samba - Apr 2010 - Samba 3.3.12/3.4.7/3.5.2 on "ancient" server stops responding/dies

Hello,

I'm trying to upgrade samba (TPTB are going to introduce win7 
workstations) on my old server (mandrake 10.0) currently running 3.0.11.
I setup a virtual machine to do some testing and I built rpms for 3.5.2 
(apart from the other problems it has this one 
https://bugzilla.samba.org/show_bug.cgi?id=7263) then for 3.4.7 and 
finally for 3.3.12.
Each one of them has problems, the most "stable" of the three is
3.3.12.
This one seems to work fine for a machine already in the domain mapping 
shares to drives, but it hangs when trying to browse the server or to 
add a machine to the domain.
E.g.

[luca at localhost samba]$ smbclient -L hp
WARNING: The "printer admin" option is deprecated
Enter luca's password:
Domain=[WETRON] OS=[Unix] Server=[Samba 3.3.12]

         Sharename       Type      Comment
         ---------       ----      -------
         print$          Disk
         privado         Disk      Directorio personal
[....]
         scanner         Disk
         IPC$            IPC       IPC Service (Samba Server 3.3.12)
         ricoh2          Printer   Equipo multifuncional con fax
         ricoh1          Printer   Impresora para eplan
[.....]
         deskjet5800     Printer   Impresora HdP en color

[here it hangs for a minute or so]

Receiving SMB: Server stopped responding
session setup failed: Call returned zero bytes (EOF)
NetBIOS over TCP disabled -- no workgroup available
[luca at localhost samba]$ smbclient -L hp
WARNING: The "printer admin" option is deprecated
Enter luca's password:
Receiving SMB: Server stopped responding
session setup failed: Call returned zero bytes (EOF)
[luca at localhost samba]$ sudo /sbin/service smb status
smbd (pid 4152 4134) is running...
nmbd (pid 4145 4144) is running...


I'm attaching the complete level 10 log, here's an extract at the time 
when smbclient hangs (when the time jumps from 14:53:17 to 14:53:57):

[2010/04/29 14:53:17,  4] smbd/uid.c:change_to_user(258)
   change_to_user: Skipping user change - already user
[2010/04/29 14:53:17,  4] rpc_server/srv_pipe_hnd.c:get_rpc_pipe(1231)
   search for pipe pnum=71ed
[2010/04/29 14:53:17,  5] rpc_server/srv_pipe_hnd.c:get_rpc_pipe(1235)
   pipe name srvsvc pnum=71ed (pipes_open=1)
[2010/04/29 14:53:17,  5] smbd/pipes.c:reply_pipe_close(309)
   reply_pipe_close: pnum:71ed
[2010/04/29 14:53:17, 10] rpc_server/srv_lsa_hnd.c:close_policy_by_pipe(244)
   close_policy_by_pipe: deleted handle list for pipe srvsvc
[2010/04/29 14:53:17,  4] rpc_server/srv_pipe_hnd.c:close_rpc_pipe_hnd(1139)
   closed pipe name srvsvc pnum=71ed (pipes_open=0)
[2010/04/29 14:53:17, 10] lib/dbwrap_tdb.c:db_tdb_fetch_locked(100)
   Locking key 7372767376632F343330
[2010/04/29 14:53:17, 10] lib/dbwrap_tdb.c:db_tdb_fetch_locked(129)
   Allocated locked data 0x0x807ad260
[2010/04/29 14:53:17, 10] lib/dbwrap_tdb.c:db_tdb_record_destr(42)
   Unlocking key 7372767376632F343330
[2010/04/29 14:53:17,  5] lib/util.c:show_msg(645)
[2010/04/29 14:53:17,  5] lib/util.c:show_msg(655)
   size=35
   smb_com=0x4
   smb_rcls=0
   smb_reh=0
   smb_err=0
   smb_flg=136
   smb_flg2=51201
   smb_tid=1
   smb_pid=4303
   smb_uid=100
   smb_mid=8
   smt_wct=0
   smb_bcc=0
[2010/04/29 14:53:57, 10] 
lib/util_sock.c:read_smb_length_return_keepalive(1187)
   got smb length of 35
[2010/04/29 14:53:57,  6] smbd/process.c:process_smb(1573)
   got message type 0x0 of len 0x23
[2010/04/29 14:53:57,  3] smbd/process.c:process_smb(1576)
   Transaction 8 of length 39 (0 toread)
[2010/04/29 14:53:57,  5] lib/util.c:show_msg(645)
[2010/04/29 14:53:57,  5] lib/util.c:show_msg(655)
   size=35
   smb_com=0x71
   smb_rcls=0
   smb_reh=0
   smb_err=0
   smb_flg=8
   smb_flg2=51201
   smb_tid=1
   smb_pid=4303
   smb_uid=100
   smb_mid=9
   smt_wct=0
   smb_bcc=0

then a little later:

