Antonio Olivares wrote:
--- On Thu, 4/30/09, Allen Kistler wrote:
I chased ConsoleKit, gnome-session, and a few others when I saw this bug on my systems. I didn't get anywhere. It's a challenging bug to chase, because it doesn't always happen. Although I think it may be more reproducible when going from runlevel 5 to runlevel 3, maybe because if something's broken, it gets killed anyway in runlevel 0 or runlevel 6 when shutdown or reboot actually shuts down or reboots.
Anyway, I've seen people ask about this bug on this list (with various subject lines, but always essentially the same thing). I've seen other people file bugs and get them marked as duplicate. I've seen people add themselves as CCs on existing bugs. Even if I'm really stupid, I'd think at least some of the people interested in this bug are pretty bright. The fact that nobody seems to have a handle on exactly what's going on is kind of scary.
For me, the power button and ACPI still team up to provide a graceful shutdown, but that seems a bit inelegant.
If you take another look at the attachment sent by Bill, the cultprit seems to be selinux not allowing shutdown. However, the updates might have fixed it? Don't know why but I have seen several of my rawhide machines doing this, but I did not bother to check what was going on and just pressed power button so that they could shutdown all the way.
I know my bad, but I see it only on two of the five machines and others do shutdown properly so it seems to me that not all users out there are experiencing the same trouble(s).
I question SELinux as the root cause, because: 1. The problem has existed for a while with no AVC denial. 2. I ran in permissive mode, anyway, and still saw the bug. 3. The bug only happens sometimes, but SELinux would be all the time.
In other words, I think SELinux is a new bug that should be fixed, and I think we still need to keep looking. I haven't done much testing in the last week, half because of the freeze making nothing new available to test. Now I've got a copy of the Preview DVD, so I'll crank it up again today or tomorrow and see what I see. I heartily encourage everyone else to do the same.
Allen Kistler (an037-ooai8@yahoo.com) said:
Anyway, I've seen people ask about this bug on this list (with various subject lines, but always essentially the same thing). I've seen other people file bugs and get them marked as duplicate. I've seen people add themselves as CCs on existing bugs. Even if I'm really stupid, I'd think at least some of the people interested in this bug are pretty bright. The fact that nobody seems to have a handle on exactly what's going on is kind of scary.
I'm having a heck of a time reproducing it - given that so far I've only seen it once in about 20 reboots, it's not really lending itself to useful debugging.
Bill
On Thu, Apr 30, 2009 at 04:32:15PM -0400, Bill Nottingham wrote:
I'm having a heck of a time reproducing it - given that so far I've only seen it once in about 20 reboots, it's not really lending itself to useful debugging.
Maybe you will find data attached to https://bugzilla.redhat.com/show_bug.cgi?id=498514 useful? Currently I have rather the opposite situation. Maybe once in 20 reboots shutdown will suddenly work.
Michal
On Thu, Apr 30, 2009 at 02:35:14PM -0600, Michal Jaegermann wrote:
On Thu, Apr 30, 2009 at 04:32:15PM -0400, Bill Nottingham wrote:
I'm having a heck of a time reproducing it - given that so far I've only seen it once in about 20 reboots, it's not really lending itself to useful debugging.
Maybe you will find data attached to https://bugzilla.redhat.com/show_bug.cgi?id=498514 useful? Currently I have rather the opposite situation. Maybe once in 20 reboots shutdown will suddenly work.
Michal
I have not seen it, I wonder if it is hardware specific, maybe a driver or something, maybe a few of the afftected people can give some details on their hardware to see if there is some common element.
--- On Thu, 4/30/09, max maximilianbianco@gmail.com wrote:
From: max maximilianbianco@gmail.com Subject: Re: Can't stop log To: "For testers of Fedora Core development releases" fedora-test-list@redhat.com Date: Thursday, April 30, 2009, 1:52 PM On Thu, Apr 30, 2009 at 02:35:14PM -0600, Michal Jaegermann wrote:
On Thu, Apr 30, 2009 at 04:32:15PM -0400, Bill
Nottingham wrote:
I'm having a heck of a time reproducing it -
given that so far
I've only seen it once in about 20 reboots,
it's not really lending
itself to useful debugging.
Maybe you will find data attached to https://bugzilla.redhat.com/show_bug.cgi?id=498514 useful? Currently I have rather the opposite
situation. Maybe once
in 20 reboots shutdown will suddenly work.
Michal
I have not seen it, I wonder if it is hardware specific, maybe a driver or something, maybe a few of the afftected people can give some details on their hardware to see if there is some common element.
--
Martin first reported the shutdown troubles here:
http://www.nabble.com/Rawhide---shutdown-not-working-to23074751.html
and posted his hardware as well.
I have encountered this on several of my rawhide machines, but not all of them :)
One which I did encounter this is :
http://www.smolts.org/client/show/pub_f15ce4e7-e658-4ea2-bd89-4f9b8b1d645e
and this one too:
http://www.smolts.org/client/show/pub_808fc0dc-61b4-4f57-a41a-62cdc710058d
The othere are OK or at least were this past Tuesday after which I have not been to see my machines and will not do so for a while :(
Regards,
Antonio
On Thu, Apr 30, 2009 at 02:19:34PM -0700, Antonio Olivares wrote:
--
Martin first reported the shutdown troubles here:
http://www.nabble.com/Rawhide---shutdown-not-working-to23074751.html
and posted his hardware as well.
