The engineer should use Repair run . A repair reruns unsuccessful tasks as part of the original job run while preserving the successful status of tasks that do not need to be executed again. In this scenario, Tasks 1 and 2 remain successful, Task 3 is rerun after its problem is corrected, and Task 4 can execute after its dependency succeeds. Run now starts a separate job run and normally executes the workflow from its beginning. Cloning a task creates another task definition and does not resume the failed run. Reset job is not the appropriate Lakeflow Jobs recovery operation. Databricks also notes that repaired tasks should use idempotent write logic where possible because a failed task might have produced partial output before failing. Therefore, option C correctly resumes the workflow without unnecessarily repeating the successful upstream tasks.

================