On this particular job we're pushing 6000 records through on one process. Simple record set in a d-base file. Simple address block is all that's in the template but the problem seems to be 32 two-page pdf files (4x6 inch postcard) that are around 3 megs each. Each record uses a unc path to variably pull one of those 2-page images (front and back). I've set "image" to create a pdf with a resolution of 600 dpi. The process has been running for about 5 hours now without seeing one file come out. I put in a splitter to output 1 file for every 1000 records.
Note that other jobs come out in a reasonable amount of time.

I'm not to sure what you mean about processes self replicating...

Thanks for your input.