Beware of short cat5/6 cables

UnixGeek

I'm not sure how common this is with Cisco equipment, but I just saw an issue pop up in my lab that I'd previously only seen with Intel NICs.

I had a known good 1' cable running from a 1750 router's Fast Ethernet interface to a 2950 switch, and was doing an HSRP lab. The interface kept flapping, and after checking everything else, I swapped out the 1' cable with a 3' one. The interface then stopped flapping.

Strangely enough, that same 1' cable works just fine with the 1750's Ethernet interface, or as a trunking port between switches (with a crossover converter).

tiersten

Hmm. Weird. 802.3u doesn't actually state a minimum cable length requirement for 100BASE-TX. I know some cable testers freak out if you have really short cables but I've never noticed it for actual network hardware.

wastedtime

I always remember hearing about that and being recommended a 3 foot minimum for attenuation but I have never seen an issue with it. I have even made loopback plugs for test purposes that would be measured in millimeters without any problems. I know NIST says something but I believe it was for cable management not an electrical/technical issue.

UnixGeek

The only time that I'd seen this behavior previously was with a Soekris board when I was doing to embedded work a few years ago. There was a quirk with the driver for the Intel 10/100 NIC that popped up with some boards, but not others. There was a software fix, but IIRC without it you had to go to something like 10' or 15' to correct it.

It's possible that there's something wrong with this particular interface, but going to a 3' cable's a good enough solution for the lab.

redwarrior

We use 1' cables all the time at work wherever we have switches stacked on top of each other or a patch panel right on top or below a switch and never have issues. I think if the cable was a little too short for the distance you're trying to span or if it's just a bad cable, then yes, but I haven't seen any issues with new, short cables.

miller811

The only time I have ever seen this issue, is when calibrating a Penta Scanner. If you used less than a 5' cable it would not calibrate properly.

I have used hundereds of 1' cables in the field and routinely use them in my home lab with no issue.

Panzer919

UnixGeek wrote: »

I'm not sure how common this is with Cisco equipment, but I just saw an issue pop up in my lab that I'd previously only seen with Intel NICs.

I had a known good 1' cable running from a 1750 router's Fast Ethernet interface to a 2950 switch, and was doing an HSRP lab. The interface kept flapping, and after checking everything else, I swapped out the 1' cable with a 3' one. The interface then stopped flapping.

Strangely enough, that same 1' cable works just fine with the 1750's Ethernet interface, or as a trunking port between switches (with a crossover converter).

Did you check the cable with a tester to see if there was a short? It might have been a good cable but maybe there was a loose wire in the rj45 connector itself and when you unplugged it, the cable became just loose enough to cause a problem.