We have an iot-core UWP, headless app that runs fine (for months, hundreds of devices) when deployed directly from Visual Studio 2015 or as an App onto the retail iot-core distribution. In order to avoid over-air-update problems caused by recent automatic iot-core updates, we are trying to get a custom oem image/ffu built and deployed to the microsoft store. However, even after walking through the documentation/examples in detail, our app is still crashing when we deploy our oem image/ffu.
UPDATE
OK, no debugger still, but I found where it crashes, now is the question why the oem-ffu behaves differently from the side-deployed code (our code is identical) Since iot-core/UWP provices no way to get the board-UUID, I use the MAC of the primary network interface. To get this, I use this http://embedded101.com/BruceEitman/entryid/676/Windows-10-IoT-Core-Getting-the-MAC-Address-from-Raspberry-Pi which requires that a webserver be running, which it normally is, otherwise the console webapp would not work. However, on the OEM-Custom-Build-Version I get a crash in this routine. I don't know where since I can't debug, but it crashes, and I get a null back, which causes my azure storage connect to crash. I do not block processing since I have a retry loop... Anyway, what is the difference or what must we do to enable this code to also work in the OEM build?
The grass roots issue is: all I really need is a unique ID for the RPi board from somewhere... which does not seem possible via C#!? See How to get the processor serial number of Raspberry PI 2 with Windows IOT
So it looks like my MAC-Address solution above was the best we can expect at the moment, but doesn't work on the oem build. Why?
