DevOps: Safe Deploys and Rollbacks
A bad deploy went out twice this quarter. Design the deployment and rollback strategy that prevents repeat incidents.
Technical Reference & Key Concepts
The Challenge: Deploys That Hurt
Twice this quarter a deploy reached production with a defect, and each rollback was manual and slow. You must own deployment safety.
Core questions to address:
- What gates should exist between 'merged' and 'all users on new version'?
- What deployment strategy would you pick and why?
- How do you detect a bad deploy fast, and how fast can you actually revert?