[Gutsy] high processor activity after logging out and then logging in again

[Pjotr12345]

The signal handler user by nautilus for logging is using non signal safe functions which creates issues, since the log are not really used anyway the easier option is to disable logging, which has already be done in hardy

[Sebastien Bacher (seb128)]

Thanks for your bug report. Please try to obtain a backtrace http://wiki.ubuntu.com/DebuggingProgramCrash and attach the file to the bug report. This will greatly help us in tracking down your problem. That's likely due to nautilus crashing. Could you look if nautilus-debug-log.txt keeps being updated?

[Daniel Hahler (blueyed)]

I think I've seen this bug just happening here, while trying to reproduce bug 148781 and bug 150454. Therefor I've started a new gnome session and when exiting from there I've noticed that nautilus was using all available CPU cycles.

nautilus-debug-log.txt keeps growing, but only with the same "debug log dumped" line:
===== BEGIN MILESTONES =====
0x8177888 2007/10/09 23:16:40.8978 (GLog): Can not calculate _NET_NUMBER_OF_DESKTOPS
0x8177888 2007/10/09 23:16:40.8979 (GLog): Can not calculate _NET_NUMBER_OF_DESKTOPS
0x8177888 2007/10/09 23:16:40.8979 (GLog): Can not get _NET_WORKAREA
0x8177888 2007/10/09 23:16:40.8980 (GLog): Can not determine workarea, guessing at layout
0x8177888 2007/10/09 23:23:20.6097 (USER): debug log dumped due to signal 11
0x8177888 2007/10/09 23:23:20.6246 (USER): debug log dumped due to signal 11
0x8177888 2007/10/09 23:23:20.6247 (USER): debug log dumped due to signal 11
0x8177888 2007/10/09 23:23:20.6248 (USER): debug log dumped due to signal 11
0x8177888 2007/10/09 23:23:20.6250 (USER): debug log dumped due to signal 11
[...]
0x8177888 2007/10/09 23:23:56.0296 (USER): debug log dumped due to signal 11
0x8177888 2007/10/09 23:23:56.0356 (USER): debug log dumped due to signal 11

[Sebastien Bacher (seb128)]

Do you use the desktop effects option? Does it happen without it?

[w.d (wouter-dijk)]

I experience the same problem. After log out sometimes I can't login at all, I can see the desktop for a split second and then the screen stays black (with a mouse pointer).
The only thing that remains is rebooting the machine, because ctrl-alt-backspace brings you to the login screen, but logging in is not possible...

[Sebastien Bacher (seb128)]

Could you get a backtrace of the crash?

[w.d (wouter-dijk)]

I'm sorry, I got so frustrated I reinstalled Gutsy! For now, I don't experience the problem anymore, but who knows... ;-)

[greenhunter (tierfreunde-hagenburg)]

on computer A nautilus isn't closed. during switch from user 1 to 2.

on my other computer B: gconf-2, evolution-data-server-1.12 and bonobo-activation-server isn't closed.

And as I said before it takes a long time to log out from one account on both computers.

both are distribution upgrades ubuntu 32 bit.

[Jean Levasseur (levasseur.jean)]

I'm having hte exact same behaviour. I'm using Ubuntu Gutsy, upgraded from gutsy-rc. I have installed Sabayon. I also disabled some services from the session-management menu (blue-tooth, at-visual) since I don't need them. Desktop effects are activated, but I get the same behaviour if they are disabled, so those are not the problem.

The process that causes the 100% CPU is Nautilus.

Steps to reproduce:
- Log in.
- Log out.
- Re-log in. At this time, the session loading is quite long. The panels are appearing after a while and are usable, but the desktop stays empty. Right-clicking on it does nothing.
- Trying to log out, the panels will freeze, and the "logout window" will appear after 1-2 minutes (long.....)

To work around this: in a terminal "killall -HUP nautilus && nautilus &"

[jiatai (cail9970)]

i don't need to switch user,
user A logout and login, i have the same problem

killall -HUP nautilus && nautilus & can fix it temporarelly

[Gareth Fitzworthington (mapping-gp-deactivatedaccount)]

I can confirm this on U7.10.

I also have some extra information.

Sebastien: For me Nautilus doesn't crash just uses 100% CPU (I think the crash reports above are different bugs). No record of crash or any details entered in any log for me.

This is how it occurs for me:

1/ startup PC.
2/ login
3/ Can do [Places->home] and quit OK. Can re-do numerous times OK.
4/ logout
5/ login
6/ Nautilus using 100% CPU (or near to).
7/ Can't do [Places->home] - no response.
8/ End Nautilus process.
9/ Can do [Places->home] but on quit Nautilus starts using 100% CPU again.
10/ End Nautilus process.
11/ logout
12/ login
13/ Can do [Places->home] and quit OK. Can re-do numerous times OK.
14/ logout
15/ login
16/ Nautilus using 100% CPU (or near to).
17/ Can't do [Places->home] - no response.
18/ End Nautilus process.
19/ Can do [Places->home] but on quit Nautilus starts using 100% CPU again.
20/ End Nautilus process.
21/ logout
..... problem repeats indefinitely on an every second login basis.

You can see from the above that Nautilus runs away with CPU time only on every second login (ie login 2,4,6...).
Login 1,3,5... all is OK.

Perhaps someone else can confirm this cycle.
Also, when Nautilus is using 100% CPU, logout doesn't work (or perhaps it would work if enough time were given).

[Gareth Fitzworthington (mapping-gp-deactivatedaccount)]

Sebastien: Sorry there is something logged to nautilus-debug-log.txt
This file fills rapidly when nautilus is 100% CPU hungry.
See attached.

[Nick B. (futurepilot)]

I'm seeing the exact same errors. Nautilus takes up 100% of the CPU and dumps a debug log file in my home folder that goes on and on (13MB!)
===== BEGIN MILESTONES =====
0x8177510 2007/10/21 14:17:07.9403 (GLog): Can not calculate _NET_NUMBER_OF_DESKTOPS
0x8177510 2007/10/21 14:17:07.9404 (GLog): Can not calculate _NET_NUMBER_OF_DESKTOPS
0x8177510 2007/10/21 14:17:07.9405 (GLog): Can not get _NET_WORKAREA
0x8177510 2007/10/21 14:17:07.9405 (GLog): Can not determine workarea, guessing at layout
0x8177510 2007/10/21 21:15:14.9326 (USER): debug log dumped due to signal 11
0x8177510 2007/10/21 21:15:14.9330 (USER): debug log dumped due to signal 11
0x8177510 2007/10/21 21:15:14.9332 (USER): debug log dumped due to signal 11
0x8177510 2007/10/21 21:15:14.9334 (USER): debug log dumped due to signal 11
0x8177510 2007/10/21 21:15:14.9337 (USER): debug log dumped due to signal 11
0x8177510 2007/10/21 21:15:14.9340 (USER): debug log dumped due to signal 11.....

If I kill Nautilus then trackerd starts going crazy and I have to kill trackerd as well.
I have seen this on 2 separate computers one with desktop effects one without.
I haven't been able to find a way to reproduce this though. Seems totally random. I'll try and get a backtrace if I can.

[Nick B. (futurepilot)]

My laptop which is not running Compiz has the same error minus the first 4 lines. So it's just a bunch of the signal 11 errors.

[Derek Chen-Becker (dchenbecker)]

Same as Nick B. This is on a laptop that worked fine with Feisty and then did an upgrade manager upgrade to Gutsy

[AdrianM (adrian-moore)]

Same as above Nautilus takes up 100% of the CPU and dumps a debug log file with new entries about 1000/min which trackerd then retries to index. Need to delete debug log and kill nautilus to get back. My setup was ok on Feisty and only happened after upgrade to gutsy. I am not running compiz.

[Eetu Huisman (eh)]

I have this same issue on my desktop machine which doesn't have compiz enabled.

[pisica (andrei-iftimovici)]

I have experienced (computer HP intel dual core 2 and Ubuntu 7.10 installed on a virgin space i.e. no upgrade) almost all the other contributors said about that bug; it seems that the bug hasn't been solved yet, so, I'll bring some precisions about its behavior:

1) logging into another account or re-logging into the same, is not relevant: the bug is still present.
2) I think there is no connection with compiz (since even when the desktop is in very bad shape, compiz works, the cube rotates, so on...), nor with beagled nor with trackerd: I have disabeled all of them at startup and the bug is still there.
3) I tried to make a script shell (called clear.sh) like this:
Code:

#!/bin/bash
sleep 3
killall -9 nautilus 1>/dev/null
nautilus&
exit

and added it to the System > Preferences > Sessions, to be taken into account when the new session starts (hoping the old nautilus process will be killed). Unfortunately, it has no effect, whatever I put 3 or 5 or 6 or etc... seconds of sleep. But it has an effect when I run it after I login in the BAD working nautilus session. (Of course it has!)

4) Still, the effect of killing and restarting nautilus is NOT complete, i.e. it doesn't kill the bug completely. Indeed,

a) It kills the bad behaviour of the desktop, so the icons are coming back on it, the panel buttons are working well and the processors are stoping their mad activity.

b) But it doesn't bring back my Disk Mounter applet from one of the panels. Nor allows me to put other such applets on panels. This is quite odd. Then, if I restart the computer and login for the first time, my disk mounter applet appears normally and the other applets I have tried to put on the panel are appearing too !

********************

I think that this is a severe bug (not a "medium" one, as it has been classified) especially for computers used by more than one person (e.g. in a university). Thank you for fixing it as soon as possible.

[James N (jnylen)]

@pisica:

There is a better workaround for this bug. Assuming you use gdm for login, open the file /etc/gdm/PostSession/Default and add the 'killall nautilus' or 'sudo killall nautilus' (I used 'pkill nautilus') before the exit statement in that script. That way nautilus will be killed as your session is exiting, and the cpu will not be eaten up when you are just sitting there at the login screen. Also, switch user / login as someone else / re-login works normally for me now.

[pisica (andrei-iftimovici)]

Right! It works now for me too, and even my Disk Mounter applet appers normally on the panel.
Thanks a lot James N !

[Sottovoce: By the way, even if one can manage it now, it still remains a bug for nautilus, right?]

[James N (jnylen)]

Yes, it is still a bug. The login/logout is not the only case that is affected by this problem - when I am logged in as myself and I run a nautilus window as root, it does not exit normally. I have to kill the process manually after closing the nautilus window.

