[Samba] netbios name resolution problem

[Russell, Toby]

Hello all,

the subject is actually my guess as to the nature of the problem.

We have two Sun machines (Sol 8), both of which offer samba shares from
within the same workgroup. They are the only members of the group. One is a
produciton system, the other a test system. The company for whom we
administer these machines has the curious situation where they can browse
the production machine, but not the test machine, unless they use the test
machine's IP address when executing e.g. net view. They come from outside
the network via their own routers to us. We can browse both machines from
within the network, and another company (who administer the databases) can
also browse both machines, using the names. The problem for our client is
there regardless of where they attempt to log in from, within their company
networks.

I have reinstalled the latest version of samba on the test machine (2.2.8a),
and on the 'working' production we have 2.2.2. We have changed who should be
masterbrowser from production to test machine and back again, but this makes
no difference. We have even placed the machines in our client's workgroup,
although the two sun boxes have no way of conversing with our client's
PDC/masterbrowser. No matter what we do, the situation stays the same. 

The problem first reared its head on the 20th May. Before that date there
were no problems with samba. And it is the usual story; all relevant parties
have not changed a thing anywhere. All we have done is to reboot the test
machine due to a httpd problem. Strangely, having finally seen a broadcast
address error and fixed it, the problem has remained, despite reboots of the
relevant PC at our client's company.

For a while the following messages showed up in log.nmbd (I have the log
level at 6 on the test machine):

[2003/05/28 12:27:23, 4]
nmbd/nmbd_packets.c:retransmit_or_expire_response_records(1664)
  retransmit_or_expire_response_records: timeout for packet id 1461 to IP *
on subnet *
[2003/05/28 12:54:52, 4]
nmbd/nmbd_packets.c:retransmit_or_expire_response_records(1664)
  retransmit_or_expire_response_records: timeout for packet id 4343 to IP *
on subnet *
[2003/05/28 13:04:53, 4]
nmbd/nmbd_packets.c:retransmit_or_expire_response_records(1664)
  retransmit_or_expire_response_records: timeout for packet id 4348 to IP *
on subnet *
[2003/05/28 13:04:53, 4]
nmbd/nmbd_packets.c:retransmit_or_expire_response_records(1664)
  retransmit_or_expire_response_records: timeout for packet id 4349 to IP *
on subnet *
[2003/05/28 13:52:41, 4]
nmbd/nmbd_packets.c:retransmit_or_expire_response_records(1664)
  retransmit_or_expire_response_records: timeout for packet id 8153 to IP *
on subnet *

However, those messages no longer seem to arise.

I have also tried setting bind interfaces only to yes, and listing the
interface in smb.conf.

On both machines smbclient returns the correct results, as does testparm.
The client can ping the test machine using the name, as well as ftp it. I
have spoken with the network people at the client's company, but they have
thrown up their hands and claimed to have changed nothing. In fact
everything works EXCEPT browsing by name on the test machine. I have been
working on this problem for days, have trawled google and am finally without
a clue as to how to proceed, other than to mail the people at the list here.
I hope you can help me.

Regards and thanks in advance

Toby Russell
(Applications Providing/UNIX)

Telefon          +49 (0) 30 72 62 3- 6669
Telefax          +49 (0) 30 72 62 3- 6689
Mobil             +49 (0) 172 6847428
E-Mail            Russell.Toby at berlindat.de
Internet          www.BerlinDat.de
_______________________________________________________

BerlinDat
Gesellschaft für Informationsverarbeitung und Systemtechnik mbH
Rohrdamm 7
13629 Berlin