NFS client slow on amd64 6.2-PRERELEASE #2

John-Mark Gurney gurney_j at resnet.uoregon.edu
Fri Oct 6 21:14:41 UTC 2006


Jeremy Chadwick wrote this message on Thu, Oct 05, 2006 at 09:08 -0700:
> The problem in that case turned out to be duplex-related.  Both
> boxes were auto-negotiating with the Cisco switch correctly, and
> indeed the Cisco labelled them as auto-100/full, but as anyone who
> is familiar with Ciscos knows, auto-negotiation on Catalysts is
> far from reliable.  Both boxes reported auto-neg and being at 100/full
> as well.  I ended up hard-setting the boxes to use 100/full, and
> set the switch ports to 100/full, then rebooted both boxes (yes,
> this is sometimes required, as driver auto-neg code is a bit tweaky);
> voila, problem fixed.

It appears that some ethernet drivers don't reset the phy (bring the
link down) when changing media (duplex setting, etc)..  This means that
if you boot w/ autoselect, and the switch autoselects to 100/full, but
then later change it to 100/full (w/ autoselect off) it will work fine..
but then if the cabel is pulled, or the switch resets, it attempts to
reautoselect, but falls back to 100/half while you are still running
100/full...

As you can imagine, it causes a very hard to track down problem since
the time it breaks is not readily apparent...

an ifconfig <iface> down; ifconfig <iface> up may help resolve this
issue if you are not sure...

I have just committed a patch to fxp0 to do this...

-- 
  John-Mark Gurney				Voice: +1 415 225 5579

     "All that I will do, has been done, All that I have, has not."


More information about the freebsd-stable mailing list