Hi @kfierce,
Are you also seeing deadlock messages in your MRP logs as well? I think the “Abandoned” and “Defunct” (what a great name for a Rock Band! LOL 
) processes are the result of Deadlocks where two processes are trying to compete for the same resources. So SQL will grant one process access to the resources it needs while I’m assuming the other process becomes “Defunct”…(at least that’s kind of my understanding of it.)
What we did to resolve the deadlock\defunct problem was we decreased the number of Schedulers we were using therefore decreasing the possibility of two processes competing for the same resources at the same time. (Thank you for your help WarrenG @ Epicor! 
) Warren mentioned that another customer of his was experiencing the same issue and when they lowered the number of schedulers they were using that fixed this issue for them as well.
We were running MRP with 4 Processes and 8 Schedulers but when we changed to using 4 Processes and 4 Schedulers it eliminated the deadlock\defunct issue.
I’d still like to see what increasing the number of schedulers by one or two does until we can find the “sweet spot” to running MRP as quickly as possible without this issue.
We never had any issues with our parts. And it never seemed to be the same parts that would get Defunct-ed (is that the appropriate form of that word?) every time. And the parts that were showing as defunct have been able to be processed successfully through MRP before - so I knew there weren’t any problems with these parts necessarily.
@GaryL, you mentioned that “Eventually the part numbers were fixed” … what did you do to fix these part numbers? Thanks!