Deploy Loop Detection
I've run into a case:
- Registration disabled
- Don't ask for name
- Using Model Match
- Deploy fails for some reason
- Reboot when Deploy completed
When these stars align, the machine "completes" deploy, but does not have a bootable OS on the internal SSD. So when it reboots, WIE (I haven't touched LIE yet) is the only valid boot target, it gets back in, skips registration and naming, finds a model match, and attempts the doomed deploy again; until someone intervenes.
Currently the reason is almost certainly my fault but I know in the future I will run into machines with faulty storage and deploying an image to them is the first point I have opportunity to discover it.
The solution I would love is that when WIE boots, the server checks whether this machine has recently completed a deploy, and if it has, bypasses Model Match, offering the normal manual interface with a warning text that it appears to have failed to deploy.
Alternatively, if it's reasonable to discern successful and failed deploys at the end, it would also be nice to have it reboot on success, exit to shell on error. I suspect a deploy validity test as comprehensive as attempting to boot, would be more complicated than detecting frequent automatic attempts to deploy.