Has anyone run into this problem with Samba?

[ME]

On Sat, 26 May 2001, Lincoln Peters wrote:
> Anyway, I now have a problem with Samba that is new since I upgraded to Red 
> Hat 7.1, but I had not noticed it until now:
> When I start Samba, smb immediately dies without any warning.  I tried 
> running testparm, and everything came out OK.  Thinking that it might be a 
> program error, I tried installing the samba RPM that ships with Mandrake, 
> and it experienced the same problem.

Some things you may have checked, but are worth mentioning:
Was smbd previously started from inetd or was it daemonized? If it was
started from inetd, is it running when you are trying to start a
daemonized copy?

Look over options like "Bind interfaces only" and then see that you
actually specified interfaces if you do set "bind interfaces only".

> Finally, I tried running 'smbd -d 10' at the command prompt and read through 
> the log file.  I have attached that log to this message.  The only part I 
> could make sense of was that it tried to bind to the IP address 0.0.0.0, 
> which was apparently already in use, and then it shut itself down.  So I 
> pinged 0.0.0.0, and each ping was returned by 127.0.0.1 (my computer).

I did not see your log file attached. There may be a size limit on the
list. I can take a look at it and see if anything stands out to me. You
can send me a copy directly if you want - can't guarantee much.

"Already in use" suggests something else is listening on that port. Check
out a new copy of "lsof" and compile (if necessary) for your kernel. If
there is a package that comes with yout stock kernel, then install that
lsof. (lsof is rather kernel dependend.) You can use it to find out what
PID is locking control to service a particular port. (Though its primary
role/job is listing open files (LiSt OpenFiles). This may help you to
verify nothing else has bound to the ports used by smbd or nmbd.

If you are feeling more ambitious, you can check to see if you have
multiple copies of samba installed in different trees. Perhaps a script
for starting a daemonized copy is referencing a different smbd than the
one that is first in your search PATH for binaries. If you have different
copies, or one copy is being started with an explicit "new" smb.conf file,
you can try using strace and | through grep for ".conf" to see what .conf
file it actually opens. (smbd -d 10 should also have told you what file it
opened, and is more likely to give you more useful information.) The
reason I suggest this is you testparm application may have an earlier
location in your search path than the actual samba installation being
called explicitly as a daemonized service or in (x)inetd.

> Anyone know what this means and/or how to fix it?

If this proves fruitless, there are a few samba mailing lists that can be
found off their web site
http://us1.samba.org/samba/support/

Thanks,
-ME