Intel NICs, Cisco, autoneg, and borken-ness

[klussier at comcast.net]

 -------------- Original message ----------------------
From: Paul Lussier <p.lussier at comcast.net>
> klussier at comcast.net writes:
> 
> > Can you try more then just ping? Try tracerouting to another network
> > to see where the traffic stops, or where it tries to go. 
> 
> If ping doesn't work, neither will traceroute, since the latter is
> built using TTL tricks of the former.  Since I can't even ping the
> gateway of the subnet I'm *on*, trying to get *off* net isn't going to
> work either.

I know that traceroute will fail, but it would at least tell you if the traffic was leaving your box. I see a lot of cases where people tell me "The server is down, I can't even ping it" when the server that they are pinging is fine. It is usually a bad route on their system, or on the server (sending the responses to the wrong place).  I also see people sending traffic out the wrong interface (which ping would tell you).

> > Try using lynx to open a web page.
> 
> To what?  I can't even get off the network I'm on?

You can't ping. If something was stopping ICMP, then http is always a decent fall back.

> > Sometimes someone does something dumb on a router like disallowing
> > ICMP, or, in Cisco's case, tuning on "ICMP-Helper", which I have
> > seen to have the same effect as blocking ICMP (not very
> > helpful...).
> 
> We checked the router/switch config several times.  It was configured
> correctly (i.e. ICMP-helper and the like were not configured).
> 
> > Also, check the cable. You can have link light with but a single
> > wire in a pair :-)
> 
> This turned out to be closer the to problem.  Bad wiring in the patch
> panel between the switch and router.  Changing to a different port
> worked just fine.

Glad to hear it. If you need anymore useless ideas, let us know :-)

C-Ya,
Kenny