Spent 3 hours chasing a dead NIC before checking the switch port light
Had a client call me out to their office in Beaverton last Tuesday because one workstation kept dropping off the network every 20 minutes or so. I did the whole dance, swapped cables, ran ipconfig renews, even pulled the card and reseated it in a different slot. Nothing stuck. Around hour three I finally look at the switch in the closet and the port light for that run is flickering like a candle in wind. Turns out one of their guys had zip tied a bundle of cables way too tight against the rack and it was yanking the RJ45 just enough to break contact. I cut the ties, rerouted the run with a little slack, and it has been solid for two weeks now. I felt like a rookie for not walking the physical layer first, honestly, since that's the rule we all know and I still skipped it. Anyone else got a checklist they actually stick to when a machine keeps dropping, or do you all just start pulling cards like I did?
That bit about the port light flickering like a candle in wind is exactly the kind of thing that makes you feel dumb after the fact. Had a buddy hit almost the same wall last year, workstation kept dropping every half hour, and he's in there swapping out the card, updating drivers, the whole nine yards for most of an afternoon. Turned out the cleaning crew had bumped the desk and the cable was hanging by a thread at the wall jack, barely making contact. He finally noticed it when he leaned on the desk and the link light blinked out right in front of him. He was so mad at himself he didn't even bill the full time. Swear that stuff humbles you faster than any weird software bug ever will. Physical layer first, every single time, and I still catch myself skipping it when I'm in a hurry.