I don't know anything about the way Nautilus is coded, but the 2 cases where this problem occurs suggest to me that nautilus expects some service to be running as it is exiting, and if that service isn't running - the process hangs and eats up all the CPU.

[doclist (dclist)]

I also experience 100% CPU usage from Nautilus after logging in userA, logging in userB, logging out userB. The offending process is owned by userB. This is new (for me) in Gutsy.

I can't seem to get a proper backtrace. This is what I get immediately after attaching through GDB.

=========
#0 0xb727d45f in fputs () from /lib/tls/i686/cmov/libc.so.6
#1 0x080e27c8 in write_string (filename=0x0, file=0x8905b48,
    str=0x1 <Address 0x1 out of bounds>, error=0x0) at nautilus-debug-log.c:446
#2 0x080e28ab in nautilus_debug_log_dump (
    filename=0x89058f0 "/home/USER/nautilus-debug-log.txt", error=0x0)
    at nautilus-debug-log.c:508
#3 0x0807e776 in dump_debug_log () at nautilus-main.c:213
#4 0x0807e7bf in sigfatal_handler (sig=11) at nautilus-main.c:258
#5 <signal handler called>
#6 0xb7657af6 in gconf_client_remove_dir () from /usr/lib/libgconf-2.so.4
#7 0xb5c2e9cd in ?? ()
   from /usr/lib/nautilus/extensions-1.0/libnautilus-actions.so
#8 0xe5000009 in ?? ()
#9 0xb5c32d74 in ?? ()
   from /usr/lib/nautilus/extensions-1.0/libnautilus-actions.so
#10 0x00000000 in ?? ()
=========

Trying to 'continue' in GDB gives the following:
====
Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread -1228183888 (LWP 22236)]
0xb7657af6 in gconf_client_remove_dir () from /usr/lib/libgconf-2.so.4
====

[Kamil Páral (kamil.paral)]

I experience the same problem quite often. People have attached logs, setting as confirmed.

[wexper (wner-linyuan)]

I experience the same problem in a worse case - when I lockscreen, and login back.
Above workarounds really worked, but all the windows and documents left open before lockscreen get lost. It's really bad.

[Nick B. (futurepilot)]

This happened again and I got a backtrace on it. However I have a feeling this won't be any good. It looks like something with pthread but I'm not sure what dbgsym I need for that. Anyone know? I will try again.

[Luiz Alberto Medaets (megahertz)]

Disable Tracker on System – Preferences – Sessions and the problem will
end.
Tracker is a desktop search tool, it has a bug, and it is launched every
time you make a search on file browser (Nautilus).
Luiz

On Wed, 2007-10-31 at 00:47 +0000, doclist wrote:
> I also experience 100% CPU usage from Nautilus after logging in userA,
> logging in userB, logging out userB. The offending process is owned by
> userB. This is new (for me) in Gutsy.
>
> I can't seem to get a proper backtrace. This is what I get immediately
> after attaching through GDB.
>
> =========
> #0 0xb727d45f in fputs () from /lib/tls/i686/cmov/libc.so.6
> #1 0x080e27c8 in write_string (filename=0x0, file=0x8905b48,
> str=0x1 <Address 0x1 out of bounds>, error=0x0) at nautilus-debug-log.c:446
> #2 0x080e28ab in nautilus_debug_log_dump (
> filename=0x89058f0 "/home/USER/nautilus-debug-log.txt", error=0x0)
> at nautilus-debug-log.c:508
> #3 0x0807e776 in dump_debug_log () at nautilus-main.c:213
> #4 0x0807e7bf in sigfatal_handler (sig=11) at nautilus-main.c:258
> #5 <signal handler called>
> #6 0xb7657af6 in gconf_client_remove_dir () from /usr/lib/libgconf-2.so.4
> #7 0xb5c2e9cd in ?? ()
> from /usr/lib/nautilus/extensions-1.0/libnautilus-actions.so
> #8 0xe5000009 in ?? ()
> #9 0xb5c32d74 in ?? ()
> from /usr/lib/nautilus/extensions-1.0/libnautilus-actions.so
> #10 0x00000000 in ?? ()
> =========
>
>
> Trying to 'continue' in GDB gives the following:
> ====
> Program received signal SIGSEGV, Segmentation fault.
> [Switching to Thread -1228183888 (LWP 22236)]
> 0xb7657af6 in gconf_client_remove_dir () from /usr/lib/libgconf-2.so.4
> ====
>
--
Megahertz
<email address hidden>

[Luiz Alberto Medaets (megahertz)]

Disable Tracker on System – Preferences – Sessions and the problem will
end.
Tracker is a desktop search tool, it has a bug, and it is launched every
time you make a search on file browser (Nautilus).
Luiz

On Sun, 2007-11-04 at 21:50 +0000, Kamil Páral wrote:
> I experience the same problem quite often. People have attached logs,
> setting as confirmed.
>
> ** Changed in: nautilus (Ubuntu)
> Status: Incomplete => Confirmed
>
--
Megahertz
<email address hidden>

[Kamil Páral (kamil.paral)]

Well, I don't know if disabling Tracker will help, but I know for sure that Nautilus takes 100% cpu. "killall nautilus" works good, apart from not getting the desktop screen and icons then. I would see this as another bug than Tracker-related.

[doclist (dclist)]

On Nov 6, 2007 6:06 AM, Luiz Alberto Medaets <email address hidden> wrote:
> Disable Tracker on System – Preferences – Sessions and the problem will
> end.
> Tracker is a desktop search tool, it has a bug, and it is launched every
> time you make a search on file browser (Nautilus).
> Luiz
>

I don't have tracker installed at all (and haven't since the first day
of installing Gutsy) and still experience the problem.

[Luiz Alberto Medaets (megahertz)]

Tracker is a desktop search tool, part of Gnome, and is loaded when you
start (Disable Tracker on System – Preferences – Sessions and the
problem will end.) or when you ask for file search.
Open System - Administration -System Monitor. Look for Trackerd on
process tab to see if it's running . If it is, right click and stop then
right click and kill.

Megahertz

07 at 00:11 +0000, doclist wrote:
> On Nov 6, 2007 6:06 AM, Luiz Alberto Medaets <email address hidden> wrote:
> > Disable Tracker on System – Preferences – Sessions and the problem will
> > end.
> > Tracker is a desktop search tool, it has a bug, and it is launched every
> > time you make a search on file browser (Nautilus).
> > Luiz
> >
>
> I don't have tracker installed at all (and haven't since the first day
> of installing Gutsy) and still experience the problem.
>
--
Megahertz
<email address hidden>

[doclist (dclist)]

On Nov 7, 2007 8:12 AM, Luiz Alberto Medaets <email address hidden> wrote:
> Tracker is a desktop search tool, part of Gnome, and is loaded when you
> start (Disable Tracker on System – Preferences – Sessions and the
> problem will end.) or when you ask for file search.
> Open System - Administration -System Monitor. Look for Trackerd on
> process tab to see if it's running . If it is, right click and stop then
> right click and kill.
>
> Megahertz

I know what Tracker is. I could have been clearer: I uninstalled
Tracker soon after upgrading to Gutsy.

[pisica (andrei-iftimovici)]

Folks, let me remind you that there is a temporary but helpful solution to the problem which has been given in this forum by James N. I quote him below:

********************
Assuming you use gdm for login, open the file /etc/gdm/PostSession/Default and add the 'killall nautilus' or 'sudo killall nautilus' (I used 'pkill nautilus') before the exit statement in that script. That way nautilus will be killed as your session is exiting, and the cpu will not be eaten up when you are just sitting there at the login screen. Also, switch user / login as someone else / re-login works normally for me now.

********************

By the way, I don't think tracker has something to do with this bug, although I saw that some problems arrived when both trackerd and beagled were started.

[Gareth Fitzworthington (mapping-gp-deactivatedaccount)]

Both of these work-arounds work for me ('Trackerd' & 'kill nautilus').

The Trackerd fix by Luiz above is particularly interesting because when Trackered is not running nautilus appears to be able to pull itself out of whatever spiral it's in. A "nautilus-debug-log.txt" is still generated by nautilus but now it rarely spans more than 1 second. I've attached an example log from one of my users. The generated log files sometimes also include extra information.

[debianmigrant (debianmigrant)]

I suffer with the same issue, and I have applied the above workaround (kill nautilus), it seems to stop the high usage of the CPU by nautilus, but it has broken the screensaver when switching users. I get a blank (white) screen, and if I type in my password, I get my desktop back. Only way I have found to fix this, is by disabling desktop effects.

[Michael Trunner (trunneml)]

Hi,

don't think uninstalling tracker is the solution. We removed tracker from our workstations on the university and even added the killall nautilus to PostSession. But the problem still exist. Okay it is better now but the problem here isn't the warm cpu, it is the traffic on the nfs-server, and the full home directory (user can't login when his home is full). Currently we can't really upgrade to gutsy, because of this bug. So i think the Importance should be higher.

[James N (jnylen)]

Even though I don't think this bug qualifies as high- or severe-importance, it is a problem with a major OS component, and the PostSession workaround seems to have undesirable side effects for some people. Just imagine if Microsoft left a bug in their Windows Explorer that caused the CPU to run around in circles non-stop.

...alright, maybe that isn't such a good analogy. But this problem really needs to be fixed.

[Sebastien Bacher (seb128)]

then somebody having the issue needs to describe how to trigger it on a new installation or to debug it, usually it's not going to work on an issue when you don't have it on your configuration

[greenhunter (tierfreunde-hagenburg)]

On my Computer following task are not closed after exiting one user account.

nautilus with eating 80% CPU
gconfd-2 sleeping
mapping-daemin sleeping
bonobo-activation-server sleeping

killing nautilus closes the 3 others.

[sja821 (sja821)]

I have a clean install of Gutsy that is having this problem. I do have thin clients running from this server, so I'm getting a lot of logins - and should be able to help debug if someone can tell me what to look for. I'm getting the same contents as in the nautilus-debug-log.txt file shown above.

[Jean-Luc Thirion (jlinho)]

Same observations as greenhunter for me.

I removed the Deskbar applet and everything is fine now.

Please tell me what happens if you remove the Deskbar applet ?

[greenhunter (tierfreunde-hagenburg)]

How do i remove the deskbar applet?

