[SGVLUG] Bad Ethernet patch cable, but passed a DC cable tester

[Dan Buthusiem]

I used a Fluke DTX to certify a few cable drops at work last summer. Even
though one computer will like a cable at a certain speed, the DTX gave
thorough detail into the physical capabilities of the wire in far more
detail than I had ever thought would have ever mattered. I felt like it
really helped me to understand the "personality" some things seem to have,
which could all be down to a kink in one of the wires, or a poor
termination at the jack. It's pretty amazing to me.
On May 4, 2013 3:58 PM, "Scott Packard" <spackard at gmail.com> wrote:

> Had an interesting issue at work with a CAT6 CMR Ethernet cable.  It was
> an old cable, CAT6, good quality, field-terminated, about 30 feet, running
> directly to a Cisco gig-E switch.  An old cable with a new PC.  It had been
> working fine with the old Sun host, until it was retired.  The PC had I/O
> issues with it, though a few PC sysadmins had just complained that
> something was wrong somewhere, but didn't know where.
>
> (I was using the VMWare's VIC on the PC to deploy an OVF template to a
> Fibre-Channel multipathed SAN, so a lot of layers to maybe have problems.)
> I figured/looked at some performance graphs in VMWare and thought the
> Ethernet I/O should be higher than 3KB/sec, traced the cable back to the
> switch and the LED on the switch was quickly flashing from green to yellow
> and back.  I replaced the cable with a new one and that fixed the problem.
>
> I wrapped up the old one and brought it home to test it was a DC-based
> tester (a Testifier by Test-Um) and all 8 wires pass.  So, I guess
> sometimes a DC-based tester won't tell you everything, or, this CMR cable
> is stiffer than your typical patch cable and it's just slightly possible
> that it may have worked its way loose a little and if I had tried just
> remating it to the PC it may have fixed it.  However, it worked well enough
> so the PC could authenticate to a domain controller, synchronize time to an
> NTP master, and look at a few web-based management sites (ESXi servers).
> It just couldn't carry a lot of traffic.
>
> Regards, Scott
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sgvlug.net/pipermail/sgvlug/attachments/20130504/464a786f/attachment.html>