Network segment dropping connectivity
Posted on 2013-05-28
Good day, experts,
I seem to be having a bit of a quandary and I was wondering if any network cable / electrical gurus might be able to shed some light on the problem that I am having.
It seems that as of late there is one particular segment of our network which just stops working out of the blue. This has happened twice, now. Here are some of the details:
- This segment uses standard CAT5e UTP cable.
- This cable serves as an uplink between 2 HP switches - a 16-port JE005A (Foundry office) and an 8-port J9449A (QA Office). I would guess that this segment is around 125 to 175 feet long.
- The run is in a fairly noisy environment (a foundry) - it currently crosses some electrical wiring and is in proximity by a few metres to a Thermfire sand reclamation machine, however activity of this machine does not seem to correlate with the failure of the network segment.
- All other connections on both switches still function normally.
When the failure first happened, the link lights went out on both switches. Moving the the cable to a different port on the switch at either end did not fix anything. The cable's conectivity tested fine, but since I had ample slack at both ends, I re-crimped both ends - the problem remained.
I replaced BOTH switches with new respectively identical backup models. The result was the same. The network uplink segment would not link up while everything else still seemed to work normally. (All devices on their respective switches could still talk to each other - the failure was strictly with the uplink segment.) I put the original switches back.
For one fleeting moment the link appeared to come back but the interface on the switches indicated that the link was at 10Mb rather than its usual 1000Mb. If I tried to ping one switch from the other, the response time was all over the map, from several hundred milliseconds to timing out. This lasted for a couple minutes and the segment link went dead again.
So I decided to experiment and swap out the 16-port in the foundry office with the new 8-port backup switch. When I did this, it didn't work at first, but when I picked up the little 8-port switch in my hands, the link came on. I set it back down, and the link went out again. The thing to note here is that when I set it down, it was directly on top of the existing 16-port swtich - metal on metal. I put a foam mat in between the two switches, and the uplink worked again. I could then uplink the this 'middle' switch to the 16-port and everything was back up and running like usual again, except I now have a third switch bridging the gap between the two originals.
A point of interest - a few days later I tried plugging the uplink back directly into the 16-port, and it worked fine again. Whatever noise or interference (I'm assuming this is what it was) had settled down the the point where it was no longer a problem for awhile. I am interested in hearing if anyone can tell from this description what the most likely cause of the problem was. I get the impression that there is some sort of grounding noise or issue of some sort at play here, but if anyone can shed some light I would appreciate the input.
I spoke to our electricians and unfortunately they do not seem to have equipment which can measure noise on the network cables here. They suggested that running an outdoor-rated cable outside of the building might help - however, is this where the problem lies? The thing that throws a wrench into the works for me is the fact that 'grounding' the one switch to the other by setting it on top of it caused the link to go down, but removing that ground restored the link to working condition. I have no idea if that says anything meaningful about the source of the problem.