After an evaluation, GNOME has moved from Bugzilla to GitLab. Learn more about GitLab.
No new issues can be reported in GNOME Bugzilla anymore.
To report an issue in a GNOME project, go to GNOME GitLab.
Do not go to GNOME Gitlab for: Bluefish, Doxygen, GnuCash, GStreamer, java-gnome, LDTP, NetworkManager, Tomboy.
Bug 325979 - nautilus uses 100% cpu after closing a ssh session
nautilus uses 100% cpu after closing a ssh session
Status: RESOLVED FIXED
Product: gvfs
Classification: Core
Component: sftp backend
1.4.x
Other Linux
: Normal major
: ---
Assigned To: gvfs-maint
gvfs-maint
Depends on:
Blocks:
 
 
Reported: 2006-01-06 13:08 UTC by Sebastien Bacher
Modified: 2013-11-14 15:01 UTC
See Also:
GNOME target: ---
GNOME version: 2.13/2.14



Description Sebastien Bacher 2006-01-06 13:08:40 UTC
This bug has been opened here: http://bugzilla.ubuntu.com/show_bug.cgi?id=21021

"When I open a ssh connection to my other computer from my laptop, everything is
fine & nautilus works as usual, but, if I close the window with the ssh
connection to my other computer & then shut down the computer, nautilus starts
consuming 100% cpu time. I assume this is because it's trying to check if the
ssh connection is still alive, or the server available, but the fact is that if
I launch top, nautilus is using 100% cpu time.

So, to reproduce, log in gnome. create a ssh connection to another computer
(with connect to another computer link from he places menu). do something in the
remote computer (for instance copy a file). Close the window with the ssh
connection & shutdown the remote computer. If you wait a bit nautlius should
start consuming 100% cpu.

I'm using breezy up to date."
Comment 1 Martin Meyer 2006-07-06 13:27:20 UTC
I've had the same issue for some time now and I've got a better way to reproduce.  Open an SSH connection to somewhere, then kill the process that opened the socket.  The socket goes to TIME_WAIT and momentarily nautilus will get extremely confused and start eating your CPU like it's candy.  Really tasty candy.

My ps output for the offending process looked like this:
<pid> pts/0    Ss+    0:00 -oForwardX11 no -oForwardAgent no -oClearAllForwardings yes -oProtocol 2 -oNoHostAuthenticationForLocalhost yes -p 22 -l <user> -s <server>

Interestingly, I don't see an actual command in there anywhere.  Just a bunch of commandline arguments.

I've got this error on Dapper Drake and two Gentoo machines of different architectures.  I'm running 2.14 on all of them.
Comment 2 Daniel Holbach 2006-11-01 12:07:09 UTC
Bug URL is now: https://launchpad.net/distros/ubuntu/+source/nautilus/+bug/27109

New comment: "Yes I can confirm this in both Dapper and Edgy. Whenever the network connection between the two hosts is terminated, nautilus on the client (the one downloading) hits 100% CPU and remains there. It's quite a serious bug because the only way to bring back down the cpu is to do a 'killall nautilus'."
Comment 3 javiermon 2006-11-05 22:03:43 UTC
FYI This still happens in gnome 2.16

Could someone confirm it?
Comment 4 Ewan Towie 2006-11-23 16:54:56 UTC
I had a big problem with Nautilus in Dapper where it would crash fairly randomly when using SSH connections.  It would do this several times a day.

I upgraded to Edgy a couple of weeks ago, and although Nautilus no longer bombs out and is forced to reload, it now consumes 100% CPU.  Again, I am still using SSH connections and find the only way to fix this is to use the system monitor to end the Nautilus process.

I would help in producing some debug output with symbols, but I am unsure how to achieve this as Nautilus no longer initiates the Bug Report tool.
Comment 5 John Spray 2007-03-11 16:05:39 UTC
Potentially related backtrace of nautilus when it was taking 100% CPU after a failed attempt to login to an SSH host.

(gdb) bt
  • #0 __kernel_vsyscall
  • #1 gettimeofday
    from /lib/tls/i686/cmov/libc.so.6
  • #2 g_get_current_time
    from /usr/lib/libglib-2.0.so.0
  • #3 g_source_get_current_time
    from /usr/lib/libglib-2.0.so.0
  • #4 g_source_get_current_time
    from /usr/lib/libglib-2.0.so.0
  • #5 g_main_context_prepare
    from /usr/lib/libglib-2.0.so.0
  • #6 g_main_context_check
    from /usr/lib/libglib-2.0.so.0
  • #7 g_main_loop_run
    from /usr/lib/libglib-2.0.so.0
  • #8 gtk_main
    from /usr/lib/libgtk-x11-2.0.so.0
  • #9 POA_Nautilus_MetafileMonitor__init
  • #10 __libc_start_main
    from /lib/tls/i686/cmov/libc.so.6
  • #11 ??

Comment 6 John Spray 2007-03-11 16:07:08 UTC
This could also be the same bug:
https://bugzilla.novell.com/show_bug.cgi?id=222490
Comment 7 Ian Ohr 2008-01-16 11:47:59 UTC
I can confirm this bug still exists in the version of Gnome that is used in Ubuntu 7.10 - Gutsy.

What information is needed to get this bug to confirmed status.
Comment 8 Ian Ohr 2008-01-16 11:51:04 UTC
I should have quoted the Gnome version in the last comment
Gnome 2.20.1

I suggest this bug be adjusted to reflect the versions of Gnome it still exists in.
Comment 9 Julien Olivier 2008-11-07 08:09:40 UTC
I just wanted to say that I still see this bug in GNOME 2.24 (from Ubuntu Intrepid), and I even have it with gedit from time to time (gedit uses 100% cpu if a file is opened through gvfs but the server doesn't respond anymore), but I don't know whether it's related.
Comment 10 Cosimo Cecchi 2010-04-15 09:45:13 UTC
-> gvfs

Users on the Launchpad bug seem to report this still applies to recent versions as well.
Comment 11 Timothy Arceri 2013-05-04 04:58:37 UTC
I can not reproduce this issue (I was trying to trigger it to help track down the cause of another upstream bug https://bugs.launchpad.net/ubuntu/+source/gvfs/+bug/377322). 

Can anyone confirm is this is still and issue with a current version?
Comment 12 Martin Meyer 2013-05-07 15:31:06 UTC
I haven't seen this issue in a couple years.
Comment 13 javiermon 2013-05-07 15:47:50 UTC
I haven't had this bug lately, not sure when it stopped so I'm not a 100% sure if it's fixed.
Comment 14 Ross Lagerwall 2013-11-14 15:01:36 UTC
This bug appears to have been fixed sometime. Closing as fixed.