I'm sure we'll keep seeing this problem for years to come... Auto negotiate does NOT work properly for a lot of NICs with older drivers. If you can, upgrade the driver on the RS6000. I know a while back IBM used the 3Com chipset in RS6000s, I don't know if that's true anymore but I bet it is. IBM may or may not have a driver that addresses this issue, in which case you are going to have to force the duplex on the RS6000 to full and do the same on the switch.
----- Original Message ----- From: "Priscilla Oppenheimer" To: Sent: Wednesday, January 30, 2002 3:38 PM Subject: Re: ethernet errors explained [7:33687] > At 09:35 AM 1/30/02, Patrick Donlon wrote: > >Positive, if you look at the show port (on the other mail) you'll see there > >are no collisions > > This side is set (or auto-negotiated) to full duplex. Receiving while > sending is not an error. This side should never report a collision. That > doesn't mean that there can't be a duplex-mode mismatch. A mismatch could > result in both sides reporting errors, just of different sorts. > > The other side could be set (or auto-negotiated) to half-duplex. You should > check if it has errors, including collisions. If the half-duplex side does > its normal CSMA/CD thing, senses no data, and happens to send while the > other side is sending, the result is a collision from the half-duplex > sender's viewpoint. The half-duplex side stops sending and backs off, in > the middle of its frame. The result is probably a runt with either an > alignment and/or Frame Check Sequence (FCS) error. The recipient receives > an errored frame, even though it can't correlate this with a collision > event. The recipient reports a runt and/or FCS or alignment error. > > Now, if you are sure that you don't have the obvious problem that everyone > is going to assume you have (duplex mismatch), and you are still seeing > alignment and FCS errors, then it's time to start investigating what else > besides collisions could damage frames. An FCS means that the FCS placed in > the frame by the sender doesn't match the FCS calculated by the recipient. > In other words, a bit got changed. An alignment error means that the frame > didn't end on an 8-bit boundary. In other words, a bit got dropped. > > Besides collisions, these errors could be caused by crosstalk, impedance > mismatch, noise, running a power generator next the cables, lightning > strikes, etc. > > Hope that makes sense. Please let us know the resolution. It will be a good > learning experience. > > Priscilla > > > >Thanks > > > > > >""Steven A. Ridder"" wrote in message > >[EMAIL PROTECTED]">news:[EMAIL PROTECTED]... > > > Are you sure switch and NIC are the same speed and duplex? Looks like > >port > > > speed/duplex mismatch. > > > ""Patrick Donlon"" wrote in message > > > [EMAIL PROTECTED]">news:[EMAIL PROTECTED]... > > > > Hi Everyone > > > > > > > > I trying to find some information on some Ethernet errors that I see > on > >a > > > > port, see the text below. The machine is an RS6000 and was > experiencing > > > > some > > > > performance problems, the NIC was set to auto negotiation and there > >were > > > > the > > > > usual errors. The port and NIC are now both fixed and the errors are > > > > increasing steadily, I've had a good search on the CCO but I can't > find > > > any > > > > explanation of what causes the errors, any advice will be appreciated > > > > > > > > Regards > > > > > > > > Patrick > ________________________ > > Priscilla Oppenheimer > http://www.priscilla.com Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=33764&t=33687 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]