Answer-first summary for fast verification
Answer: Tasks B and C will be skipped; however, any individual operations within Task A that successfully committed before the failure will remain in the Lakehouse.
In Databricks Workflows, task dependencies and data persistence follow these specific behaviors: 1. **Task-Level Dependencies:** Downstream tasks (B and C) only trigger if all upstream dependencies (Task A) return a `SUCCESS` status. If Task A fails, the scheduler marks B and C as `SKIPPED` and they never begin execution. 2. **No Job-Wide Transactions:** Databricks does not support atomic transactions across multiple tasks. Each task (and each write operation within a task) commits independently to the Delta Lake transaction log. 3. **Persistence of Committed Data:** If Task A contains multiple commands and some of those commands successfully wrote data to a Delta table before an error occurred, those commits are durable. Databricks does not perform an automatic rollback of previously successful commits within a failed task. Therefore, Task A's failure stops the downstream workflow, but it does not undo the work already finalized in the Lakehouse.
Author: LeetQuiz Editorial Team
Ultimate access to all questions.
No comments yet.
A Databricks Job is configured with three tasks: Task A has no dependencies, while Tasks B and C are configured to run concurrently only after Task A completes successfully. If Task A fails during a scheduled run, which of the following describes the resulting state of the job and the Lakehouse?
A
All tasks in the workflow must complete successfully before any changes are committed to the Lakehouse, ensuring a global transaction across the job.
B
Task A will automatically roll back all of its changes, and Tasks B and C will still attempt to execute as scheduled.
C
Tasks B and C will be skipped; however, any individual operations within Task A that successfully committed before the failure will remain in the Lakehouse.
D
The failure of Task A triggers a workspace-wide rollback of all commits to ensure Lakehouse consistency across the dependency graph.
E
Task A is guaranteed not to have committed any data because the task failed, and Tasks B and C will be skipped.