Tell me about a pipeline change or migration in AWS that went wrong and how you handled it.

S: During a migration of our CI/CD pipeline from on‑prem to AWS CodePipeline, the new pipeline failed to trigger builds after a configuration change. T: I was responsible for ensuring the migration completed on schedule. A: I first isolated the issue by reproducing the failure locally, then examined CloudWatch logs and CodePipeline execution history. I discovered that a missing IAM permission caused the pipeline to abort. I updated the role, added a rollback plan, and communicated the issue to the team via Slack and a brief stand‑up. I also set up a canary deployment to validate changes before full rollout. R: The pipeline was restored within 2 hours, the rollback plan prevented downtime, and we documented the fix to avoid future regressions. The incident also led to a review of our IAM policies, improving overall security posture.