Apparently the problem was "it's very big and complicated with over a MEEELION LINES of CODE!". No, I'm not kidding,
that is the official reason.
Is this the message NATS wants to deliver - running something right on the edge of manageability that's so complex it can barely be understood? Blaming code size is a very strong indication of design, implementation, testing and/or management flaws - most of which will come back to roost in NATS' own eves.
I don't know the details of the system, nor would I want to speculate about stuff on which I'm that ignorant (would like to know what happened, though)... but I do know something about disaster
PR in the tech world...