samba - Oct 1999 - Corrupted Excel files, oplock_break(905) errors

Hi there everyone!

I'm using Samba 2.0.5a-19990721 on a RedHat 6.0 machine with Linux kernel
2.2.12. My system generally works fine, however recently I've had some
complaints about some files being corrupted. The latest and best-handled
to enable a debugging happened at around the same time. Both clients were
using Windows95 OSR2 machines with Excel 97 SP-2. Two users reported the
problem and they were using separate machines. Nobody was "messing
around"
with the server (though I noticed SIGHUPs in the log files!!! I don't know
why they were there, maybe someone could explain why Samba gets regular
SIGHUPs and if these could cause damage?).

Here is the log of the first user. Please pardon the excessive
log-quoting. I'm hoping to help you all help me out. :-)

NOTE: Please cc your reply to me <jijo@leather-collection.com> because I
cannot subscribe myself to the list. I will, however, regularly check the
archives on the web for responses to my query. Thank you very very much!!!

---------------------------------------------------------------
[1999/10/20 16:53:27, 0] smbd/oplock.c:oplock_break(905) oplock_break
resend

[1999/10/20 16:53:37, 0] smbd/oplock.c:oplock_break(905) oplock_break
resend

[1999/10/20 16:53:47, 0] smbd/oplock.c:oplock_break(905) oplock_break
resend

[1999/10/20 16:53:54, 0] smbd/oplock.c:oplock_break(915) oplock_break: end
of file from client oplock_break failed for file Delivery
Confirmation-Update.xls (dev = 309, inode = 2385923).

[1999/10/20 16:53:54, 0] smbd/oplock.c:oplock_break(992) oplock_break:
client failure in break - shutting down this smbd.

[1999/10/20 16:53:54, 1] smbd/service.c:close_cnum(557) sales-1 (0.0.0.0)
closed connection to service ce

[1999/10/20 16:53:54, 1] smbd/service.c:make_connection(521) sales-1
(192.168.0.18) connect to service ce as user jta (uid=508, gid=250) (pid
3052)

[1999/10/20 16:54:07, 0] smbd/oplock.c:oplock_break(905) oplock_break
resend

[1999/10/20 16:54:17, 0] smbd/oplock.c:oplock_break(905) oplock_break
resend

[1999/10/20 16:54:27, 0] smbd/oplock.c:oplock_break(905) oplock_break
resend

[1999/10/20 16:54:37, 0] smbd/oplock.c:oplock_break(922) oplock_break:
receive_smb timed out after 30 seconds. oplock_break failed for file
Delivery Confirmation-Update.xls (dev = 309, inode = 2385923).

[1999/10/20 16:54:37, 0] smbd/oplock.c:oplock_break(992) oplock_break:
client failure in break - shutting down this smbd.
---------------------------------------------------------------

The log of the second user. Very very similar error codes!!!

---------------------------------------------------------------

[1999/10/20 16:33:01, 0] smbd/filename.c:print_stat_cache_statistics(108)  
stat cache stats: lookups = 5, hits = 4, misses = 1, stat cache was
80.000000% effective.

[1999/10/20 16:49:49, 0] smbd/oplock.c:oplock_break(905) oplock_break
resend

[1999/10/20 16:49:59, 0] smbd/oplock.c:oplock_break(905) oplock_break
resend

[1999/10/20 16:50:09, 0] smbd/oplock.c:oplock_break(905) oplock_break
resend

[1999/10/20 16:50:19, 0] smbd/oplock.c:oplock_break(922) oplock_break:
receive_smb timed out after 30 seconds. oplock_break failed for file
Reports/jostatus.xls (dev = 309, inode = 3321862).

[1999/10/20 16:50:19, 0] smbd/oplock.c:oplock_break(992) oplock_break:
client failure in break - shutting down this smbd.

