That’s a shame. I was thinking the Linux OS logs might have been able to tell you something about why that particular job was not executing correctly. I do not have a 2,000+ installation, but have been very curious about this as its answer will shape how we use FreePBX/Asterisk over the long-term.
I’ve found other posts before that seem to hint at 2,000 being the “limit”, even with decent server specs.