SOA-C03 - Deployment, Provisioning, and Automation - Section 3.5

Implement deployment strategies and use third-party tools such as Terraform and Git to automate resource deployment.

Implement blue/green and rolling deployment strategies to reduce downtime and risk during application releases, and use Terraform and Git to version and automate infrastructure changes. Distinguish when a blue/green cutover offers safer rollback compared to an in-place rolling update.

Blue/green deploymentRolling deploymentTerraformGit

Practice question for this objective

Free sampleDeployment, Provisioning, and Automationhard

An update to a CloudFormation stack failed and CloudFormation attempted to roll back, but the stack is now stuck in UPDATE_ROLLBACK_FAILED because one resource could not be returned to its previous state. The team needs the stack to reach a stable, updatable status again so they can resume normal deployments, ideally after fixing the offending resource by hand. What is the appropriate recovery step?

  • AFix the failed resource manually, then call continue update rollback, optionally skipping that resource, so the stack returns to UPDATE_ROLLBACK_COMPLETE. Correct
  • BDelete the whole stack and recreate it from the template, since UPDATE_ROLLBACK_FAILED leaves no supported path back to a usable state.
  • CRun drift detection on the stack so CloudFormation repairs the failed resource and moves the stack out of the UPDATE_ROLLBACK_FAILED status.
  • DCreate a change set against the failed stack so CloudFormation recalculates the rollback and clears the UPDATE_ROLLBACK_FAILED status automatically.
Recover a stack stuck in UPDATE_ROLLBACK_FAILED by fixing the resource and calling continue update rollback, skipping it if required. UPDATE_ROLLBACK_FAILED means CloudFormation could not restore at least one resource to its pre-update state. The supported recovery is to correct the underlying problem on that resource manually and then issue continue update rollback, optionally listing the resource to skip so CloudFormation stops trying to roll it back. The stack then settles into UPDATE_ROLLBACK_COMPLETE and becomes updatable again, avoiding any need to delete and rebuild it.

Why A is correct: Continue update rollback is the supported recovery for UPDATE_ROLLBACK_FAILED; correcting the resource out of band and then resuming, skipping it if needed, lets the rollback finish and the stack reaches a stable updatable status.

Why B is wrong: Deleting and recreating destroys healthy resources and risks data loss for a recoverable condition, and it is unnecessary because CloudFormation provides a continue rollback path out of UPDATE_ROLLBACK_FAILED.

Why C is wrong: Drift detection only reports differences between live resources and the template and never modifies resources or stack status, so it cannot move a stack out of UPDATE_ROLLBACK_FAILED.

Why D is wrong: A change set cannot be executed against a stack in UPDATE_ROLLBACK_FAILED because the stack is not in an updatable state, so it neither recalculates the rollback nor clears the failed status.

See more SOA-C03 practice questions, answers explained.

More in this domain

Back to all Deployment, Provisioning, and Automation objectives, or the SOA-C03 cert hub.

Examworthy is not affiliated with or endorsed by Amazon Web Services. Original, blueprint-aligned practice material only.