cifs implementation on samba 3.0 a20

Christopher R. Hertel crh at
Mon Nov 4 19:02:01 GMT 2002

On Mon, Nov 04, 2002 at 11:44:55AM +0100, Lu Jianliang wrote:
> Hi,
> We have the samba configured to use the wins for register/release the
> netbios names (#1b, #1c, #00, etc.), but after we stopped the samba, so it
> should do an explicite release to the wins server,  the names were remain in
> the wins (checked using nmblookup).
> Now I will to check the frames on the wire to verify if it conform to RFC
> 1001/1002.
> Have anyone had this problem?

Samba does not conform to RFC1001/1002.  Samba conforms to Microsoft's 
implementations (which also do not conform to RFC1001/1002).

Even so, Samba should send release notifications to the WINS server *if*
is is shut down correctly.  If we are not doing that, then we have a
(small) problem.

I've been getting some interesting information from a friend showing that
the W2K WINS implementation does a *very* bad job with Name Service
packets.  Related to some work on jCIFS that Mike is doing, he's shown me
that W2K WINS sends malformed Negative Name Query Response messages.  In
some cases, the reply does not contain an Answer Record (which is required
by the RFCs).  In others, if there is an expired entry in the WINS
database, W2K WINS will send an incorrect Negative Name Query Response
containing the IP address from the expired entry.  The IP should *not* be

So, at one level, the problem for Samba is to handle the badly-formed 
messages being sent by Windows.  It is a long-shot, but it *may* be that 
nmblookup is reporting results based on these incorrect messages from 
WINS.  I would need an Ethereal trace, or a system to test against, to 
find out more.

Chris -)-----

Samba Team --     -)-----   Christopher R. Hertel
jCIFS Team --   -)-----   ubiqx development, uninq.
ubiqx Team --     -)-----   crh at
OnLineBook --    -)-----   crh at

More information about the samba-technical mailing list