Certified Data Engineer Associate Practice Test 35
Question 340
At a food delivery company, a silver customer dimension occasionally contains null customer IDs that should never reach billing gold tables. The dataset `vehicle_arrivals` is in scope, and downstream dashboards cannot accept fabricated values. Before release, which acceptance test would actually demonstrate that the root problem is controlled?
- Inject a null customer ID and verify the exception record is visible but absent from trusted gold
- Run only a normal, unchanged `vehicle_arrivals` sample and review null key rate.
- Review a platform-level availability report for `vehicle_arrivals` without testing whether a null business key breaks reliable joins and can create untraceable billing records.
- Check that the intended configuration exists for `vehicle_arrivals` and that a healthy run succeeded, without inducing the earlier failure.
Explanation
Correct Answer
A:Inject a null customer ID and verify the exception record is visible but absent from trusted gold. This reproduces the scenario-dependent condition, verifies null key rate, rejected-record sample, silver-to-gold counts, and quality rule events, and directly tests whether a null business key breaks reliable joins and can create untraceable billing records is resolved. Observed results, not intent, determine approval.
Incorrect Answers
B:Run only a normal, unchanged `vehicle_arrivals` sample and review null key rate. A normal sample does not challenge a null business key breaks reliable joins and can create untraceable billing records.
C:Review a platform-level availability report for `vehicle_arrivals` without testing whether a null business key breaks reliable joins and can create untraceable billing records. Aggregate availability does not establish the behavior of affected `vehicle_arrivals` records. It cannot replace the direct exercise: inject a null customer ID and verify the exception record is visible but absent from trusted gold.
D:Check that the intended configuration exists for `vehicle_arrivals` and that a healthy run succeeded, without inducing the earlier failure. Installed configuration and a healthy run are insufficient to prove the failure mode is gone. The original issue was a null business key breaks reliable joins and can create untraceable billing records.
Question 341
At an education technology firm, a gold revenue table occasionally drifts from verified silver totals after a new transformation deployment. The dataset `payment_decisions` is in scope, and raw records must remain available for replay. What evidence should the engineer collect to establish the cause rather than a coincidental symptom?
- Examine only a normal successful run of `payment_decisions` and trend source and gold aggregate totals without recreating the fault.
- Correlate source and gold aggregate totals, distinct business keys across affected and unaffected `R034-CASE-069` runs.
- Review aggregate cluster availability and job completion for `payment_decisions` without correlating the observed error path.
- Compare the saved configuration with the signed change ticket for `payment_decisions` while omitting source and output records.
Explanation
Correct Answer
B:Correlate source and gold aggregate totals, distinct business keys across affected and unaffected `R034-CASE-069` runs.. By joining source and gold aggregate totals, distinct business keys, change timestamp, and quality assertion results across the affected and unaffected conditions, the investigation directly tests the recorded mechanism: a successful job status alone cannot establish transformation correctness when totals have diverged. It also sets a measurable baseline for inject a duplicated invoice into silver and confirm the publication gate catches the changed revenue total.
Incorrect Answers
A:Examine only a normal successful run of `payment_decisions` and trend source and gold aggregate totals without recreating the fault. A healthy-run trend can be useful, but it does not expose a successful job status alone cannot establish transformation correctness when totals have diverged. The failing-path source and gold aggregate totals, distinct business keys, change timestamp, and quality assertion results has to be correlated before attributing cause.
C:Review aggregate cluster availability and job completion for `payment_decisions` without correlating the observed error path. A successful aggregate platform state can coexist with a successful job status alone cannot establish transformation correctness when totals have diverged. This check never observes the affected `payment_decisions` records or their causal path.
D:Compare the saved configuration with the signed change ticket for `payment_decisions` while omitting source and output records. A desired configuration and approval ticket are not runtime proof. In this case they omit source and gold aggregate totals, distinct business keys, change timestamp, and quality assertion results, the direct means of distinguishing a successful job status alone cannot establish transformation correctness when totals have diverged.
Question 342
At an education technology firm, a gold revenue table occasionally drifts from verified silver totals after a new transformation deployment. The dataset `payment_decisions` is in scope, and raw records must remain available for replay. Which platform configuration or data-engineering change is most defensible?
