Question 730 A scheduled run does not begin because the job already has its permitted maximum concurrent executions. Where should the engineer investigate? Check only the cron expression while ignoring the number of currently active runs Only the Delta file retention period The job's maximum concurrent runs and run queueing behavior Only the data catalog's metadata search index Explanation Correct Answer C:Concurrency limits can prevent or delay new executions until capacity is available. Incorrect Answers A:A correct schedule can still be blocked or queued when the job has reached its maximum concurrent-run setting. B:Retention settings do not determine how many job runs execute simultaneously. D:Catalog discoverability does not set job execution concurrency.
Question 731 A nightly notebook normally ends within fifteen minutes but occasionally hangs for hours without an exception. Which operational setting protects the job from an unbounded wait? Reduce the job schedule frequency while leaving each individual task execution unbounded Configure a task execution timeout appropriate to the workload Disable all task run history Increase the source dataset retention period Explanation Correct Answer B:A timeout makes prolonged task execution fail predictably rather than occupying resources indefinitely. Incorrect Answers A:Running the job less often cannot stop an individual hung execution; a task-level timeout bounds each run. C:Removing logs would make diagnosing the hang harder without bounding execution. D:Retention does not bound the duration of a running notebook task.
Question 732 A notebook task calls an API that intermittently returns 503, and the contract permits safe retries. What configuration most directly improves recovery? A permanently increased data table retention setting A finite retry count with an appropriate retry interval An unlimited tight loop without a delay An unrelated new dashboard schedule Explanation Correct Answer B:Bounded retries allow transient errors to clear while avoiding uncontrolled repeated requests. Incorrect Answers A:Data retention does not recover transient service availability errors. C:Unbounded immediate retries can amplify overload and never terminate. D:Dashboard refreshes do not retry the notebook's failed API request.
Question 733 A continuous workflow is still running, but its streaming backlog crosses an agreed threshold. Which alert captures degradation before a hard failure? Subscribe only to deployment PR notifications while collecting no runtime streaming backlog metric Only a terminal failure notification Only a success notification at run completion A streaming backlog notification with the threshold configured Explanation Correct Answer D:A backlog warning signals accumulation even when the workflow has not yet failed. Incorrect Answers A:Repository events describe source changes, not a live streaming job whose backlog is accumulating despite remaining in a running state. B:A terminal failure signal misses backlog growth during otherwise ongoing execution. C:A continuous job may run indefinitely without reaching a completed-success event.
Question 734 A daily job usually takes 12 minutes, but the team wants to be notified when any run exceeds 40 minutes even if it later succeeds. Which option fits? Only a successful-completion email Only a source-file arrival trigger An expected-duration threshold and Duration Warning notification Only a schema update notification Explanation Correct Answer C:The duration warning is tied to exceeding the configured limit, regardless of ultimate run success. Incorrect Answers A:A completion email need not disclose that the run exceeded its operational duration threshold. B:The trigger controls start conditions but does not monitor execution duration. D:A schema event does not measure the runtime of a job execution.
Question 735 A job auto-retries failing tasks; operators need an alert each time a task fails, not just on terminal job failure. What should they configure? Only notifications when all tasks succeed Only job-level failure notifications Alert only when a scheduled job finishes successfully, without inspecting task failures Failure notifications on the individual task Explanation Correct Answer D:Task-level notifications capture task failures even when the wider job retries and eventually succeeds. Incorrect Answers A:Success alerts conceal intermediate task failures that operators explicitly want to see. B:Job-level alerts are not sent for individual failed attempts that are retried. C:Successful job completion notifications will not provide an event on every failing task during intermediate retries.
Question 736 An enterprise wants Lakeflow Jobs failure events sent to a preapproved HTTPS incident endpoint. What prerequisite should be verified? The incident endpoint is stored as a plain-text cell in every notebook Every table in the catalog is made publicly readable The job owner deletes all failure logs before the call An administrator-configured system notification destination for the endpoint Explanation Correct Answer D:System destinations must be configured at the workspace level before jobs use them for supported event notifications. Incorrect Answers A:Notebook cells do not create an authorized system notification destination. B:Data access expansion is neither necessary nor appropriate for a job alert integration. C:Deleting logs obscures the event and cannot establish webhook delivery.
Question 737 A scheduled job's notebook succeeds for its developer but fails as the configured production service principal with an access-denied error on a Unity Catalog table. What should be checked? Make the job owner a workspace administrator without checking the run-as principal’s table grants The job run-as identity's effective catalog, schema and table privileges Only the file extension of the notebook Change the job owner to the developer while leaving the service principal as run-as Explanation Correct Answer B:The scheduled execution must have the data permissions required under its configured runtime identity. Incorrect Answers A:Workspace ownership does not imply the runtime identity has the Unity Catalog SELECT privileges needed by the scheduled task. C:Notebook naming does not grant Unity Catalog read permissions. D:Ownership and runtime identity are distinct; assigning a different owner does not grant the service principal the missing catalog access.
Question 738 A finance pipeline stops running after the original employee who configured it leaves and their user access is disabled. What governance choice would reduce this operational dependency? Move the entire data warehouse into a personal folder Reinstate the former employee's access indefinitely Use an appropriately privileged, governed service principal as the production run identity Grant all workspace users ownership of the job Explanation Correct Answer C:A stable service identity keeps the workflow authorization separate from an individual employee's account. Incorrect Answers A:Personal storage increases the same dependency rather than resolving it. B:Keeping a departed user active creates an access-control risk and a fragile dependency. D:Broad ownership is unnecessary and increases control risk.
Question 739 A manual backfill reprocesses 2026-03-01 while the current-day scheduled job is also running. Which input/output design minimizes accidental overwrites? Use the current run date implicitly as the overwrite partition for every backfill Pass an explicit partition date and use date-scoped idempotent writes Use the current wall-clock date for both runs Drop the complete target table before each backfill Explanation Correct Answer B:A parameterized partition boundary ensures the backfill writes only its intended date and can be safely retried. Incorrect Answers A:The actual execution date can differ from the requested historical business date; it must be passed explicitly. C:Both executions would target today and risk mixing or overwriting unrelated partitions. D:Full-table drops destroy unaffected historical periods.
Question 740 A job is configured to start when a UC source table is updated, but the business needs to process only rows where status=approved. Which separation is sound? Assume the trigger only fires for approved status values automatically Use a table update trigger as if it supported status=approved row predicates Delete nonapproved records from the source table to control the trigger Use the table update trigger to start the job and filter approved rows in the task Explanation Correct Answer D:The table update event indicates new table activity; the notebook transformation enforces business row selection. Incorrect Answers A:An update trigger is not a business-row predicate for the entire table. B:Table update events indicate table changes; they do not replace the filter logic inside the task. C:Deleting source records is unnecessary and damages data fidelity.