[sja821 (sja821)]

I have taken tracker out of the sessions preferences, but have not yet
removed the package. What is the name of the package to make sure I
remove the proper one? (It shouldn't be in use anyway, but it is worth
a shot)

On 11/19/07, greenhunter <email address hidden> wrote:
> How do i remove the deskbar applet?
>
> --
> [Gutsy] high processor activity after logging out and then logging in again
> https://bugs.launchpad.net/bugs/150471
> You received this bug notification because you are a direct subscriber
> of the bug.
>

[greenhunter (tierfreunde-hagenburg)]

ups,

I don't use the deskbar aplpet at all under no accounts and no computers.

[Jean-Luc Thirion (jlinho)]

Sorry greehunter, but it solved the problem clearly for me. If I re-add
Deskbar to my Gnome Panel the problem appears again.

2007/11/19, greenhunter <email address hidden>:
>
> ups,
>
> I don't use the deskbar aplpet at all under no accounts and no
> computers.
>
> --
> [Gutsy] high processor activity after logging out and then logging in
> again
> https://bugs.launchpad.net/bugs/150471
> You received this bug notification because you are a direct subscriber
> of the bug.
>

[Pjotr12345 (computertip)]

Even if you don't use Deskbar Applet, it is still active als long as it resides in your panel. Therefore it can still be the culprit of this nasty bug.

I haven't had this bug since I severely limited the search functions of Deskbar Applet (in it's preferences): it now only searches files on my hard disk.

[Jean Levasseur (levasseur.jean)]

Well I've unstalled Deskbar (sudo apt-get remove *deskbar*) from my computer and the problem is still there, so I guess the problem is not with Deskbar but maybe with one of its plugin...

[Pjotr12345 (computertip)]

Possibly the bad guy is Tracker. Tracker is a plugin in Deskbar applet. An hour ago, I've completely removed Tracker from my system with Synaptic, and I'm curious if the problem is over now.

Maybe you could try this, too, and report here if it has been a success?

[doclist (dclist)]

On Thu, 2007-11-22 at 10:28 +0000, Pjotr12345 wrote:
> Possibly the bad guy is Tracker. Tracker is a plugin in Deskbar applet.
> An hour ago, I've completely removed Tracker from my system with
> Synaptic, and I'm curious if the problem is over now.
>
> Maybe you could try this, too, and report here if it has been a success?

I do not have Tracker installed and still experience the problem.

[Pjotr12345 (computertip)]

@doclist: I would like some clarification from you, because this is an important matter. Tracker is installed in the standard installation, so everybody has tracker automatically. So if I understand you correctly, you already removed tracker before, and still have the problem?

[Marcel Schaal (marcelschaal)]

Hi Pjotr,

I explicitly removed tracker, but nautilus still stresses the processor.

[doclist (dclist)]

On Thu, 2007-11-22 at 21:25 +0000, Pjotr12345 wrote:
> @doclist: I would like some clarification from you, because this is an
> important matter. Tracker is installed in the standard installation, so
> everybody has tracker automatically. So if I understand you correctly,
> you already removed tracker before, and still have the problem?
>

That's right.

[Pjotr12345 (computertip)]

Tracker is definitely *not* the bad guy here. Nearly a week ago, I completely removed Tracker, using Synaptic, and also Deskbar Applet. Yet I occasionally still have the problem...

This is really a bad bug, in urgent need of repair.

[Jean-Luc Thirion (jlinho)]

You say: "I occasionally still have the problem..."
Did you experience this problem each time before removing Deskbar and
Tracker ?

2007/11/27, Pjotr12345 <email address hidden>:
>
> Tracker is definitely *not* the bad guy here. Nearly a week ago, I
> completely removed Tracker, using Synaptic, and also Deskbar Applet. Yet
> I occasionally still have the problem...
>
> This is really a bad bug, in urgent need of repair.
>
> --
> [Gutsy] high processor activity after logging out and then logging in
> again
> https://bugs.launchpad.net/bugs/150471
> You received this bug notification because you are a direct subscriber
> of the bug.
>

[Pjotr12345 (computertip)]

No, before the removal of Tracker and Desbar applet it was also occasionally. So nothing has changed.

This is a bug that probably doesn't affect people who have only one user account on their PC's. But for people with two or more user accounts on the same machine, this is really very bad.

[Tres Finocchiaro (tres-finocchiaro)]

FYI - The following helped me in my efforts:

nvidia-settings >> Uncheck Sync to VBlank
sudo killall compiz.real

If compiz.real has been restarted, logins seem to work fine.

Dell Precision 450 Dual CPU, nVidia, Desktop Effects, Ubuntu 7.10.

Worth mentioning, this occurred after installing the AWN desktop manager. It may have existed before, but I don't think so. I compiled it from source, and had a TON of dependencies. :beer:

I will add the line to "/etc/gdm/PostSession/Default" as described above.

-Tres

[Tres Finocchiaro (tres-finocchiaro)]

Comfirmed. Adding "killall compiz.real" to "/etc/gdm/PostSession/Default" before "exit" fixes this issue on my computer. Thank you for the hints everyone.

-Tres

[Kjetil Thuen (kjetil-thuen)]

Killing compiz does not help for me. Killing nautilus does.

[tomten (fredrik-corneliusson)]

I can confirm this issue (nautilus uses 100% CPU on one core) on my system. It's a clean installation of Gutsy dual booted with Win XP on a separate partition.

System:
MB: Asus A8N Deluxe with NV chipset.
AMD 64 X2 4200
2Gigs
ATI Radeon 1600
Running the Vesa X driver, so no compiz or 3D running at all.

Sometimes I also get the error that Gnome theme manager failed to start and I have to re-login to get the correct theme. But I don't know if it's related.

Did not have any of these this issues in Edgy.

[Tres Finocchiaro (tres-finocchiaro)]

I can confirm that Nautilus has a 100% cpu bug. I noticed it after creating a shortcut "gksu nautilus /usr/share/icons". It hasn't happened since, but it was the culprit of some strange Firefox behavior on my end. Perhaps a good general fix is to kill both! :)

Edit:
sudo gedit /etc/gdm/PostSession/Default

Before "exit", put:
killall compiz.real
killall nautilus

Its like crashing out Explorer in windows, right!? J/k. Again, great feedback here!

-Tres

[Pjotr12345 (computertip)]

FatButtLarry's solution appears to work for me: I only need to add "killal nautilus" to that file, and no more problems. Thanks, Larry!

In short:

In the terminal:
sudo gedit /etc/gdm/PostSession/Default

Now add "killall nautilus" just above "exit 0". The file reads then as follows:

#!/bin/sh

PATH="/usr/bin:$PATH:/bin:/usr/bin"
OLD_IFS=$IFS

gdmwhich () {
  COMMAND="$1"
  OUTPUT=
  IFS=:
  for dir in $PATH
  do
    if test -x "$dir/$COMMAND" ; then
      if test "x$OUTPUT" = "x" ; then
        OUTPUT="$dir/$COMMAND"
      fi
    fi
  done
  IFS=$OLD_IFS
  echo "$OUTPUT"
}
killall nautilus
exit 0

[nicoladimaria (info-nicoladimaria)]

fresh install of ubuntu 7.10, no compiz nor tracker nor deskbar applet.
same problem

[Sebastien Bacher (seb128)]

nautilus (1:2.20.0-0ubuntu8) hardy; urgency=low

  * debian/patches/18_disable_signal_handler.patch:
    - Don't use the logging code signal handler it's buggy and makes nautilus
      being stuck and eat cpu on crash (LP: #150471)

 -- Sebastien Bacher <email address hidden> Thu, 29 Nov 2007 18:34:22 +0100

[Kamil Páral (kamil.paral)]

What about Gutsy? It happens quite often and maybe it is not security hole, but it's quite annoying and troublesome. Why not to put it into recommended updates repository?

[James N (jnylen)]

@Kamil:

I think the /etc/gdm/PostSession/Default fix is acceptable for Gutsy (the idea is to kill any remaining nautilus processes at logout). The only issue I still have is that I can't run nautilus normally as a root user, but I just do that in a terminal now and press Ctrl+C when I'm done.

[wvengen (wvengen)]

I had the same problem:
* Start nautilus
* Close nautilus in gui
* Find nautilus still running, dumping the following line to ~/nautilus-debug-log.txt ad infinitum:
    0x8177510 <date> <time> (USER): debug log dumped due to signal 11

I applied Sebastien's patch on Gutsy and can confirm that my problem is solved, thanks!
I too would suggest that this is backported to Gutsy when no problems occur.

[Sebastien Bacher (seb128)]

Could you not use the bug tracker to suggest doing ugly workarounds to gdm and take the user discussion somewhere else? The bug tracker is about solving correctly technical issues

[Jordan Erickson (lns)]

I can also confirm this on 2 seperate Ubuntu Gutsy (64-bit) installs as LTSP servers.

Since these are LTSP servers (and serve many clients), GDM is not a factor in my equation (LTSP uses LDM instead, and the 100% CPU Nautilus processes are from people logged into thin clients).

Help would be greately appreciated as I have to manually go in and kill rampant Nautilus processes daily after complaints of slowness on the whole LTSP network.

[Jordan Erickson (lns)]

I should also add that trackerd is *not* running on either of my servers, as a troubleshooting step from earlier posts. Compiz/Desktop effects are also not in the equation.

[Jordan Erickson (lns)]

nautilus (1:2.20.0-0ubuntu8) hardy; urgency=low

  * debian/patches/18_disable_signal_handler.patch:
    - Don't use the logging code signal handler it's buggy and makes nautilus
      being stuck and eat cpu on crash (LP: #150471)

---

Is there any way this will be ported to Gutsy? I have a hard time believing this is a "low" urgency, as I have the same result in multiple live X/LTSP servers, which slows the powerful machines to a crawl when nautilus misbehaves (daily)...

[Sebastien Bacher (seb128)]

The change will be considered for gutsy once it has get testing on hardy

[Rhomboid (rhomboid)]

This doesn't appear to be fixed. I just installed 7.10 today. I had a working desktop for about 10 minutes and then installed the available updates. Now, after rebooting, when I log in I get a blank desktop for about 5 seconds and then the screen goes completely black except for the mouse (cursor still moves) and I never get a desktop. There was a lot of disk activity so I installed iostat and the box is writing to nautilus-debug-log.txt at about 8M/sec. While I never get an interactive desktop, the machine is usable from the command line on an alternate terminal.

The box is just a P4 3GHz with an Nvidia 6800GT and 2GB RAM.