- Approve gold because the notebook task exited successfully; use that approach for `payment_decisions`.
- Run reconciliations on key aggregates and row-level invariants, fail the affected publication on material mismatch
- Delete the previous gold snapshot before computing comparison metrics; use that approach for `payment_decisions`.
- Round all totals aggressively until the difference disappears; use that approach for `payment_decisions`.
Explanation
Correct Answer
B:Run reconciliations on key aggregates and row-level invariants, fail the affected publication on material mismatch. This directly addresses a successful job status alone cannot establish transformation correctness when totals have diverged. The design should be accepted only after inject a duplicated invoice into silver and confirm the publication gate catches the changed revenue total; the related operational boundary is to block material mismatches while retaining diagnostic data to fix a broken transformation.
Incorrect Answers
A:Approve gold because the notebook task exited successfully; use that approach for `payment_decisions`. Successful execution does not prove that aggregate business measures reconcile. For `payment_decisions`, this cannot remove the causal limitation: a successful job status alone cannot establish transformation correctness when totals have diverged.
C:Delete the previous gold snapshot before computing comparison metrics; use that approach for `payment_decisions`. Removing a baseline makes it harder to identify and reverse the regression. For `payment_decisions`, this cannot remove the causal limitation: a successful job status alone cannot establish transformation correctness when totals have diverged.
D:Round all totals aggressively until the difference disappears; use that approach for `payment_decisions`. Rounding can mask a genuine allocation or duplication defect. For `payment_decisions`, this cannot remove the causal limitation: a successful job status alone cannot establish transformation correctness when totals have diverged.
Question 343
At an education technology firm, a gold revenue table occasionally drifts from verified silver totals after a new transformation deployment. The dataset `payment_decisions` is in scope, and raw records must remain available for replay. A teammate proposes: "Approve gold because the notebook task exited successfully." Which technical limitation is the decisive reason not to approve that specific proposal?
- Removing a baseline makes it harder to identify and reverse the regression. This would apply to another option for `payment_decisions`.
- The dashboard cached the prior result; no DataFrame operation or underlying data requires correction.
- Successful execution does not prove that aggregate business measures reconcile. The proposal does not repair `payment_decisions`.
- Rounding can mask a genuine allocation or duplication defect. This would apply to an alternative change for `payment_decisions`.
Explanation
Correct Answer
C:Successful execution does not prove that aggregate business measures reconcile. The proposal does not repair `payment_decisions`. This identifies the known limitation of the exact proposed action. Successful execution does not prove that aggregate business measures reconcile. The underlying problem remains a successful job status alone cannot establish transformation correctness when totals have diverged; Run reconciliations on key aggregates and row-level invariants, fail the affected publication on material mismatch instead targets that failure.
Incorrect Answers
A:Removing a baseline makes it harder to identify and reverse the regression. This would apply to another option for `payment_decisions`. This limitation belongs to the different alternative "Delete the previous gold snapshot before computing comparison metrics". The stem calls for the specific failure of "Approve gold because the notebook task exited successfully".
B:The dashboard cached the prior result; no DataFrame operation or underlying data requires correction. The case provides a concrete failing data or control path; a mere alternate theory does not address a successful job status alone cannot establish transformation correctness when totals have diverged.
D:Rounding can mask a genuine allocation or duplication defect. This would apply to an alternative change for `payment_decisions`. This failure applies to the separate proposal "Round all totals aggressively until the difference disappears".
Question 344
At an education technology firm, a gold revenue table occasionally drifts from verified silver totals after a new transformation deployment. The dataset `payment_decisions` is in scope, and raw records must remain available for replay. What risk-and-cost compromise should the owner accept instead of the proposed shortcuts?
- Delete the previous gold snapshot before computing comparison metrics; monitor results only after launch of `payment_decisions`.
- Round all totals aggressively until the difference disappears; monitor results only after launch of `payment_decisions`.
- Block material mismatches while retaining diagnostic data to fix a broken transformation
- Approve gold because the notebook task exited successfully; monitor results only after launch of `payment_decisions`.