[2010/04/29 14:53:57,  5] lib/util_sock.c:read_socket_with_timeout(928)
   read_socket_with_timeout: blocking read. EOF from client.
[2010/04/29 14:53:57, 10] smbd/process.c:receive_smb_raw_talloc(280)
   receive_smb_raw: NT_STATUS_END_OF_FILE
[2010/04/29 14:53:57,  3] smbd/process.c:smbd_process(1952)
   receive_message_or_smb failed: NT_STATUS_END_OF_FILE, exiting
[2010/04/29 14:53:57,  5] lib/gencache.c:gencache_shutdown(93)
   Closing cache file
[2010/04/29 14:53:57,  5] libsmb/namecache.c:namecache_shutdown(81)
   namecache_shutdown: netbios namecache closed successfully.
[2010/04/29 14:53:57,  3] smbd/sec_ctx.c:set_sec_ctx(324)
   setting sec ctx (0, 0) - sec_ctx_stack_ndx = 0
[2010/04/29 14:53:57,  5] auth/token_util.c:debug_nt_user_token(522)
   NT user token: (NULL)
[2010/04/29 14:53:57,  5] auth/token_util.c:debug_unix_user_token(548)
   UNIX token of user 0
   Primary group is 0 and contains 0 supplementary groups
[2010/04/29 14:53:57,  5] smbd/uid.c:change_to_root_user(370)
   change_to_root_user: now uid=(0,0) gid=(0,0)
[2010/04/29 14:53:57, 10] lib/dbwrap_tdb.c:db_tdb_fetch_locked(100)
   Locking key 49442F343330342F3130
[2010/04/29 14:53:57, 10] lib/dbwrap_tdb.c:db_tdb_fetch_locked(129)
   Allocated locked data 0x0x807b3248
[2010/04/29 14:53:57, 10] lib/dbwrap_tdb.c:db_tdb_record_destr(42)
   Unlocking key 49442F343330342F3130
[2010/04/29 14:53:57,  3] smbd/connection.c:yield_connection(31)
   Yielding connection to
[2010/04/29 14:53:57, 10] lib/dbwrap_tdb.c:db_tdb_fetch_locked(100)
   Locking key D0100000FFFFFFFF0000
[2010/04/29 14:53:57, 10] lib/dbwrap_tdb.c:db_tdb_fetch_locked(129)
   Allocated locked data 0x0x807b4190
[2010/04/29 14:53:57, 10] lib/dbwrap_tdb.c:db_tdb_record_destr(42)
   Unlocking key D0100000FFFFFFFF0000
[2010/04/29 14:53:57,  3] smbd/server.c:exit_server_common(980)
   Server exit (normal exit)

Since the problems are similar for the 3 versions of samba, I suspect 
that one of the old system libraries is the cause of the problem:

