Lots of input errors...

[Shawn Ramsey]

> >>
> >> 1) You have a bad cable.  CRC and framing errors are usually a result
of
> > bad
> >> media.
> >> 2) You should set the port to auto/auto, not 100/Full hard set, unless
you
> >> have specifically set that up in rc.conf.  If one side is hard set and
the
> >> other is set to auto, you will see lots of runts because of a duplex
> >> mismatch.  I think the RFC states you have to go to half duplex if you
are
> >> autosensing and the other side doesn't respond, which is the case if
it's
> >> hard set.
> >
> > Ok.. It is set hard to 100BT/FD on both ends. I don't know if you saw
this,
> > but since resetting the stats a few hours ago, it shows no errors at all
on
> > their end, but were still getting errors on our end... The CRC and frame
> > errors could have been from a media mismatch when things were first
setup.
> > Im not sure if the stats have ever been reset. Assuming the errors stay
at
> > zero on their end, any ideas? Am going to try another server/router, and
if
> > things look better, start replacing hardware on this router...
> >
>
> I would say the physical wire is probably bad.  Seeing unidirectional
errors
> in this case wouldn't be uncommon; one of the pair of the receive wires
may
> have issues.  Have you swapped the cable?  Most of the time you won't see
> framing errors related to duplex mismatching.
>
> Also, although this sounds contrary to sane thinking about hard coding of
> duplex and speed, the Cisco will do better autosensing if you are using an
> Intel ethernet card on the server.  I'm not sure why, but I've seen this
> happen in the past.

For what its worth, changing the NIC fixed it. I've got an onboard rl0,
called "the worst ethernet controller ever made" in the source to atleast
170Mb total, I think it will go higher, thats just our internet traffic, I
haven't done any local tests yet to see if we can max it out... The
controller in question was a 3COM 3c980C-TXM.