Explanation
Correct Answer
C:Block material mismatches while retaining diagnostic data to fix a broken transformation. This explicitly bounds the compromise while still addressing a successful job status alone cannot establish transformation correctness when totals have diverged. Acceptance relies on inject a duplicated invoice into silver and confirm the publication gate catches the changed revenue total rather than simply declaring the change complete.
Incorrect Answers
A:Delete the previous gold snapshot before computing comparison metrics; monitor results only after launch of `payment_decisions`. Removing a baseline makes it harder to identify and reverse the regression. Monitoring after deployment does not make this approach a safe bounded tradeoff for `payment_decisions`.
B:Round all totals aggressively until the difference disappears; monitor results only after launch of `payment_decisions`. Rounding can mask a genuine allocation or duplication defect. Monitoring after deployment does not make this approach a safe bounded tradeoff for `payment_decisions`.
D:Approve gold because the notebook task exited successfully; monitor results only after launch of `payment_decisions`. Successful execution does not prove that aggregate business measures reconcile. Monitoring after deployment does not make this approach a safe bounded tradeoff for `payment_decisions`.
Question 345
At an education technology firm, a gold revenue table occasionally drifts from verified silver totals after a new transformation deployment. The dataset `payment_decisions` is in scope, and raw records must remain available for replay. Which direct test could falsify the proposed fix instead of only confirming deployment?
- Inject a duplicated invoice into silver and confirm the publication gate catches the changed revenue total
- Run only a normal, unchanged `payment_decisions` sample and review source and gold aggregate totals.
- Review a platform-level availability report for `payment_decisions` without testing whether a successful job status alone cannot establish transformation correctness when totals have diverged.
- Check that the intended configuration exists for `payment_decisions` and that a healthy run succeeded, without inducing the earlier failure.
Explanation
Correct Answer
A:Inject a duplicated invoice into silver and confirm the publication gate catches the changed revenue total. This reproduces the scenario-dependent condition, verifies source and gold aggregate totals, distinct business keys, change timestamp, and quality assertion results, and directly tests whether a successful job status alone cannot establish transformation correctness when totals have diverged is resolved. Observed results, not intent, determine approval.
Incorrect Answers
B:Run only a normal, unchanged `payment_decisions` sample and review source and gold aggregate totals. A normal sample does not challenge a successful job status alone cannot establish transformation correctness when totals have diverged.
C:Review a platform-level availability report for `payment_decisions` without testing whether a successful job status alone cannot establish transformation correctness when totals have diverged. Aggregate availability does not establish the behavior of affected `payment_decisions` records. It cannot replace the direct exercise: inject a duplicated invoice into silver and confirm the publication gate catches the changed revenue total.
D:Check that the intended configuration exists for `payment_decisions` and that a healthy run succeeded, without inducing the earlier failure. Installed configuration and a healthy run are insufficient to prove the failure mode is gone. The original issue was a successful job status alone cannot establish transformation correctness when totals have diverged.
Question 346
At a construction materials supplier, a transient REST call fails intermittently and the entire overnight job stops without retry or delay. The dataset `supplier_orders` is in scope, and operators cannot perform manual daily launches. Which correlated observation is the strongest basis for the change decision?
- Correlate run attempt count, retry intervals across affected and unaffected `R034-CASE-070` runs.
- Inspect normal-period success metrics for the stated dataset without reproducing the source error
- Review completed run status and aggregate cluster health without record-level correlation
- Compare the saved configuration with the signed change ticket for `supplier_orders` while omitting source and output records.
Explanation
Correct Answer
A:Correlate run attempt count, retry intervals across affected and unaffected `R034-CASE-070` runs.. By joining run attempt count, retry intervals, transient error codes, and downstream completion across the affected and unaffected conditions, the investigation directly tests the recorded mechanism: a recoverable transient error should be retried within a controlled budget instead of aborting the whole DAG immediately. It also sets a measurable baseline for inject a transient first-attempt failure and confirm limited retries succeed; inject a persistent failure to ensure final alert.