#ldd /usr/sbin/smbd
	linux-gate.so.1 =>  (0xffffe000)
	libldap.so.2 => /usr/lib/libldap.so.2 (0x40021000)
         liblber.so.2 => /usr/lib/liblber.so.2 (0x40055000)
	libgssapi_krb5.so.2 => /usr/lib/libgssapi_krb5.so.2 (0x40061000)
	libkrb5.so.3 => /usr/lib/libkrb5.so.3 (0x40075000)
	libk5crypto.so.3 => /usr/lib/libk5crypto.so.3 (0x400ea000)
	libcom_err.so.2 => /lib/libcom_err.so.2 (0x40112000)
	libresolv.so.2 => /lib/libresolv.so.2 (0x40115000)
	libcups.so.2 => /usr/lib/libcups.so.2 (0x40127000)
	libssl.so.0.9.7 => /usr/lib/libssl.so.0.9.7 (0x40145000)
	libcrypto.so.0.9.7 => /usr/lib/libcrypto.so.0.9.7 (0x40177000)
	libnsl.so.1 => /lib/libnsl.so.1 (0x4027a000)
	libcrypt.so.1 => /lib/libcrypt.so.1 (0x4028d000)
	libpam.so.0 => /lib/libpam.so.0 (0x402ba000)
	libacl.so.1 => /lib/libacl.so.1 (0x402c3000)
	libattr.so.1 => /lib/libattr.so.1 (0x402ca000)
	libcap.so.1 => /lib/libcap.so.1 (0x402ce000)
	libdl.so.2 => /lib/libdl.so.2 (0x402d2000)
	libpopt.so.0 => /lib/libpopt.so.0 (0x402d5000)
	libtalloc.so.1 => /usr/lib/libtalloc.so.1 (0x402dd000)
	libtdb.so.1 => /usr/lib/libtdb.so.1 (0x402e4000)
	libwbclient.so.0 => /usr/lib/libwbclient.so.0 (0x402f0000)
	libc.so.6 => /lib/tls/libc.so.6 (0x402fb000)
	libsasl2.so.2 => /usr/lib/libsasl2.so.2 (0x40443000)
	/lib/ld-linux.so.2 => /lib/ld-linux.so.2 (0x40000000)

these come from:

openldap 2.1.25
kerberos5 1.3
e2fsprogs 1.34
cups 1.1.20
openssl 0.9.7c
glibc 2.3.3
pam 0.77
attr 2.4.14
acl 2.2.22
popt 1.8.2
sasl 2.1.15

(libtalloc, libtdb, libwbclient are the ones that come with samba 3.3.12 
here).

and, minus libtalloc, libtdb and libwbclient, are the same libraries 
used by the currently working 3.0.11
Any one of those seem suspect?

Bye
-- 
Luca Olivetti
Wetron Automatizaci?n S.A. http://www.wetron.es/
Tel. +34 93 5883004 (Ext.133)  Fax +34 93 5883007
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: smb.conf
URL:
<http://lists.samba.org/pipermail/samba/attachments/20100429/97907afc/attachment.ksh>

En/na Luca Olivetti ha escrit:
> I'm attaching the complete level 10 log, here's an extract at the
time
The complete log has been scrubbed by the list.
It's available here:

http://www.wetron.es/samba/log.txt

Bye
-- 
Luca Olivetti
Wetron Automatizaci?n S.A. http://www.wetron.es/
Tel. +34 93 5883004 (Ext.133)  Fax +34 93 5883007

En/na Luca Olivetti ha escrit:
> Each one of them has problems, the most "stable" of the three is
3.3.12.
> This one seems to work fine for a machine already in the domain mapping 
> shares to drives, but it hangs when trying to browse the server or to 
> add a machine to the domain.
I now tried both to rebuild 3.0.11 (to see if there is some problem with 
my build system), and it still works fine, and 3.3.0 (trying to bisect 
when the problem was introduced) and it still fails the same way with a 
slight difference:
> E.g.
> 
> [luca at localhost samba]$ smbclient -L hp
> WARNING: The "printer admin" option is deprecated
> Enter luca's password:
> Domain=[WETRON] OS=[Unix] Server=[Samba 3.3.12]
> 
>         Sharename       Type      Comment
>         ---------       ----      -------
>         print$          Disk
>         privado         Disk      Directorio personal
> [....]
>         scanner         Disk
>         IPC$            IPC       IPC Service (Samba Server 3.3.12)
>         ricoh2          Printer   Equipo multifuncional con fax
>         ricoh1          Printer   Impresora para eplan
> [.....]
>         deskjet5800     Printer   Impresora HdP en color
> 
> [here it hangs for a minute or so]
And here, after the minute, instead of this
> 
> Receiving SMB: Server stopped responding
> session setup failed: Call returned zero bytes (EOF)
> NetBIOS over TCP disabled -- no workgroup available
I get this:

Domain=[WETRON] OS=[Unix] Server=[Samba 3.3.0]

         Server               Comment
         ---------            -------
         HP                   Samba Server 3.3.0

         Workgroup            Master
         ---------            -------
         WETRON               HP


And smbd dies (with 3.3.12 it also died but not until the next query).

I'm going to try earlier versions.

Bye
-- 
Luca Olivetti
Wetron Automatizaci?n S.A. http://www.wetron.es/
Tel. +34 93 5883004 (Ext.133)  Fax +34 93 5883007