Microsoft DP-700 Practice Test 44

Question 430

A pipeline resumes after a transient failure, then duplicates rows because its target is append-only. What explains this?

A retry-safe Copy activity always deduplicates by business key when the target is append-only.
The duplicates prove a nightly Delta OPTIMIZE job did not execute.
The retry reexecuted a logical batch that had already written some rows, and the sink had no idempotent key/transaction boundary.
A source SQL query must use a new random transaction ID for each retry.

Explanation

Correct Answer

At-least-once retry behavior can repeat input; a blind append treats rerun records as new.

Incorrect Answers

Appending the same batch on retry repeats its rows unless idempotence or checkpointed keys are implemented.

Compaction changes files, not logical duplicated transactions inserted by retry.

A stable business key is needed to recognize retries; changing IDs would prevent deduplication.

Question 431

A pipeline is updated to use idempotent transaction processing. What test proves the retry fix?

Run the pipeline once successfully and assume its second attempt cannot duplicate data.
Count the total sink rows without reconciling distinct transaction IDs.
Verify only that retry policy is enabled without exercising an actual timeout/restart.
Force a failure after a partial or completed write, rerun the same batch and compare distinct business keys and full target totals before and after.

Explanation

Correct Answer

A deliberately repeated interval validates that the target converges without gaps or duplicates.

Incorrect Answers

A single delivery never tests replaying the same business records after partial failure.

Total count alone cannot distinguish legitimate new transactions from repeated attempts.

Retry configuration isn't evidence that target writes are idempotent under repeated input.

Question 432

After a producer lakehouse reorganizes a Delta table folder, a dependent OneLake shortcut fails with PathNotFound. An operator suspects a lost permission, but direct reads of the new producer path work under the consumer identity. What should incident remediation prioritize?

Confirm the old shortcut target locator, retarget it to the new producer path, and then rerun the least-privilege read to distinguish path repair from an access-grant problem.
Increase unrelated Spark executor count to recreate the missing source.
Delete source access logs so the shortcut appears healthy.
Assume every shortcut retains a physical data copy after source deletion.

Explanation

Correct Answer

A PathNotFound response and successful direct consumer read of the new source location indicate a stale locator rather than lost authorization; retargeting followed by a real read verifies both boundaries.

Incorrect Answers

Additional compute cannot make an outdated shortcut locator point to the moved data. It does not implement the corrective control requested in this incident.

Auditing does not repair a missing target folder or access control. It does not implement the corrective control requested in this incident.

Shortcuts are references; consumers can fail when the source target disappears. It does not implement the corrective control requested in this incident.

Question 433

A OneLake shortcut query now returns PathNotFound after a producer moved its source table. The same consumer can read the new producer folder directly, and no AccessDenied error appears. Which root cause is best supported by this evidence?

A shortcut automatically follows any moved source Delta folder by matching table display names.
The shortcut still resolves its previous stored target path, while the producer has moved the data; the failure is a stale locator rather than a proven permission revocation.
An empty shortcut query proves the consumer lakehouse deleted all physical source rows.
A Fabric domain reassignment transparently repairs all old source path references.

Explanation

Correct Answer

Direct access at the new location rules against a blanket source grant failure. The shortcut metadata must be retargeted because PathNotFound on its old path identifies a locator mismatch.

Incorrect Answers

Stored target paths can become stale when source files move; shortcut metadata may require retargeting.

The reference can break without copying/deleting source data; inspect its resolved target.

Domain changes don't retarget a shortcut to a reorganized OneLake folder.

Question 434

After a stale-target repair, a OneLake shortcut is discoverable and the PathNotFound error is gone, but the intended analyst still receives AccessDenied. What validation must occur before the incident can be closed?

Check only that the shortcut's icon still appears in the consumer workspace.
Query an unrelated destination lakehouse table rather than the repaired source shortcut.
Confirm the corrected shortcut locator, then test the exact shortcut query under the affected analyst and reconcile source/target OneLake grants until both path resolution and authorization succeed.
Test with a privileged workspace administrator but not the intended consumer identity.

Explanation

Correct Answer

Resolving PathNotFound is only one layer; an AccessDenied response after retargeting requires checking effective identity/target and shortcut-side permissions under the actual analyst before closure.

Incorrect Answers

Metadata presence does not prove the underlying target path is valid or readable.

An unrelated successful read says nothing about the broken cross-workspace reference.

Admin access may hide new path-level permissions that still block the actual consumer.

Question 435

A source REST connector starts returning HTTP 429 while a Fabric Copy activity runs many parallel requests. How should the pipeline respond?

Grant all readers Fabric Admin to bypass a source API quota.
Disable all activity retries after a single transient 429 without checking policy.
Increase Copy parallelism to the maximum when the source is already throttling.
Reduce source concurrency as appropriate and configure bounded retry with increasing delay or honored Retry-After semantics for transient throttling.

Explanation

Correct Answer

Rate-limit responses indicate request pressure; selective backoff gives the upstream service time to recover without a retry storm.

Incorrect Answers

Workspace role escalation does not change the upstream connector’s rate limits. It does not implement the corrective control requested in this incident.

No retry may cause unnecessary failure; a bounded selective backoff is the safer design when permitted. It does not implement the corrective control requested in this incident.

Higher request concurrency can worsen 429 errors and reduce effective throughput. It does not implement the corrective control requested in this incident.

