AZ-400 domain - 53% of the exam

Design and Implement Build and Release Pipelines

Design and Implement Build and Release Pipelines is 53% of the AZ-400 exam. These are the objectives it covers, each with practice questions, with every answer explained.

The domain in numbers

158
Practice questions
13
Objectives
85
Medium
73
Hard

Where people go wrong

  • Using a human approval where an automated check is needed

    The wrong answers reach for an approval or a named reviewer whenever a deployment must be gated, even when the requirement says nobody should click. An approvals check suits sign-off by a person or group. Automated needs map to other environment checks: an Azure Monitor alerts check for telemetry health, an Invoke REST API check for an approved change record held in an external system, and a business hours check for a permitted time window.

    See it in a sample question
  • Overwriting or deleting a published package version

    The wrong answers try to change a version that already exists in an Azure Artifacts feed: rebuilding and re-pushing it, deleting and re-uploading it, unlisting it and publishing again, or promoting it to a view in the hope of replacing its contents. A published version is immutable and its number cannot be reused. Corrected code ships under a new version such as 4.2.1, and promoting a version into a view only changes which consumers resolve that exact build.

    See it in a sample question
  • Confusing canary, ring, blue-green and A/B strategies

    The wrong answers swap the rollout strategies. A canary sends a small percentage of live traffic to the new build and judges it on health against a baseline. A ring deployment promotes through stable audience cohorts with a soak at each ring. Blue-green is one all-at-once cutover between two environments, reverted by switching back. An A/B test runs two variants at the same time for random segments and compares a business metric such as conversion.

    See it in a sample question
  • Putting a trigger in the block for a different event

    The wrong answers use a block that listens for another event. A trigger block governs pushes to branches, a pr block governs pull requests that target a branch and reports a status check, schedules runs on a clock, and a resources pipelines trigger starts a run when another pipeline completes. Path filters then narrow a push trigger, such as including services/payments/* or excluding docs/*.

    See it in a sample question

Objectives in this domain

What to study

The official documents this domain's practice questions cite most.

Sample questions from this domain

Free sampleDesign and Implement Build and Release Pipelineshard

Before promoting a build to the staging "environment", a pipeline must confirm that a downstream ticketing system has recorded an approved change record, querying that external system over HTTPS and proceeding only when it returns an approved status. The check must run unattended on every release. Which environment check type fits this requirement?

  • AAn "Invoke REST API" check that calls the ticketing system endpoint and passes the run only when the returned payload reports the change record as approved, evaluated automatically each release. Correct
  • BA required reviewers approval that lists the change management team, so they look up the ticket in the external system and approve the run once the record is confirmed as approved.
  • CAn "Azure Monitor" alerts check pointed at the ticketing resource, so the pipeline reads fired alerts from the system and proceeds only while the change record alert is inactive.
  • DA business hours check on the environment, so the pipeline holds the deployment until a permitted release window opens and assumes the change record was approved during that window.
Use an Invoke REST API environment check to gate a deployment on an external system response without manual review. The Invoke REST API check calls a configured HTTPS endpoint and evaluates its response, passing or failing the run based on the returned data. This lets a pipeline confirm an external change record automatically on every release, which an approval check, an Azure Monitor check, or a time window cannot do.

Why A is correct: The Invoke REST API check calls an external HTTPS endpoint and gates the run on the response, letting the pipeline confirm the approved change record automatically without any human step.

Why B is wrong: A reviewers approval also blocks promotion, but it relies on a person manually checking the ticket, whereas the requirement is an unattended automated call to the external system on every run.

Why C is wrong: The Azure Monitor check reads alert rules on Azure resources, not arbitrary ticketing endpoints, so it cannot query a third-party change system for an approved status as required.

Why D is wrong: A business hours check only restricts when a run may proceed and never queries the ticketing system, so it cannot confirm that an approved change record actually exists.

Free sampleDesign and Implement Build and Release Pipelinesmedium

A library is published to "Azure Artifacts" at version 1.7.0 and several downstream services already depend on it. A maintainer discovers the published 1.7.0 contains a packaging defect. The team wants consumers to stop selecting 1.7.0 while preserving an audit record and not breaking builds that have already restored it. Which action best meets these constraints?

  • AUnlist or deprecate version 1.7.0 in the feed, so new resolutions avoid it while the version stays available for builds that already reference it and the record is retained. Correct
  • BPermanently delete version 1.7.0 from the feed, so no consumer can resolve the defective build and the bad version disappears from the package list entirely.
  • COverwrite version 1.7.0 in place with a corrected binary under the same number, so consumers automatically receive the fixed build the next time they restore it.
  • DLower the feed retention count so older versions including 1.7.0 are pruned automatically, so the defective build ages out without any manual intervention.
Choose deprecation or unlisting to retire a defective package version while preserving history and existing builds. Deprecating or unlisting a package version removes it from default resolution so new dependency restores skip it, while the version remains in the feed for any build that already references it and its history stays intact. This meets the goals of stopping new use, keeping an audit record, and not breaking existing consumers, unlike deletion or an impossible in-place overwrite.

