Bug 3284 - A lot of SMB errors, crash wriet files and freez client Linux
Summary: A lot of SMB errors, crash wriet files and freez client Linux
Status: RESOLVED INVALID
Alias: None
Product: Samba 3.0
Classification: Unclassified
Component: smbclient (show other bugs)
Version: 3.0.20b
Hardware: x86 Linux
: P3 critical
Target Milestone: none
Assignee: Samba Bugzilla Account
QA Contact: Samba QA Contact
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2005-11-25 05:15 UTC by Charlie G Mentorez
Modified: 2007-01-11 10:50 UTC (History)
2 users (show)

See Also:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Charlie G Mentorez 2005-11-25 05:15:56 UTC
System setup.
We using samba for all our system servers.
All installation is based on the sewdish Linux dist CRUX by Per Liden. 

1. Central Linux fileserver with samba
2. All linux applications-server using samba client. As eMail server, Http Apache and MySQL etc.
3. All windows servers access from/to the central linux samba server.

All datafiles have always been stored on the central fileserver.
Which includ http files as MySQL database files.

History.
We set up the basic system during 2001 and since then it have work very good.
We have during a day up to 30-50 Mbs throuput to hour linux fileserver.
this have never cause any problem during the last 4 years there have not not been a single error.

Problem.
Because some of the system now went old we have upgrade them both harware and linux system and other releatde functions.
To the new installtion we use Crux ISO 2.1 with 2.6.14.2 alt 2.6.13.2 and udev
To this we installed samba 3.0.20b
This server is also using apache 2.0, php, and MySQL 5

When we now started up and using the new system with samba 3.0.20b and 2.6.14.2 kernel we got a lot of strange error. All other old application-server still working on both linux and windows to our central linux fileserver.

The error you can see bellow.

An other problem which never have happned befor is MySQL database files will be crash writed on the fileserver. All files for MySQL is used via symlink's. And
this have bean working in 4 years with the old SAMBA, not an single error before. For temp solution we now have the databases on the local system. But all other files for webpages still recide on the central server.

As we can see in the test we have made there is no general communication errors. Nothing is reported to be wrong. 
We have tested with differn NIC's to make sure the problem not are related to that.

:Errors.
Nov 25 12:52:39 ndc-fox kernel: smb_lookup: find nya/sparbanken.gif failed, error=-5
Nov 25 12:52:42 ndc-fox kernel: smb_add_request: request [c823b680, mid=61080] timed out!
Nov 25 12:53:17 ndc-fox kernel: smb_add_request: request [c823ba80, mid=486] timed out!
Nov 25 12:54:35 ndc-fox kernel: smb_add_request: request [c823ba80, mid=2489] 

timed out!
Nov 25 12:55:56 ndc-fox kernel: smb_add_request: request [d5498080, mid=3801] timed out!
Nov 25 12:58:05 ndc-fox kernel: smb_add_request: request [c8419280, mid=7832] timed out!
Nov 25 12:59:24 ndc-fox kernel: smb_add_request: request [c8419280, mid=9032] timed out!
Nov 25 12:59:25 ndc-fox kernel: smb_add_request: request [c8419d80, mid=9085] timed out!
Nov 25 12:59:25 ndc-fox kernel: smb_lookup: find spel/.htaccess failed, error=-5
Nov 25 12:59:57 ndc-fox kernel: smb_add_request: request [c8419d80, mid=10489] timed out!
Nov 25 13:00:06 ndc-fox ntpdate[17426]: sendto(0.0.0.1): Invalid argument
Nov 25 13:00:09 ndc-fox last message repeated 3 times
Nov 25 13:00:10 ndc-fox ntpdate[17426]: no server suitable for synchronization found
Nov 25 13:01:11 ndc-fox kernel: smb_add_request: request [c8419d80, mid=12829] timed out!
Nov 25 13:01:11 ndc-fox kernel: smb_lookup: find utskick/.htaccess failed, error=-5
Nov 25 13:02:00 ndc-fox kernel: smb_add_request: request [c8419280, mid=13490] timed out!
Nov 25 13:02:52 ndc-fox kernel: smb_receive_header: short packet: 0
Nov 25 13:03:22 ndc-fox kernel: smb_add_request: request [c23e8d80, mid=15303] timed out!
Nov 25 13:03:22 ndc-fox kernel: smb_add_request: request [c23e8e80, mid=15304] timed out!
Nov 25 13:03:33 ndc-fox kernel: smb_add_request: request [c23e8c80, mid=15633] timed out!

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

Nov 25 11:31:53 ndc-fox kernel: smb_add_request: request [ca259580, mid=7732] timed out!
Nov 25 11:31:59 ndc-fox kernel: smb_add_request: request [ca259280, mid=7781] timed out!
Nov 25 11:32:07 ndc-fox kernel: smb_receive_header: short packet: 0
Nov 25 11:32:08 ndc-fox kernel: smb_proc_readX_data: offset is larger than SMB_READX_MAX_PAD or negative!
Nov 25 11:32:08 ndc-fox kernel: smb_proc_readX_data: -35 > 64 || -35 < 0
Nov 25 11:32:25 ndc-fox kernel: smb_add_request: request [ca259880, mid=8714] timed out!
Nov 25 11:32:37 ndc-fox kernel: smb_add_request: request [ca259280, mid=9210] timed out!
Nov 25 11:32:37 ndc-fox kernel: smb_add_request: request [ca259a80, mid=9211] timed out!
Nov 25 11:32:37 ndc-fox kernel: smb_add_request: request [ca259580, mid=9212] timed out!
Comment 1 Charlie G Mentorez 2005-11-25 05:27:32 UTC
To check if the problem could be cuased because of the the old central server have an older samba version. We have set up an new linux fileserver based on same setup as in client linux. But we get the same type of error on the client. No error on the fileserver side.

