Very odd cable problem

[usuallyconfused]

After some weeks of head scratching, I have just come back after spending my Sunday morning solving a slow network problem. Basically, I removed every patch cable and just had two in a hub - one from the server and one from a PC.

I then posted an entry in a financial package on the PC, which accesses the database on the server, and checked the time it took. It normally takes about 1.5 seconds, but users have been complaining of it taking over 20s per post.

The problem has turned out to be two patch cables and by swapping cables about, I can prove that there is abolutely no doubt about it and in every area, this is a box standard ethernet setup.

However, when I check each cable with a tester, they are wired correctly and the contact appears to be good. Very curious!

Has anyone come across this before?

 

[Servamatic]

Just out of curiosity, what is the order of the conductor colors in the patch cords?

Richard S. Anderson, RCDD

http://www.servamatic.net

 

[usuallyconfused]

O/W, O, G/W, B, B/W, G, Br/W, Br.

However, it seemed so unlikely, that I went back after lunch and tried it all again and one or two of my 'bad' cables worked and one or two of my 'good' ones didn't.

Dispite my original certainty on where the problem was, I think it more likely now that the hub is the problem, with some intermitent fault that has hidden itself by coincidence, and my assumption that it either works or it doesn't.

I will get another one tomorrow and try it. Unfortunately, the kind of equipment you would need to test a hub properly is not the kind of thing most people have lying around.

 

[Ghostin]

one of the things to keep in mind about testing patch cables is the length. They should be a minimum of 3 ft, but even then, they can give faulty test results. Are they good quality cords or the low cost ones from overseas? That might be an issue as well.

 

[AvayaNovice]

Hmmm, yeah I'd try another hub. If you're having mixed results with patch cord change outs... then perhaps it's your problem.

Was this once a good installation, now gone bad? It's much more likely that you'd have hardware failure with an active device, than passive.

 

[skip555]

why not replace the hub with a switch ?

lots of benifits of a switch over a hub with little or nop more $$$$ outlay

 

[usuallyconfused]

one of the things to keep in mind about testing patch cables is the length. They should be a minimum of 3 ft, but even then, they can give faulty test results..."

The patch cables are all about 4" - 6" long, just enough to span the gap between patch panel and hub. Are you saying these should be 3ft long?

"Was this once a good installation, now gone bad?" Yes.

"Why not replace the hub with a switch ?" Well, it is a switch actually - a Netgear FS524 - I just get used to calling it a hub as a generic term. But you are right there is almost no difference in price and I have noticed that some manufacturers, including Netgear, have almost stopped making hubs at all - they are all switches.

 

[AvayaNovice]

I'd certainly replace the hardware first.

Patch cables can be 1' long, that doesn't matter... what's important is that the entire run is long enough. Some devices have trouble with short run lengths (I don't know the exact standard, if there is one, etc.) but if you've got a horizontal run that's 100 feet long, and then a 1' patch cord, your entire length is 101 feet... the 1' patch length has no place in the matter.

Good luck, let us know how it goes.

Netgear products are good for SOHO users... but I find them to be a little flakey, I've had several netgear SOHO routers go bad over the years. I'm a little bit more trusting of Nortel, Cisco, or even Linksys.

 

[orypecos]

How are u testing the cables?

 

[usuallyconfused]

Using a simple Belkin cable tester. It sends a current down each strand in turn and a second passive unit has 8 LEDs which should be lit in order.

 

[Ghostin]

One of the reasons you are getting errors on your tester when you test patch cables is the fact that those testers such as Belkin, Microscanner, Seimons, ... require minimum cable lengths of 5m to make an accurate test.

 

[orypecos]

If u have a ohm meter I think that I would test with that to see if one wire has high ohm reading.
If it was a long cable length that would be hard to replace u could use a megameter to check for interferance between wires.

 

[Bobg1]

Since you have found that it really does not seem to make any difference which cables you are using..

"However, it seemed so unlikely, that I went back after lunch and tried it all again and one or two of my 'bad' cables worked and one or two of my 'good' ones didn't."

Check and see what you are getting for connections.
If you have a switch you should be able to set what it can do. If it is set to auto, you might want to hard set it to what you require.
It sounds to me like you might be running auto and getting
100 half or 10 full connections, either of which can be a dog (slow).

 

[Servamatic]

Ghostin

The minimum length only applies to the TDR (length) function, for wiremap there is no minimum length. I routinely use my Microscanner for wiremapping patch cords.

Are these field made or factory made patch cords? If these are field made, are you sure you have the right plugs for the type of cable? Plugs for stranded cable will give you intermittent faults when used on solid conductor cable.

That said, I tend to agree with those that are questioning the switch itself, based on info provided so far.

Richard S. Anderson, RCDD

http://www.servamatic.net

 

[Ghostin]

Richard,

Are we talking about the same item. The Fluke microscanner for about $400? I thought there was a mimimum on it, but if not, I stand corrected. I know the siemons tester does strange things on the shorter patch cords.

 

[usuallyconfused]

Servamatic - some of the patch cables I made myself, others have been there for some time, but appear to have been made up on site. To be honest, I use whatever Cat5 cable happens to be around and use RJ45 connectors from a catalogue or the electrical shop round the corner, which is usually where the cable comes from.

Sorry if I sound naive, but I was not aware of different plugs for different cable.

 

[Ghostin]

If you are going to make your own cords, start with some quality connectors and cable. Cheap connectors can cause lots of problems with reliability. As inexpensive as patch cords are now, it is far more cost effective to buy them pre-made and tested.

 

[Servamatic]

Ghostin
You are absolutely right about patch cords, they are way to inexpensive and guaranteed from manufacturers to make your own...

As for testers, I can't speak about the Siemons tester, but I do have the Microscanner from Microtest (before Fluke bought them) and it works flawlessly on wiremap, but when it comes to TDR it has problems with short cables, as do all TDR's.

Usuallyconfussed
I would pick up some factory patch cords and replace all the homemade ones first. If you have the wrong plugs now, that could very easily clear up your problems. 3' patch cords should be less than $2.00 each.


Richard S. Anderson, RCDD

http://www.servamatic.net

 

[usuallyconfused]

Thanks for all your help, guys.

It turns out to have been a two-fold problem. I have replaced the switch, which was faulty, and also replaced two of the patch leads which were also faulty.

It all works fine now!

Thanks again.