My Dexcom sometimes gives me some trouble on long runs. This is less likely to happen if you put it in a baggie. I think that's often a source of trouble as a little moisture gets into the port.
As Bill King wrote about earlier in this thread, there can also be problems caused by a little moisture getting between the sensor and the transmitter. Twice in the past three weeks I've had my Dexcom stop giving readings, then I pressed down hard on the transmitter for a while, and within a mile it was showing reasonable numbers again.
I'm glad you're daring to wear your pump on race day.
I'm not sure you've ever expressed that you understand about meal boluses before a long run. Eating well three hours before your start rather than as you leave the house, and more importantly bolusing more than 3 hours before, would make a big difference in the insulin in your system at the start of your run.
If your blood sugar is dropping in the first hour of your run, it's not because you're wearing your pump. It's because of insulin that was already in your body at the start.
Concerning the desire to have a sandwich, I think it could totally work. There's just no more time to experiment before race day, so think about it for your next marathon. :-)
We didn't always have gels and gatorade, and they're not a requirement. Many excellent performance are achieved with cookies, gummi bears, pbjs, orange juice, defizzed coke, koolaid...
I ran a good, hilly 14.5 yesterday, and felt really great. Beautiful weather here. I'm looking forward to little 10 mile race on October 5. It will be fun to do a race I've done enough training for.
Good luck at the races!