fix: Detect deployment rollback after service stabilization

[NahutabDevelop]

Issue #, if available:

Fixes #191.

Description of changes:

Problem

When a service has the deployment circuit breaker with rollback enabled, a failed deployment triggers a rollback. The service becomes stable again on the previous task definition, and the action reports success — even though the deployment actually failed.

Fix

After waitUntilServicesStable, verify the deployment actually succeeded:

1. Before updateService, capture the current PRIMARY deployment ID
2. From the updateService response, capture the new PRIMARY deployment ID
3. If the IDs differ (a new deployment was created), call describeServices after stability and check:
   • Deployment not found → rolled back by circuit breaker → fail
   • rolloutState: FAILED → fail
   • Otherwise → success

This uses deployment ID comparison instead of task definition ARN comparison, which correctly handles force-new-deployment , EBS volume changes where the task definition doesn't change across deployments, and anything that would create a deployment that is added to ECS logic in the future.

No new API permissions required — uses the existing ecs:DescribeServices permission.

Testing

• Updated existing unit tests to include deployment data in mocks
• Added unit tests for: rollback detection, FAILED rolloutState, and no-new-deployment skip
• Used my own AWS account for E2E validation with a Fargate service with circuit breaker + rollback enabled, deploying a broken image and confirming the action correctly fails

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[wanis-fahmy]

Hey @omkhegde @amazreech, can someone please release this in GitHub so we can make use of it?
many thanks 🙏

[NahutabDevelop]

It's been released now! @wanis-fahmy 😃