This is a wonderful problem for Simulation. have you met WinBUGS or @Risk? Delightfully useful programs.
1) It should be obvious that your system is good until you have the 8th failure (assuming your stored replacements are in working order). It is only then that you rely on getting a component back from the repair shop.
2) Unfortunately, you are running four components simultaneously, so your risk of present failure is nowhere near the three months for which you may be hoping.
3) Are the failures independent? For example, old strings of lights used to go very bad very quickly right after the first light burned out. The mehanism was the spread of voltage of the burned out light over the remaining lights. This created increasingly increased voltage over remaining lights as subsequent lights burnd out.