Incorrect Answers
B:Inspect normal-period success metrics for the stated dataset without reproducing the source error. A healthy-run trend can be useful, but it does not expose a recoverable transient error should be retried within a controlled budget instead of aborting the whole DAG immediately. The failing-path run attempt count, retry intervals, transient error codes, and downstream completion has to be correlated before attributing cause.
C:Review completed run status and aggregate cluster health without record-level correlation. A successful aggregate platform state can coexist with a recoverable transient error should be retried within a controlled budget instead of aborting the whole DAG immediately. This check never observes the affected `supplier_orders` records or their causal path.
D:Compare the saved configuration with the signed change ticket for `supplier_orders` while omitting source and output records. A desired configuration and approval ticket are not runtime proof. In this case they omit run attempt count, retry intervals, transient error codes, and downstream completion, the direct means of distinguishing a recoverable transient error should be retried within a controlled budget instead of aborting the whole DAG immediately.
Question 347
At a construction materials supplier, a transient REST call fails intermittently and the entire overnight job stops without retry or delay. The dataset `supplier_orders` is in scope, and operators cannot perform manual daily launches. What implementation should the engineer select for the failing workflow?
- Add unbounded recursive retries inside the notebook code; use that approach for `supplier_orders`.
- Mark the task successful after any HTTP error; use that approach for `supplier_orders`.
- Configure bounded retries with sensible retry intervals on the failing Lakeflow Jobs task and retain failure visibility
- Retry every unrelated downstream task without isolating the failing call; use that approach for `supplier_orders`.
Explanation
Correct Answer
C:Configure bounded retries with sensible retry intervals on the failing Lakeflow Jobs task and retain failure visibility. This directly addresses a recoverable transient error should be retried within a controlled budget instead of aborting the whole DAG immediately. The design should be accepted only after inject a transient first-attempt failure and confirm limited retries succeed; inject a persistent failure to ensure final alert; the related operational boundary is to tolerate a finite number of transient failures while limiting delay and exposing persistent outages.
Incorrect Answers
A:Add unbounded recursive retries inside the notebook code; use that approach for `supplier_orders`. Unbounded retries can hide persistent outages and consume compute indefinitely. For `supplier_orders`, this cannot remove the causal limitation: a recoverable transient error should be retried within a controlled budget instead of aborting the whole DAG immediately.
B:Mark the task successful after any HTTP error; use that approach for `supplier_orders`. Treating failure as success allows dependent tasks to use incomplete data. For `supplier_orders`, this cannot remove the causal limitation: a recoverable transient error should be retried within a controlled budget instead of aborting the whole DAG immediately.
D:Retry every unrelated downstream task without isolating the failing call; use that approach for `supplier_orders`. Broad reruns waste work and obscure the original transient failure. For `supplier_orders`, this cannot remove the causal limitation: a recoverable transient error should be retried within a controlled budget instead of aborting the whole DAG immediately.
Question 348
At a construction materials supplier, a transient REST call fails intermittently and the entire overnight job stops without retry or delay. The dataset `supplier_orders` is in scope, and operators cannot perform manual daily launches. A teammate proposes: "Mark the task successful after any HTTP error." Which technical limitation is the decisive reason not to approve that specific proposal?
- Unbounded retries can hide persistent outages and consume compute indefinitely. This would apply to another option for `supplier_orders`.
- The upstream notebook succeeded, so the task graph and dependent job configuration cannot explain the outcome.
- Broad reruns waste work and obscure the original transient failure. This would apply to an alternative change for `supplier_orders`.
- Treating failure as success allows dependent tasks to use incomplete data. The proposal does not repair `supplier_orders`.
Explanation
Correct Answer
D:Treating failure as success allows dependent tasks to use incomplete data. The proposal does not repair `supplier_orders`. This identifies the known limitation of the exact proposed action. Treating failure as success allows dependent tasks to use incomplete data. The underlying problem remains a recoverable transient error should be retried within a controlled budget instead of aborting the whole DAG immediately; Configure bounded retries with sensible retry intervals on the failing Lakeflow Jobs task and retain failure visibility instead targets that failure.
Incorrect Answers
A:Unbounded retries can hide persistent outages and consume compute indefinitely. This would apply to another option for `supplier_orders`. This limitation belongs to the different alternative "Add unbounded recursive retries inside the notebook code". The stem calls for the specific failure of "Mark the task successful after any HTTP error".