===== BEGIN MILESTONES =====
0x8177510 2007/12/20 18:50:17.8199 (USER): debug log dumped due to signal 11
0x8177510 2007/12/20 18:50:17.8280 (USER): debug log dumped due to signal 11
0x8177510 2007/12/20 18:50:17.8389 (USER): debug log dumped due to signal 11
0x8177510 2007/12/20 18:50:17.8529 (USER): debug log dumped due to signal 11
0x8177510 2007/12/20 18:50:17.8561 (USER): debug log dumped due to signal 11
0x8177510 2007/12/20 18:50:17.8600 (USER): debug log dumped due to signal 11
0x8177510 2007/12/20 18:50:17.8640 (USER): debug log dumped due to signal 11
0x8177510 2007/12/20 18:50:17.8680 (USER): debug log dumped due to signal 11
0x8177510 2007/12/20 18:50:17.8721 (USER): debug log dumped due to signal 11
0x8177510 2007/12/20 18:50:17.8887 (USER): debug log dumped due to signal 11
0x8177510 2007/12/20 18:50:17.8971 (USER): debug log dumped due to signal 11
0x8177510 2007/12/20 18:50:17.9054 (USER): debug log dumped due to signal 11
0x8177510 2007/12/20 18:50:17.9137 (USER): debug log dumped due to signal 11
0x8177510 2007/12/20 18:50:17.9221 (USER): debug log dumped due to signal 11
0x8177510 2007/12/20 18:50:17.9304 (USER): debug log dumped due to signal 11
0x8177510 2007/12/20 18:50:17.9387 (USER): debug log dumped due to signal 11
0x8177510 2007/12/20 18:50:17.9471 (USER): debug log dumped due to signal 11
0x8177510 2007/12/20 18:50:17.9554 (USER): debug log dumped due to signal 11
0x8177510 2007/12/20 18:50:17.9637 (USER): debug log dumped due to signal 11
0x8177510 2007/12/20 18:50:17.9720 (USER): debug log dumped due to signal 11
0x8177510 2007/12/20 18:50:17.9804 (USER): debug log dumped due to signal 11
0x8177510 2007/12/20 18:50:17.9887 (USER): debug log dumped due to signal 11
0x8177510 2007/12/20 18:50:17.9970 (USER): debug log dumped due to signal 11

[Kamil Páral (kamil.paral)]

It is fixed in 8.04, not in 7.10. Since this bug is reported against 7.10, I think it should *not* be marked as "Fix released".

[greenhunter (tierfreunde-hagenburg)]

this bug don't let me update to gutsy about 20 Computers, I care for.

[schiemanski (schiemanski)]

I've got exact the same problem.I have only one user account on my laptop and I have removed Tracker two months ago and the problem still exist. But not every day so it is difficult to find out where it come from.

[Richard Ayotte (rich-ayotte)]

WARNING: There are side effects to the following workaround so use with caution.

Here's another method to clean up a users session after they log out.

Make sure you have the slay program installed and in /etc/gdm/PostSession/Default add before exit 0:

/usr/sbin/slay -clean $USER

if you are certain that nautilus is the only app hanging, instead use:

/usr/bin/killall nautilus -user $USER

[Psykotik (linux-ikiru)]

Any news regarding this bug, affecting so many among us? Any patch for gutsy users?

[Jean-Luc Thirion (jlinho)]

I had the 100% processor problem today and it reproduced 3 times today.

I did nothinh special, the only 2 used applications in this session were: nautilus and totem.
I just clicked much around the music files in nautilus and played a lot of music. 10 minutes later, the CPU becomes crazy and "top" indicates nautilus is eating 100% of my CPU.

This bug seems now more important to me.

[Jordan Erickson (lns)]

I am still having this (horrible) issue as well, at multiple LTSP sites (which effectively slows the entire lab down). I would greatly appreciate a backport to Gutsy. It has effectively turned one of my customers away from Linux all together - and they're strongly considering moving their computer lab back to Windows.

[Sebastien Bacher (seb128)]

Opening a gutsy task. Does anybody has an easy way to trigger the issue and could try if the patch works correctly on gutsy?

[Jordan Erickson (lns)]

I would think, opening up a Gnome session, navigating in Nautilus, doing
other misc. tasks, and then logging out would cause the bug to trigger -
for me, it's not EVERY time, but definitely more times than not.

Sebastien Bacher wrote:
> Opening a gutsy task. Does anybody has an easy way to trigger the issue
> and could try if the patch works correctly on gutsy?
>
> ** Changed in: nautilus (Ubuntu Gutsy)
> Importance: Undecided => High
> Status: New => Confirmed
>
> ** Changed in: nautilus (Ubuntu Gutsy)
> Assignee: (unassigned) => Ubuntu Desktop Bugs (desktop-bugs)
>

--
Jordan Erickson
Owner, Logical Networking Solutions
http://www.logicalnetworking.net
707-636-5678

Latest LNS Blogs - http://blogs.logicalnetworking.net

 RFID Passports coming in 2008
 Peru orders 260,000 OLPC laptops for kids, Mexico 50,000
 CNet to Microsoft: Ditch Vista

[Rhomboid (rhomboid)]

In my case the repro was simply:

1) Install Ubuntu 7.10 from CD.
2) Log in to desktop GUI and apply available updates from notification icon.
3) Reboot and never get an interactive desktop upon GDM login while nautilus log thrashes (text console logins work fine).

My is/was slightly different because it was triggered simply by logging in, not logging out and back in or by switching users. However, the symptoms were the same.

I installed Debian Etch because I need a stable desktop to get work done. I was genuinely surprised this wasn't getting attention in a supported *release* version, thanks.

[Tres Finocchiaro (tres-finocchiaro)]

Worth noting, 7.10 at work (Dual Processor Xeon) does this, but 7.10 at home (AMD64) I haven't seen it yet.

Might be because I haven't run updates at home (can't remember).

-Tres

[Jordan Erickson (lns)]

Interesting... The sites that I'm experiencing this on are dual CPUs
(Dual Core-2 Xeon 1.6GHz)...One is AMD64 build of Gutsy, the other i386
build. I haven't experienced it on any UPGRADED machines, only new Gutsy
(Desktop) installs (with server kernels).

FatButtLarry wrote:
> Worth noting, 7.10 at work (Dual Processor Xeon) does this, but 7.10 at
> home (AMD64) I haven't seen it yet.
>
> Might be because I haven't run updates at home (can't remember).
>
> -Tres
>

--
Jordan Erickson
Owner, Logical Networking Solutions
http://www.logicalnetworking.net
707-636-5678

Latest LNS Blogs - http://blogs.logicalnetworking.net

 RFID Passports coming in 2008
 Peru orders 260,000 OLPC laptops for kids, Mexico 50,000
 CNet to Microsoft: Ditch Vista

[Nick B. (futurepilot)]

The easiest way I've been able to reproduce it is login to one account, log out, login to another account, logout and keep repeating the process. Eventually it will just happen.

[Jean-Luc Thirion (jlinho)]

For me the easiest way to reproduce it is:
- The "User Switcher Applet" should be added to the panel if you have
removed it. The bug occurs also on a fresh install.
- Log in in one account
- Work a little bit with nautilus
- Log out
- Log in on an another account
- Nautilus is now 100% CPU

If i remove the "UserSwitcher Applet", the bug occurs rarely on my account.
But my brother manages to repeat the bug each time he works on his account,
playing with nautilus and totem.

I use a 32 bits Ubuntu; with an Intel Core 2 Quad Q6600.

How can I apply the patch manually ?

Thanks for help*
*

[Gligor Horia (gligorhoria)]

i got this bug on kiwilinux (ubuntu 7.10), after installing apache2 and php, i could not figure out wich one did it but i think this is an older bug that needs to be fixed! PS: after i closed nautilus, another application went 100% cpu but i forgat wich one, i closed it to fast, is there any way i can view logs on my system monitor so i cand trace that out wile it is still fresh? Oh and another thing, if what i say is right, shouldn't this thread be posted as ubuntu 7.10 bugs?

ty, i just hope i-m not wasting your time, i know that's verry important for all of us!

[Tres Finocchiaro (tres-finocchiaro)]

If you killed it from command line, try "history".

-Tres

[Gligor Horia (gligorhoria)]

nope, i got to the system monitor first...

hope i don't uppset any 1 for duble posting on this, i red the instructions and i don-t think any atachments will make any deal on this subject, after reinstalling apache2 and php5 (aka apt-get remove, apt-get install (no dependencies erased) ) it seams to be working fine...hmm...

[Tres Finocchiaro (tres-finocchiaro)]

No worries. I don't think my affected system has php installed. Cheers.

[manzano (manzano-vocamen)]

Hi guys,
Since Nautilus is linked with MANY things, there are many reasons for Nautilus to crash.
I spent hours to track this one and removed a lot of stuff before pushing the right triger for me.
Let me ask a question to everyone:
Do you use to have the 'list view' with 'smallest icon size possible' by default??
If yes, try to increase it to 'small' and give it a try.
When I'll have time, I'll write a full article about it on my unbuntu blog, at blog.vocamen.com, let me know there if it worked for you.

[manzano (manzano-vocamen)]

oops, I forgot to add:
remove all content of subfolders in .thumbnails (in normal, large and fail folders)

[manzano (manzano-vocamen)]

"- Log in in one account
- Work a little bit with nautilus
- Log out
- Log in on an another account
- Nautilus is now 100% CPU"

Good chances are, in this case, it's tracker

[manzano (manzano-vocamen)]

Sorry, second thought of my last comment, it could be many thing in Nautilus too.

[Tres Finocchiaro (tres-finocchiaro)]

If you mean, do we show detailed view, yeah, I usually do by default. I haven't loaded the station back up at work in a while. killall nautilus works for now lol...

-Tres

[Alen (cshadow)]

It's not so complicated for me: all I have to do is open nautilus on the secondary screen (LCD TV) and close it - it remains among the other processes with 100% CPU usage. Usually there are two users logged on on the machine and no one can logout without nautilus to go wild :-)
Removed tracker a long time ago since it would go crazy when nautilus was killed...
Should I remove user switcher? I t tends to crash xserver from time to time...

[manzano (manzano-vocamen)]

I had this problem for 2-3 days, everytime I started Nautilus (within same session, login-logout, etc).
Based on my logs (different error messages), I found other bugs related to nautilus and tried many different things.
It was fixed when I changed in a Nautilus window (after waiting 5-10 minutes for the hang to stop):
Edit>Preferences>Views>List View Default>Default zoom level>75%
(or gconf-editor
apps>nautilus>list_view>default_zoom_level>small if Nautilus cannot never starts)
and removed 8000+ thumbnails created in .thumbnails subfolders (but I didn't removed the folders) to get back to my prefered list view at 25% (smallest).

