[olug] victory and defeat w/Linux router @work

[Daniel Linder]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

<quote who="Charles Bird">
> Yep, cant get signal from cross over or straight on the
> 100m leg

When you say "signal" do you mean a link-light, a tone-generator, a
continuity tester, or something else?

Neither the tone-generator nor continuity tester prove much on an ethernet
connection other than you have connectivity from pin-to-pin.

I have seen a number of cases where the link-light test is just about as
worthless.  Most NICs just check for a voltage across certain pins -- in
most cases the NIC on the far end will bring up the DC-voltage enough to
turn on the remote sides link light, but once you start sending the
modulated signal across the wire then the RF component of the wiring comes
into play.

Here is the testing methodology I would try.  If posible, can you bring
the complete cable back to your desk?  This will help eliminate a lot of
unknowns.  If so, then try options 1 and 2...
1: Using either a continuity tester or checking the colors on the
individual wires, ensure that your wiring is correct.  Since a lot of
cable manufacturers have gone to the muted coloring on the strands, I like
the continuity tester -- that ensures there is an eletronic connection
from pin-to-pin.

2: Once the pin-outs are verified, try plugging each end into known good
devices (PC and switch, or PC-to-PC if it is a crossover cable).

If you are able to bring up a good connection with this, then you'll have
to put the cable back in place and continue troubleshooting.

3: With the cable in place, do a visual inspection of the cable route.
 = Does it pass over/under/near any flourescent or high-voltage lighting?
 = How about near any non-grounded electrical frames or machines?
 = Does it pass near power cords or high-voltage powerlines?
 = Near any other electronic equipment that might have a large transformer
inside it (photo copiers, laser printers, etc)?

If all this checks out, you might have to purchase/beg/borrow/steal a
decent cable tester (Fluke, WireScope, etc) to do in-depth testing.

At my previous job (network engineer at IPRevolution.net) we had a
WireScope testing tool.  It was really nice, but I believe it had a cost
of around $20,000 for the base system.  It had two smart ends that plugged
into the RJ45 and the auto-test feature would test all aspects of the
cabling and give a certification as to the level the cable works (10bT,
100bT, 1000bT, Fiber, etc).  These high-end tools will do the indepth
testing you might need, but are quite costly if you don't do this on a
regular basis.

If this it turns out that RF is causing the problem, or the run is going
through an electronically uncontrolled environment (i.e. outside, in a
manufacturing plant, etc) you might want to look at running a fiber
connection.  Purchasing a 100+ ft fiber jumper (~$150.00), and two
Fiber-to-100bT transceiver/media-converter (~$60.00) will be in the
$300-$350 ballpark, and the benefit is that the bump to Gig speeds would
just require a replacement of the trancievers.

Good luck!

Dan

Found this tranciever on TigerDirect for $60.00:
http://biz.tigerdirect.com/applications/SearchTools/item-details.asp?EdpNo=548196&CatId=203

- - - - -
"I do not fear computer,
I fear the lack of them."
 -- Isaac Asimov

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.1 (GNU/Linux)

iD8DBQFBbqNnNiBNyqUzGb8RAmDdAJ9w6wKgtMnhG9bDK1dYB1DyA9R9HQCfXlLa
Nu7KvJsM/IuVCb7VGVnwa8M=
=Os1u
-----END PGP SIGNATURE-----