[tlug] Re: Networking problem

["Jonathan Q" <jq@example.com>]

On 9/12/2002, "Stephen Lee" <sl@example.com> wrote:


>It really sounds like an autonegotiation problem, or configuration
>problem on "their" end.

I agree.  Your answers to my series of questions indicate the problem is
not on your subnets, but exists somewhere between the uplink ports of your
hubs. The only things in between there are the data center's layer 3 switch
and the data center's uplink cables.  The only part of your equipment that
could be involved are the two hubs, and you can exonerate them by doing as
Stephen suggests (even using your own machines, if they won't bring laptops
in and/or you can't have an interruption to your web server) or by swapping
them out with a known-good spare hub.

Do you know if the switch is a Cisco?  Received wisdom states that running
auto-negotiation on Cisco is not terribly reliable, even when both ends of
the link are Cisco equipment.  I've seen several examples of that in the
wild myself.

>I'd suggest you try fixing a machine on the 'A' subnet to 100M/half
>duplex, autonegotiation off and directly attaching it to the uplink,
>then check if the speed changes.  Then try 100M/full, auto off.  If
>neither talks at all then the far end might have been
>misconfigured/misdetected as 10Mbps.

Yeah, what he said :-)

If you have problems with that, or even if you don't, you might thereafter
ask them to lock their switch ports at 100-full.  Assuming your hubs have
100-full-capable uplinks (junk them if they don't), that should solve any
auto-neg problems.


GPG key: DF12B4EF (5399 C834 3ABB C3AF 610C  5345 D5D6 E6EA DF12 B4EF)
gpg --keyserver pgp.mit.edu --recv-keys  DF12B4EF