We need to just set aside a time for me to come by if that could be possible. These issues be crazy.
If you’re in the Venice office, that would be quite easy. Let me know if there are some days in the near future that might work for you, and I’ll see about getting things organized here.
In case it’s worth considering, I wanted to mention my running theory that the instability might be related to another thread about memory usage in the Monitor (see this post and the one immediately after it). It seems that Deadline tends to load job documents fairly naively (that is, all at once), so it’s possible that various types of operations on large job objects are just causing it to crap out.
Well, I’ll be in LA for SIGGRAPH from the 8th to the 14th. I’ll ask about squeezing in a visit and e-mail you about it.
Lemme know if you get a chance to give the MONO_THREADS_PER_CODE a shot. You’re probably right about the memory usage, but you’ll see the Modo stack trace say something about ‘null reference in ThreadWait’ or some such for this specific problem.