OpenSUSE 11.3 fresh install issues during updating phase

English Install/Boot/Login
1

Hi Everyone,

I’ve been working through a fresh installation of OpenSUSE 11.3 on a new box, and managed to track down the root cause of all the issues I’ve been running into.

I can install OpenSUSE 11.3 with out issue, and can update all the software via the update applet or through the software management interface to bring them up to current versions with the exception of the kernel itself.

Every time I try to update the kernel via the update applet or via the software management, it locks the system up. It can download the kernel delta file, and “create” a merged RPM of the new kernel, but during the subsequent install phase it locks the system solid.

I’ve attempted to update the kernel in init 3 (zypper up) and 5 (via software management or update applet) levels, and have not yet tried it in single user mode

The hardware is as follows:

  • Core i7-950 3.2 Ghz Intel CPU
  • 12 Gbytes DDR3, triple channel memory, clocked at 1066 Mhz
  • 1 x Intel X25M Rev 2 SSD, 80 Gbyte Boot Drive
  • 6 x Seagate 7200.12 1 TB HDDs, configured in RAID 6
  • GByte GA-X58A-UD7 Mobo

Most BIOS features are enabled, except for the “No execute” feature, as I thought it could be one of the causes, but it appears to have no change on if I can update the kernel or not.

The current installed kernel is:

2.6.34-12-desktop #1 SMP PREEMPT 2010-06-29 02:39:08 +0200 x86_64 x86_64 x86_64 GNU/Linux

And I’m looking to bring the kernel up to the 2.6.34.4 update that is in the system update. Thus far, in all cases the lock up as corrupted the kernel itself and the system fails to reboot.

I’m looking for options that I’ve not yet thought of, and can provide more details if they are needed.

Thanks in advance.

2

So this does seem like a strange problem that only the kernel can not be updated. I must say that every time I see someone with a SSD, there are strange problems up to being unable to install openSUSE at all. I might make a couple of suggestions.

First, I would modify how YaST installs kernels so that your original working kernel is not removed and thus still able to boot your system. Second, I might even consider installing kernel 2.6.35 just to see if the SSD works any better with a newer kernel. I have a thread which talks about both of these changes. Check it out here:

Unable To Boot after updating the Kernel

Thank You,

3

I’ve made the changes you’ve recommended and have noted that the issue seems to be more around hwinfo’s execution vs the kernel itself. I managed to get a “hung” system, but only the installation part hung, the rest was responsive, and via a ps listing I noted that hwinfo was active, and using 99.9 percent CPU - not bad for “4” cores + 4 hyperthread channels :/. However, killing the hwinfo process did complete the total hang, and a reboot was required.

I can remove the “new” kernel with out issue via rpm -e, but all installations seem to fail even though the RPM database seems to be updated, as does the menu.lst file for grub, but each time I try an install, it’ll hang, force a reboot, and the expected result of a kernel failing to boot for the “new” kernel is the end result, however with your recommended changes I can reboot into the old kernel.

I do not doubt there are issues with SSD’s, and all my other OpenSUSE 11.3 systems do not seem to have the kernel update issue this one does, wondering if the issue is in fact centered around the hwinfo process as stated above, because even running that manually cause a complete lock.

4

Well, I wonder if it is a problem with the kernel 2.6.34. For instance, I have a USB 3 drive and it can not be mounted in a fstab file using kernel 2.6.34, but they work just fine using kernel 2.6.35. A second part of the link was on how to install the latest kernel. I would not try using 2.6.36, but I think a try with 2.6.35 might be in order for you.

S.A.K.C - SuSE Automated Kernel Compiler

If you can not get through the compile, that is a problem. You need to install all of the development tools for base, kernel & C/C++. Open YaST / Software Management / View / Patterns then look for the the ones I just mentioned. Do this before you try to install the latest kernel. Stay away from the Partitioner and anything that scans your disks.

This method of installing a new kernel takes a long time but it does not remove any other Kernels from your system.

Thank You,

5

On 2010-10-11 03:36, cbowlby wrote:
>
> I’ve made the changes you’ve recommended and have noted that the issue
> seems to be more around hwinfo’s execution vs the kernel itself. I
> managed to get a “hung” system, but only the installation part hung, the
> rest was responsive, and via a ps listing I noted that hwinfo was
> active, and using 99.9 percent CPU - not bad for “4” cores + 4
> hyperthread channels :/. However, killing the hwinfo process did
> complete the total hang, and a reboot was required.

