{{ examPhase === 'setup' ? 'Round setup' : examPhase === 'result' ? 'Round results' : 'Round progress' }}

{{ summaryValue }}

{{ summaryLine }}

Build a DOP-C02 practice round

Unofficial AWS Certified DevOps Engineer – Professional (DOP-C02) practice · current DOP-C02 HTML guide verified September 11, 2026. Independently authored; not affiliated with or endorsed by AWS.

36 original English scenarios cover selected operational concepts across all six domains. This is untimed single-answer practice, sampled uniformly without replacement. The real exam has 75 questions in 180 minutes, including multiple-choice and multiple-response questions. This bank does not reproduce full-exam breadth or prove hands-on competence.

Official guide and format · Official preparation resources

Blueprint weights: SDLC 22%; IaC 17%; resilience 15%; monitoring 15%; incident response 14%; security 17%. Each domain has six authored questions here. These weights are study context, not sampling weights or score multipliers. Scenario difficulty is uncalibrated. AWS recommends two or more years of AWS operations experience; no prior certification is required.

Optional. Blank uses a stable default. The first 128 UTF-16 units of a custom seed are used, without splitting characters. The active round keeps its own seed for replay.

{{ examSetupValidation }}

{{ answeredCount }} of {{ questionCount }} answered · {{ Math.round(progressPercent) }}%

{{ progressStatusLine }}

Question {{ examCursor + 1 }} of {{ questionCount }}

{{ currentQuestion.stem }}

{{ currentQuestion.accessible_clue }}

Answer recorded. Corrections appear when you view results.

Review this practice round

{{ resultLead }}

DOP-C02 · guide verified 2026-09-11 · original bank 2026.09.12. This score describes only the items answered. It is not an AWS scaled score, a pass prediction, or evidence of certification. AWS and its service names are trademarks of Amazon.com, Inc. or its affiliates.

{{ readinessReviewTitle }}

{{ readinessReviewLine }}

{{ card.label }}
{{ card.value }}
{{ card.note }}

Missed answer review

  1. {{ item.prompt }}

    Your answer: {{ item.selected_answer_label }}

    Correct answer: {{ item.correct_answer_label }}

    {{ item.explanation }}

    {{ item.source.label.split(' — ')[0] }} · {{ item.source.version }}

Domain performance

Correct and missed answers by official domain for this round. Domains absent from the chart were not sampled. Use the missed-item sources to review that domain, then practise a focused set and check the full official guide.

Loading domain-performance chart…

{{ examChartStatus }}

{{ examChartExportStatus }}

Answer ledger

Question ID Prompt Topic Your answer Correct answer Outcome Explanation Source Copy
{{ row.question_id }} {{ row.prompt }} {{ row.category_label }} {{ row.selected_answer_label }} ({{ row.selected_answer_id }}) {{ row.correct_answer_label }} ({{ row.correct_answer_id }}) {{ row.outcome_label }} {{ row.explanation }} {{ row.source ? `${row.source.label.split(' — ')[0]} · ${row.source.version}` : 'Official guide' }} {{ sourceUrl(row.question_id) }}

{{ examLedgerExportStatus }}

AWS CloudOps Engineer Associate Exam Prep Practise AWS CloudOps SOA-C03 with 30 original questions and sourced explanations that help identify gaps across five operational domains.
AWS Developer Associate Exam Prep Practise AWS Developer Associate DVA-C02 with 28 original questions, source-linked explanations and focused retries of missed answers.
CI Failure Log Analyzer Find actionable clues in a failed CI log, map them to pipeline stages, and mask common secret patterns before sharing the evidence.
GitHub Actions Matrix Generator Build a GitHub Actions matrix workflow with accurate job expansion, exception rules, cache choices, and review checks before you commit.

A deployment can finish successfully while customers still face errors. The application may be unhealthy, a database change may be incompatible with the previous release, or a repeated event may trigger the same recovery action twice. Professional DevOps work connects software delivery to operational checks, controlled permissions and tested recovery procedures.

The AWS Certified DevOps Engineer – Professional exam, DOP-C02, concerns provisioning, operating and managing distributed systems on Amazon Web Services (AWS). Its subject extends from preparing a release to observing its behavior and responding when something goes wrong. Recognizing a service name is useful, but the important judgment is whether its configured behavior meets the stated requirement.

