Question 150 A source-object ingest job has the correct target schema but fails because the cloud storage credential is expired. Which fix is most direct? Grant a reporting user dashboard edit permission Renew or repair the approved source-storage authentication and test access Rewrite the downstream gold aggregation SQL Increase the number of ingestion retries without repairing source credentials Explanation Correct Answer B:A source authentication failure must be resolved at the storage access boundary before loading can succeed. Incorrect Answers A:Dashboard permissions cannot renew storage identity credentials. C:Aggregation logic does not repair expired source credentials. D:Repeating a request with expired authentication cannot restore authorization; the credential must first be renewed or repaired.
Question 151 A new source contains millions of immutable CSV records, and a first load is followed by hourly increment batches. Which data-control property prevents replayed files from doubling counts? Using a separate notebook each hour without state Change filename patterns before every rerun so the same records appear to be new source files Keeping a permanent all-purpose cluster alive Idempotent file ingestion tracking paired with business-key reconciliation where needed Explanation Correct Answer D:File-level progress avoids reloads; business-level dedup is still necessary if distinct files repeat events. Incorrect Answers A:New notebooks alone do not track loaded files or duplicate business IDs. B:Renaming previously loaded records undermines file-level deduplication and can cause repeated ingestion. C:Continuous compute does not itself implement idempotency.
Question 152 A migration plan assumes that selecting Auto Loader alone guarantees all source files will remain readable forever. Which external constraint matters? Auto Loader creates unlimited source backups by default Assume that the streaming checkpoint retains a permanent copy of every original file Source storage retention and permissions independent of the ingestion checkpoint Depend solely on streaming checkpoint history after the cloud provider expires source objects Explanation Correct Answer C:Progress tracking does not preserve source objects removed by cloud retention rules or lost storage grants. Incorrect Answers A:An ingestion mechanism does not automatically make immutable lifetime backups of source objects. B:A checkpoint stores processing state, not an independent archive of source payloads; deleted or expired cloud files may be unavailable on replay. D:Checkpoints track processing state, but cannot guarantee an object remains present or readable under separate storage lifecycle rules.
Question 153 An ingestion task previously loaded an object successfully but gets a new duplicate file with different name and identical payload. What must be understood? The new file cannot be loaded because its bytes are identical Assume different object names always represent different business transactions Increasing data partition size resolves duplicates File identity tracking may treat the new name as a new file, so logical dedup is separate Explanation Correct Answer D:COPY INTO or Auto Loader file tracking is not equivalent to content- or business-record deduplication across renamed copies. Incorrect Answers A:File-level ingestion generally tracks source identities, not a universal business content uniqueness contract. B:File-name deduplication cannot identify the same customer record resent under a new file name; business-key handling is separate. C:Partition sizing affects execution but not whether two files contain the same invoice.
Question 154 A source feed must preserve unknown attributes for compliance, and a new incompatible type is encountered. Which pipeline outcome is safer than silent loss? Change only the downstream query refresh frequency while still dropping unexpected raw fields Retain the raw payload or rescued field and route it through controlled validation Drop offending records without reporting counts Overwrite valid existing records with null Explanation Correct Answer B:Preserving unexpected data supports traceability and explicit conversion decisions. Incorrect Answers A:More frequent reporting cannot recover source attributes discarded at ingestion; preserving raw or rescued content is essential. C:Silent omission violates the data completeness and audit requirement. D:Destroying good data does not repair the new source type.
Question 155 A lakehouse receives orders via both an API connector and daily backup files. Both paths can contain the same order IDs. What is required for a consolidated silver table? Turn off one connector without evaluating coverage Deduplicate within each ingestion path independently but never reconcile cross-source order IDs Use source-aware identifiers and deterministic reconciliation rules across both inputs Assume each ingestion path is independently idempotent so cross-path duplicates are impossible Explanation Correct Answer C:Cross-source record identity and ordering avoid duplicate or inconsistent order state. Incorrect Answers A:Disabling a source may lose updates not covered by the other path. B:Per-source deduplication cannot eliminate duplicates that appear across both an API feed and backup files. D:Per-path retry safety does not prevent the same logical order arriving through two channels.
Question 156 A provider connector supports only a daily full export, but the business requires 10-minute changes. Which decision is responsible? Increase the number of downstream reporting queries while keeping a 24-hour source extraction cadence Configure a ten-minute job schedule that still uses a once-per-day updated source export Identify that the connector cadence cannot meet the SLA and evaluate a suitable alternative source pattern Schedule the same daily extraction every 10 minutes and claim new changes Explanation Correct Answer C:An unsupported refresh frequency is a real design constraint, not a problem to conceal with downstream scheduling. Incorrect Answers A:More consumers cannot make a daily source connector emit new data at the required ten-minute interval. B:Running extraction more frequently cannot produce changes absent from the source connector until its daily update. D:More frequent job starts cannot force a source connector to provide unavailable changes.
Question 157 A pipeline needs repeatable ingestion of local reference data but is currently pointed to an engineer's temporary laptop file path. Which correction is needed? Keep the input in the user's personal workspace folder and depend on the notebook owner staying active Increase Spark task retries without moving the file Stage the file in durable governed storage available to the scheduled compute principal Ask the engineer to keep the laptop awake forever Explanation Correct Answer C:Production runs need a stable approved source reachable independently of an individual laptop. Incorrect Answers A:A personal workspace file may be inaccessible to production identities and is not the same as a durable governed source path. B:Retries cannot reach a personal path unavailable from cloud compute. D:A personal workstation is an unreliable production data dependency.
Question 158 A team must reconcile a supplier's daily file count and total monetary amount against what landed in bronze. Which acceptance test is meaningful? Compare source file bytes with bronze storage bytes without reconciling monetary values Use only the number of bronze rows without reconciling the partner’s monetary total Compare only the job's last successful execution date Compare source manifest count and amount totals with committed bronze counts and totals Explanation Correct Answer D:Independent totals can expose missing, duplicate or misparsed records before curation. Incorrect Answers A:Different compression and Delta layout can change byte totals, so this does not prove records or amounts were imported correctly. B:A count can match while monetary values differ because of truncation, reprocessing or parse errors; both controls are required. C:Success status alone does not prove data values reconcile.
Question 159 A source API returns malformed JSON occasionally, but the ingest must not re-request already committed good pages when one bad page is corrected. What should it retain? Only a global job restart counter Keep only the latest failed-page response in notebook output without a durable error ledger Committed page/cursor state with an error ledger for failed payloads Retry the full API export from page one on every bad page without a persisted cursor Explanation Correct Answer C:Separating progress from parse failures enables selective retry of bad pages without resetting good committed extracts. Incorrect Answers A:A counter does not identify which exact source pages committed or failed. B:Transient display output does not establish which pages were successfully committed and which malformed pages still need replay. D:Re-fetching all prior pages can repeat committed data and does not reliably isolate failed payloads for correction.
Question 160 A source publishes data as gzip-compressed CSV files. The ingestion strategy must correctly interpret the content. What does the design need to account for? Only the Delta table history retention policy Set the reader to parse uncompressed text and rely on notebook retries for the gzip payload Only whether SQL keywords are uppercase The file format and supported compression handling in the selected reader Explanation Correct Answer D:The reader must decode compressed CSV payloads before interpreting delimiters and typed values. Incorrect Answers A:History retention affects stored versions, not source decompression. B:Retries repeat the same incompatible parsing behavior; the reader needs support for the actual compression and file format. C:SQL casing does not decode gzip data.