Question

A team uses CDK Pipelines (the CodePipeline construct in the aws-cdk-lib/pipelines module) to deploy a multi-stage application. Developers add a new deployment stage to the pipeline definition in code and merge to the main branch. When the existing pipeline runs, they expect the new stage to appear, but they are concerned about how the pipeline picks up structural changes to its own definition. Which behavior of CDK Pipelines correctly explains how the new stage becomes part of the running pipeline?

Accepted Answer

The pipeline includes a self-mutation step that runs 'cdk deploy' on the pipeline stack itself early in execution, updating the pipeline (adding the new stage) before the application stages run.

Answer

The new stage is only added after an engineer manually runs 'cdk deploy' from a workstation; CDK Pipelines cannot modify its own structure during execution.

Answer

CodePipeline automatically detects the source change and re-synthesizes the pipeline on every action, so no dedicated mutation step is required.

Answer

The new stage is applied only on the next pipeline run after the current one completes, because CloudFormation cannot update a stack while its pipeline is executing.