Dell PE 1950 bce NICs revisited

[Josh Paetzel]

On Wednesday 29 November 2006 10:43, Scott Long wrote:
> Josh Paetzel wrote:
> > I've been using 6.1-R on a PE1950 for some time now.  The stock
> > bce driver doesn't work at all.  I dug up a driver off the web
> > (0.9.6) that worked fine with my workload (basically all TCP) but
> > in talking to Scott I discovered that UDP traffic was a problem
> > for this driver. Some time ago I upgraded to the driver in
> > -STABLE.  This driver also appears to work fine, but about once a
> > day I get the following:
> >
> > Nov 29 01:16:47 server2 kernel:
> > bce1: /usr/src/sys/dev/bce/if_bce.c(5000): Watch
> > dog timeout occurred, resetting!
> > Nov 29 01:16:47 server2 kernel: bce1: link state changed to DOWN
> > Nov 29 01:16:48 server2 kernel: bce1: link state changed to UP
> >
> > Pretty minor complaint, as it doesn't really affect the operation
> > of the box, but I suppose it's a sign that there's still some
> > work to be done.
> >
> > $FreeBSD: src/sys/dev/bce/if_bce.c,v 1.2.2.6 2006/10/24
> >
> > As I write this I see there has been some further work on the
> > driver, so I think I'll upgrade it and see what happens.
>
> It's possible that rev 1.2.2.6 will fix what you're seeing.  If
> not, then it might be the general issue with the watchdog framework
> being unreliable that we addressed with the em driver.  If so, then
> it's mostly harmless.
>
> Scott
>

The string I pasted is what I was using.  Rebuilt using:
/repoman/r/ncvs/src/sys/dev/bce/if_bce.c,v 1.2.2.6.2.1
this morning.  I'll give it a few days and see how it's working.  I 
think either way it's harmless.  It happens about once a day, and 
strangely enough it's almost always in periods of very low load.

-- 
Thanks,

Josh Paetzel