First off, the computer involved:
Intel STL2 server motherboard
2 x Intel PIII 1GHz CPUs, matched pair
4x512MB Compaq PC133 ECC SDRAM
Hiper 425W PSU (model HPU-4S425)

First issue is, that any drive ATTACHED to the IDE channel, except for the main drive which is a Seagate U6 20GB, fails Seatools diagnostics. I tried a separate controller originally (an ST Labs RAID controller) and the drives failed tests, on the main IDE channel they fail too. The drives are all used so they could just be faulty, but I'm not sure.

Secondly, this machine ran fine for a good while, occasional (INTENTIONAL) reboots but I'm pretty sure it ran uninterrupted for at least a few days, maybe a week. Then one day, it just hung - no display, not responsive to anything. Had to flick the power switch, after which it booted just fine.

I ran Memtest today to see if some bad RAM was at fault. It ran for just over 6 and a half hours, then hung, not going any further. RAM and CPU heatsinks were cool to the touch, so I decided it wasn't just taking a LONG time. Memtest passed once, and was a third of the way through a second TEST at this point.

My current dilemma is, what's causing these problems? It's not overheating.
I'm going to test all the drives in my main machine using the bootable version of Seatools on Friday, as well as running Memtest on each RAM module in the AFFECTED machine individually.

Any ideas for me here?Update: the drives that tested bad on the server machine also tested bad on mine. Guess it was just bad luck, and they are all failing.
Memtest tested out just fine too - had to run the RAM in pairs of 512MB modules, but reach pair tested perfectly for 3 passes.