[c-nsp] Ethernet debugging...MTU?

[Matthew Bynum]

On 10/31/07, Gert Doering <gert at greenie.muc.de> wrote:
>
> Hi,
>
> On Mon, Oct 29, 2007 at 03:28:36PM -0400, Andy Dills wrote:
> > incrementing error counters anywhere on any layer. So I started trying
> > different things and discovered the issue I mentioned in the email, as I
> > have to assume once we resolve that issue the overall issue of packet
> loss
> > will be fixed.
>
> Is the link clean if you "flood ping" it?  Send out 1500 byte packets
> "back to back"?  Like running multiple
>
> cisco(enable)# ping $remote size 1500 repeat 1000
>
> in parallel...
>
> It *could* be a duplex issue (one part of the overall link being half-
> duplex and its peer side being full-duplex).  Or some component being
> flakey when more than a few packets are seen.
>
> gert



I've got an issue now with a 6748 GE blade and ANY 100mbps connections (auto
or hard coded 100 full).  Simply pinging the SVI where the bytes on the wire
are more than 1514 shows packet loss, no errors anywhere.  We've got a TAC
case open on it, and had to get a 6500 series genius on the line.  He wound
up digging into the Rohini, Janus, and Pinnacle ASICs and gathering a ton of
statistics.  That was last week, and as of right now we haven't gotten any
updates.  Could be related to your issue....

Matt