Question 436

A pipeline immediately retries HTTP 429 errors at full parallelism and the failure rate increases. What is the causal defect?

The retry strategy adds traffic faster than the source rate limit can recover, creating repeated throttling.
The 429 status indicates invalid SQL schema columns at the Copy sink.
The pipeline should increase parallel request count to overcome source throttling.
The source API will recover faster when all retry waits are set to zero.

Explanation

Correct Answer

Increasing-delay backoff and throughput limits are more appropriate than immediate unbounded reissue of denied requests.

Incorrect Answers

HTTP 429 is source rate limiting; repeated full-speed retries worsen the service's request budget.

More simultaneous requests can amplify 429 responses instead of respecting the upstream limit.

Immediate repeats disregard retry-after/backoff and can extend saturation under the same quota.

Question 437

After tuning a throttled Copy job, what validation is relevant?

Verify only that the target row count increased once under low concurrency.
Run a representative load and compare successful transfer rate, 429 counts, retry delays and total completion time under the source quota.
Check Copy's sink write speed but ignore source HTTP 429 and Retry-After metrics.
Declare success after one transient 200 response and skip end-to-end data reconciliation.

Explanation

Correct Answer

A controlled rate/latency comparison demonstrates the connector now respects the service’s limits while meeting transfer needs.

Incorrect Answers

More rows on one run don't prove peak throughput or absence of source throttling on retries.

The upstream API is the observed bottleneck, so source response metrics must be included.

One successful request cannot establish bounded retries and complete unique-source record transfer.

Question 438

A Fabric ingestion run reports Succeeded but the target lakehouse table still has zero rows. What must be inspected?

Change only the target table display name to match the source.
Increase warehouse query timeout when zero rows were ever written.
Inspect the Copy/Dataflow effective destination item and path, write mode, committed row count, and any staging or transactional finalization.
Disable destination activity statistics after the first green run.

Explanation

Correct Answer

A green activity can be consistent with zero selected rows or writing to a different sink; actual target-state investigation is required.

Incorrect Answers

Renaming without updating the actual write path does not move staged files into that table. It does not implement the corrective control requested in this incident.

More query time cannot return nonexistent target data. It does not implement the corrective control requested in this incident.

Suppressing metrics removes information needed to diagnose source-to-sink mismatches. It does not implement the corrective control requested in this incident.

Question 439

A data pipeline wrote 10,000 rows to `Files/staging` while the operator queries an empty Delta table under `Tables`. Why?

A successful write to `Files/staging` automatically creates a registered Delta table in `Tables`.
The SQL analytics endpoint needs V-Order optimization before it can show any rows.
The source Copy job must have written zero bytes because the table is empty.
The activity output path differs from the table being queried, so successful file staging did not materialize rows into the intended Delta table.

Explanation

Correct Answer

OneLake Files and Tables are distinct destinations; a successful write to Files is not equivalent to Delta table ingestion.

Incorrect Answers

Files storage and managed Delta Tables registration are distinct; the queried target was not populated.

Physical ordering doesn't move staged Files rows into a Delta table or register them.

The pipeline wrote 10,000 rows elsewhere; empty table contents do not refute the successful Files write.

Question 440

After repairing a zero-row destination mismatch, select TWO pieces of evidence that prove data landed correctly.

Check only the number of Files objects created, without querying the target Delta table.
Query a different, preexisting SQL endpoint table as evidence of the repaired path.
Check the activity output for successful committed rows written to the intended destination path or table.
Independently query the exact target Delta table and reconcile known source keys and row counts.
Inspect the pipeline's success icon without comparing actual source and destination row samples.

Explanation

Correct Answers

Execution details should identify the actual sink and nonzero write outcome.

A physical target read verifies the committed data exists where downstream consumers expect it.

Incorrect Answers

File object count cannot prove the intended managed table was populated.

An unrelated table can be nonempty even when this job still writes to the wrong destination.

Success shows the configured path executed, not that data appeared in the promised table.

Implementing Data Engineering Solutions Using Microsoft Fabric (DP-700) Practice Tests

Microsoft DP-700 Practice Test 42 - 10 Questions (Last Reviewed 23-Sep-2026)
Microsoft DP-700 Practice Test 43 - 10 Questions (Last Reviewed 22-Sep-2026)
Microsoft DP-700 Practice Test 44 - 10 Questions (Last Reviewed 21-Sep-2026)
Microsoft DP-700 Practice Test 45 - 10 Questions (Last Reviewed 20-Sep-2026)
Microsoft DP-700 Practice Test 46 - 10 Questions (Last Reviewed 19-Sep-2026)

Pass Your Microsoft DP-700 Exam Easy!

Question 430

Explanation

Correct Answer

Incorrect Answers

Question 431

Explanation

Correct Answer

Incorrect Answers

Question 432

Explanation

Correct Answer

Incorrect Answers

Question 433

Explanation

Correct Answer

Incorrect Answers

Question 434

Explanation

Correct Answer

Incorrect Answers

Question 435

Explanation

Correct Answer

Incorrect Answers

Question 436

Explanation

Correct Answer

Incorrect Answers

Question 437

Explanation

Correct Answer

Incorrect Answers

Question 438

Explanation

Correct Answer

Incorrect Answers

Question 439

Explanation

Correct Answer

Incorrect Answers

Question 440

Explanation

Correct Answers

Incorrect Answers

Implementing Data Engineering Solutions Using Microsoft Fabric (DP-700) Practice Tests