B:The upstream notebook succeeded, so the task graph and dependent job configuration cannot explain the outcome. The case provides a concrete failing data or control path; a mere alternate theory does not address a recoverable transient error should be retried within a controlled budget instead of aborting the whole DAG immediately.
C:Broad reruns waste work and obscure the original transient failure. This would apply to an alternative change for `supplier_orders`. This failure applies to the separate proposal "Retry every unrelated downstream task without isolating the failing call".
Question 349
At a construction materials supplier, a transient REST call fails intermittently and the entire overnight job stops without retry or delay. The dataset `supplier_orders` is in scope, and operators cannot perform manual daily launches. Which decision preserves the stated data contract while keeping the change scope proportionate?
- Retry every unrelated downstream task without isolating the failing call
- Tolerate a finite number of transient failures while limiting delay and exposing persistent outages
- Mark the task successful after any HTTP error
- Add unbounded recursive retries inside the notebook code; monitor results only after launch of `supplier_orders`.
Explanation
Correct Answer
B:Tolerate a finite number of transient failures while limiting delay and exposing persistent outages. This explicitly bounds the compromise while still addressing a recoverable transient error should be retried within a controlled budget instead of aborting the whole DAG immediately. Acceptance relies on inject a transient first-attempt failure and confirm limited retries succeed; inject a persistent failure to ensure final alert rather than simply declaring the change complete.
Incorrect Answers
A:Retry every unrelated downstream task without isolating the failing call. Broad reruns waste work and obscure the original transient failure. Monitoring after deployment does not make this approach a safe bounded tradeoff for `supplier_orders`.
C:Mark the task successful after any HTTP error. Treating failure as success allows dependent tasks to use incomplete data. Monitoring after deployment does not make this approach a safe bounded tradeoff for `supplier_orders`.
D:Add unbounded recursive retries inside the notebook code; monitor results only after launch of `supplier_orders`. Unbounded retries can hide persistent outages and consume compute indefinitely. Monitoring after deployment does not make this approach a safe bounded tradeoff for `supplier_orders`.
Question 350
At a construction materials supplier, a transient REST call fails intermittently and the entire overnight job stops without retry or delay. The dataset `supplier_orders` is in scope, and operators cannot perform manual daily launches. What verification provides evidence of correct behavior under the original adverse condition?
- Check that the intended configuration exists for `supplier_orders` and that a healthy run succeeded, without inducing the earlier failure.
- Run only a normal, unchanged `supplier_orders` sample and review run attempt count.
- Review a platform-level availability report for `supplier_orders` without testing whether a recoverable transient error should be retried within a controlled budget instead of aborting the whole DAG immediately.
- Inject a transient first-attempt failure and confirm limited retries succeed; inject a persistent failure to ensure final alert
Explanation
Correct Answer
D:Inject a transient first-attempt failure and confirm limited retries succeed; inject a persistent failure to ensure final alert. This reproduces the scenario-dependent condition, verifies run attempt count, retry intervals, transient error codes, and downstream completion, and directly tests whether a recoverable transient error should be retried within a controlled budget instead of aborting the whole DAG immediately is resolved. Observed results, not intent, determine approval.
Incorrect Answers
A:Check that the intended configuration exists for `supplier_orders` and that a healthy run succeeded, without inducing the earlier failure. Installed configuration and a healthy run are insufficient to prove the failure mode is gone. The original issue was a recoverable transient error should be retried within a controlled budget instead of aborting the whole DAG immediately.
B:Run only a normal, unchanged `supplier_orders` sample and review run attempt count. A normal sample does not challenge a recoverable transient error should be retried within a controlled budget instead of aborting the whole DAG immediately.
C:Review a platform-level availability report for `supplier_orders` without testing whether a recoverable transient error should be retried within a controlled budget instead of aborting the whole DAG immediately. Aggregate availability does not establish the behavior of affected `supplier_orders` records. It cannot replace the direct exercise: inject a transient first-attempt failure and confirm limited retries succeed; inject a persistent failure to ensure final alert.