Question 290 A workflow has three independent source ingestions and a final reconciliation step. All three source ingestions must be complete, even if one fails, before reconciliation produces a failure report. Which Run if mode is suitable? At least one succeeded (success in any branch) At least one failed (failure in any branch) All succeeded (all branches successful) All done (regardless of task outcome) Explanation Correct Answer D:Reconciliation waits for all upstream tasks to finish, including failed ones, so it can report their statuses. Incorrect Answers A:A single successful source may complete while another is still running. B:This would skip the report when all sources succeeded, although reconciliation is still required. C:One failure would prevent the requested reconciliation report.
Question 291 A job reads source files then runs three independent transformations. One transformation has a slow dependency unrelated to the others. What DAG design reduces unnecessary waiting? Chain every task serially regardless of data needs Allow each transformation to depend only on its actual required inputs Place the slowest transformation first in every branch Start all transformations before their required source load Explanation Correct Answer B:Minimal truthful dependencies permit parallel work without forcing unrelated transformations to wait. Incorrect Answers A:Unnecessary edges lengthen the critical path. C:Artificial serial ordering does not eliminate the underlying dependency costs. D:Removing true dependencies causes incomplete-input races.
Question 292 A job must execute a small notification after upstream completion whether it succeeded or failed. A team sets it to At least one succeeded. What is the problem? At least one succeeded always fires after total failure At least one succeeded causes all upstream tasks to rerun At least one succeeded automatically edits table privileges A notification could be skipped when every upstream dependency fails Explanation Correct Answer D:At least one succeeded is false if there were no successful tasks; All done is better for completion notifications. Incorrect Answers A:The condition explicitly requires some success. B:Run if conditions determine downstream eligibility, not upstream reruns. C:Workflow status conditions do not grant catalog rights.
Question 293 A pipeline is driven by source table updates, but its job also contains an expensive quality check that should be skipped when the updated table has no relevant records. What mechanism is needed? Check the relevant record count and branch the DAG using an If/else condition Run all expensive tasks unconditionally after every trigger Assume every table update contains relevant business records Trigger the job on table updates but always run the expensive downstream quality task Explanation Correct Answer A:An event starts the job, while a data-aware condition decides whether the costly downstream check is needed. Incorrect Answers B:Unnecessary work violates the stated conditional requirement. C:Updates can be unrelated to the intended transformation. D:The update trigger starts a job, but cannot decide to skip an internal branch based on its computed relevant-row count.
Question 294 A team uses Lakeflow Jobs to run SQL and notebook transformations, and needs a visual record of which branch caused a publish delay. What should it inspect? Inspect only the final job completion time without task dependency timings The current run DAG with individual task durations and dependency status Inspect only source-control commit history without the execution DAG Only the catalog storage credential list Explanation Correct Answer B:Task graph timing reveals the slow prerequisite on the publish critical path. Incorrect Answers A:Overall duration cannot identify the delayed branch when the workflow contains multiple dependent tasks. C:Commit lineage describes code versions but does not show which particular task in this run delayed publication. D:Storage credentials do not summarize DAG branch durations.
Question 295 A job processes an array of 24 store IDs with a For each task. One store fails its processing while others succeed. What should the operator assess? Rename the governed catalog without reconciling business identifiers across source records Iteration-level failure handling and the job's retry/error policy for that store Assume one failed store implies all others had no output Delete every successful store output before investigation Explanation Correct Answer B:Each nested iteration can have a different outcome; failure handling should identify and rerun only the relevant work where supported. Incorrect Answers A:Catalog naming does not identify duplicate business entities or resolve inconsistent records in the source data. C:Independent iterations may have completed successfully. D:Destroying good work is not required for a single iteration failure.
Question 296 A batch job needs to run a notebook that creates silver tables, and later execute a Databricks SQL query building a gold report. Which DAG arrangement is best? Notebook task upstream, SQL task downstream with success dependency SQL query first and notebook second Independent parallel tasks with no shared-data coordination Only a dashboard task with no transformation Explanation Correct Answer A:This sequences transformations so the gold query reads committed silver results. Incorrect Answers B:The gold query may run before required silver data exists. C:Parallel execution can race against changing silver output. D:Refreshing visualization does not perform the silver transformation.
Question 297 A team wants an automated workflow to start only when source files appear in an approved UC volume. What must it avoid assuming about the trigger? That source-path permissions remain necessary for the running job That event-driven execution can reduce empty scheduled runs That a file-arrival trigger is supported for monitored UC locations That a file-arrival event alone makes downstream ingestion duplicate-proof Explanation Correct Answer D:A trigger initiates execution; the ingestion code still needs safe file/record progress handling. Incorrect Answers A:The job must still be authorized to read its source files. B:It can reduce polling when source arrivals are sporadic. C:Supported UC file-arrival monitoring is the intended trigger mechanism.
Question 298 A Lakeflow Job uses a notebook task whose output is needed as an input for a downstream conditional branch. What is the right communication pattern? Read the upstream notebook return string without publishing a named task value Change notebook cell descriptions to TRUE or FALSE Set a task value upstream and reference it in the If/else task Rely on a notebook variable that exists only in an unrelated interpreter Explanation Correct Answer C:Task values provide supported runtime data flow from notebooks to conditional job tasks. Incorrect Answers A:A notebook return message is not interchangeable with a task value explicitly referenced by the downstream If/else condition. B:Descriptions are not runtime Boolean outputs for the job graph. D:An in-process variable is not automatically shared across separate job tasks.
Question 299 An engineering team needs to run the same unit-test notebook across a dynamic list of table names supplied at runtime. Which job control provides direct parameterized repetition? Define a fixed group of notebook tasks for each table name For each task over the supplied list A single file arrival event A static job name containing all tables Explanation Correct Answer B:For each can execute a nested task for each dynamic table argument. Incorrect Answers A:Static task copies do not iterate over the dynamic runtime list of tables without workflow edits. C:A trigger starts a job, not a loop over table names. D:Names do not create executions for separate parameters.
Question 300 A Lakeflow Jobs dashboard refresh depends on a gold table refresh, but the gold pipeline may take between 3 and 40 minutes. What scheduling method avoids a fixed-time race? An explicit task dependency on gold pipeline completion Always start both at midnight Always start the dashboard exactly five minutes later Schedule the dashboard refresh a fixed five minutes after the gold job begins Explanation Correct Answer A:Dependency-based scheduling waits for actual completion rather than guessing a static delay. Incorrect Answers B:Simultaneous starts can read stale data. C:A five-minute delay can be too short for a 40-minute pipeline. D:A fixed offset can run too early when gold takes forty minutes; completion dependency is required.