System CPU maxes out at arbitrary times

[Kyle]

System: Ubuntu 11.10

At arbitrary times, my laptop's CPU will max out and the system will come to a halt and it becomes almost unresponsive. Simple actions like opening a tty and logging in take 3 minutes+. I've yet to identify a culprit. A screenshot from htop during this unresponsive time is attached and can also be found here: http://i.imgur.com/t5ca0.png.

I would say about 75% of the time when this happens, I have a virtual machine running in Virtual Box and the machine is minimized. It also happens when the machine is in use, but not as often. The other 25% of the time it will happen without having a virtual machine running, so I've ruled that out as the cause. Most of the time Google Chrome is also running, but I've had this occur without Google Chrome running, as well.

I always notice, however, that the process "/usr/lib/erlang/erts-5.8.3/bin/beam.smp ..." is running when this happens. Killing this process does not cause the system to recover, however.

Whatever causes the hang up causes all the processes to hang. Most of the time I have to manually power cycle my laptop to resolve the issue (on restart, most of the time I have to let the laptop cool because the CPU won't let the system boot because it's over-heated). Sometimes I can kill Virtual Box or Chrome or another process taking up much of the CPU and the system will recover.

I've had this issue occur after long periods of uptime and also within 5 minutes of restarting my computer.

This has been an issue ever since I upgraded to 11.10. I'm currently using the classic gnome session, but I've had this happen in the other UI's as well.

[Vadim Rutkovsky (roignac)]

Could you please try instructions in https://bugs.launchpad.net/ubuntu/+source/erlang/+bug/458453/comments/33 to lower 'beam.cmp' process usage?
---
Ubuntu Bug Squad volunteer triager
http://wiki.ubuntu.com/BugSquad

[Kyle (kyleflanagan)]

I will try that and report back my findings.

[Kyle (kyleflanagan)]

After doing as Vadim suggested, I notice a slight improvement. Now when my computer locks up, the beam.smp process doesn't take up as much CPU, but it's still present. Again, I'm not sure if this is actually the cause of the issue.

I can confirm that this happens when virtualbox is running and when it's not.

[Otto Kekäläinen (otto)]

When the systems is slow again, could you please run in a terminal "top" and copy-paste the first 15 lines or so here that we can see what it shows?

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Michael Stucki (mstucki)]

I am not exactly sure if I have the same problem, but it sounds very much like it.
Since I reinstalled my machine with Ubuntu 13.04, VirtualBox became almost unusable for me.

Whenever I start a machine (or when I restore a saved machine) the load of the host machine gets extremely high (> 30) and the machine reacts very slow, if at all.

I tried this with VirtualBox 4.2 as well as 4.1.

I am next going to run memtest to find out if the memory is corrupt, however this would be surprising as the problem never happened before I switched to Ubuntu 13.04...

[Michael Stucki (mstucki)]

I checked and according to memtest86+, my memory is fine.

However, the problem still occurs and renders my VirtualBox completely unusable.

My next attempt will be to install a newer kernel.

[Michael Stucki (mstucki)]

I just switched to the new kernel image from http://kernel.ubuntu.com/~kernel-ppa/mainline/ (current version is v3.9-raring).
So far, everything works smooth and feels a lot better than before.

If the machine will hang again, I'll report back here, but otherwise I assume that a kernel update will fix the problem!

[Michael Stucki (mstucki)]

Forget about it, the kernel update didn't help. Same situation again like before, after a while the CPU went over 100% usage, the system got slower and slower until it was freezing completely.

What I tried so far:
- VirtualBox in 3 different versions (4.2.10 Ubuntu package, 4.2.12 package for Ubuntu from Oracle's repository, 4.1 using the generic installation archive from virtualbox.org)
- Kernel 3.8.0-10-generic from stock Ubuntu 13.04
- Kernel 3.9.0 from kernel.ubuntu.com (see above)
- disabling all VBox options like PAE/NX, VT-x (incl. disabling it in the BIOS), IO-APIC
- tried with 2 CPUs, 1 CPU, 50% CPU usage, 100% CPU usage ...
- changing the Chipset from PIIX3 to ICH9 and back
- removed kernel modules kvm_intel and kvm (read somewhere that this might be the culprit, but it also didn't help)

Nothing helped, the problem came back all the time. Sometimes just during booting the VM, otherwise after a few minutes of work....

I am really out of ideas here. What can I do to find out what is freezing my machine? How can I prevent it?