I understand that killing Nautilus is a quick fix and it worked for me too (at least when I didn't need Nautilus), but I wanted to go further to find why this bug. I found that a faulty thumbnail can lead to a non working Nautilus. I couln't check each of 8000 created thumbnails and since Nautilus re-builds thumbnails when it needs it ...
In addition if you kill Nautilus at gdm session's end, you shouldn't be able to save your session state for it. You might loose this function.

[manzano (manzano-vocamen)]

As promised I wrote a full article about tracking a bug in Ubuntu with this bug as example of investigation.
see my first post above for the web address.

[Manuel Iglesias Alonso (glesialo)]

I think nautilus has done its thing in front of my very nose!

First I want to confirm I have noticed all the symptoms mentioned here:
-Occassionally nautilus grabs almost 100% CPU.
-When I kill it, trackerd in turn grabs CPU.
-There are more chances nautilus goes berserk if it is run as root (gksu).
-Nautilus writes enormous nautilus-debug-log.txt files with the useless message:
  0x8177510 2008/01/14 11:18:03.2623 (USER): debug log dumped due to signal 11
Note: My system also has dual processor (Pentium IV hyperthreading) and, when nautilus grabs the CPU, it alternates using both CPUs.

Let's go to the crime's scene:
-I had two nautilus' windows open one as normal user (manolo) and another as root using my own version of gksu: XSu.
-One of the windows was showing /home/manolo/.local/share/applications, the other (running as root) /home/manolo2/.local/share/applications.
-I had copied a file from first to second window.
-In a console (running as root), I then ran 'chown -R manolo2:manolo2 /home/manolo2/.local' to change ownership of the file just copied.
-Immediately the root-running nautilus' window went blank.
-I ran gnome-system-monitor and nautilus was using 80-90% CPU alternately from each CPU.
-I found a nautilus-debug-log.txt in /root with current date-time.
-In the root console, I killed the berserk nautilus instance with 'kill -9 29473'.

Next follows the relevant part of .xsession-errors:
{
#### XSu-29054. BEGIN.
XSu-29054.CmdLine: >XSu DefaultTerminalEmulator<
#### XSu-29291. BEGIN.
XSu-29291.CmdLine: >XSu DefaultTextEditor<
XSu-29291: (gedit:29368): GnomeUI-WARNING **: While connecting to session manager:
XSu-29291: Authentication Rejected, reason : None of the authentication protocols specified are supported and host-based authentication failed.
#### XSu-29291. END.
#### XSu-29395. BEGIN.
XSu-29395.CmdLine: >XSu DefaultFileSystemBrowser /<
#### XSu-30705. BEGIN.
XSu-30705.CmdLine: >XSu echo | DefaultTextEditor $COMMON_ETC_DIR/CommonSettings<
XSu-30705: Command to run, 'echo | DefaultTextEditor /home/common/etc/CommonSettings', not found. Aborting.
#### XSu-30705. END.
#### XSu-31072. BEGIN.
XSu-31072.CmdLine: >XSu DefaultTextEditor $COMMON_ETC_DIR/CommonSettings<
XSu-31072: (gedit:31149): GnomeUI-WARNING **: While connecting to session manager:
XSu-31072: Authentication Rejected, reason : None of the authentication protocols specified are supported and host-based authentication failed.
#### XSu-31072. END.

** (gnome-system-monitor:31202): WARNING **: SELinux was found but is not enabled.

** (gnome-system-monitor:31496): WARNING **: SELinux was found but is not enabled.

** (gnome-system-monitor:31496): CRITICAL **: Could not run gksu_run("kill -s 19 29473") : Password prompt canceled.

XSu-29395: (nautilus:29473): GnomeUI-WARNING **: While connecting to session manager:
XSu-29395: Authentication Rejected, reason : None of the authentication protocols specified are supported and host-based authentication failed.
XSu-29395: (gedit:29778): GnomeUI-WARNING **: While connecting to session manager:
XSu-29395: Authentication Rejected, reason : None of the authentication protocols specified are supported and host-based authenticatio...

[manzano (manzano-vocamen)]

Manolo,
You errors messages are XSu related, you should try the same thing with gksu, not XSu.
As explained, think to open xsession-errors BEFORE reproducing the bug and reload it each time it produce a new message.
Good luck in your investigation!

[Manuel Iglesias Alonso (glesialo)]

Ref: My previous comment.

Sorry I did not explain myself. XSu messages in .xsession-errors mean there were no error messages from nautilus. It went crazy silently.
These are normal messages and they appear every time nautilus starts (Launched by XSu).{
(nautilus:29473): GnomeUI-WARNING **: While connecting to session manager:
Authentication Rejected, reason : None of the authentication protocols specified are supported and host-based authentication failed.
}

Sad to tell, I have tried several times to reproduce the fault without any success :-(

[Manuel Iglesias Alonso (glesialo)]

Even if I could not reproduce the fault by copying a file from normal_user_nautilus's window to root_user_nautilus's window and then changing ownership of the copied file (and the directory where it resides) with 'chown -R user:user Dir', I got some (interesting?) error messages from the root_user_nautilus:
{
#### XSu-6185. BEGIN.
XSu-6185.CmdLine: >XSu DefaultFileSystemBrowser /<
XSu-6185: (nautilus:6260): GnomeUI-WARNING **: While connecting to session manager:
XSu-6185: Authentication Rejected, reason : None of the authentication protocols specified are supported and host-based authentication failed.
XSu-6185: (nautilus:6260): Gtk-CRITICAL **: gtk_tree_model_sort_get_value: assertion `VALID_ITER (iter, tree_model_sort)' failed
XSu-6185: (nautilus:6260): GLib-GObject-CRITICAL **: g_object_set_property: assertion `G_IS_VALUE (value)' failed
XSu-6185: (nautilus:6260): GLib-GObject-CRITICAL **: g_value_unset: assertion `G_IS_VALUE (value)' failed
XSu-6185: (nautilus:6260): Gtk-CRITICAL **: gtk_tree_model_sort_get_value: assertion `VALID_ITER (iter, tree_model_sort)' failed
XSu-6185: (nautilus:6260): GLib-GObject-CRITICAL **: g_object_set_property: assertion `G_IS_VALUE (value)' failed
XSu-6185: (nautilus:6260): GLib-GObject-CRITICAL **: g_value_unset: assertion `G_IS_VALUE (value)' failed
XSu-6185: (nautilus:6260): Gtk-CRITICAL **: gtk_tree_model_sort_get_value: assertion `VALID_ITER (iter, tree_model_sort)' failed
XSu-6185: (nautilus:6260): GLib-GObject-CRITICAL **: g_object_set_property: assertion `G_IS_VALUE (value)' failed
XSu-6185: (nautilus:6260): GLib-GObject-CRITICAL **: g_value_unset: assertion `G_IS_VALUE (value)' failed
XSu-6185: (nautilus:6260): Gtk-CRITICAL **: gtk_tree_model_sort_get_value: assertion `VALID_ITER (iter, tree_model_sort)' failed
XSu-6185: (nautilus:6260): GLib-GObject-CRITICAL **: g_object_set_property: assertion `G_IS_VALUE (value)' failed
XSu-6185: (nautilus:6260): GLib-GObject-CRITICAL **: g_value_unset: assertion `G_IS_VALUE (value)' failed
XSu-6185: (nautilus:6260): Gtk-CRITICAL **: gtk_tree_model_sort_get_value: assertion `VALID_ITER (iter, tree_model_sort)' failed
XSu-6185: (nautilus:6260): GLib-GObject-CRITICAL **: g_object_set_property: assertion `G_IS_VALUE (value)' failed
XSu-6185: (nautilus:6260): GLib-GObject-CRITICAL **: g_value_unset: assertion `G_IS_VALUE (value)' failed
XSu-6185: (nautilus:6260): Gtk-CRITICAL **: gtk_tree_model_sort_iter_next: assertion `tree_model_sort->stamp == iter->stamp' failed
XSu-6185: (nautilus:6260): Gtk-CRITICAL **: file /build/buildd/gtk+2.0-2.12.0/gtk/gtktreeview.c: line 5928 (validate_visible_area): assertion `has_next' failed.
XSu-6185: There is a disparity between the internal view of the GtkTreeView,
XSu-6185: and the GtkTreeModel. This generally means that the model has changed
XSu-6185: without letting the view know. Any display from now on is likely to
XSu-6185: be incorrect.
XSu-6185: (nautilus:6260): Eel-WARNING **: "nautilus-metafile.c: metafiles" hash table still has 3 elements at quit time (keys above)
XSu-6185: (nautilus:6260): Eel-WARNING **: "nautilus-directory.c: directories" hash table still has 3 elements at quit time (keys above)
#### XSu-6185. END....

[Tres Finocchiaro (tres-finocchiaro)]

I often run file browser as root (gksu, sudo, etc), and I believe mouting an NTFS volume does this too... This is a great point that you add. It may be common grounds...

-Tres

[Manuel Iglesias Alonso (glesialo)]

I did it again!

This time I used (like the first time) directories full of *.desktop files.
After copying and changing ownership of a bunch of files (procedure described in earlier messages), nautilus went crazy again. It did not happen the first time, I had to repeat the trick 3 or 4 times.

I suppose desktop files are particularly difficult to show: nautilus has to read contents to show 'Name' and 'Icon' instead of the real file name and default document type icon.

Next follows the relevant section of .xsession-errors (Lines not coming from the XSu instance which launched the root_user_nautilus edited out):
{
#### XSu-12288. BEGIN.
XSu-12288.CmdLine: >XSu DefaultFileSystemBrowser /<
XSu-12288: (nautilus:12367): GnomeUI-WARNING **: While connecting to session manager:
XSu-12288: Authentication Rejected, reason : None of the authentication protocols specified are supported and host-based authentication failed.
XSu-12288: ** (nautilus:12367): WARNING **: No description found for mime type "x-special/socket" (file is ".gnome-system-monitor.manolo"), please tell the gnome-vfs mailing list.
XSu-12288: /home/common/bin/DefaultFileSystemBrowser: line 590: 12367 Killed $Application $Options "$@"
#### XSu-12288. END.
}
As you can see nautilus went crazy again without giving any error messages.

