For me, it had to have been SABOTAGE!
Seriously, the problems that have affected me (and the solutions taken when designing my third incarnation of a reef tank two years ago) were:
1. Topoff failure!
2. GFCI Failure!
3. Tank Overheating!
4. CO2 Rate Failure!
A topoff failure, in my mind, is the most common means of failure in a reef tank. One solution I have taken to this problem is to use a Tunze Level Controller. But even then, I have had plenty of plumbing failures causing a leakage of system water. The Tunze level controller doesn't know otherwise, and simply tries to replace the loss of water with RO/DI water. This can cause a hyposalinity problem. My solution is to have the top off pump located in a well that contains only about 6 gallons of RO/DI water. No this means I must manually transfer the water to this reservoir about every 5 days, but I'd rather do that then run the risk of a hyposalinity crash. Even if there is a failure, 6 galons makes up only 6.7% of the total system volume, which would drop the salinity from 1.025 to a little above 1.023, which has been shown in my system to have little effect on the life.
A GFCI failure also ocurred, usually in conjunction with a plumbing failure. My solution to this was to build a power box where the tank is run off of 5 GFCIs in parallel. Critical systems are spread across different GFCI circuits. If one GFCI trips, the other four keep running (unless they trip). If the GFCI with the AquaController on it trips, then it registers a power failure and notifies me. Its also really important not to buy an external GFCI like the ones on an extension cord. My experience with these is that if the power shuts off (no GFCI trip), the GFCI must be manually reset. This is terrible, because if your tank loses power even for a second, when the power comes back on, your tank will still be without power. I also have my auto air bubbles plugged in on the GFCI responsible for tank circulation, so if that GFCI trips, the tank will be aertaed.
Tank overheating is a major problem with MHs and summer. I run a chiller, but if the chiller should fail, or my tank fans fail on a day when the chiller is not plugged in (in winter), my controller will start shutting MH lights off. It is unlikely the tank will get much above 84F without lights on in a single day. Moreover, the overheating triggers an alarm, which prompts the controller to contact me by e-mail. To prevent histeresis, when the lights are shut off due to heat, they stay off for the day and cannot turn back on until the next day. The lights shut off sequentially, so the corals will have as much light for as long as possible.
Finally, I've experienced a CO2 failure, from being too dramatic with bubble rate adjustment and overshooting my target. Always adjust your bubble counter slowly. Also, put your CO2 solenoid on an x10 module so it can be shut down if the pH drops too low.
In short, I think having a smallish topoff reservoir, putting all your eggs in different baskets GFCI speaking, and using controllers are some good precautionary measures. Still, failures are inevitable.
Matt