[1999/10/20 16:50:19, 1] smbd/service.c:close_cnum(557) acctg-2 (0.0.0.0)
closed connection to service jomaster

[1999/10/20 16:50:19, 1] smbd/service.c:make_connection(521) acctg-2
(192.168.0.14) connect to service jomaster as user rose (uid=511, gid=250)
(pid 2977)

[1999/10/20 16:50:31, 0] smbd/oplock.c:oplock_break(905) oplock_break
resend

[1999/10/20 16:50:41, 0] smbd/oplock.c:oplock_break(905) oplock_break
resend

[1999/10/20 16:50:51, 0] smbd/oplock.c:oplock_break(905) oplock_break
resend

[1999/10/20 16:51:01, 0] smbd/oplock.c:oplock_break(922) oplock_break:
receive_smb timed out after 30 seconds. oplock_break failed for file
Reports/jostatus.xls (dev = 309, inode = 3321863).

[1999/10/20 16:51:01, 0] smbd/oplock.c:oplock_break(992) oplock_break:
client failure in break - shutting down this smbd.
---------------------------------------------------------------

Again I thank you all for the time and energy! :-)

   [        Jijo Sevilla         ]
   [ jijo@leather-collection.com ]
   [ eFax # :  +1 (707) 988-0887 ]

> Date:   Wed, 20 Oct 1999 18:54:45 -0200
> From:   Andreas Hasenack <unixach@ez-poa.com.br>
> To:     plussier@ne.arris-i.com,
>         "Paul L. Lussier" <plussier@ne.arris-i.com>
> Cc:     Multiple recipients of list SAMBA <samba@samba.org>
> Subject: Re: REPEAT: hasn't anyone used smbclient linux->linux?
> Message-ID: <99102018575000.02774@maestro.hasenack.fam.br>
> Content-Type: text/plain
> MIME-Version: 1.0
> Content-Transfer-Encoding: 8bit
>
> Em qua, 20 out 1999, Paul L. Lussier escreveu:
> > In a message dated: Wed, 20 Oct 1999 17:40:49 -0200
> > Andreas Hasenack said:
> >
> > >NFS's authentication is weak (it's host based). With Samba
you get
also user
> > >authentication. With pam_smb you could even replace N[IY]S[+] for
the user
> > >auth part.
> >
> > True, but the basic Unix permissions are user *and* group based.
Proper
> > configuration of these, combined with the host based auth make NFS a
better
> > choice IMO than SMB.  In addition, with NFS and netgroups, you can
restrict a
> > user access to anything based on both the username *and* the host
combined,
> > which smb can't do.
>
> I think I agree with you. I just don't like the concept that, with the
root
> password of a *client* machine, one can su to any local user and thus
> invalidate the user authentication part. OK, one shouldn't give the
root
> password away, but I don't like this concept: a client machine being
able to
> look at any file (but root owned ones) on a *server* machine.
You may have missed this little gem. Set the stickybit for ALL of your
share points: ex: "chmod  +t  /home". This will cause the root
accounts
on remote systems to be treated as a non-root account like nobody
(60001), noaccess (60002), or access denied (-1) depending the options
specified in your share statement. Linux is supposed to 'root squash'
(change client root uid's to 65534) by default. remove
"no_root_squash"
from your export file.
Examples:
    Solaris 2.6/NIS+ environment set anon=-1 in /etc/dfs/dfstab:
share -F nfs -o anon=-1 -o rw /share.
    Linux/NIS env set anonuid=-1 in /etc/exports:     /home
(rw,anonuid=-1,anongid=-1)

- Jim Mulholland
    Periphonics Corporation

[snip]
> [1999/10/20 16:54:27, 0] smbd/oplock.c:oplock_break(905)
> oplock_break resend
> 
> [1999/10/20 16:54:37, 0] smbd/oplock.c:oplock_break(922)
> oplock_break: receive_smb timed out after 30 seconds.
> oplock_break failed for file Delivery Confirmation-Update.xls
> (dev = 309, inode = 2385923). 
What is the behavior with an Excel file *without* a space in the 
name?

Steve

******************************************************************
Stephen L Arnold                      http://www.rain.org/~sarnold
#include <std_disclaimer.h>
******************************************************************

Federico Sevilla III wrote:
> 
> Hi there everyone!
> 
> I'm using Samba 2.0.5a-19990721 on a RedHat 6.0 machine with Linux
kernel
> 2.2.12. My system generally works fine, however recently I've had some
> complaints about some files being corrupted. The latest and best-handled
Glad I am not the only one! I am seeing the same behaviour for Excel
files, and in one case, Word docs as well. Samba 2.0.5a on Solaris 2.6.

