RJ45 problems 2

[OldSkoolSkater]

I came across a problem today where the PCs at work would not ping ip addresses on the network (network connectivity was really bad for all apps and browsing). Managed to track it down to a network point that had cores 2 & 7 open circuit, punched them back in and the network was fine.
My question is why an o/c on core 2 would cause such a major problem. Any ideas?

Cheers

Steve - Network Coordinating in the UK

 

[chiefred]

Are you talking about pinouts on a RJ-45? If you are, then the ethernet pins are 1-2 and 3-6.

 

[WaitronUnit]

Steve -

[A] pathing, or ? And what's going across, just reg. data or additional streams? For eg., it might make a difference if it were set up as a config for POS and it was being doubly used (ugh) for voice.

 

[Servamatic]

Since pin 2 is required for ethernet to work, having an open would certainly cause problems.
Like chief said, you need 1,2 and 3,6 for 10/100 ethernet.

Richard S. Anderson, RCDD

http://www.servamatic.net

 

[OldSkoolSkater]

Thanks for all your replies gents.
I am aware of the pinouts for RJ45, so perhaps shouldn't have mentioned pin 7 as it wasn't relevant.
As for traffic, it is pure data without any VOIP, etc.

I wouldn't expect an o/c on pin 2 to cause widespread network problems like it did. Surely it would affect just that one network point??

What alerted me to the fact was that the switchport (3com switch) it was plugged into was showing a solid light and not flickering in time with all other 23 switchports. Pinging IP addresses from any PC connected to that switch resulted in latency or timeouts, and if I unplugged the patch lead of the dodgy network point the ping times dropped to <10ms which is the norm.

So, back to my original question - how can an o/c on pin2 on one network point affect all users connected to a switch - I'm confused??

Cheers

Steve - Network Coordinating in the UK

 

[Servamatic]

For the same reason any fault will slow down an Ethernet network....Ethernet is a CD protocol or Collision Detection, when it detects a collision it will wait to resend. Do that enough times , lets say because 1 line is continually reporting troubles, it will slow the network down, have several lines doing it, you will crash the network.

That is an overly simplistic explanation, but hopefully this will help explain what happened.


Richard S. Anderson, RCDD

http://www.servamatic.net

 

[SkipCox]

Richard,

I think that's an outstanding explanation. Much better than my "you can't have all these &%#$*@up machines connected and expect the network to function".

Skip

http://home.comcast.net/~skip9/index.html

http://scc.i8.com

 

[Servamatic]

LMAO thanks Skip

Richard S. Anderson, RCDD

http://www.servamatic.net

 

[AvayaNovice]

Hmmm.

I don't mean to rock the boat as usual.

But...

There's one really weird thing about switches and hubs Richard... collisions are only related to half duplex communication on a hub. Not full duplex communication on a switch.

There is no such thing as a collision on a switch, which is why they're much quicker... and also allow isolation of network problems because they don't use broadcasting.

That being said. I can't begin to support that analogy. You can have a bunch of messed up ports on an ethernet switch, and have a network that's operating flawlessly. That's part of your layering.

So again, I'm sticking with facts here. Network collisions on an ethernet network only related to hubs, not switches. Our poster is talking about switches.

 

[OldSkoolSkater]

Sorry for the delay in replying chaps, your thoughts are greatly appreciated.

I tend to agree with AvayaNovice in that since we are using switch technology on full duplex, CSMACD would not apply. Also, the port problem should be isolated to that one port instead of flooding all switchports.

This is why I posted originally though - I understand the theory of it all but this real-life scenario seems to have proved out differently.
Most confusing!

Cheers

Steve - Network Coordinating in the UK

 

[Servamatic]

Once again you are not exactly correct Avaya....

Filtering of packets, and the regeneration of forwarded packets enables switching technology to split a network into separate collision domains. Regeneration of packets allows for greater distances and more nodes to be used in the total network design, and dramatically lowers the overall collision rates. In switched networks, each segment is an independent collision domain. In shared networks all nodes reside in one, big shared collision domain.

There are two reasons for switches being included in network designs. First, a switch breaks one network into many small networks so the distance and repeater limitations are restarted. Second, this same segmentation isolates traffic and reduces collisions relieving network congestion. It is very easy to identify the need for distance and repeater extension, and to understand this benefit of switching. But the second benefit, relieving network congestion, is hard to identify and harder to understand the degree by which switches will help performance. Since all switches add small latency delays to packet processing, deploying switches unnecessarily can actually slow down network performance.

The above is from:

http://www.lantronix.com/learning/tutorials/switching.html

You still have collisions on a switch but they are treated differently.
Since many cable people sometimes confuse a switch and a hub I posted an accurate explanation of one possibility.

