Question

A team runs a CodePipeline with a Source stage (CodeCommit), a Build stage (CodeBuild), and a Test stage. The Test stage provisions a full ephemeral test environment using a CloudFormation stack, runs end-to-end tests against it, and must then destroy the environment regardless of test outcome to avoid ongoing charges. Currently they use a single CodeBuild action whose buildspec creates the stack, runs tests, and deletes the stack in the same phase. When tests fail, the build exits immediately and the CloudFormation stack is left running, accumulating cost. Which change most reliably ensures the environment is always torn down?

Accepted Answer

Move the stack deletion commands into the buildspec's finally block under the post_build phase so they execute whether or not preceding commands succeed.

Answer

Add a CloudWatch alarm on the CloudFormation stack cost and trigger a Lambda function to delete the stack when charges exceed a threshold.

Answer

Enable automatic rollback on the pipeline so a failed Test stage reverts all resources created during the execution.

Answer

Set the CodeBuild action's timeout lower so the build terminates faster and CodeBuild automatically cleans up any resources it created.