Hi @surajitDE

Currently, there is no out of the box feature to achieve that. What you can do, is to try to implement notebook logic that in case of error will cancel for each task run using REST API or python sdk:

- use /api/2.1/jobs/runs/cancel endpoint passing run_id of for_each task that I want to cancel.
Cancel a run | Jobs API | REST API reference | Databricks on AWS