[Samba] Weird samba problems / more info

[Peter Eckhardt]

I have a level 10 debug of the smbd/oplock process_local_message problem
but can still not figure out what causes the unknown UDP message command 
code. I checked the source but could only see that this is an invalid 
opcode command.

Anyone a clue what is happening here ?
Might wrong opcode commands crash samba ?

Peter


[2003/09/04 11:50:12, 3] smbd/process.c:switch_message(685)
   switch message SMBreadbraw (pid 12660)
[2003/09/04 11:50:12, 4] smbd/uid.c:change_to_user(119)
   change_to_user: Skipping user change - already user
[2003/09/04 11:50:12, 3] smbd/reply.c:reply_readbraw(2368)
   readbraw fnum=5697 start=53248 max=4096 min=0 nread=4096
[2003/09/04 11:50:12, 10] smbd/fileio.c:seek_file(51)
   seek_file (LEXWARE/Buchhalterpro/daten/Framework/lxframe.dll): 
requested pos = 53248
, new pos = 53248
[2003/09/04 11:50:12, 10] smbd/fileio.c:read_file(128)
   read_file (LEXWARE/Buchhalterpro/daten/Framework/lxframe.dll): pos = 
53248, size = 4
096, returned 4096
[2003/09/04 11:50:12, 5] smbd/reply.c:reply_readbraw(2372)
   readbraw finished
[2003/09/04 11:50:13, 10] smbd/process.c:receive_message_or_smb(265)
   receive_message_or_smb: oplock_message is waiting.
[2003/09/04 11:50:13, 10] smbd/process.c:async_processing(118)
   async_processing: Doing async processing.
[2003/09/04 11:50:13, 5] smbd/oplock.c:receive_local_message(108)
   receive_local_message: doing select with timeout of 1 ms
[2003/09/04 11:50:13, 5] smbd/oplock.c:process_local_message(351)
   process_local_message: Got a message of length 23 from port (53)
[2003/09/04 11:50:13, 0] smbd/oplock.c:process_local_message(418)
   process_local_message: unknown UDP message command code (4c94) - 
ignoring.
[2003/09/04 11:50:13, 10] 
lib/util_sock.c:read_smb_length_return_keepalive(559)
   got smb length of 90
[2003/09/04 11:50:13, 6] smbd/process.c:process_smb(845)
   got message type 0x0 of len 0x5a
[2003/09/04 11:50:13, 3] smbd/process.c:process_smb(846)
   Transaction 737 of length 94
[2003/09/04 11:50:13, 5] lib/util.c:show_msg(275)
   size=90
   smb_com=0x8
   smb_rcls=0
   smb_reh=0
   smb_err=0
   smb_flg=0
   smb_flg2=0
[2003/09/04 11:50:13, 5] lib/util.c:show_msg(281)
   smb_tid=3
   smb_pid=8499
   smb_uid=100
   smb_mid=29185
   smt_wct=0
[2003/09/04 11:50:13, 5] lib/util.c:show_msg(291)
   smb_bcc=55
[2003/09/04 11:50:13, 10] lib/util.c:dump_data(1541)
   [000] 04 5C 4C 45 58 57 41 52  45 5C 42 55 43 48 48 41  .\LEXWAR E\BUCHHA




-----

Hi,

some days ago i sent a mail regarding a small samba installation
with periodically dying nmbds. Since the customer didn't call
again regarding these problems I had the opinion that everything is
fine ....

Today the customer called and mentioned that there are additional
problems. He had fixed the nmbd problem everyday by rebooting ....

Now the smbds also start to die. Again no pattern. No way to
reproduce.

But there were strange entries in the logs

linux smbd[8311]: smbd/oplock.c process_local_message(418)
linux smbd[8311]: process localmessage: unknown UDP message command code 
(e393) ignoring

with different command codes i.e. 1e73, 2b94 and others.

After some of these messages have arrived the smbd processes start to
panic with internal errors. one after the other.

I could locate the machine which is sending the UDP messages. And we 
will check what happens when that machine is down tomorrow.

But anyway, does anyone has a clue what these unknown UDP message codes
may be?

Thanks a lot
Peter
-- 

dadi-linux                       www.dadi-linux.de

Peter Eckhardt                   Fon: +49 6071 951256
Weberstr. 36B                    Fax: +49 6071 951257
64846 Groß-Zimmern               peter.eckhardt at dadi-linux.de