Across the software development life cycle (SDLC), teams build and test changes, identify the revision approved for release, and decide how it reaches production. Continuous integration and continuous delivery (CI/CD) automate much of this work. Automation does not eliminate approval requirements. A notification that a release occurred cannot substitute for a decision that must happen before deployment.

Infrastructure as code (IaC) describes infrastructure in a form that can be reviewed and deployed repeatedly. Two different questions arise when infrastructure changes. A proposed update needs review before execution, while an existing environment may need checking for drift, meaning differences from its declared configuration. Confusing those questions can lead to inspecting the wrong evidence before a consequential change.

Recovery planning adds a tradeoff between resources kept running and work left until an incident. A recovery-time objective (RTO) is the maximum acceptable interruption of service. The recovery strategy must account for data, application startup, capacity and traffic routing, rather than merely naming a backup location.

Recovery approach Available before recovery Work still required
Backup and restore Backups of the required data and supporting configuration. Restore data and establish the infrastructure and application needed to serve requests.
Pilot light Replicated critical data and a minimal operational core. Start or provision application capacity, scale it and redirect traffic.
Warm standby A functioning application in another Region at reduced capacity. Scale the environment and redirect production traffic.

Keeping more of the recovery environment operational can reduce the work needed during an outage, but a strategy name does not guarantee a recovery time. Exercises must establish whether the actual workload can meet its objective. Similarly, redeploying an earlier application revision is not automatically a reversal of every file change or database operation performed by the failed release.

Automated incident response must also tolerate repeated requests. Idempotency means that repeating the same logical operation does not produce an additional unintended effect. For example, retrying an incident event should not create another ticket or repeat a destructive change merely because the event was delivered again. Monitoring, permission controls and recovery procedures need to work together without assuming that detecting a problem has already fixed it.

AWS recommends two or more years of experience operating AWS environments, together with development-life-cycle and programming or scripting experience. A prior certification is not a prerequisite. Scenario practice can reveal misunderstood requirements, but a correct selection alone does not demonstrate that someone can safely deploy, secure or recover a production workload.

How to Use This Tool:

Choose mixed practice to look for gaps or a focused domain to revisit a subject. The bank contains 36 original English scenarios. This is independently authored, unofficial practice, with no AWS affiliation or endorsement.

  1. Choose Domain / practice focus and Question count. The default is six questions from All six domains. Choose an offered count, up to 36 for mixed practice or six for one domain. Changing focus resets an unavailable count to six.

  2. Choose Feedback timing. After each answer, the default, reveals the correction, explanation and supporting source before you continue. At the end records each selection without revealing correctness until the completed review. Both modes are untimed and use one answer from four choices.

  3. Enter an optional Replay seed under Advanced, then select Start exam practice. A blank seed uses a stable default rather than automatically creating a different round. Surrounding spaces are ignored. Long entries are shortened at a whole character. The length allowance holds 128 basic letters or digits; many emoji take two places.

    To reproduce a round, keep the seed, domain, question count and feedback timing unchanged while the question bank remains unchanged. Changing feedback timing can change question or option order, even with the same seed.

  4. Select the single best answer for the stated requirements. Selecting an option immediately records and locks it in either feedback mode. Use Next question to continue, then View results after the final answer. Questions cannot be skipped through this workflow.

    End round returns to setup and clears the active attempt. It does not produce a partial results review.

  5. Review Missed answer review and the Answer ledger for the requirement your selection failed to satisfy. Retry missed starts a new round containing only the incorrect items and can therefore contain just one question. Replay same seed repeats the active round with its saved seed. Try a new seed changes the seeded round but does not guarantee unseen questions. Use Change setup to choose a different or broader practice focus.

Interpreting Results:

The main result is the number correct divided by the number of questions in that round. It describes those selected scenarios, not professional readiness. Perfect round means that every selected answer matched the authored key. It is not a certification, a passing forecast or evidence of mastery.

