rsalles,AT,rsnetservices,DOT,com,DOT,br, CIPE <cipe-l,AT,inka,DOT,de>|
Re: [Fwd: Re: tcpdump - laptop]|
James Knott <james.knott,AT,rogers,DOT,com>|
Fri, 12 Sep 2003 12:50:10 -0400|
Renato Salles wrote:
[2003/09/11 21:59:45, 0] nmbd/nmbd.c:main(783)
Netbios nameserver version 2.2.3a started.
Copyright Andrew Tridgell and the Samba Team 1994-2002
[2003/09/11 21:59:45, 0] nmbd/nmbd_subnetdb.c:create_subnets(240)
create_subnets: No local interfaces !
Did you see that?
An example of a running smb.conf here:
# Configure Samba to use multiple interfaces
# If you have multiple network interfaces then you must list them
# here. See the man page for details.
interfaces = 192.168.2.0/24 127.0.0.0/16 192.168.10.0/24 192.168.3.0/24
I didn't know about that. I'll have to give it a try. It certainly
appears to be a likely suspect. In OS/2, when using NetBIOS over
TCP/IP, you only have to bind it to one adapter (it doesn't even have to
be a real one). I had assumed the same to be true here. I just looked
at the smb.conf man page and found:
"This option allows you to override the default network interfaces list
that Samba will use for browsing, name registration and other NBT
traffic. By default Samba will query the kernel for the list of all
active interfaces and use any interfaces except 127.0.0.1 that are
As I read this, it should use any available interface that's broadcast
capable. Perhaps this may have something to do with it. As I recall,
ppp links are not normally broadcast capable.
My netmasks are correct, 192.168.3.0 is the VPN tunnel range. The others
are niks interconnected to win32 LAN's.
Well, it seems your samba is not running at all! It doesn't surprise me
you see nothing at with "nmap".
That was my mistake, I should have had smbd running, to be consistent
with previous posts. However, smbd is the server and it's not necessary
to be running it, in order to use the client. As I mentioned in earlier
notes, sharing files from the notebook works fine. It's going in the
other direction, sharing from the desktop system at home that's causing
the problem. It's very strange in that it also occurs with shares from
my OS/2 system and only via dial up. If I connect via wireless or
directly to the input of my firewall, the problem doesn't occur. In all
cases, the input to the firewall is via eth0. Nothing is changed on the
firewall or desktop system, when changing connection methods.
Given that OS/2 shares also have the problem, it doesn't seem likely
that it's a fault within the Samba server. Also, NFS shows similar
problems. FTP works fine.
Given that it only happens with dial up, and not other connections that
also pass through eth0, it's not likely to be a firewall issue.
Since everything is passing through an encrypted tunnel, it's unlikely
that anything is being filtered by either ISP, though lost or delayed
packets remain a possibility.
The problem is asymmetrical, even though identical software is used on
all Linux systems. This again causes me to suspect lost or delayed
packets. One difference, is that the packets from the server pass over
a 100 Mb lan, before entering the VPN on the firewall. The data passing
from the notebook, after entering the VPN, is sent directly to the ppp
connection, which implies it's less likely to have packets lost or
delayed, simply because it can't send them as fast as the server.
If I could get CIPE working on XP, that might offer some more clues.
All in all, it's a very *WEIRD* problem.
[2003/09/11 21:59:45, 0] nmbd/nmbd.c:main(861)
ERROR: Failed when creating subnet lists. Exiting.
[2003/09/11 21:59:45, 0] smbd/server.c:main(698)
smbd version 2.2.3a started.
The error on the last line of log.nmbd looks interesting.
Here's the last entry for log.nmbd on the server:
[2003/09/11 21:58:10, 0]
Unable to find the Domain Master Browser name HOME<1b> for the
Unable to sync browse lists in this workgroup.
For sure, there is not even a host connected running SMB-CIFS protocol.