Originally Posted by Sdruvss
That is where fault can emerge: Arbitration
Every system that has redundancy, on some point has arbitration. It's impossible to project a redundant system without arbitration. On some point, there is a decision beeing made by a circuit
Correct. The technical term is "metastability". As I seem to recall saying earlier in the thread, see the paper by Lamport and Palais in the 1970's, and you'll also find a lecture at HP by Charles Molnar, thought by some to be the inventor of the personal computer, and probably the first person to observe metastability problems in HW. Lamport also wrote a "popular" paper on The Buridan Problem (Buridan's Ass: the issue has indeed been known theoretically since the 14th century).
If you want to know the chances of a metastability phenomenon happening in everyday use, there was a reasonable set of MIT lectures by Devadas on the topic, but they seem to have disappeared (even the topic "operating systems" seems to have disappeared from MIT's curriculum. They must know something we don't. Or maybe it's that Bill G was a Harvard student....).
But in two words: extremely rare. In a redundant system which was built to be sensitive to such failures, not in the expected lifetime of the system.
PBL