Intel Pro/1000 PT - Cable Diagnostics Anomaly

[imported_Redwood]

I've got two Intel Pro/1000 PT NICs installed in a machine running Windows Server 2003. I've got the latest version of the drivers (12.4 - 2/2008).

Here's the issue... When I go into the device properties, and use the Intel diagnostics, I get a result that the signal is good, but the cable is bad. However, I have another Pro/1000 PT installed on another machine running XP, and this one shows flying colors for both tests. I switched the cable to the server NICs, so we have the exact same cable, connected to the same port on the same switch - but the server NICs report poor cable quality. They are able to connect through these cables, but it concerns me that the tests show up with such poor results - and I have some other data transfer symptoms that might be explained by a high error count.

I find it hard to believe that both NICs (which are only a month old) are defective. So, perhaps there's something about Server 2003 that causes this? Perhaps it needs something tweaked in the driver settings?


Anyone familiar with these symptoms and can shed some light?

 

[drebo]

I'd say you've got some bad cables. It happens.

 

[Crusty]

Are your cables a direct connection to the switch or is there a wall drop + patch panel involved?

 

[imported_Redwood]

The reason I'm not inclined to think that we're talking about a bad cable, is because it tests out with flying colors on the same model NIC in a different system. If the cable was bad, it should test bad everywhere.

And it's a direct connection to the switch - nothing in the way.

 

[spidey07]

You can't say that about bad cables - just because it's good on one doesn't mean it is good. It's either a bad cable or bad drivers or bad nic.

 

[imported_Redwood]

Originally posted by: spidey07
You can't say that about bad cables - just because it's good on one doesn't mean it is good. It's either a bad cable or bad drivers or bad nic.

If a cable has poor electrical characteristics, that's not going to disappear when we plug it somewhere else. If we were talking about different results from different model NICs, I could understand some deviation, because they won't necessarily test the same way. I could also understand some deviation if we were talking about a minor difference in the two tests - but it's showing almost at top quality on one, and almost at bottom quality on the other. And when I switch cables, the tests are identical. But unless all the cables are bad, or unless both brand new NICs are bad, we're looking at something else. I think.

 

[spidey07]

Originally posted by: Redwood

Originally posted by: spidey07
You can't say that about bad cables - just because it's good on one doesn't mean it is good. It's either a bad cable or bad drivers or bad nic.

If a cable has poor electrical characteristics, that's not going to disappear when we plug it somewhere else. If we were talking about different results from different model NICs, I could understand some deviation, because they won't necessarily test the same way. I could also understand some deviation if we were talking about a minor difference in the two tests - but it's showing almost at top quality on one, and almost at bottom quality on the other. And when I switch cables, the tests are identical. But unless all the cables are bad, or unless both brand new NICs are bad, we're looking at something else. I think.

All I can say is I've been doing this a very long time and you can't know for certain if a cable is good unless you scan it with a cable certifier.

Try just using a patch cable between the machines and re-rerun your tests. I assume these are not home-made cables.

Also notice that you have two different operating systems so it could still be driver.

 

[imported_Redwood]

Originally posted by: spidey07

All I can say is I've been doing this a very long time and you can't know for certain if a cable is good unless you scan it with a cable certifier.

Try just using a patch cable between the machines and re-rerun your tests. I assume these are not home-made cables.

Also notice that you have two different operating systems so it could still be driver.

Well, it's the same NIC driver download that was installed on both machines (it supports both OS's). And the cables are a mix - some of them are commercial cables, and some I crimped myself.

I don't have a dedicated cable analyzer, so that's out, unfortunately. As soon as I get the chance, I'll try the patch cable.

 

[Crusty]

Throw out any of your self crimped cables, they will only cause you frustrations. Like Spidey said, just because the two computers report different things about teh cable does not prove it's an okay cable. In fact, since one computer reported it bad I would assume it's bad and replace it. Replacing the cable is FAR cheaper then spending hours of time trying to troubleshoot the problem.

 

[NickOlsen8390]

I know you guys always say to throw out self made cables and replace them with commercial terminated cables.
I crimp all my own cables, and the place i work for crimps all there own cables at the clients location (were a ISP)
And we never have a problem with them. I don't have one commercial crimped cable in my house. and they all work great and run gigabit all day long.

 

[Crusty]

Originally posted by: NickOlsen8390
I know you guys always say to throw out self made cables and replace them with commercial terminated cables.
I crimp all my own cables, and the place i work for crimps all there own cables at the clients location (were a ISP)
And we never have a problem with them. I don't have one commercial crimped cable in my house. and they all work great and run gigabit all day long.

The difference is that the ISP trains their techs to properly crimp the cables with the correct lengths and pressures for a proper termination. Most other people don't have the tools to do that properly every time.

 

[cmetz]

>you can't know for certain if a cable is good unless you scan it with a cable certifier.

Seconded.

NickOlsen8390 (et al.), we've had this debate time and time again so I'm not going to bother rehashing it. I'll just say this: you will eventually come to understand why us old farts believe this is such a bad belief.

 

[RebateMonger]

I LOVE it when clients have hand-crimped cables. I get to charge them for hours of labor while I troubleshoot their networking problems!

Seriously, the last time I crimped a cable was in 2002. Now I carry $1 apiece molded cables and give them away to my clients.

 

[spidey07]

Originally posted by: NickOlsen8390
I know you guys always say to throw out self made cables and replace them with commercial terminated cables.
I crimp all my own cables, and the place i work for crimps all there own cables at the clients location (were a ISP)
And we never have a problem with them. I don't have one commercial crimped cable in my house. and they all work great and run gigabit all day long.

I'm sorry and I don't mean to insult but to warn others.

Your employer is completely incompetent. That's not an insult, it just means they don't know what they are doing.

 

[NickOlsen8390]

I understand.
But we have never had a problem with our cables.
The only time we use commercial cables is when we buy the one foot patch cables because we dont want to hand crimp 48 one foot cables.

 

[ch33zw1z]

Dont sweat it Nick. Many of us have crimped hundreds of cables that test perfect with a Fluke tester, including myself. The way I see it, if someone can't manage to properly crimp a cable, then they should be buying prefabs..

 

[drebo]

The main thing is that it generally costs MORE to crimp a cable than it does simply to buy one.

The cables we buy at my company cost between $0.35 (for a 3-foot) and $0.85 (for a 15 foot). It would cost my cabler (who could probably do just as good a job) more than that, just in the time it would take, to crimp the same cables. So, now the benefit is two-fold...we get pre-certified cables so we can blame someone else, and they're cheaper.

I don't think he'd really like it all that much if we made him crimp cables anyway.