Question

A company deploys a Java application to a fleet of 20 EC2 instances behind an Application Load Balancer using AWS CodeDeploy with an in-place deployment and a CodeDeployDefault.OneAtATime configuration. The team wants to ensure that if a newly deployed instance fails its application-level health validation, the deployment stops immediately and the already-deployed instances are automatically reverted to the last known good revision without manual intervention. Which combination of configuration changes best achieves this requirement?

Accepted Answer

Add a ValidateService lifecycle hook script in appspec.yml that exits with a non-zero code on failure, and enable automatic rollback on deployment failure in the CodeDeploy deployment group.

Answer

Switch to a CodeDeployDefault.AllAtOnce configuration and rely on the ALB health checks to remove unhealthy instances from the target group.

Answer

Add a BeforeAllowTraffic hook that runs integration tests and configure a CloudWatch alarm on CPU utilization to trigger a rollback.

Answer

Enable a blue/green deployment on the same instances and set the original instances to terminate immediately after traffic shifts.