By the step, I mean each copy, each filter, each columns, or each sink. I have a hard time figuring it out because I only know how to re-run the whole pipeline but not the individual steps inside pipeline. The re-run of each step is to determine th intermittent errors.
You can use Activity Policies available only for execution activities (data movement and data transformation activities)
This property includes a timeout and retry behavior.
For more information, see official MS doc on Activity policy