A computer problem I encountered recently that I resolved today, but ended up being pretty weird:
I visited a customer about a week ago whose (Ivy Bridge era) desktop PC sometimes wouldn't start. Switch it on, nothing appears on the screen, no beeps, but the computer stays running. During the visit, the computer didn't act up for me at all, so it was just a case of searching for evidence. One interesting point was a live kernel event that pointed to the graphics card (144 I believe). Otherwise, everything seemed OK. I updated the graphics card driver in case the failure occurred at the point where a high-resolution display is drawn (the customer was in the habit of switching off the monitor manually after shutting down the computer, so they have to manually switch it back on again while the computer boots). One other detail is that the i5 processor in the customer's series was the P type so minus integrated graphics.
Fast forward to today's appointment. I brought along a spare second-hand graphics card, but while the computer did its usual problematic symptom with the customer's graphics card, it gave a beep sequence for mine (6 beeps, HP desktop, standards for beep sequences be damned, along with the stupid HP proprietary screws). I decided to take the computer home as I had more cards there, but I was worried that maybe HP got the idea of having a graphics card whitelist in place (not that I've ever encountered one, but I knew that graphics card worked). When I got the computer home, I tried the customer's graphics card in a spare PC, and sure enough it gave the same problematic symptom as in their PC (yay, at least there's some more evidence of my theory about the graphics card).
Long story short: It turns out my spare graphics cards are unlikely to be UEFI compatible. I know one of them isn't (HD Radeon 6670), and as the other two 8400GS's are older, I guess they're not either, hence the beep sequence. Luckily, I had a spare Ivy Bridge processor, I put that in there (also had to clean out a HSF from a PC that belongs to a smoker... not pleasant), which worked. Except now the computer was complaining that the processor fan wasn't working despite spinning it up for a split second before posting the error. I ended up connecting the processor fan to the chassis fan header and vice versa which worked fine. I also had a brief panic that when I removed the first DVI onboard socket cover, it was DVI-D (the customer was relying on a DVI-I to VGA adapter for their monitor). Weirdly, the board has a second on-board DVI socket, which luckily was DVI-I.