OpenSUSE 11.1 freezing

[brazooka]

Dear OpenSUSE forum members:

I would like to count on your assistance providing some light on the causes of the following problem:

I built a new computer a few weeks ago with the following configuration:

Processor: AMD Phenom II X4 940
Motherboard: Asus M4A78
Memory: Corsair Dominator 4GB (2 x 2GB) DDR2 1066
Video Card: PCie Nvidia GeForce 9500GT (1GB)
Storage: Two 500GB HD configured in SW RAID 1

Memory tests run without any problems.

I successfully installed OpenSUSE 11.1 (x84_64, KDE4.1), run the updates (no factory updates though) and installed the nvidia drivers from the yast repository.

Then problems started: the machine started freezing (no mouse, no keyboard, no high CPU) randomly for no apparent reason. The system logs show no alert.

First I thought the problem could be the nvidia drivers and I uninstalled them. The freeze still happened with the drivers shipped with OpenSUSE.

Then I started reading the forums trying to find some assistance. I changed iommu and memaper kernel parameters during boot as suggested in some forum messages without success. I also enabled and disabled the "memory hole remap" option in the bios to check if it would make a difference, but it didn't.

At this point, suspecting a HW problem I added a third HD and installed Ubuntu 9.0.4 (AMD64). Gladly, it has been running for days now without a single freeze. So I am excluding the HW defect as a possible cause of the freezes.

Then I reinstalled OpenSUSE 11.1 all over, with the same configuration. The problem persisted.

After analyzing the logs of the booting process of both distributions I came noticed some differences, but their analysis is beyond my level of expertise with Linux, and besides they are from different kernel "versions", i.e., Ubuntu is using kernel 2.6.28 whereas OpenSUSE is using 2.6.27.

Can comparing the boot messages from both distributions help clarify this problem? I could post them here.

I could ignore this problem, start using Ubuntu, but the thing is I really love this distribution and have been using it since 1999. I am crossing my fingers I can come back using it as I did in the past ten years.

[caf4926]

Your comment suggest that maybe if you didn't update after install it would run fine? Do you think that is a correct assumption?

Just say you install 11.1 install the kernel-source, gcc, make,.........and so on, so you can build the nvidia driver manually.
Try it and see how it is.
It's possibly the kernel update that's the issue.

Actually if you have 11.1 installed now and it's updated, you can roll back the kernel to the release version.

[brazooka]

caf4926:

I have just completed rolling back the kernel to the 11.1 release version, i.e., 2.6.27.7-9.

To exclude the nvidia driver influence, I uninstalled it, reverting the the original driver of the installation DVD.

In less than an hour OS freezed, just like it happened before.

[caf4926]

Have you tried booting with these arguments:

acpi=off noapic

add them to your default bootline manually first.

On Thu August 6 2009 09:06 am, brazooka wrote:

>
> Dear OpenSUSE forum members:
>
> I would like to count on your assistance providing some light on the
> causes of the following problem:
>
> I built a new computer a few weeks ago with the following
> configuration:
>
> Processor: AMD Phenom II X4 940
> Motherboard: Asus M4A78
> Memory: Corsair Dominator 4GB (2 x 2GB) DDR2 1066
> Video Card: PCie Nvidia GeForce 9500GT (1GB)
> Storage: Two 500GB HD configured in SW RAID 1
> Memory tests run without any problems.
>
> I successfully installed OpenSUSE 11.1 (x84_64, KDE4.1), run the
> updates (no factory updates though) and installed the nvidia drivers
> from the yast repository.
>
> Then problems started: the *machine started freezing *(no mouse, no
> keyboard, no high CPU) randomly for no apparent reason. The *system
> logs show no alert*.
>
> First I thought the problem could be the nvidia drivers and I
> uninstalled them. The freeze still happened with the drivers shipped
> with OpenSUSE.
>
> Then I started reading the forums trying to find some assistance. I
> changed iommu and memaper kernel parameters during boot as suggested in
> some forum messages without success. I also enabled and disabled the
> "memory hole remap" option in the bios to check if it would make a
> difference, but it didn't.
>
> At this point, suspecting a HW problem I added a third HD and
> *installed Ubuntu 9.0.4* (AMD64). Gladly, it has been *running for
> days* now without a single freeze. So I am excluding the HW defect as a
> possible cause of the freezes.
>
> Then I reinstalled OpenSUSE 11.1 all over, with the same configuration.
> The problem persisted.
>
> After analyzing the *logs of the booting process* of both distributions
> I came noticed some differences, but their analysis is beyond my level
> of expertise with Linux, and besides they are from different kernel
> "versions", i.e., Ubuntu is using kernel 2.6.28 whereas OpenSUSE is
> using 2.6.27.
>
> *-Can comparing the boot messages from both distributions help clarify
> this problem? -*I could post them here.
>
> I could ignore this problem, start using Ubuntu, but the thing is *I
> really love this distribution* and have been using it since 1999. I am
> crossing my fingers I can come back using it as I did in the past ten
> years.
>
>
brazooka;