I forgot:
The Linux client time to time just freez. And must be powerd off and rebooted.
Comment 2 Charlie G Mentorez 2005-11-25 05:34:13 UTC
smb.conf

[global]
   workgroup = MXD
   server string = MXD-FOX Samba Server
   hosts allow = 192.168.117. 127.
   interfaces =  eth0 192.168.117.60
   log file = /var/log/samba/log.%m
   max log size = 50
   security = domain
   encrypt passwords = yes
   socket options = TCP_NODELAY SO_RCVBUF=8192 SO_SNDBUF=8192
   name resolve order = wins lmhosts bcast
   wins server = 192.168.117.15
   dns proxy = no
Comment 3 Charlie G Mentorez 2005-11-25 05:37:22 UTC
samba_login (mount)
#!/bin/sh
#
# /etc/rc.d/samba_login: start/stop mount/umount
#

user=mxd-xxx
passwd=xxxxxxxxxx

if [ "$1" = "start" ]; then

#works
        smbmount //MXD-LNX-FS1/ihotell  /samba/mxd-lnx-fs1/ihotell -o username=$user,password=$passwd,fmask=777,dmask=777
#works
        smbmount //MXD-LNX-FS1/appsrv   /samba/mxd-lnx-fs1/appsrv -o username=$user,password=$passwd,fmask=777,dmask=777

elif [ "$1" = "stop" ]; then
        umount /samba/mxd-lnx-fs1/ihotell
        umount /samba/mxd-lnx-fs1/appsrv
else
    echo "usage: $0 start|stop"
fi

# End of file
Comment 4 Charlie G Mentorez 2005-11-25 06:04:33 UTC
Sorry but I missunderstood the mening of Summary. I could not understand that was the subject on the problem. So beacuse of that I only write "smb"
That was the summary for me. But subject or summary as it is called here shold ofcurse be more spcific, like: A lot of SMB errors, and crash wriet files and freez server
Comment 5 Gerald (Jerry) Carter (dead mail address) 2005-11-28 05:52:52 UTC
Your bug report is about smbfs.  We don't maintain that kernel code.
The normal answer on 2.6 kernels is to try the cifs fs.
Comment 6 Charlie G Mentorez 2005-11-30 06:54:21 UTC
(In reply to comment #5)
> Your bug report is about smbfs.  We don't maintain that kernel code.
> The normal answer on 2.6 kernels is to try the cifs fs.

So you meen the new SAMBA dont support SMB file handling to NT4 any more.. That is really odd then. Becaus I have try to use cifs. But that will not work.

And the old Linux fileserver using older SAMBA version..  Hey.... What is this.. If I need to upgrade one machine with an new version do I then need to update all my 40 servers in the system. In this case the MS have a better comabillty to there older OS.

Or is there any way to get latest SMAB with CIFS work with an old SAMBA server.
I can't upgrade the SAMBA server right now becuase the whol system is depending on that.

How can this case be RESOLVED INVALID. I still have the problem. Or is this just bug for system using the same version through the whol system. There is no system that can change everthing during one night.
 


 
Comment 7 Gerald (Jerry) Carter (dead mail address) 2005-11-30 07:04:18 UTC
You filing a bug with the wrong people.  You need to file a 
bug with the kernel developers that maintain that code.
Is is marked as invalid because you filed a bug with the 
wrong people.  It would be the same if you filed a bug in 
MS Word with us.  We can only fix our bugs.  If you can 
prove that this is a Samba bug and not an smbfs issue, 
then we can reopen it.
Comment 8 Daniel Kabs 2007-01-11 10:24:21 UTC
(In reply to comment #5)
> Your bug report is about smbfs.  We don't maintain that kernel code.
> The normal answer on 2.6 kernels is to try the cifs fs.

I've seen this anwser of yours in several different bugs, e.g. bug #2096 and bug #4002 .

I was also confused about what is samba.org about, whether it would include the clients or not. The samba source comes with code for smbmnt. The web page details about smbmount and similar stuff. Thusly I guessed, samba.org would also provide the code for smbfs. Obviously I guessed wrong.

Could you please elaborate who the maintainer of cifs and smbfs is, please? Maybe a short note to this effect on in the FAQ or in "What is Samba?" on samba.org would reduce confusion.

Cheers
Daniel


Comment 9 Daniel Kabs 2007-01-11 10:46:12 UTC
And for everybody out there searching for a solution to the 

Nov 25 13:02:00 ndc-fox kernel: smb_add_request: request [c8419280, mid=13490]
timed out!

problem, follow Jerry's suggestion and use cifs instead.

I think as of kernel 2.6.19, smbfs has been declared deprecated in favor of cifs. smbfs is broken, don't use it any more.
Comment 10 Gerald (Jerry) Carter (dead mail address) 2007-01-11 10:50:44 UTC
The cifs kernel code is hosted at http://linux-cifs.samba.org/.  The smbfs kernel code has no maintainer and is up for removal soon from what I understand.