The last line shows when I killed the rogue nautilus process (It was originally invoked with a switchboard shell script: DefaultFileSystemBrowser) from the root terminal.

The root_user_nautilus' window went blank like the first time.

Th mail account I used to register (several years ago) is chockful with spam. If you want details please use glesialoAtgmailDotcom.

Saludos,
Manolo.

[manzano (manzano-vocamen)]

Where do you have a file called .gnome-system-monitor.manolo ??

I can't reproduce manolo's bug with gksu (i.e everything went fine when copying files in .local/share/applications between users and chown them to the other user).

I have bad news Manolo:
I did a little bit of 'investigation' on xsu:
It's not updated since around 2003 (and of course, Nautilus evolved since this time)
xsu is removed from debian (2004) and might be removed from hardy (https://bugs.launchpad.net/ubuntu/+source/xsu/+bug/181495)
I'm sorry Manolo but it looks like you'll need to switch to another su, I encourage you to test your procedure with gksu.

[Manuel Iglesias Alonso (glesialo)]

Thanks for your comments Manzano.

My .local/share/applications has 233 *.desktop files. Perhaps that is the difference. It must be quite a job for nautilus to show and update them all.

It seems somebody else thought of the same name. I have written XSu myself. It is a bash shell using 'Xdialog' and 'expect' packages. I have used if for quite a long time (way before Gutsy and nautilus' craziness) without any problems.

I encourage any other reader to try to reproduce the fault in the way I have. The number of *.desktop files to show may be relevant: You can copy, several times, the files you have until you have quite a bunch.
Note: Don't mess your $HOME/.local directory make a copy somewhere else.

Best regards,
Manolo.

[Manuel Iglesias Alonso (glesialo)]

By the way '.gnome-system-monitor.manolo' is in /home/manolo. I guess it has been produced by gnome-system-monitor.

[Jordan Erickson (lns)]

I'd like to reiterate that it my instance of 100% CPU utilization of Nautilus, it is a brand new install of Gutsy. I was able to reproduce it the first time the first created user logged out (with extremely minimal usage of Gnome, not including the actual browsing through the Nautilus file manager window).

[Gligor Horia (gligorhoria)]

Jordan Erickson wrote:
> I'd like to reiterate that it my instance of 100% CPU utilization of
> Nautilus, it is a brand new install of Gutsy. I was able to reproduce it
> the first time the first created user logged out (with extremely minimal
> usage of Gnome, not including the actual browsing through the Nautilus
> file manager window).
>
>
I think the bug description is verry accurate: log in , log out (that's
all u have to do...)

[Manuel Iglesias Alonso (glesialo)]

I can reproduce nautilus fault any time I want.

This is the procedure:
Preparation {
  -Copy, to a temporary directory, 2 copies of $HOME/.local
    ~/Temp/.local1
    ~/Temp/.local2
  -Make sure ~/Temp/.local?/share/applications have al least 250 *.desktop files (Make copies to increase number).
  -Open a nautilus window (normal user) in ~/Temp/.local1/share/applications
  -Open a nautilus window (root user) in ~/Temp/.local2/share/applications
  -Open a root terminal.
  -Run gnome-system-monitor
}
Repeat 1 .. 3 times {
  -In root terminal run 'chown -R user2:user2 ~/Temp/.local2' (user2: registered user, not owner of ~/Temp/.local1).
  -Select a group of files (5 or 6) in normal_user_nautilus' window for copying (Ctrl C).
  -Move to root_user_nautilus' window and copy files (Ctrl V).
!!-Very often nautilus goes crazy now -Check with gnome-system-monitor.
  -Repeat in root terminal 'chown -R user2:user2 ~/Temp/.local2'
!!-Very often nautilus goes crazy now -Check with gnome-system-monitor.
}

Please try it in your systems and inform.
Notes:
  -At least 250 *.desktop files in ~/Temp/.local?/share/applications.
  -Repeat at least 3 times.

Saludos,
Manolo.

[Manuel Iglesias Alonso (glesialo)]

I can reproduce nautilus fault any time I want.

Sorry, small mistake because of hasty typing.
Here it goes again:

This is the procedure:
Preparation {
  -Copy, to a temporary directory, 2 copies of $HOME/.local
    ~/Temp/.local1
    ~/Temp/.local2
  -Make sure ~/Temp/.local?/share/applications have al least 250 *.desktop files (Make copies to increase number).
  -Open a nautilus window (normal user) in ~/Temp/.local1/share/applications
  -Open a nautilus window (root user) in ~/Temp/.local2/share/applications
  -Open a root terminal and run 'chown -R user2:user2 ~/Temp/.local2' (user2: registered user, not owner of ~/Temp/.local1)..
  -Run gnome-system-monitor
}
Repeat 1 .. 3 times {
  -Select a group of files (5 or 6) in normal_user_nautilus' window for copying (Ctrl C).
  -Move to root_user_nautilus' window and copy files (Ctrl V).
!!-Very often nautilus goes crazy now -Check with gnome-system-monitor.
  -Repeat in root terminal 'chown -R user2:user2 ~/Temp/.local2'
!!-Very often nautilus goes crazy now -Check with gnome-system-monitor.
}

Please try it in your systems and inform.
Notes:
  -At least 250 *.desktop files in ~/Temp/.local?/share/applications.
  -Repeat at least 3 times.

Saludos,
Manolo.

[manzano (manzano-vocamen)]

OK,

I discovered that my investigation wasn't finished when I red the first comment of this bug and followed the links deeply, it is very important: https://bugs.launchpad.net/ubuntu/+source/nautilus/+bug/150471/comments/1

So following Ubuntu advises, I made scripts that debug everything, from one simple window, see screenshot and download them from http://blog.vocamen.com/2008/01/18/100-cpu-and-stops-before-debug-a-solution/

I hope it will help you finding Nautilus bug and others.
Have a nice week end
Phil

[Kyle M Weller (kylew)]

ok I have this issue as well, here is some lsof usage of the user "kim" on my system after logging her off, keep in mind all processes should be killed after logoff, hopefully this will shed some light...
kyle@ubuntu:~$ sudo lsof | grep -i kim >>lsof-nautilus.txt
kyle@ubuntu:~$ mousepad lsof-nautilus.txt
output attached........................................
kyle@ubuntu:~$ w
 18:35:44 up 57 min, 2 users, load average: 3.32, 2.91, 2.68
USER TTY FROM LOGIN@ IDLE JCPU PCPU WHAT
kyle tty9 :0 18:08 0.00s 1:16m 0.14s x-session-manager
kyle pts/0 :0.0 18:08 0.00s 0.01s 0.00s w
btw my cpu usage is 100%

[Kyle M Weller (kylew)]

i can re-create this issue by having myself logged in and have any user log in then log off, their processes arent killed and nautilus is at 100 % cpu usage here is a syscall trace screenshot attached, dont know if it shows anything to diagnose problem

[Gligor Horia (gligorhoria)]

Kyle M Weller wrote:
> i can re-create this issue by having myself logged in and have any user
> log in then log off, their processes arent killed and nautilus is at
> 100 % cpu usage here is a syscall trace screenshot attached, dont know
> if it shows anything to diagnose problem
>
> ** Attachment added: "strace.png"
> http://launchpadlibrarian.net/11446926/strace.png
>
>
don-t get me rong but didn-t we astablished that nautilus has a bug and
we should figure out what the problem is, let-s find similarities, or we
will not go any where

1. is it buged out of the box?
did u install aditinal software, is it because of un update to some
library, let-s think a litle more constructive, not just confirm the
bug, the bug was already confirmed!

[Alen (cshadow)]

It is bugged out of the box...
Maybe I'm wrong, but it happened on my machine on the fresh install and also after upgrading from feisty.
All I have to do is start nautilus from "Places" menu on the secondary screen and then close it.
Or log off one of the users. Simple as that: start/close or logoff.

[Kyle M Weller (kylew)]

it is indeed bugged out of the box, its a shame, same bug on all 15 computers at my work, and 3 ubuntu pc's at my home, just login from another account while one is logged in and then log off the second user and 100% cpu on all 18 computers I operate... Its pretty sad to have this bug for so long without a fix. Everyone is effected, ubuntu is meant for multple users but unfortunately we cannot take advantage of the linux operating system, THIS SHOULD BE LABELED CRITICAL!

[Jean-Luc Thirion (jlinho)]

Does everybody here have a Core 2 XXX CPU ?

I have a Core 2 Quad CPU. Maybe hardware has an influence for this bug? I
just can't imagine that everybody who installs Ubuntu 7.10 has such an
enormous bug (The 100% CPU Nautilus bug appears on a fresh Gutsy install).

Also I am not using a 64 bits install, just a 32 bits one.

2008/1/20, Alen <email address hidden>:
>
> It is bugged out of the box...
> Maybe I'm wrong, but it happened on my machine on the fresh install and
> also after upgrading from feisty.
> All I have to do is start nautilus from "Places" menu on the secondary
> screen and then close it.
> Or log off one of the users. Simple as that: start/close or logoff.
>
> --
> [Gutsy] high processor activity after logging out and then logging in
> again
> https://bugs.launchpad.net/bugs/150471
> You received this bug notification because you are a direct subscriber
> of the bug.
>

[greenhunter (tierfreunde-hagenburg)]

This problem is indeed critical.

i have this problem on old AMD CPUs.
like a Mobile AMD Sempron2.8Ghz

[Jordan Erickson (lns)]

I thought at first the similarities were dual-core CPUS - I'm having
this issue on HP Proliant ML370 (G5) servers - 2x dualcore Xeon 1.6GHz
(i386 *and* AMD64 installs). They are all Ubuntu LTSP servers that serve
entire labs. I have seen Nautilus hog 100% of the CPU when I am the only
one logged in (and then out), after the first login after installation.
I don't have to launch an actual filebrowser window for it to come up
(Nautilus loads automatically to show desktop icons, etc. anyway).

What's changed in Gutsy? Well, Compiz (which I don't use on my servers),
desktop-applet/tracker (which I have removed)... Is only happens when
someone is logging out. There has to be some sort of cleanup/other
routine that is triggering it.

I agree it can't be SO huge that *everyone* is experiencing this issue.
 So here is my setup, hopefully people who are having this issue can
post their setups as well so we can find some commonalities.