Domain performance shows counts of correct and missed answers, not an official weighted domain score. A domain absent from the chart was not sampled. More misses in a domain can also reflect more questions from that domain, so compare the correct and missed counts together. The ledger supplies the prompt, your answer, the expected answer, explanation and source needed to investigate a specific mistake.

The official DOP-C02 examination has a standard duration of 180 minutes and 75 questions, including questions requiring multiple selections. Of these, 65 affect the score and 10 are unidentified unscored questions. AWS reports a scaled score from 100 to 1,000, with 750 as the minimum passing score. A 75% practice result is not an AWS score of 750, and no conversion from this practice score to the official scale is provided.

The question bank is dated September 12, 2026 and identifies an official-guide check on September 11, 2026. These are content snapshot dates, not the exam’s launch date or a promise of continuous updates. Difficulty is uncalibrated, and six authored questions per domain cannot establish full syllabus coverage. Use missed-item sources to resolve the specific misunderstanding, then consult the current official guide and practise the corresponding work. Repeated success with familiar questions is not independent evidence of broader competence.

Technical Details:

Single-answer scoring measures agreement with the best answer for a stated scenario. Each question has one keyed answer and three authored alternatives. Every correct choice earns one point, while an incorrect choice earns zero. There is no partial credit, negative marking, difficulty adjustment or official-domain score multiplier.

Correct answers and round progress

Accuracy expresses the score as a percentage of all questions in the round. Progress measures how many questions have been answered, regardless of whether those answers were correct.

S =C Accuracy (%) =100×CN Progress (%) =100×AN M =A−C

N is the round’s question count, A is the answered count, C is the correct count, S is the points earned and M is the number missed. Possible points equal N, so the score ranges from zero to N. Unanswered questions remain in the accuracy denominator but are not counted as incorrect or included in a missed-only retry.

The visible score fraction uses exact integer counts. Accuracy keeps the unrounded ratio. The progress label rounds to the nearest whole percent, with an exact half-percent tie rounding upward. Completion depends on every question being answered, not on the rounded progress label or a minimum accuracy percentage.

Condition Result meaning
A < N The round is incomplete. Unanswered questions do not become misses, and the completed review is not ready.
A = N and C = N Perfect round. There are no incorrect items to retry.
A = N and C < N Start with the missed answers. At least one item is available for focused review and retry.

Domain coverage and sampling

Official blueprint percentages describe the distribution of scored examination content. The practice bank instead contains six questions in each domain. Selection is uniform and unweighted within the chosen pool, without replacement, so an item does not repeat within a round.

Official domain Official weight Selected authored topics
SDLC Automation 22% Manual approvals, test reports, release protection and deployment rollback.
Configuration Management and IaC 17% Change sets, drift, organization baselines and resource retention.
Resilient Cloud Solutions 15% Application health, queue backlog and disaster recovery.
Monitoring and Logging 15% Host telemetry, organization trails, log metrics and composite alarms.
Incident and Event Response 14% Undelivered events, recovery runbooks, remediation and repeated events.
Security and Compliance 17% Permission limits, temporary credentials, secret rotation and audit integrity.

A six-question mixed round is not guaranteed to contain one question from each domain. Selecting all 36 questions includes the entire authored bank, with six from each domain, but still does not reproduce the official weighting or examination breadth. A different seed can change selection and order without introducing any new content beyond this bank.

Release gates, previews and rollback

A CodePipeline manual approval action stops progression at the configured release boundary. In the approval-timeout scenario, the account’s default seven-day timeout applies and there is no action-specific override. An unanswered approval therefore fails when that timeout expires. Seven days is not a universal maximum: the documented action override permits 5 to 86,400 minutes, inclusive, with 86,400 minutes equal to 60 days. Timeout failure neither grants approval nor promises to reverse earlier successful stages.

CloudFormation change sets describe a proposed update and require execution before those changes are applied. Drift detection compares actual configuration with supported, explicitly declared properties of the existing template. It does not inspect every resource or unspecified default, and detection alone does not repair a deviation. Reviewing a change set also does not guarantee that the subsequent update will succeed.

CodeDeploy rollback redeploys an earlier application revision as a new deployment. Automatic rollback must be configured for the relevant failure or monitoring condition. On instance deployments, cleanup of installed files does not necessarily remove files created by lifecycle scripts. Database migrations and other external effects likewise require deliberate recovery handling rather than an assumption that application rollback restores the entire machine or data state.

