Lesson 21 / 21
Monitoring & Rollback
A deploy isn't done when it ships — it's done when it's confirmed healthy.
Watch the deploy, not just the pipeline
A green pipeline only proves the code built and passed tests — it says nothing about production. After deploy, watch error rates, latency, and key business metrics for a window before calling it 'done'.
Rollback strategies
When something breaks, the fastest fix is usually to roll back to the last known-good artifact — not to debug live in production. Blue-green makes this instant; rolling deployments roll the batches back the same gradual way.
Automate the trigger
Mature pipelines wire health checks and error-rate alerts to trigger an automatic rollback, instead of waiting for a human to notice and act at 2 a.m.