-- 
Gang Warily

Federico Sevilla III wrote:
> I'm using Samba 2.0.5a-19990721 on a RedHat 6.0 machine with Linux
kernel
> 2.2.12. My system generally works fine, however recently I've had some
> complaints about some files being corrupted. The latest and best-handled
> to enable a debugging happened at around the same time. Both clients were
> using Windows95 OSR2 machines with Excel 97 SP-2. Two users reported the
> problem and they were using separate machines. Nobody was "messing
around"
> with the server (though I noticed SIGHUPs in the log files!!! I don't
know
> why they were there, maybe someone could explain why Samba gets regular
> SIGHUPs and if these could cause damage?).
> 
> Here is the log of the first user. Please pardon the excessive
> log-quoting. I'm hoping to help you all help me out. :-)
Could you try downloading the 2.0.6-pre3 snapshot and testing
to see if it fixes the problem. We have done extensive work
in this area on this issue.

Cheers,

	Jeremy Allison,
	Samba Team.

-- 
--------------------------------------------------------
Buying an operating system without source is like buying
a self-assembly Space Shuttle with no instructions.
--------------------------------------------------------

I've posted here about this problem before.

The problem has to do with Windows clients abandoning an existing TCP
connection to the SMB server due to a timeout (the threshold is 45
seconds), then re-connecting to the server.

Now, with Samba what happens is that the old smbd, for whatever reason,
hangs around and neither it nor the kernel realize that its TCP
connection to the client is dead, so that the old smbd sleeps waiting
for input from the client.

Meantime, the client reconnected and a new smbd starts and the the
client attempts to re-establish the locks, oplocks, share modes that it
had held prior to reconnecting and the new smbd may wait for those locks
to be given up (by the old smbd) or, in the case of oplocks, it will
send an oplock_break to the old smbd (but the old smbd will not catch
the oplock_break as it is sleeping).

The solution that I use is as follows:

 - set the kernel SO_KEEPALIVE timer to 10 minutes (instead of the
   default 2 hours)

 - set the Samba 'keepalive' config parameter to 5 minutes

 - use the 'root preexec' and 'root postexec' Samba config
parameters
   with a script that detects and kills old, stale smbd processes when
   clients reconnect. When a stale smbd is killed it releases all
   locks/oplocks/share modes and the new smbd can proceed.

I've posted about this to this list twice before.

It's becoming a FAQ.

I've still not seen a complete, correct analysis of why the Solaris
kernel and smbd fail to notice the dead TCP connection and myself and
others have speculated here before about the source of the bug (NT?
Solaris?). At any rate, the preexec/postexec stale smbd detector script
that I use does the trick.

Nico
-DISCLAIMER: an automatically appended disclaimer may follow. By posting-
-to a public e-mail mailing list I hereby grant permission to distribute-
-and copy this message.-

This message contains confidential information and is intended only 
for the individual named.  If you are not the named addressee you 
should not disseminate, distribute or copy this e-mail.  Please 
notify the sender immediately by e-mail if you have received this 
e-mail by mistake and delete this e-mail from your system.

E-mail transmission cannot be guaranteed to be secure or error-free 
as information could be intercepted, corrupted, lost, destroyed, 
arrive late or incomplete, or contain viruses.  The sender therefore 
does not accept liability for any errors or omissions in the contents 
of this message which arise as a result of e-mail transmission.  If 
verification is required please request a hard-copy version.  This 
message is provided for informational purposes and should not be 
construed as a solicitation or offer to buy or sell any securities or 
related financial instruments.

Hi,

I have problems with samba 2.0.5a amd my 3com 3c905b in 100 mbit mode:
writing performance: about 7,5 MB/s (ide hdd !!)
reading performance: about 0,2 MB/s (I can't explain !!)

Can anybody help me ? Is there a spezial option that can help me ?


Thank you all



Where will you go tomorrow ?


-------------- next part --------------
HTML attachment scrubbed and removed