- 30. RUNTIME OF TEST SUITE > 15 Min. tricky Context Switch is expensive for Engineers
- 31. HEROKU PREBOOT $ heroku labs:enable -a myapp preboot
- 32. OUR FUTURE PLANS Better Release Notes Email based on finished stories in Tracker Statistics in Pull Request to understand the change based on compiled JS, CSS size, test suite build time, # of database queries, etc. Engineering Dashboard: See how a deploy changes business #s