Richard S. Anderson, RCDD

http://www.servamatic.net

 

[Servamatic]

Avaya, since I don’t have your email address, I will post this here....
If you are going to try and stick with the facts as you say, as least get your facts right before you correct someone.
If you read the page I posted, you will find switches can REDUCE collisions, they do not eliminate them.
Also as it says in my post above from Lantronix installing a switch in the wrong situation may actually slow your network.
So get your facts right before you correct people.


Richard S. Anderson, RCDD

http://www.servamatic.net

 

[daronwilson]

One thing that would make much of the different 'beliefs' easier is if it were prefaced with "it is my understanding" or "i think" or "in my experience" etc. It is when a statement comes out that boldly has no exception that I start to question.

Another great value is to research your issue and post a supporting statement from a reputable source. Of course you may find that these sources have errors and even have disputes, but at least it gives everyone else something to go read and understand your direction.

This one will force me to read a bit, because it has a logic issues. A collision is two transmitters transmitting at the same time on the same channel. By definition, a full duplex channel has two discrete paths that do not interfere with each other. Hence, logically, there can be no collisions unless there were multiple transmitters on the same channel.

That said, perhaps on the backplane of the switch, or the backbone portion, or somewhere there must be a mechanism for incurring collisions. Maybe it is a terminology issue, maybe they are technically something different when somewhere inside the switch.

Normally I think the switch partitions that port off if it is bad, however, I have seen bad NIC's do this when they go into a transmit mode and flood the network. A better quality switch can detect this and isolate it, but I'm not sure that every switch can do this.

Always learning something new, that's what it is all about folks.

It is only my opinion, based on my experience and education...I am always willing to learn, educate me!
Daron J. Wilson, RCDD
daron.wilson@lhmorris.com

 

[AvayaNovice]

I'll have to post the link, but that is a fact according to Cisco. That's one of the major reasons for going to a switched network: isolation, no collisions, and full duplex communication.

You'll also notice that a switch has no facility for indicating collisions, because they don't exist.

It is m goal to post correct information.

It is also widely known that internet webpages are not always correct, have little or no editing, etc. So I'm going to post documents from reputable sources:

http://www.cisco.com/application/pd...ps5062/c1225/ccmigration_09186a008014db75.pdf

http://www.cs.virginia.edu/~cs458/slides/module16-lanswitchingV2.pdf

http://www.stanford.edu/class/cs244a/handouts/H11_Bridging.ppt

Please inform me where it says that an ethernet switch "reduces" collissions, but does not prevent them.

Yes, colissions can occur if using a hub on a switched port, but if you've got one end device per port, then collissions are in fact impossible. Please read the above if you question any of that.

There is such a thing as a "switching hub" which can still have those problems, but this is not in fact a true switch.

I find no support evidence that switches reduce collisions. If you could provide a reputable source for such information, that would be fine, and I'll be quiet.

"Ethernet is a CD protocol or Collision Detection, when it detects a collision it will wait to resend."

Yes, when using a hub. No when using a switch. Our poster is using a switch. Perhaps it is a switched hub, and then you'd be correct.

If you were correct, then one small network segment could take down an entire WAN if everything was switched... which is not true, because of laying, VLAN segmenting, etc. I just want to post the correct information, nothing personal.

 

[franklin97355]

As long as the uplink ports do not become congested, collisions do not occur and no packets are lost.

 

[AvayaNovice]

Franklin,

I agree with you, but those collisions are not real, they're software based (read a little further down) from the switching sending everything over one port. And it's pretty rare, it has to do with the amount of headroom you have.

I'll have to find the material from cisco's training material, they cover it well for CCNP.

 

[Servamatic]

The Encyclopedia of Networking has something to say on this as well...

http://www.linktionary.com/s/switching.html

From the page:
Network switching reduces or removes the sharing of the network and the problems that result from sharing, such as contention (when computers wait to use the cable), collisions (when two systems attempt to use the cable at the same time), and delays caused by contention and collisions.

They stop well short of outright saying that switches ELIMINATE ALL collisions.


Richard S. Anderson, RCDD

http://www.servamatic.net

 

[AvayaNovice]

Cisco doesn't.

Cisco's documentation clearly states that full duplex communication on a switched network cannot have a collision unless the layering is using an uplink architecture, or if there are hubs somewhere in the mix.

It CLEARLY states that.

 

[Servamatic]

What you posted is a brief which usually means it is a marketing piece...that said, it is full of caveats:

"an Ethernet switch usually has an internal backbone that is much faster than the combined speed of all the ports, thereby eliminating collisions"
"As long as the uplink ports do not become congested, collisions do not occur and no packets are lost."

Anyways, I am tired of debating this issue, I have read enough documentation to prove that a switch can not completely eliminate collisions. Everything I have read from non-manufacturers says they REDUCE collisions.

Richard S. Anderson, RCDD

http://www.servamatic.net

 

[AvayaNovice]

Then you're fine with being incorrect.