I encountered the problem myself only a few times, not anything life threatening

I was stating the actual problem, wich has been complained about numerous times in this very thread, why do you think they are switching away from the current backend...
mytrezor.com is down again...
Yeah, it was down for few minutes; current backend is driving me nuts. It's failing more and more often and I'll celebrate when I'll "rm -rf" it! We implemented heavy automatic checks and watchdog, so at least it can recover itself after few minutes of downtime.
In meantime we're working hard on replacing current backend by opensource Insight block explorer, but it requires significant refactoring of network API in myTREZOR.