you could use some DO NOTOK before DO rerun to notice everytime the job gets error and execute again so you notice if the error is recurrent and stop the infinite bucle [manually] that you are creating
Figured I would just use a max rerun of 2 which would only rerun the job once. The job already fails but exits with COMPSTAT=1 which it does for many types of errors but I only want to rerun in the case of two different types.