Server: Ubuntu 7.10 (A mix of i386 & AMD64, server & desktop installs,
all with latest linux-image-server kernels, all acting as LTSP servers
for i386 clients). All on HP Proliant ML370 G5 servers, 8GB RAM (using
PAE in the kernel to address all 8GB), 146GB RAID 1 arrays (2 SAS
disks). All of them running Gnome with the latest updates.

All instances of 100% CPU hogging happens with users who have not
modified Nautilus settings in any way (or have even opened filebrowser
windows). It's definitely stock, out of the box behavior. It doesn't
happen every time, but definitely at least 30-50% of the time, whenever
someone logs out.

HTH,
Jordan

jlinho wrote:
> Does everybody here have a Core 2 XXX CPU ?
>
> I have a Core 2 Quad CPU. Maybe hardware has an influence for this bug? I
> just can't imagine that everybody who installs Ubuntu 7.10 has such an
> enormous bug (The 100% CPU Nautilus bug appears on a fresh Gutsy install).
>
> Also I am not using a 64 bits install, just a 32 bits one.
>
> 2008/1/20, Alen <email address hidden>:
>> It is bugged out of the box...
>> Maybe I'm wrong, but it happened on my machine on the fresh install and
>> also after upgrading from feisty.
>> All I have to do is start nautilus from "Places" menu on the secondary
>> screen and then close it.
>> Or log off one of the users. Simple as that: start/close or logoff.
>>
>> --
>> [Gutsy] high processor activity after logging out and then logging in
>> again
>> https://bugs.launchpad.net/bugs/150471
>> You received this bug notification because you are a direct subscriber
>> of the bug.
>>
>

--
Jordan Erickson
Owner, Logical Networking Solutions
http://www.logicalnetworking.net
707-636-5678

Latest LNS Blogs - http://blogs.logicalnetworking.net

 RFID Passports coming in 2008
 Peru orders 260,000 OLPC laptops for kids, Mexico 50,000
 CNet to Microsoft: Ditch Vista

[Jordan Erickson (lns)]

Oh man...Ok sorry everyone, I'm not sure why my brain farted and I
forgot that there is already a fix for this (released in Hardy, not
backported to Gutsy).

---
nautilus (1:2.20.0-0ubuntu8) hardy; urgency=low

  * debian/patches/18_disable_signal_handler.patch:
    - Don't use the logging code signal handler it's buggy and makes
nautilus
      being stuck and eat cpu on crash (LP: #150471)
---

Everyone who is having this issue needs to go to the actual bug report
in Launchpad, at:

https://bugs.launchpad.net/ubuntu/gutsy/+source/nautilus/+bug/150471

Make an account if you don't already have one, and post a comment on the
bug requesting a backport to Gutsy. That way, the developers will see
how many people this bug is currently affecting in Gutsy and hopefully
will shift some momentum to it to backport it. =)

- Jordan

Jordan Erickson wrote:
> I thought at first the similarities were dual-core CPUS - I'm having
> this issue on HP Proliant ML370 (G5) servers - 2x dualcore Xeon 1.6GHz
> (i386 *and* AMD64 installs). They are all Ubuntu LTSP servers that serve
> entire labs. I have seen Nautilus hog 100% of the CPU when I am the only
> one logged in (and then out), after the first login after installation.
> I don't have to launch an actual filebrowser window for it to come up
> (Nautilus loads automatically to show desktop icons, etc. anyway).
>
> What's changed in Gutsy? Well, Compiz (which I don't use on my servers),
> desktop-applet/tracker (which I have removed)... Is only happens when
> someone is logging out. There has to be some sort of cleanup/other
> routine that is triggering it.
>
> I agree it can't be SO huge that *everyone* is experiencing this issue.
> So here is my setup, hopefully people who are having this issue can
> post their setups as well so we can find some commonalities.
>
> Server: Ubuntu 7.10 (A mix of i386 & AMD64, server & desktop installs,
> all with latest linux-image-server kernels, all acting as LTSP servers
> for i386 clients). All on HP Proliant ML370 G5 servers, 8GB RAM (using
> PAE in the kernel to address all 8GB), 146GB RAID 1 arrays (2 SAS
> disks). All of them running Gnome with the latest updates.
>
> All instances of 100% CPU hogging happens with users who have not
> modified Nautilus settings in any way (or have even opened filebrowser
> windows). It's definitely stock, out of the box behavior. It doesn't
> happen every time, but definitely at least 30-50% of the time, whenever
> someone logs out.
>
> HTH,
> Jordan
>
>
>
> jlinho wrote:
>> Does everybody here have a Core 2 XXX CPU ?
>>
>> I have a Core 2 Quad CPU. Maybe hardware has an influence for this bug? I
>> just can't imagine that everybody who installs Ubuntu 7.10 has such an
>> enormous bug (The 100% CPU Nautilus bug appears on a fresh Gutsy install).
>>
>> Also I am not using a 64 bits install, just a 32 bits one.
>>
>> 2008/1/20, Alen <email address hidden>:
>>> It is bugged out of the box...
>>> Maybe I'm wrong, but it happened on my machine on the fresh install and
>>> also after upgrading from feisty.
>>> All I have to do is start nautilus from "Places" menu on the secondary
>>> screen and then close it.
>>> Or log off one of ...

[Jordan Erickson (lns)]

Yikes. Nevermind again, I didn't know this was being copied to the bug
report already.

My signal-to-noise ratio is getting worse. ;)

Jordan Erickson wrote:
> Oh man...Ok sorry everyone, I'm not sure why my brain farted and I
> forgot that there is already a fix for this (released in Hardy, not
> backported to Gutsy).
>
>
> ---
> nautilus (1:2.20.0-0ubuntu8) hardy; urgency=low
>
> * debian/patches/18_disable_signal_handler.patch:
> - Don't use the logging code signal handler it's buggy and makes
> nautilus
> being stuck and eat cpu on crash (LP: #150471)
> ---
>
> Everyone who is having this issue needs to go to the actual bug report
> in Launchpad, at:
>
> https://bugs.launchpad.net/ubuntu/gutsy/+source/nautilus/+bug/150471
>
> Make an account if you don't already have one, and post a comment on the
> bug requesting a backport to Gutsy. That way, the developers will see
> how many people this bug is currently affecting in Gutsy and hopefully
> will shift some momentum to it to backport it. =)
>
>
> - Jordan
>
>
> Jordan Erickson wrote:
>> I thought at first the similarities were dual-core CPUS - I'm having
>> this issue on HP Proliant ML370 (G5) servers - 2x dualcore Xeon 1.6GHz
>> (i386 *and* AMD64 installs). They are all Ubuntu LTSP servers that serve
>> entire labs. I have seen Nautilus hog 100% of the CPU when I am the only
>> one logged in (and then out), after the first login after installation.
>> I don't have to launch an actual filebrowser window for it to come up
>> (Nautilus loads automatically to show desktop icons, etc. anyway).
>>
>> What's changed in Gutsy? Well, Compiz (which I don't use on my servers),
>> desktop-applet/tracker (which I have removed)... Is only happens when
>> someone is logging out. There has to be some sort of cleanup/other
>> routine that is triggering it.
>>
>> I agree it can't be SO huge that *everyone* is experiencing this issue.
>> So here is my setup, hopefully people who are having this issue can
>> post their setups as well so we can find some commonalities.
>>
>> Server: Ubuntu 7.10 (A mix of i386 & AMD64, server & desktop installs,
>> all with latest linux-image-server kernels, all acting as LTSP servers
>> for i386 clients). All on HP Proliant ML370 G5 servers, 8GB RAM (using
>> PAE in the kernel to address all 8GB), 146GB RAID 1 arrays (2 SAS
>> disks). All of them running Gnome with the latest updates.
>>
>> All instances of 100% CPU hogging happens with users who have not
>> modified Nautilus settings in any way (or have even opened filebrowser
>> windows). It's definitely stock, out of the box behavior. It doesn't
>> happen every time, but definitely at least 30-50% of the time, whenever
>> someone logs out.
>>
>> HTH,
>> Jordan
>>
>>
>>
>> jlinho wrote:
>>> Does everybody here have a Core 2 XXX CPU ?
>>>
>>> I have a Core 2 Quad CPU. Maybe hardware has an influence for this bug? I
>>> just can't imagine that everybody who installs Ubuntu 7.10 has such an
>>> enormous bug (The 100% CPU Nautilus bug appears on a fresh Gutsy install).
>>>
>>> Also I am not using a 64 bits install, just a 32 bits one.
>>>
>>> 2008/1/20, Alen <email address hidden>:
>>>> It is bugged out of...

[Pjotr12345 (computertip)]

We absolutely need a backport to Gutsy. This is a critical and totally unacceptable bug, that chases many people away from Ubuntu. No way this can remain unfixed. It's very unprofessional not to fix it, as well.

[Kyle M Weller (kylew)]

Pjotr12345, you are totally right, this is totally unacceptable, my company is already thinking of switching back to windows because of this issue. It is indeed very unprofessional as well.

[Kyle M Weller (kylew)]

By the way the issue can be re-created with Pentium 4 Hyperthreaded 3.0 ghz pc's with speed step technology, as well as P4 HT's w/o speed-step

[Jordan Erickson (lns)]

Wow - unacceptable, unprofessional?

Somehow I think relevant information regarding when, how and with what
equipment would be worth more than simply throwing stones. This *is*
open source, guys. The community helps improve itself by working together.

- Jordan

---
Kyle M Weller wrote:
> Pjotr12345, you are totally right, this is totally unacceptable, my
> company is already thinking of switching back to windows because of this
> issue. It is indeed very unprofessional as well.
>

--
Jordan Erickson
Owner, Logical Networking Solutions
http://www.logicalnetworking.net
707-636-5678

Latest LNS Blogs - http://blogs.logicalnetworking.net

 RFID Passports coming in 2008
 Peru orders 260,000 OLPC laptops for kids, Mexico 50,000
 CNet to Microsoft: Ditch Vista

[Psykotik (linux-ikiru)]

Please people, calm down. You do not pay for this service, remember this fact.

If you want to run a flavor which is designed to be used in professional area, stay on LTS (ubuntu 6.06, Dapper Drake).

I'm also worried by this critical bug, but I can see devs are aware and working on it.

Do not expect a professional (understand here "correction by a week) service when you do not pay for it. Such childish comments do not help ubuntu team.

[Tres Finocchiaro (tres-finocchiaro)]

I have to agree with Jordan here. If Ubuntu is such a high priority for your company, you may want to consider working with the repositories more closely, paying for support, or sticking with the more supported versions.

