[From the comp.risks digest:] Bug in the network: a real spider +33)388412674 <"Nick BROWN " (Tel> 18 Oct 1996 16:14:03 +0200 About five years ago we were the first on our block to have an Ethernet segment implemented by a pair of 20 mW infra-red lasers. We had been promised that it was pretty reliable; as long as we could see the other building, so could the laser. And so it turned out; even though we spanned a canal in the foggiest town in France, we managed 98% uptime during the first few months (in the winter). Only on the foggiest days did we lose the connection, until about 10 each morning and from 4 each evening, for about 5 days total. Came the spring, and we started to get breaks in the connection at what seemed to be random times. Eventually we worked out that it was happening half an hour after sunrise and half an hour after sunset. We thought it might be the sun in line with the laser, but the layout wasn't right for that. We changed every component in the network. Until... One day, I noticed a spider's web connecting the laser box to the wall. Something must have flipped to On in my brain's background processing section, because about three days later (I run background processing on a very low priority), it dawned on me. I went up to the roof at the appropriate time and watched the laser carefully. Sure enough... a spider appeared, walked across the web, climbed into the transmission lens, and sat there (upside down), apparently eating its dinner, while a voice in my ear called "the link's down again !". We got a cloth, a box of Q-tips, and an aerosol of insecticide, put them in a box marked "Network Hardware Maintenance Kit", and set up a weekly rota for cleaning the laser, and had no more problems. I dined out on the story for several months afterwards. Nick Brown, Strasbourg, France (Nick.Brown@dct.coe.fr)