samba - Mar 2002 - tdb((NULL)): tdb_brlock failed on Samba 2.2.3a startup

Hi all,.

I'm getting this:-

[2002/03/08 10:26:28, 8] lib/util.c:(1274)
  fcntl_lock 6 34 0 1 2
[2002/03/08 10:26:28, 8] lib/util.c:(1314)
  fcntl_lock: Lock call successful
[2002/03/08 10:26:28, 5] tdb/tdbutil.c:(475)
  tdb((NULL)): tdb_brlock failed (fd=7) at offset 4 rw_type=2 lck_type=6
[2002/03/08 10:26:28, 5] smbd/connection.c:(156)
  claiming  0
[2002/03/08 10:26:28, 0] tdb/tdbutil.c:(475)
  tdb(/usr/local/samba/var/locks/connections.tdb): expand_file write of 1024
failed (Invalid argument)
[2002/03/08 10:26:28, 0] smbd/connection.c:(188)
  claim_connection: tdb_store failed with error IO Error.
[2002/03/08 10:26:28, 3] lib/util_sock.c:(798)
  bind succeeded on port 139


on startup of smbd (2.2.3a) with debug level 100.


And there-after everything tdb-related seems to fail, which means I
can't connect to Samba at all; well, I can, but I get:-

	SMBtconX failed.  ERRSRV - ERRnoresource (No resources currently
	available for request.)

and nmbd is having a hard time too.

This is on IRIX 5.3 config-ed with:-

	./configure --exec-prefix=/usr/local/samba/2.2.3a \
	--mandir=/usr/local/samba/2.2.3a/man \
	--with-configdir=/usr/local/samba/etc \
	--with-codepagedir=/usr/local/samba/etc

I'm recompiling Samba now so as to (try to) include the 'errno'
value in
the tdb_brlock error message, but I was wondering if ayone has seen
anything like this before. (just finished compile, errno=13 (EACCESS))

Also, should I be worried about the  ((NULL)) where the filename should
be?

The directory in which the connections.tdb (et al.) are being stored is
physically local to the UNIX box, is owned and writeable by root. The
files are created, and the 'smbd.pid' file is created, written to and
locked OKay by smbd (proved by running a second smbd and seeing it fail
to lock the pid file because the first smbd has it locked).

(And side issue, hows does one get the debug timestamps and sourcecode
filenames on a 'smbd -i' invocation?)

(Second side issue, is it possible for tdb debug messages to have the
real source code reference instead of all coming from the same place
(line 475) of tdb_util.c ?)

A quick mailing list search has shown nothng similar.  Also 'tdbtool'
can
create tdbs, but not 'insert' nor 'store' items into them. (Test
file
/tmp/f ) (error message 'insert failed')

	Help?

                               Mac
          Assistant Systems Adminstrator @nibsc.ac.uk
                        dmccann@nibsc.ac.uk
   Work: +44 1707 654753 x285      Everything else: +44 7956 237670 (anytime)

Some further info about:-
>To: samba@samba.org
>From: Mac <dmccann@nibsc.ac.uk>
>Subject: [Samba] tdb((NULL)): tdb_brlock failed  on Samba 2.2.3a startup
>Date: Fri, 8 Mar 2002 11:24:23 +0000 (GMT)
>
>
>I'm getting this:-
>
>[2002/03/08 10:26:28, 8] lib/util.c:(1274)
>  fcntl_lock 6 34 0 1 2
>[2002/03/08 10:26:28, 8] lib/util.c:(1314)
>  fcntl_lock: Lock call successful
>[2002/03/08 10:26:28, 5] tdb/tdbutil.c:(475)
>  tdb((NULL)): tdb_brlock failed (fd=7) at offset 4 rw_type=2 lck_type=6

I can stop this error (brlock failed) by starting smbd first (and nmbd
second) but the next tdb error :-

>[2002/03/08 10:26:28, 5] smbd/connection.c:(156)
>  claiming  0
>[2002/03/08 10:26:28, 0] tdb/tdbutil.c:(475)
>  tdb(/usr/local/samba/var/locks/connections.tdb): expand_file write of 1024
failed (Invalid argument)
>[2002/03/08 10:26:28, 0] smbd/connection.c:(188)
>  claim_connection: tdb_store failed with error IO Error.

still occurs just the same.

Dunno if this is significant.


Is it now mandatory to start smbd first?



                               Mac
          Assistant Systems Adminstrator @nibsc.ac.uk
                        dmccann@nibsc.ac.uk
   Work: +44 1707 654753 x285      Everything else: +44 7956 237670 (anytime)