And the results from the staggered start:
The first two were no problem no losses (maybe that's fixed?) but they were on the correct schedule. Of course I could not test my "delayed startup of system" theory since I have no idea when the system got loaded. So I still need to test that part of the theory. The last set - the ones started at 10:50 however had accelerated production in the sense that the entire run got done. by the time I checked. Weather it was finished immediately or part way through the day - no clue.
The next test is to have all 6 processors start at 10:50. We'll see the results if there's no extended downtime in 40 minutes or so.