Yesterday, one of our larger basestations decided to give issues. It wouldn’t respond to arp requests, or ping. When I added a static arp entry, it replied with a bunch of DUPs. But still, the thing was unmanageble. Service was not really affected though, since packets were forwarded just like it should. Troubleshooting is a bitch this way :)

Today, 2 technicians went on site to check things out. After swapping around some cables and switchports (and some elevator trips of 12 minutes each) and documenting old stuff in the process, we finally found the problem. For some reason, the basestation decided to forget its IP confuguration.. And, when I took a closer look, the “management” address is somehow connected / assigned to each terminal (client) as well. At least that explains why we got a shitload of dupes (because each terminal would respond to that IP. Recalculated that, the amount of dupes matched the amount of terminals :) And, it also explains why I could still reach the “management” address from my home: it was my own antenna responding. Kind of figures, with the 1ms replies.

Let’s see how long it’ll hold this time, I fear we might have to replace the puppy. Which might be tricky, given the fact that it’s on one of the hardest places to get to, on top of the chimney :)