I have encountered this on several of my rawhide machines, but not all of them :)
One which I did encounter this is :
http://www.smolts.org/client/show/pub_f15ce4e7-e658-4ea2-bd89-4f9b8b1d645e
and this one too:
http://www.smolts.org/client/show/pub_808fc0dc-61b4-4f57-a41a-62cdc710058d
The only common element that jumps out at me is that all the machines have wireless interfaces. A source of trouble with linux if ever there was one, however I have just been bitten by this bug for the first time now,make that twice. This box doesn't have wireless so there goes that idea. Its the xserver i'm guessing, it hangs when I logout from ssh too but only if I use an app like gthumb or firefox. If I only run emacs then no problems it just logs right out. Reminds me to follow up on the other thread since FF works now. Anyone know what emacs does different from a graphical presentation standpoint than say FF or gThumb?It would seem that it does something the others do not when it closes its window. Since this isn't a problem from the console or shutting down from gdm, I would guess Xserver issue but I don't recall if anyone mentioned their desktop enviro or not, is this affecting KDE and XFCE as well? Anyway its just a guess since I could fit what I know about thr xserver on a business card in large fonts :^)
On Fri, May 01, 2009 at 12:59:56PM -0400, max wrote:
On Thu, Apr 30, 2009 at 02:19:34PM -0700, Antonio Olivares wrote:
--
Martin first reported the shutdown troubles here:
http://www.nabble.com/Rawhide---shutdown-not-working-to23074751.html
and posted his hardware as well.
I have encountered this on several of my rawhide machines, but not all of them :)
The only common element that jumps out at me is that all the machines have wireless interfaces.
The one from which https://bugzilla.redhat.com/show_bug.cgi?id=498514 originated does not have wireless. It is a desktop machine with a rather old x86_64 mobo and a radeon graphics card.
Michal
max wrote:
On Thu, Apr 30, 2009 at 02:19:34PM -0700, Antonio Olivares wrote:
--
Martin first reported the shutdown troubles here:
http://www.nabble.com/Rawhide---shutdown-not-working-to23074751.html
and posted his hardware as well.
I have encountered this on several of my rawhide machines, but not all of them :)
One which I did encounter this is :
http://www.smolts.org/client/show/pub_f15ce4e7-e658-4ea2-bd89-4f9b8b1d645e
and this one too:
http://www.smolts.org/client/show/pub_808fc0dc-61b4-4f57-a41a-62cdc710058d
The only common element that jumps out at me is that all the machines have wireless interfaces. A source of trouble with linux if ever there was one, however I have just been bitten by this bug for the first time now,make that twice. This box doesn't have wireless so there goes that idea. Its the xserver i'm guessing, it hangs when I logout from ssh too but only if I use an app like gthumb or firefox. If I only run emacs then no problems it just logs right out. Reminds me to follow up on the other thread since FF works now. Anyone know what emacs does different from a graphical presentation standpoint than say FF or gThumb?It would seem that it does something the others do not when it closes its window. Since this isn't a problem from the console or shutting down from gdm, I would guess Xserver issue but I don't recall if anyone mentioned their desktop enviro or not, is this affecting KDE and XFCE as well? Anyway its just a guess since I could fit what I know about thr xse
rver on a business card in large fonts :^)
No, I see it on one laptop, one server, one little Shuttle firewall box, and KVM virtual machines. I see it in both 32 and 64 bit versions. What mine have in common is that they are running GNOME, nothing more.
THIS JUST IN: I installed the new stuff which came in overnight, kernel, initscripts, selinux policy, plymouth, lots of possibly related stuff. Since the upgrade shutdown and reboot have worked flawlessly. tested in the VM (never rebooted before), the laptop (64 bit) and the firewall (32 bit).
Smells pretty good to me. Confirmation please? Or am I just lucky?
On Tue, May 05, 2009 at 04:46:04PM -0400, Bill Davidsen wrote:
No, I see it on one laptop, one server, one little Shuttle firewall box, and KVM virtual machines. I see it in both 32 and 64 bit versions. What mine have in common is that they are running GNOME, nothing more.
THIS JUST IN: I installed the new stuff which came in overnight, kernel, initscripts, selinux policy, plymouth, lots of possibly related stuff. Since the upgrade shutdown and reboot have worked flawlessly. tested in the VM (never rebooted before), the laptop (64 bit) and the firewall (32 bit).
Smells pretty good to me. Confirmation please? Or am I just lucky?
Looking good here too...
On Thu, Apr 30, 2009 at 4:32 PM, Bill Nottingham notting@redhat.com wrote:
I'm having a heck of a time reproducing it - given that so far I've only seen it once in about 20 reboots, it's not really lending itself to useful debugging.
Incomplete GUI shutdown happens about 20% of the time on my Acer Aspire One.
Random reproducibility is the worst kind. (The last non-hardware random issue I recall was with mdadm in a race condition)
Jason Farrell wrote:
On Thu, Apr 30, 2009 at 4:32 PM, Bill Nottingham <notting@redhat.com mailto:notting@redhat.com> wrote:
I'm having a heck of a time reproducing it - given that so far I've only seen it once in about 20 reboots, it's not really lending itself to useful debugging.Incomplete GUI shutdown happens about 20% of the time on my Acer Aspire One.
Random reproducibility is the worst kind. (The last non-hardware random issue I recall was with mdadm in a race condition)
-- Jason
I can reproduce this regularly. GUI shutdown has worked twice in 40+ cycles.
TK009
On Thu, Apr 30, 2009 at 03:18:02PM -0500, Allen Kistler wrote:
Antonio Olivares wrote:
If you take another look at the attachment sent by Bill, the cultprit seems to be selinux not allowing shutdown.
I question SELinux as the root cause, because:
- The problem has existed for a while with no AVC denial.
- I ran in permissive mode, anyway, and still saw the bug.
- The bug only happens sometimes, but SELinux would be all the time.
I concur. I have selinux off on my test system and I got hit.
Michal
-
Allen Kistler -
Antonio Olivares -
Bill Davidsen -
Bill Nottingham -
Jason Farrell -
max -
Michal Jaegermann -
TK009