Signals, delivery failures and safe remediation

CloudWatch Logs metric filters turn matching incoming log events into metric observations. Creating a filter does not backfill metrics from previously stored logs. A missing historical count therefore does not prove that an error never occurred; that investigation requires examining the historical logs.

An EventBridge rule target’s dead-letter queue retains events that could not be delivered for later diagnosis and possible reprocessing. The supported queue is an Amazon Simple Queue Service (Amazon SQS) standard queue, not a first-in, first-out queue. It must be in the same AWS Region as the rule and have the required permissions. Retaining a failed event does not itself repair its destination or complete its business operation.

AWS Config automatic remediation can begin from an older compliance snapshot, including when a resource has already been corrected. A recovery runbook should check current state before changing it and safely leave an already-correct resource unchanged. For repeated Lambda events, durable, concurrency-safe recognition of the logical operation helps prevent duplicate side effects. A temporary in-memory record from one invocation is not a reliable history across execution environments.

Permission limits and audit evidence

A service control policy (SCP) limits available permissions for affected principals. It does not grant access by itself. Applicable identity or resource policies must still allow the operation, and explicit denies must be respected. SCPs do not affect users or roles in an organization’s management account, so a member-account control must not be assumed to cover that account.

CloudTrail log-file integrity validation provides signed digest evidence for checking whether delivered log files changed or disappeared. Enabling digest delivery and actually validating the files are separate actions. Integrity validation is also not a write-prevention control. Those distinctions matter when a scenario asks for evidence of modification rather than prevention of modification.

Queue-backlog calculations

Queue-scaling scenarios distinguish total demand from demand per running instance. Backlog per instance divides the visible queue by available instance count. The documented starting target divides acceptable queue latency by average message-processing time.

B =QI Btarget =Lt

Q is the number of visible messages and I is the number of in-service instances. L is acceptable queue latency in seconds and t is average processing time in seconds per message for one instance. Both backlog quantities are expressed as messages per instance. The divisions require positive instance count and processing time. These equations explain the authored scenarios, not a live assessment of an AWS workload, and the target is not a guarantee of worst-case queue latency.

Privacy and Saved Attempts:

Question selection and scoring run in your browser, but the attempt is also encoded in the page address for restoration. That address carries the round information, replay seed, responses and current position. Encoding is not encryption. Someone receiving the link can restore the included attempt and inspect its answers.

Attempt addresses can remain in browser history and can reach the site when opened or reloaded. Clearing the active attempt does not erase previously copied links or history entries. Exported ledgers also contain answers, and the study document includes the replay seed. Keep credentials and personal information out of seed text and treat shared attempts as disclosed study records.

Chart and document features can load external resources, and opening a reference contacts its publisher. Browser scoring therefore should not be read as a promise of zero network activity or complete offline operation.

An invalid or incompatible saved attempt can show Saved attempt unavailable and return to setup. Start a new round with the available settings rather than interpreting that warning as a zero score. If a warning says the current attempt could not be saved in the URL, keep the page open to continue it; the address may not restore that progress.

Worked Examples:

Finishing a mixed round and retrying two misses

Choose six questions from All six domains with feedback At the end. After two selections, progress displays 33% regardless of correctness, and corrections remain hidden. Suppose the completed round contains four correct answers and two incorrect answers. View results shows 4/6, equivalent to approximately 66.7% accuracy. Selecting Retry missed creates a two-question round. Answering both correctly gives 2/2 for that retry, not an amended 6/6 result for the original attempt. Review the two explanations as evidence of what was corrected, rather than treating the smaller retry as a new broad assessment.

Choosing a backlog target from processing time

Select the six-question Resilient Cloud Solutions set to include its queue-target scenario. Each instance processes a message in an average of 0.2 seconds, and acceptable queue latency is 20 seconds. Dividing 20 by 0.2 gives a starting target of 100 messages per in-service instance. Selecting that answer earns one point when the correction is revealed. Multiplying the two times would give the wrong quantity. The explanation should lead to evaluating the target against actual queue behavior, not assuming that every message will complete within 20 seconds.

References: