[CentOS] losing NFS connection

Mon Apr 25 14:47:26 UTC 2005
Marc Powell <marc at ena.com>


> -----Original Message-----
> From: centos-bounces at centos.org [mailto:centos-bounces at centos.org] On
> Behalf Of Joshua Baker-LePain
> Sent: Monday, April 25, 2005 7:59 AM
> To: CentOS mailing list
> Subject: RE: [CentOS] losing NFS connection
> 
> On Sun, 24 Apr 2005 at 2:53pm, Marc Powell wrote
> 
> > > Anyone any idea what is wrong here?
> >
> > Just a thought but have you hard-coded speed and duplex all the way
> > through? Don't trust auto-negotiation.
> 
> Just a dissenting opinion here on that last bit.  No less a source
than
> Donald Becker often advises *strongly* against disabling
auto-negotiation.
> Yes, some switches *cough*Cisco*cough* historically did it very badly.
> But that was then, not now.  And, AIUI, it's actually part of the gbe
> standard.

Sure, let me rephrase my original wording -- "I don't trust
auto-negitiation." =) I say that for two reasons --

I work with a very large network consisting of tens of thousands of
machines connecting to a hodge-podge of switches and routers from many
different vendors. It is my personal experience that auto-negotiation
does not result in optimum or even compatible speed/duplex settings
often enough to be trustworthy. I've experienced this as recently as
last week with brand new Cisco equipment and Dell computers running
CentOS 3.4 (100/Full on one end, 100/Half on the other). I've seen the
problem with Alteon and Foundry equipment recently as well. It may be
part of the standard but as anyone who's been around a while knows, each
vendor's interpretation of a standard can vary enough to be problematic.

Second, as an administrator I want to do everything in my power to make
sure the devices I manage are going to run smoothly. Why leave something
as simple but as problematic as speed/duplex settings to chance or trust
when it is a simple task to force it to be that which works best in the
networking environment that the device lives in? For example, if
auto-negotiation comes up with 100/half, 10/full or 10/half and every
other local device is 100/full or vice-versa, the network is operating
at less than peak efficiency and that can also result in odd problems.

--
Marc