[Samba] Error messages every 32min after Win2k security update

[Marc Sherman]

I'm running Samba 3.0.10 from Debian Sarge (3.0.10-1).  Samba is 
configured as the primary browse master for the network, and there are 
two windows boxes that access shares from the samba box, one win2k and 
the other XP Pro.

Earlier this week, I installed a number of patches to the win2k box via 
windows update.  I'm having trouble finding out exactly what patches 
were installed, but there are a stack of .log files with identical 
timestamps in my WINNT directory that might be the complete list:
KB873339.log
imsins.log
KB885836.log
comsetup.log
iis5.log
KB885835.log
ocgen.log
ockodak.log

Ever since I installed those patches, I've been getting periodic errors 
in my samba logs.  They occur every 32 minutes, but only when the Win2k 
machine is on:

Jan  1 14:06:17 pyloric smbd[25844]: [2005/01/01 14:06:17, 0] 
lib/util_sock.c:get_peer_addr(1000)
Jan  1 14:06:17 pyloric smbd[25844]:   getpeername failed. Error was 
Transport endpoint is not connected
Jan  1 14:06:17 pyloric smbd[25844]: [2005/01/01 14:06:17, 0] 
lib/util_sock.c:read_socket_data(384)
Jan  1 14:06:17 pyloric smbd[25844]:   read_socket_data: recv failure 
for 4. Error = Connection reset by peer

I've googled for these error messages, and found a bunch of cases where 
people have them showing up quite frequently, and causing performance 
problems, but that doesn't seem to be the case here.  I was able to 
trigger the same messages once on accessing a share from the win2k 
machine, but it didn't feel like my performance was impacted at all. 
One of the hits I found suggested that "use sendfile" might be causing 
this, but that's defaulting to off already in my install, so it's not 
the problem.

Any idea what could be causing this, or what I could do to fix it?

Thanks in advance,
- Marc