I have Samba 2.06 on RedHat 6.0 as a client in the domain called
"workgroup". It has Domain logins disabled, and is set so that it
cannot
become a local browse master.
The domain controller on the network is a SCO Openserver 5.0.2 running
Advanced File and Print Server 3.5.2 (with the name scadis). The only
protocol on the network is TCP/IP and there is no DNS or WINS server
available.
The problem I am having since bringing the Samba box on-line is that every
few hours, other machines on the network start giving "Domain Controller
not
available" messages. This disrupts network function for a while, but some
minutes later, everything returns to normal again. I have found that running
nbtstat -c on one of the NT4WS machines on the network before and during one
of the "blackouts" gives me this result:
Before:
33 <20> UNIQUE 192.168.102.33 480
SCADIS <03> UNIQUE 192.168.102.48 -1
SCADIS <00> UNIQUE 192.168.102.48 -1
SCADIS <20> UNIQUE 192.168.102.48 -1
workgroup <1C> UNIQUE 192.168.102.48 -1
workgroup <1B> UNIQUE 192.168.102.48 -1
During:
SCADIS#19388 <20> UNIQUE 192.168.102.48 120
33 <20> UNIQUE 192.168.102.33 480
SCADIS <03> UNIQUE 192.168.102.48 -1
SCADIS <00> UNIQUE 192.168.102.48 -1
SCADIS <20> UNIQUE 192.168.102.48 -1
WORKGROUP <1C> GROUP 192.168.102.48 660
workgroup <1C> UNIQUE 192.168.102.48 -1
workgroup <1B> UNIQUE 192.168.102.48 -1
It seems to me that it is this 3rd-last entry that is causing a problem, but
I am not fluent enough in NETBIOS to see exactly what is going on here. My
current theory is that the Samba machine is broadcasting periodically to
find the domain controller, and the reply is picked up by the NT machine.
This disrupts its knowledge of who the domain controller is until this entry
times out.
This matches the symptoms, and running nbtstat -R on the NT machine seems to
bring an instant cure.
Thoughts on the cause and the cure are more than welcome.
Marc Cosgrove