Why A is correct: Deprecating or unlisting hides a version from normal selection yet keeps it restorable and recorded, which satisfies stopping new use of 1.7.0 without deleting history or breaking existing pinned builds.

Why B is wrong: Hard deletion removes the audit record the team wants kept and can break consumers that lack a cached copy, so it overshoots the requirement to preserve history and avoid breakage.

Why C is wrong: Published package versions are immutable in the feed, so overwriting 1.7.0 is not permitted, and even if it were it would silently change a version that consumers trust to be fixed.

Why D is wrong: Retention policies prune by age and count rules and may remove versions other than the defective one, so they are an imprecise tool that cannot reliably target 1.7.0 while keeping its record.

Free sampleDesign and Implement Build and Release Pipelineshard

A high-traffic checkout service is moving to a risky new payment library. The team wants the new build to receive only a small slice of live production traffic at first, so real customer behaviour and error rates can be observed before the change reaches everyone. They want to grow that slice in steps and halt automatically if failures rise. Which deployment strategy fits this goal?

  • AA blue-green deployment that runs the new build on a parallel environment and cuts all production traffic across in a single switch once smoke tests on the green environment pass.
  • BA recreate deployment that stops every instance of the current version and then starts the new version, so all production traffic moves to the new payment library at the same moment.
  • CA canary deployment that routes a small percentage of live traffic to the new build, compares its error rate against the baseline, and increases the percentage in steps while rolling back if failures climb. Correct
  • DA rolling deployment that replaces production instances in batches at full traffic weight, so each batch serves the new build to its normal share of users until every instance is updated.
Select a canary deployment when a change must reach a small slice of live traffic first and ramp up based on observed health. A canary deployment directs a controlled, small fraction of production traffic to the new version and uses live error and latency signals to decide whether to widen exposure or abort. This metric-driven graduated rollout limits blast radius in a way that all-at-once strategies such as blue-green or recreate cannot.

Why A is wrong: Blue-green gives a clean instant cutover and easy rollback, but it sends all users to the new build at once, so it does not provide the small graduated slice of real traffic the team wants to observe first.

Why B is wrong: A recreate strategy is the simplest rollout but it causes downtime and exposes every customer to the new library immediately, which is the opposite of the gradual, observable exposure the requirement demands.

Why C is correct: A canary releases the new build to a small share of real users, lets you compare its health against the stable baseline, and grows or aborts the rollout based on metrics, which matches the graduated, fail-safe exposure described.

Why D is wrong: A rolling update limits how many instances change at once but each updated instance still takes its full traffic share, so there is no deliberate small canary slice or metric-gated comparison before wider exposure.

Free sampleDesign and Implement Build and Release Pipelinesmedium

A YAML pipeline in "Azure Pipelines" builds a large monorepo, but the team wants a continuous integration run to start only when files under the "services/payments" folder change on the main branch. Pushes touching other folders should not trigger it. Which trigger configuration achieves this?

trigger:
  branches:
    include:
      - main
  paths:
    include:
      - services/payments/*
  • ARemove the trigger block entirely so the pipeline relies on the default behaviour, which the team can later refine to watch only the payments folder on main.
  • BKeep a branch include for main but add a schedule that runs hourly, so the payments build happens regularly regardless of which folders changed in the repository.
  • CAdd a pull request trigger that includes the main branch and the payments path, so the run starts whenever a pull request modifies files under that folder.
  • DConfigure the trigger with a branch include for main and a path include for "services/payments", so only commits to main that touch that folder start the run. Correct
Scope a continuous integration trigger with branch and path filters so only relevant commits on the target branch start a run. Azure Pipelines evaluates the trigger's branch filters and path filters together for continuous integration. A branch include of main with a path include of the payments folder means a run starts only when a pushed commit to main changes a file matching that path, leaving unrelated folder changes alone.

Why A is wrong: Omitting the trigger leaves the default continuous integration on all branches and paths, so pushes to unrelated folders would still start the run, which is what they want to avoid.

Why B is wrong: A schedule fires on a clock and ignores which paths changed, so it both runs when payments is untouched and misses the requirement to trigger on payments changes.

Why C is wrong: A pull request trigger fires on pull request events, not on commits pushed to main, so it answers a different scenario than the continuous integration push described.

Why D is correct: Combining a branch filter with a path include scopes continuous integration to commits on main that modify the named folder, matching the requirement precisely.

Other domains in this exam

See also the AZ-400 cert hub, the study guide, and the cheat sheet.

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