Ubuntu
apparmor package

system bogs down when apparmor is running

Bug #552225 reported by Martin Pool
8
This bug affects 1 person
Affects Status Importance Assigned to Milestone
apparmor (Ubuntu)
Invalid
Undecided
Unassigned

Bug Description

Binary package hint: apparmor

This may or may not be a dupe of bug 458299. I don't have the kernel error message quoted there.

After I updated and rebooted my Lucid machine today, everything seemed very laggy: there's a fraction-of-a-second delay between each keypress and it showing up in firefox, gnome-terminal, etc. top doesn't show a high load average or anything using an unusually large fraction of CPU. This seems like a regression from lucid as of a couple of days ago.

Robert suggested "sudo service apparmor stop" and that does immediately fix the problem.

grace% sudo service apparmor start
 * Starting AppArmor profiles Skipping profile in /etc/apparmor.d/disable: usr.bin.firefox
Warning: found usr.sbin.ntpd in /etc/apparmor.d/force-complain, forcing complain mode
                                                                                                  [ OK ]

and the problem comes back. It persisted for over an hour so I don't think this is due only to overhead from apparmor starting up.

ProblemType: Bug
DistroRelease: Ubuntu 10.04
Package: apparmor 2.5-0ubuntu2
ProcVersionSignature: Ubuntu 2.6.32-18.27-generic 2.6.32.10+drm33.1
Uname: Linux 2.6.32-18-generic x86_64
NonfreeKernelModules: nvidia
Architecture: amd64
Date: Wed Mar 31 12:39:20 2010
EcryptfsInUse: Yes
ProcEnviron:
 PATH=(custom, no user)
 LANG=en_AU.utf8
 SHELL=/bin/zsh
SourcePackage: apparmor

Revision history for this message
Martin Pool (mbp) wrote :
Revision history for this message
Robert Collins (lifeless) wrote :

what happens if you reboot - did the kernel get updated when you updated this morning?

Revision history for this message
Martin Pool (mbp) wrote :

Actually having experimented a bit more, I'm not sure that this is in fact an apparmor problem: it's still slower than a couple of days ago even with apparmor disabled. Sorry this is so vague.

Revision history for this message
John Johansen (jjohansen) wrote :

Martin,

can you try booting with apparmor=0 as a kernel parameter at boot. This will completely disable AppArmor so that it never initializes.

Revision history for this message
Martin Pool (mbp) wrote : Re: [Bug 552225] Re: system bogs down when apparmor is running

On 31 March 2010 12:50, Robert Collins <email address hidden> wrote:
> what happens if you reboot - did the kernel get updated when  you
> updated this morning?

I booted this morning only long enough to update, but during that time
I didn't notice any problems. After I rebooted into "Linux grace
2.6.32-18-generic #27-Ubuntu" I started having this problem. I
upgraded this morning to

2010-03-31 09:55:55 status installed linux-generic 2.6.32.18.19

I haven't changed any packages since rebooting.

--
Martin <http://launchpad.net/~mbp/>

Revision history for this message
Martin Pool (mbp) wrote :

That seems to help.

- Martin

On 31/03/2010 1:50 PM, "John Johansen" <email address hidden> wrote:

Martin,

can you try booting with apparmor=0 as a kernel parameter at boot. This
will completely disable AppArmor so that it never initializes.

--
system bogs down when apparmor is running
https://bugs.launchpad.net/bugs/552225
You received t...

Revision history for this message
John Johansen (jjohansen) wrote :

Martin,

can you elaborate on how this helps, where you notice it, etc.

Also can you reboot with AppArmor enabled and rerun aa-status and attach its output along with the the kern.log again.

Also can you try,
making symlinks in /etc/apparmor.d/disable/ to the /usr/sbin/mysqld and /usr/bin/evince profiles in the /etc/apparmor.d/ directory. This will disable those profiles by default. After doing this I suggest a clean reboot to test if this has any affect.

Revision history for this message
Martin Pool (mbp) wrote :

Actually this seems to have just gone away with current updates. :-|

Changed in apparmor (Ubuntu):
status: New → Invalid
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.
Subscribing...

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.