Subject: Re: `ifconfig wm3 ip4csum` creates bad cksums via tcpdump
To: Peter Eisch <firstname.lastname@example.org>
From: Charlie Allom <email@example.com>
Date: 05/23/2007 02:14:10
On Tue, May 22, 2007 at 11:33:43AM -0500, Peter Eisch wrote:
> On 5/22/07 11:24 AM, "Charlie Allom" <firstname.lastname@example.org> wrote:
> > On Tue, May 22, 2007 at 07:50:34AM +1000, Daniel Carosone wrote:
> >> On Sat, May 19, 2007 at 07:17:43PM +0100, Charlie Allom wrote: >>>
> % ifconfig wm3 tcp4csum udp4csum tso4 ip4csum >>> [[wait 34 seconds!
> % poor sockets ]] >> >> That seems weird. Can you determine whether
> % this is some kind of name >> resolution problem for, say, 'tcp4sum',
> % or really time spent waiting >> in the driver to set the flags?
> > other systems don't do it with different cards?
> > would it matter how much traffic is flowing through it?
> > does it link down - then link up? could be the cisco switch?
> This looks like spanning-tree took your port down for 30 seconds after
> the nic reset. The nic will reset when you apply something like that
> as well as if you do a tcpdump on that interface without the -p flag.
> You can avoid such things in the future if you enable the
> spanning-tree portfast option on that interface, or just do it prior
> to bringing the interface up at boot time.
%Warning: portfast should only be enabled on ports connected to a single
host. Connecting hubs, concentrators, switches, bridges, etc... to this
interface when portfast is enabled, can cause temporary bridging
Use with CAUTION
%Portfast has been configured on GigabitEthernet0/13 but will only
have effect when the interface is in a non-trunking mode.
this time changing an option only paused for a microsecond.