The original 11.1 kernel had a broken inotify module. This may not apply to
you, but if you are running an IMAP or Samba server it may. See:

https://bugzilla.novell.com/show_bug.cgi?id=463372

This should have been corrected with a kernel update.
--
P. V.
"We're all in this together, I'm pulling for you." Red Green

[brazooka]

It took a while to happen again, but passing "acpi=off noapic" to the kernel during the boot did not solve the problem. Interestingly, after these changes top and meminfo reported less memory, around 3.2MB.

BTW, I am not using IMAP or Samba.

Should I try compiling 2.6.28 from the source? What could be the unwanted effects of using a different kernel version from the 11.1 release?

Thanks for the help!

[lwfinger]

brazooka wrote:
> It took a while to happen again, but passing "acpi=off noapic" to the
> kernel during the boot did not solve the problem. Interestingly, after
> these changes top and meminfo reported less memory, around 3.2MB.
>
> BTW, I am not using IMAP or Samba.
>
> Should I try compiling 2.6.28 from the source? What could be the
> unwanted effects of using a different kernel version from the 11.1
> release?

The main differences will be the absence of the splash screen while
booting, a couple of warnings that the kernel does not support certain
ioctl's, and no AppArmor. In other words, no real differences.

I do kernel testing as my daily routine - mostly for wireless drivers
and support code. Once I do an installation, my first step is to build
and install the latest kernel version. Once that is complete, it is
extremely rare for me to boot the standard kernel.

[caf4926]

This probably sounds a bit lame. But my Box as you see in my signature: I failed to get it working well with 11.1 - for me it just rebooted at random. Sadly I ended up without openSUSE on my Box and still have Fedora 10.

The hardware you have should make for a nice unit if it run properly and I can't see why it shouldn't.

It's possible .xsession-errors might have some enlightenment. That's in hidden /home/user/

[brazooka]

As I mentioned in the first post, Opensuse 11.1 root file system was running on top of a raid array (SW Raid1) and kept freezing, whereas Ubuntu 9.0.4 was running on a single HD and did not freeze.

In the past days I decided to install Ubuntu on the raid array, expecting the have a stable system, but I found out that it also freeezes! Even before installing the nvidia proprietary drivers.

My current plan is to try Fedora 11 tonight on the raid array and possibly Opensuse 11.2 after that, just to check if still freezes.

caf4926, thanks for all the suggestions and the feedback.

[nickelarse]

I'm also getting system freezes with OpenSuSE 11.1 on a RAID 0 array. Installed onto a single hard drive it runs flawlessly. Before 11.1, I was running 10.3 for years (or however long it was since it came out) with no problem on the exact same RAID array. Windows runs flawlessly on the same RAID array too. As with you, I get no errors in the logs. The kernel panics at any time between one hour and 2 days after boot, with the num lock and caps lock lights flashing. Alt+SysRq+B doesn't reboot it.

It seems like there is a problem with the RAID driver in the kernel, possibly, but I've got no idea where to go next!