MLA-C01 Practice Question 67: CI/CD Approval Gate
AWS Certified Machine Learning Engineer – Associate (MLA-C01) · Test 7 · question 7 of 10
An ML team deploys a model-serving stack through AWS CodePipeline whenever code merges to main. Its release policy has three rules. Every change must first be deployed to a staging SageMaker AI endpoint and pass automated integration tests against that endpoint. A failed test must stop the release. The ML lead must explicitly sign off before anything reaches production, and must only ever be asked to approve a build that has already passed its tests.
Which stage order meets the policy?
Show answer and explanation
Answer:
CodePipeline runs stages in order, and a failed action stops the pipeline execution. A CodeBuild test action placed right after the staging deployment therefore blocks every later stage when a test fails. A manual approval action pauses the pipeline until someone approves or rejects it. A rejection, or no response within seven days, counts as a failed action. Placing the approval *after* the tests means the ML lead only ever sees builds that passed, and placing it *before* the production deployment means nothing reaches production without sign-off (option A).
- B is wrong because the ML lead approves before staging is deployed or tested, so the lead can be asked to approve a build that later fails its tests. - C is wrong because the approval comes before any tests run, and the integration tests run against production *after* it is already deployed, so a failing build reaches production. - D is wrong because production is deployed before the tests or the sign-off, which breaks every rule in the policy.
Reference: Add a manual approval action to a stage - AWS CodePipeline