Current Behavior:

Currently, Harness does not support re-running a failed stage from an existing pipeline execution if the pipeline or manifest definition has changed after the execution started. The re-run is validated against the current pipeline definition, and changes can prevent the stage from being re-executed.

Customer Impact:

Customers may need to start a new pipeline execution and repeat already completed stages such as DEV → UAT → PROD, resulting in additional deployment time and operational overhead.

Requested Enhancement:

Allow users to re-run a failed stage using the original execution snapshot/definition, even if the pipeline or manifests have been modified after the execution started.

The expected behavior is:

Existing execution continues to use the definition captured when it started.

Failed stages can be re-run using that original definition.

Pipeline/manifest changes are applied only to new executions.