[morzh]

Well, this was not exactly my fault but:

In the year 1986 I was servicing a Soviet-made (well, Soviet-Polish; the CPU was made in Ukraine and the rest of Mera-CAMAC computer was made in Poland) mini-computer, a PDP11-34 clone. The processor designation was Electronika 100-25. (This is its front panel).

   

The processor was mounted on rails in its stand and could be rolled out or back in. The boards were inserted on one side and the back was wire-wrap type.
Whoever have seen wire-wrap panels, knows that the pins of the connectors, that were wrapped with wire, were very rigid. And very sharp at their ends.

So, upon having serviced it (taking boards out, dusting the board edge connectors, cleaning the edges with some good quality grain alcohol) I rolled it back. It so happened that near the wire-wrap pins there was a cable that powered the cooling fans (220V of course). The cable was not jacketed - just the rubber-insulated wires. 
The  wire-rap pins grazed it as the processor was moving back, and the computer stopped.

This computer (whopping 256kBytes of semiconductor DRAM, freshly installed to replace 4 cubes of 64kBytes each of core memory) was powering the whole research facility.

Then there came my boss. His first reaction was very emotional.
I explained what happened.
So, while he realized I was not at fault, but then he tole me that I would have to fix it.

Suffices it to say, the pins were some handshake signals on the UNIBUS. So, every device connected was affected.
I spent 2 weeks pulling every board, diagnosing it, replacing the driver chips.
Some nights too. No weekends.

I fixed it.

It was fun.