It goes against my grain to say this but that's a software problem not a hardware problem.
(Never trust a DS Rev from October 39, 2008)
It probably comes down to what what was running in the cRIO at what time.
(RSL: solid, jaguar: blink) I'd say that at first what was running on the cRIO code didn't properly enable the PWM outputs you had the Jaguars hooked to.
(RSL: blink, DS: No code) Second, the code eventually failed to maintain the watchdog, or just plain ended execution.
Everything fine and working, the startup code seems to be fine, but you may have a timing condition that leaves a window for the Watchdog to fail and PWMs to fail to initialize.
Are you running parallel processes as in independent LabVIEW loops that could cause synchronization issues where something can get used before it's been properly initialized?