Cisco vs 100Mbit (Re: [SunRescue] Re: Woah, a PCI Ultra 1)

[Gregory Leblanc]

> -----Original Message-----
> From: Bjrn Ramqvist [mailto:brt at osk.sema.se]
> Sent: Sunday, March 12, 2000 11:12 PM
> To: rescue at sunhelp.org
> Subject: Cisco vs 100Mbit (Re: [SunRescue] Re: Woah, a PCI Ultra 1)
> 
> 
> Jonathan [no, I don't write for /.] Katz wrote:
> > 
> > Our U1/170E (Creators) would fall back to 100HDX way too easily when
> > connected to our high-end Bay 10/100 auto-sensing switches. 
> The Bay's
> > autosensing was "flawless" with the PCMCIA 3COM NICs used in the
> > Windows-user's laptops and with the Ciscos we had 
> segmenting our test
> > networks. I've also seen the hme NICs on an e4k have 
> difficulty talking to
> > a Cisco Catalyst which had its port hard-set at 100Mbit/FDX 
> because it
> > "auto-negotiated" to 100HDX.
> 
> Odd. I have exactly the same problem on both R10k and R12k 
> Octanes, and
> the Origin200... I just leave that port to "autonegotiate", and that
> does the magic. If I somehow "force" the Catalyst into 
> 100FDX, it's just
> like you're saying.
> Apparently, SGI has it's way of fixing this, just change a little line
> in the kernel-files.

I don't think this is really specific to Cisco's gear, I've had similar
problems several times, especially with embeded NICs.  Something tells me
that some of these devices aren't obeying the "auto-detect" spec, whatever
that is.  
	Greg