Hi All,
When they refer to long they often mean 'real long'. I once looked at a faulty LAN in
this case the servers were falling of the end. They got smmart moved the servers out
of the computer room and into the middle of the physical LAN this helped slightly.
Eventually the rang me and I had a look. Straight away I saw late collision type
things (I picked up on the CRC's and Fragments). A cable scan showed 450+ meters on a
10Base2 segment.
What had happened was someone tied two segments together with a bit of thin cable
about 50 metres long. Also the site used AMP outlets and the spare fly leads were
still inserted in the sockets.
A repeater and removal of about 100 meters of cable fixed the issue.
These days you won't see this type of problem.
Teunis
Hobart, Tasmania
Australia
On Monday, February 26, 2001 at 01:19:29 PM, John Neiberger wrote:
> I don't remember this thread, but I wanted to chime in. This one time
> (at band camp) we had a file server connected to a hub, but someone set
> the server to full duplex. This was wreaking all sorts of havoc on the
> LAN. I noticed the large number of late collisions but I didn't know
> what that indicated. Thanks to someone on this list, I checked the
> duplex settings and voila, that was it.
>
> I've read many times that late collisions are often caused by extra
> long ethernet cables, but I've never experienced that. I have, however,
> experienced the duplex-caused late collisions many times. I have to
> keep a close eye on the LAN guys around here. <g>
>
> >>> "Kevin Wigle" <[EMAIL PROTECTED]> 2/26/01 12:59:01 PM >>>
> Group,
>
> An update on that late-collision issue I brought to the list a while
> back.
>
> Finally got to talk to a tech with my ISP today and we worked through
> the
> circuit.
>
> It seems the half-duplex / full-duplex answer wins the prize.
>
> At first they tried to get me to verify my router's settings and as I
> have
> done many times before, a sh int e0/1 indicated that the interface was
> not
> full-duplex.
>
> But he wanted me to give a command to change it to half-duplex "just to
> see
> what happens".
>
> But I suggested he do it on his end first - "just to see what
> happens".
>
> In the meanwhile we were monitoring the router interface with sh int
> and
> observing console errors.
> The console was constantly spewing out transmit errors - late
> collision.
>
> The sniffer was seeing significant alignment errors.
>
> Anyway, he "does something" and immediately the console stops
> scrolling
> errors.
>
> amazing..........
>
> So, we're going to stress this circuit a bit before letting them close
> the
> ticket.
>
> It seems they paid more attention when we said we had a sniffer on the
> line.
>
> thanks for all the responses!
>
> Kevin Wigle
>
>
>
> _________________________________
> FAQ, list archives, and subscription info:
> http://www.groupstudy.com/list/cisco.html
> Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
>
>
>
> _________________________________
> FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
> Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
>
>
--
www.tasmail.com
_________________________________
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
