A Databricks Job, consisting of 5 tasks with linear dependencies, typically completes in about an hour. However, it recently failed after the first 3 tasks succeeded, due to an issue in the 4th task, causing the 5th task to be skipped. What is the most efficient way to repair this job run to ensure success while minimizing execution time and compute resources?