Bug 3150 - Emacs 21.3 locks up randomly when editing a text file across a Samba connection on another Linux machine
Summary: Emacs 21.3 locks up randomly when editing a text file across a Samba connecti...
Status: RESOLVED INVALID
Alias: None
Product: Samba 3.0
Classification: Unclassified
Component: File Services (show other bugs)
Version: 3.0.20a
Hardware: x86 Linux
: P3 major
Target Milestone: none
Assignee: Samba Bugzilla Account
QA Contact: Samba QA Contact
URL: http://polard.com/stefan/samba_failur...
Keywords:
Depends on:
Blocks:
 
Reported: 2005-10-07 03:57 UTC by Stefan Viljoen
Modified: 2005-10-08 01:57 UTC (History)
0 users

See Also:


Attachments
Emacs screen on lockup when attempting to save (124.74 KB, image/jpeg)
2005-10-07 03:58 UTC, Stefan Viljoen
no flags Details
This is what is reported by smbstatus on the remote Rh9 machine (181.93 KB, image/jpeg)
2005-10-07 04:00 UTC, Stefan Viljoen
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description Stefan Viljoen 2005-10-07 03:57:44 UTC
I am editing files from a Linux machine (FC3 - kernel 2.6.12) on another Linux
machine (RH9 - kernel 2.4.8) using Emacs 21.3, with Samba 3.0.20a running on
both machines. Sometimes, when saving a file, Emacs will lock up -after- saving
the file on the Samba share under discussion. See
http://polard.com/stefan/samba_failure_01.jpg for a screenshot.

Examining the file on the Rh9 remote machine reveals that the file and changes
made to it is correctly saved by Emacs - it is just as if Emacs can't release
the file again after saving to it, or is not allowed to do so by the remote Samba.

Executing smbstatus on the remote machine gives

http://polard.com/stefan/samba_failure_02.jpg

on which I have highlighted the "unknown-please report!" line.

The only way to fix the lockup is to kill Emacs locally on the FC3 box, kill and
restart Samba on the remote RH9 Linux box, and then REBOOT the local FC3 machine
- there is no way that I can discern of disconnecting from such a locked share
on the local machine before or after restarting the remote samba instance.

Can you please assist or clarify?

Here is the relevant parts of the smb.conf on the RH9 (remote) machine:

[global]

netbios name = server
workgroup = GOEDEHOOP10
security = user
log file = /var/log/samba.log
log level =  1
socket options = TCP_NODELAY IPTOS_LOWDELAY SO_RCVBUF=16384 SO_SNDBUF=16384
wins support = yes
domain logons = no
logon drive = f:
logon home = \\server\%U
os level = 99
preferred master = yes
local master = yes
hosts allow = 169.254.255.10 169.254.255.20 169.254.255.30 169.254.255.40
169.254.255.50 127.0.0.1 169.254.255.60 169.254.255.70 169.254.255.80
encrypt passwords = yes
browseable = yes
lanman auth = yes
lm announce = yes

[STEFAN_DEVEL_DOME]
path = /usr/local/apache2/htdocs/php/dome_stefan
guest ok = no
writeable = yes
create mode = 0777
directory mode = 0777
browseable = yes
public = yes
username = stefan
read list = stefan
write list = stefan
valid users = stefan

The connection script used to connect from the FC3 machine with 3.0.20a to the
RH9 machine with 3.0.20a:

smbmount "//server/stefan_devel_dome" /mnt/polarserver_dome  -o
username="stefan",password="stefan1",uid=rylan,dmask=0775,fmask=0775

I have also noticed that even if Emacs 21.3 on the local FC3 machine is NOT
currently locked up after a -failed- save over on the Rh9 machine,

http://polard.com/stefan/samba_failure_02.jpg

is -STILL- displayed by smbstatus... with the "please report" request...
Comment 1 Stefan Viljoen 2005-10-07 03:58:37 UTC
Created attachment 1492 [details]
Emacs screen on lockup when attempting to save

This is the screen displayed by Emacs when it has locked up on attempting to
save to the Rh9 machine over the share.
Comment 2 Stefan Viljoen 2005-10-07 04:00:59 UTC
Created attachment 1493 [details]
This is what is reported by smbstatus on the remote Rh9 machine

This is displayed by smbstatus no matter if an Emacs lockup has just occurred
on the FC3 machine, or if Emacs is still responsive on the FC3 machine, this is
displayed as well with the `please report` request.
Comment 3 Gerald (Jerry) Carter (dead mail address) 2005-10-07 04:33:19 UTC
we don't manage the smbfs code.  You are better 
off testing the cifs fs client code which is under active development.
Comment 4 Jeremy Allison 2005-10-07 09:09:24 UTC
The share modes database looks corrupt in that screenshot. Can you check is smbd
has crashed please ?
Jeremy.
Comment 5 Stefan Viljoen 2005-10-08 01:57:59 UTC
Hi Jeremy, 
 
No, SMBD is still running fine. I can connect and work with other shares 
besides the one that is locked up in the Emacs 21.3 save session... 
 
Thanks for the reply!