Why Did My System Lock Up?

[topcat247]

Earlier today, I tried to move the mouse and nothing happened. I couldn't use Alt-Tab to switch to another task. I couldn't use Ctrl-Atl-F1 to get a console. Ctrl-Atl-Del got me nothing. I noticed the disk access light was lit solid and the only thing I could do was press the reset button and start over.

So what happened? I looked at /var/log/messages just before the reboot and didn't see anything of note.

Where can I look to find clues about what happened?

I'm running F16 on an AMD box with 8GB of memory. I tend to have lots of windows open on several different desktops as I use different desktops for different projects. Which means I hate to reboot.

uname -r tells me 3.3.5-2.fc16.x86_64 and yum update says I'm all up to date.

I'm usually running Firefox, VirtBox, Skype, Eclipse, Dropbox, Banshee, NixNote and Gkrell.of course all at the same time. This doesn't happen very often but it has happened before. If I knew what caused it I might be able to fix it and keep it from happening again.

Thanks in advance for your feedback.

TopCat

[stevea]

My first suspicion is virtbox. There have been reports of it's driver scribbling at incorrect pages in the kernel. That's not proof - but I'd remove the loadable driver for vbox for a test.

[topcat247]

The system lockup doesn't happen very often so I wouldn't want to disable anything right now.

What I'm looking for are the steps to follow to provide forensic data which hopefully indicates what was happening before the lockup occurred.

How do I configure my system to log this data so I can review it next time?

Thanks,
TopCat

[Gareth Jones]

This is difficult. If the problem is a kernel panic, then unless the events which lead to it are significant enough to be logged and happen long enough before, they'll never make it to /var/log/messages. It's possible that X (the graphics/mouse/keyboard layer) crashed rather than the kernel though.

My advice, if looking through the logs doesn't yield anything, would be to learn how to use Magic SysRq: http://en.wikipedia.org/wiki/Magic_sysrq. In Fedora you need to enable that feature (it's not secure for machines with remote access) by setting "kernel.sysrq = 1" in /etc/sysctl.conf. This will perhaps give you a safer way to reboot, and maybe even avoid rebooting or get debugging info.

[topcat247]

Last night my system just froze while I was using it. Since it was late, I went to bed. This morning it was working fine and I could see from the logs it had not rebooted.

What I also found in the logs, about the time it froze up was a seires of messages about a page fault error from a VirtBox vm. I'll enable Magic Sysrq. It's looks like a nice thing to have.

If I have any breakthroughs, I'll post again.

Thanks,
TC