Network wiring problem - weird one!

[Andy Burns]

The limiting factor in that case is more likely to be the size of the MAC address tables in the switches.

[teddysnips]

The only thing missing in this feast of information is how the individual pins are numbered on the socket. I can find how they're numbered on the cable, but not the socket.

Edward

[Adrian]

[e-mail] gurgled happily, sounding much like they were saying:

A two-second google finds...

[teddysnips]

Edward

[Mike Clarke]

No, we've eliminated that one. His Netgear DG834 has autosensing ports that can cope with this.

[The Natural Philosopher]

Yes. Thats the great thing about switches. They reconstitute to digital, and then retransmit as analogue. You will get a delay through each one, which makes short packet traffic a bit slower though.

[Adrian]

[e-mail] gurgled happily, sounding much like they were saying:

No, it's the plug. You've already been given the colour codes for the cable that correspond with which socket pin.

D'you think that pin 1 on the cable might correspond to pin 1 on the socket...?

[Steve Firth]

Disconnect all patch leads. Connect the laptop directly to the router using one of the patch leads. Ping the router. If that passes then get another patch lead and connect that directly to the router and the laptop and ping the router.

With luck you now have two patch leads that you know are OK.

Patch one port on the router to the patch panel using one of the patch leads.

Connect the laptop to the appropriate socket using the other patch lead.

Ping the router.

If that fails remove the faceplates and check wiring integrity.

Call the f****it who installed it and get him to fix it.

[Dave Plowman (News)]

The URL I gave shows this.

[Mike Clarke]

Most sockets have numbered and colour coded labels beside the connectors. It's a bit hard to identify the labels from your rather dark photo but I've put a copy on with the levels adjusted to make things more visible. The wires obscure most of the labels and numbers but it looks like the terminals are numbered and the colour codes for both EIA 568A and 568B are shown. The connections should follow one of these colour schemes, i.e. terminals 1 & 2 (top right) should be either the green pair or the orange pair. If you use the orange pair on 1 &

2 then the green pair should go to 3 & 6, which I think is the bottom left.

My guess is that things should work if you offer your mate a pint to come back with his punchdown tool and swap the green/white and orange/white striped wires. But poke around among the wires first to see the numbers to check this.

[teddysnips]

[...]

Many thanks to all of you who have tirelessly helped in the quest to allow my peri-teenage son to watch unsuitable YouTube videos in the privacy of his bedroom. I'm EXTREMELY happy to report that, thanks to your nuggets of information, there is one very red-faced network engineer swapping Green/White and Orange/White wires even as I speak. Job done.

Thanks again

Edward

[Owain]

Just wait till he discovers xtube. ;-)

Owain

[John Weston]

[e-mail] wrote: =20

=20 From Mike's picture. it looks more like phone cable than cat5e to me but=20 that could be because all the twists have been undone - but they don't=20 look tightly twisted back at the insulation cut - is it really Cat5e=20 cable?=20

I'm now sure the pairs are split onto the incorrect pins of the socket.=20 I can see the "1" by the wh/gr and I can't work out how the opposite=20 corner connector could be the other half of the pair.

I suggest you get your friend back, asking him to do it properly and=20 show him

The=20 wiring of the socket is at the top. (and you'd think someone at=20 surrey.ac.uk would spell correctly...) Don't forget to check he cuts off=20 the tiny piece of each wire at the end that was used in the punch-down=20 connector, so he is connecting to a fresh piece of wire - and get him to=20 fit a tie-wrap as a strain relief.

--=20 John W To mail me replace the obvious with co.uk twice

[dennis]

Not the way it was designed. It still used IP and/or MPLS to route/switch between areas. You could get a nice 16+2 port switch chip that supported MPLS.

The best thing was how everyone keeps telling me MPLS is a core protocol when it looks like it could be very useful at the edge to do things like switching different service types down different paths.

Another idea I see being copied is the idea of having different physical ports dedicated to different services so that you don't have to delve into the protocols to set priorities e.g. you plug your BT (IP) phone into ports

1, video into port 2, ... , internet port 4, etc. each with a different plug. I never did like the idea that you needed to take a protocols apart to set priorities.

[dennis]

The photo has pin numbers/colours on. I can't see them very well because of all the wires. You could post a pic of each end if they are different.

[Dave Liquorice]

You could assume that but it is not a safe assumption. BT type telephone connectors have arse about face numbering between the scoket and the plug.

[Adrian]

"Dave Liquorice" gurgled happily, sounding much like they were saying:

That's BT, though...

[Ron Lowe]

Yes indeed, one typo, as has already been ack'd. Of course we don't use pin 3 twice over.

My original post is correct, concerning what wires go where.

R.

[Ron Lowe]

More or less.

Try to maintain a 'tree' structure, and avoid any loops in the topology. ( ie there being 2 different routes between any 2 switches, for example running a cable from A to B in your example, when there's already a route via R )

Smarter switches will handle looped topology correctly, ( and indeed will use this for redundancy ) using the 'spanning tree protocol', but cheap switches will probably not, and this can result in odd behaviour where some packets accelerate in a clockwise direction, colliding with an anti-clockwise beam of packets. The resulting black hole will eat the server room.

[Dave Osborne]

Er, well then, you'd be wrong to expect that.

The TIA 568B standard (almost universal for UK structured cabling installations) for an RJ45 connector is based upon the Western Electric Colour Code, as follows

1. White/Orange
2. Orange or Orange/White
3. White/Green
4. Blue or Blue/White
5. White/Blue
6. Green or Green/White
7. White/Brown
8. Brown or Brown/White

In all cases where two colours are shown, the first colour is the main colour and the second colour is the tracer (or stripe) colour.