I don't think "calming down" is the right advice though. Your enthusiasm is great. Demanding free support is putting that enthusiasm in the wrong direction. I think you'll find that if you budget the money to switch to windows, you'll get an immediate solution to switching users, along with a whole other can of worms not worth discussing here.

-Tres

[Pjotr12345 (computertip)]

I have to admit that my previous reaction was a bit over the top, for which I apologize.

Nevertheless, I am annoyed that this confirmed critical bug isn't fixed yet, although I posted it more than 3 months ago (October 8, 2007).

Also it irritates me, that apparently somehow the decision was made to fix it only for Hardy Heron (next release) and not for Gutsy Gibbon (current release). This bug bears the stamp "fixed" now, but it is *not* fixed.

Greetz, Pjotr.

[Alen (cshadow)]

Did anyone notice some other simptoms beside nautilus hogging all the CPU?
Killing it in /etc/gdm/PostSession/Default solved the logout problem for me. Today after returning from work I noticed (again) slow user switching, slow application startup - it took 20 s to maximize pan with a lot of disk activity. Swap usage 1.5/2 GB :-)
Sometimes nautilus from the other user hangs "silently", not using any CPU, but eating a lot of RAM

alen@Amber:~$ free -m
             total used free shared buffers cached
Mem: 2027 1976 50 0 17 300
-/+ buffers/cache: 1658 368
Swap: 2047 1574 473

  PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
14755 elly 15 0 648m 519m 11m S 0 25.6 2:13.74 nautilus

So, until now i have:
- nautilus hanging every time i open it on the secondary screen
- when logging off users (solved)
- random hanging with 0% CPU usage but eating RAM
Not to mention other bugs...people I'm switching back to feisty until hardy comes out, still have it on other partition...

[Psykotik (linux-ikiru)]

to the Ubuntu team: may I (may we) help you to resolve this bug? Do you need more informations?

[amias (amias)]

noticed the same thing and solved it by adding a killall -9 nautilus to /etc/gdm/PreSession/default and /etc/gdm/PostSession/default . This prevents nautilus from chewing cycles when people are not logged in
instead of just waiting untill someone logs in again. This also works if X dies randomly or someone does
Ctrl+Alt+Backspace.

Beware that if you use xdmcp or some other way of having more than one session at once then other nautilui
will be killed by this . Nautilus should restart automatically for any effected user so this might not be as bad as
it seems , ymmv .

How about a patch ubuntu peeps ?

[Kim Pepper (kim-pepper)]

Deleting .thumbnails folder and restarting fixed the problem for me.

[Jean Levasseur (levasseur.jean)]

@ Sebastien Basher:
Thank you very much Sebastien. I was about to post my PPA's link, in which I've put your patched version of Nautilus to make it available for wide testing, for I'm using it since the beggining of December without an issue on my Gutsy machine, but I guess I wont have to do that. Anyway, if you think is a good idea to do so, please advise me.

[Sebastien Bacher (seb128)]

Thank you Jean, that's not required, users should rather try the gutsy update when it'll be available

[Martin Pitt (pitti)]

Ah, this will stop creating the 'nautilus-debug-log.txt' files in user's home directory? I think that's a regression we can live with, we can always ask folks to click on the apport .crash file (which we should get instead now). Approved, please upload.

[Sebastien Bacher (seb128)]

right, neither the ubuntu triagers nor upstream has really used this log so it should be no issue, I've already uploaded the update

[Martin Pitt (pitti)]

Accepted into gutsy-proposed, please test and give feedback here. Can someone come up with a reproducible test case?

[Nick Fishman (bsdlogical)]

Like Jordan, I've also bin bitten by this bug on a dual LTSP server network. We also have NFS home directories, which compounds the slowness.

I just applied the updates for nautilus and nautilus-data from gutsy-proposed. I'll see what happens over the next few days.

[Sebastien Bacher (seb128)]

you need to restart nautilus to get the new version running

[Jordan Erickson (lns)]

Thank you!!! Just saw it download from the ropes.

=) =) =)

Sebastien Bacher wrote:
> you need to restart nautilus to get the new version running
>

--
Jordan Erickson
Owner, Logical Networking Solutions
http://www.logicalnetworking.net
707-636-5678

Latest LNS Blogs - http://blogs.logicalnetworking.net

 RFID Passports coming in 2008
 Peru orders 260,000 OLPC laptops for kids, Mexico 50,000
 CNet to Microsoft: Ditch Vista

[Psykotik (linux-ikiru)]

I applied the updates for nautilus, nautilus-data and libnautilus-exension1 from gutsy-proposed (release 7.1). Unfortunately, no change; nautilus runs, and whenever I quit, it crashes and the process uses a high amount of the processor.

Reproducing it is quite easy: open a terminal, run "sudo nautilus", and close. The processor will be highly solicited. The only way to have back its ressources, is to kill the process.
I think it only go through this behaviour once, afterwhat it works normally. However, the close reports always a "segmentation fault (core dumped)" into the terminal.

Find in attachment the crash report. Be warned I cannot open it, don't know why the reports regarding nautilus can't be opened. I also attach a crash resulted from opening a second user session.

[Tres Finocchiaro (tres-finocchiaro)]

Great level of detail Psykotic. :)

[Sebastien Bacher (seb128)]

the change is not to solve crasher but to get nautilus not being stuck on the log when there is one

[Psykotik (linux-ikiru)]

Since it was an "artificial crash" (when talking about the first one), a consequence of nautilus being stuck, I thought it could be of use.

I was wrong. If you need something else...

[Sebastien Bacher (seb128)]

the bug seems to mix different issues. Does anybody still get nautilus creating a nautilus-debug-log.txt on crash when using the update?

[Jean-Luc Thirion (jlinho)]

I applied the patch. No nautilus-debug-log.txt up to now.

[Ronald van Engelen (ronalde)]

After applying the update yesterday on our LTSP-server there are no hanging processes anymore.

[Martin Pitt (pitti)]

Thanks for the testing so far. Positive feedback in the sense of "I applied the update and everything still works as normal" is also appreciated.

[greenhunter (tierfreunde-hagenburg)]

applied

and no high cpu usage so far on single und multi user desktops with nautilus.
No nautilus debug file anymore.

thx a lot.

[Psykotik (linux-ikiru)]

Nobody experiments a nautilus crash since the patch when doing a

sudo nautilus

??

[Psykotik (linux-ikiru)]

(sorry, when doing a sudo nautilus, AND closing nautilus.)

[Alen (cshadow)]

I applied the update last night and it works for now: no more hanging (and high CPU usage) nautilus when opened & closed on secondary screen.
As for sudo nautilus, usually I don't do that, but here is the result: upon closing I get "Segmentation fault (core dumped)".
But nautilus doesn't get hung any more...it crashes nicely instead :-)

[Martin Pitt (pitti)]

What did the previous version do for you if you called it through
sudo?

--
Martin Pitt | http://www.piware.de
Ubuntu Developer (www.ubuntu.com) | Debian Developer (www.debian.org)

My 5 bugs today: #185273 #190947 #193494 #192786 #172792
Do 5 a day - every day! https://wiki.ubuntu.com/5-A-Day

[Psykotik (linux-ikiru)]

It used also to crash (don't remember if a .crash was created, though) but the processor was additionnaly under a high activity, and unable to close.

Nautilus process was needed to be killed, to unload the processor.

[Martin Pitt (pitti)]

Thanks. So that's hardly a regression then and the new package seems better. Thanks for confirming!

[Alen (cshadow)]

The same thing happened here. But my biggest problem was logging off another user or closing nautilus on the secondary screen.
Fixing /etc/gdm/PostSession/Default solved the first problem, and this update seems to have fixed the other. I never had the problem with large nautilus-debug-log.txt

[Jean-Luc Thirion (jlinho)]

No nautilus-debug-log.txt anymore. Thanks a lot. sudo nautilus and closing
does not lead to a crash for me. Si that's another issue.

[DickeyWang (invariance)]

Is this fix for Hardy only or it is also for Gutsy? My Gutsy system is up to date, and I just got another "nautilus-debug-log.txt" in my home directory 5 minutes ago.

[Martin Pitt (pitti)]

DickeyWang [2008-02-26 23:09 -0000]:
> Is this fix for Hardy only or it is also for Gutsy? My Gutsy system is
> up to date, and I just got another "nautilus-debug-log.txt" in my home
> directory 5 minutes ago.

It is already in Hardy. For gutsy it is in -proposed at the moment,
where it is tested by a wider audience. I guess you don't have
proposed updates enabled? (System -> Administration -> Software
Sources -> Updates).

[debianmigrant (debianmigrant)]

Fix has been working for me for several days, now with no evident side effects. Thank you!

[Martin Pitt (pitti)]

Copied to gutsy-updates.

[gimme5 (mauriciootta)]

I have the same problem on Gutsy amd64 on a Core2Duo with nVidia Go7600 256Mb and no nautilus installed

I can run and see pidgin, the calculator, evolution
but no Firefox or terminal

[gimme5 (mauriciootta)]

Oops... i did had nautilus installed.... dããã

disabling the effects
and
apt-get purge nautilus
apt-get install nautilus
/etc/init.d/gdm restart

seems to have resolved the black desktop problem.... but still no icons on menu and no buttons on windows

[gimme5 (mauriciootta)]

maybe a shot in the /dev/null ... but I installed everything again, and I see a bunch of lines like this:
/var/lb/scrollkeeper/oc/scrollkeeper_extended_cl.xml:2777: Parse error : Extra
content at the end of the document
</sect>
^
/var/lb/scrollkeeper/oc/scrollkeeper_extended_cl.xml:2777: Parse error : Extra
content at the end of the document
</sect>
^
/var/lb/scrollkeeper/oc/scrollkeeper_extended_cl.xml:2777: Parse error : Extra
content at the end of the document
</sect>
^

on the 199 updates I have listed after this fresh install...

[Schnitty (brownpr0)]

running hardy here and saw same errors as gimme5 described, which on reboot gives me a black screen with no control bars/panels. the only reason i'm able to submit this bug is because pidgin starts correctly and i can tell it to open my gmail from there. adding the killall nautilus to PostSession/Default did not do anything for me. i'm thinking perhaps gimme5 and i are seeing a different issue than this bug, but i did see the same list of lines he described here and now can't use my nautilus at all. :(