How to handle map internal errors in distributed mode for csv file iteration

I am working with Step Function, and I have a MAP type step to which I pass an S3 path in which there is a csv on which the MAP has to iterate. In each loop of the map, a script is executed with the csv record. If any step inside the MAP fails, I want the whole MAP to fail, so I can handle the error from outside the MAP.

This is the step flow visualization:

This is the step flow configuration:

This is the Maps Error handling:

For now, the internal step are failing but the error is not being detected by MAP, so we cannot handle it by the email error.