Samba WINS server killing NMBD process
John E. Malmberg
malmberg at Encompasserve.org
Fri Aug 2 14:23:27 GMT 2002
Alder <ue191 at victoria.tc.canada> wrote:
> Samba 2.0.6 on VMS Alpha 7.2 using TCPIP Services for OpenVMS version 5.0
> I've posted earlier about this problem, and have tried enabling higher
> logging levels to track down the problem, but am still unable to ferret
> out the problem. Basically, adding the line "wins support=yes" to my
> SMB.CONF file results in the NMBD process starting, then dying a few
> seconds later (after an election?).
Try running the NMBD process interactively instead of detached, and see
if it writes anything useful on the console.
The starting of the NMBD process is not keyed to anything in the
SMB.CONF. It is started by a command file prior to starting the rest of
SAMBA and run as a detached process (in 2.0.6).
I do not know what 2.2.4 does, I have not looked at the source.
But it appears that NMBD will need to become a TCP/IP started process in
order to avoid deadlocks in the 2.2.x and later streams from my previous
investigations. See the list archives for details.
> In NMB.LOG after a restart of the NMBD process, I see the first
> reference to WINS_SERVER when its "subnet" is created:
> making subnet name:WINS_SERVER_SUBNET Broadcast address:0.0.0.0 Subnet
> After that, I see each of the various "names" - shares and whatnot - on
> the LAN being added to the WINS database, i.e.:
> add_name_to_subnet: Added netbios name *<00> with first IP 192.168.0.3
> ttl=0 nb_flags=40 to subnet WINS_SERVER_SUBNET
> [snipped more add_name_to_subnet lines]
> add_name_to_subnet: Added netbios name WHATEVER<03> with first IP
> 192.168.0.1 ttl=510662 nb_flags= 4 to subnet WINS_SERVER_SUBNET
> After some more names are added to other subnets, the log file ends with
> these lines:
> [2002/08/01 19:52:14, 3]
That would indicate a problem with the logging system. It stopped in the
middle of an entry.
My guess is there is an ACCESS VIOLATION lurking around their somewhere,
indicating a bad pointer was passed to the snprintf() routine.
> Note the empty entry at the end where the process stops. What should I
> be seeing in the log if the WINS server was working? Can someone
> running the Samba WINS server perhaps dig into NMBD.LOG and post what
> should follow the last line, please.
> Anyone have any suggestions for where I go next to look for the reasons
> behind this behaviour?
You might want to look at NMBD_WORKGROUPDB.C for that debugging
statement in the SAMBA UNIX source. The 272 should help you find it.
wb8tyw at qsl.network
Personal Opinion Only.
More information about the samba-vms