I wonder if you can install the kernel manually via “rpm” on CLI (use rpm install, not upgrade: so
that you keep both kernels). That would not use hwinfo.

We need to know if the problem is in the kernel or in the installer, or both. I feel it is the
installer.

Then I would report in Bugzilla the problem with the installer.

–
Cheers / Saludos,

Carlos E. R.
(from 11.2 x86_64 “Emerald” at Telcontar)

6

Who decided to remove this important & useful feature from the graphic installation? Shoot them.
Nvidia 5500FX & Belinea LCD 2225 S1W 22.2" 1680x1050@60Hz monitor.
Just wanted to break that previous thread & ask the question in title. If you update, you keep the display/screen settings but a fresh install, with auto config turned off doesn’t allow you to change the ‘best guess’ set up. I tried writing xorg.conf & 50-monitor.conf files a few times, which sent it into text only. After installing the Nvidia driver/inteface it restricted it to 640x480 & less I assume (maybe wrongly) the Nvidia install worked off the Xserver vga settings, hence the problem. I understand people have had problems setting up the Belinea monitor, which is why the ability to change settings during installation, if only to a generic LCD, is so useful. I’m not a very good writer of code but if somebody could find a way of re introducing this feature I’m sure a lot of people would be much happier. Things like that have a detrimental effect on encouraging SuSE linux newbies or users who follow both the light & dark sides of PC use. To remove a method of sorting a problem is just plain stupid.

7

I’ve confirmed that I can compile the kernel from scratch, but the install still fails due to the following:

test@test:~> ps aux | grep hwinfo
root      2949 99.6  0.0 406316  4532 tty1     R+   14:34   2:01 /usr/sbin/hwinfo --framebuffer
1000      3046  0.0  0.0   7668   808 pts/0    S+   14:36   0:00 grep hwinfo

I’ve not yet had it completely lock, so I will let it sit out there for a bit to see if it will finish, who knows, maybe I’ve lost patience with it… :>

8

Hi Carlos,

It seems to not matter if I’m in init 3 or 5, in both cases the following code call is made:

I’ve confirmed that I can compile the kernel from scratch, but the install still fails due to the following:

test@test:~> ps aux | grep hwinfo
root      2949 99.6  0.0 406316  4532 tty1     R+   14:34   2:01 /usr/sbin/hwinfo --framebuffer
1000      3046  0.0  0.0   7668   808 pts/0    S+   14:36   0:00 grep hwinfo

it is getting CPU time, but I suspect it’s stuck somewhere, so the root cause seems to be more hwinfo vs the kernel being installed from an RPM, or compiled then installed.

9

I’ve left it run for about 20-30 minutes with no progress made, and attaching an strace to the process confirms it has hung, as there’s no code execution taking place. It is simply sitting there doing nothing.

10

cbowlby, I never ask you if you tried the kernel failsafe mode or if you might have tried the kernel option nomodeset to see if the problem might be video related somehow. When I see the term --framebuffer, I am thinking video problem.

Thank You,

11

On 2010-10-11 20:06, cbowlby wrote:
>
> Carlos E. R.;2236383 Wrote:

> I’ve confirmed that I can compile the kernel from scratch, but the
> install still fails due to the following:

  • Install it via rpm on CLI and paste the output, if any.

  • Keep that strace and open a bugzilla with it.

–
Cheers / Saludos,

Carlos E. R.
(from 11.2 x86_64 “Emerald” at Telcontar)

12

Jdmcdaniel3, I had tried it before within failsafe mode, I still had the vga=0x345 kernel option defined (it was a default option). When I removed that, I noticed that nomodeset was already defined. I let the system boot into X, bring the system back to init 3 mode, and was able to compile and install with out issue. I’m about to test if that kernel (the compiled 2.6.35.7) will work.

13

Carlos,

Ok, I will do that later tonight when I have more time, only had enough time to test a kernel build before work, and will do this test once I get home.

14

carlos,

I’ve managed to get some output from an RPM install via the command line, and am in the process of creating a bugzilla, will update this post with the URL once done.

15

Bug report posted:

https://bugzilla.novell.com/show_bug.cgi?id=645986

16

As an aside, installing the RPM via failsafe mode at init 3 worked with out issues, this lends weight to the problem being the:

hwinfo --framebuffer

command

edit, however, now the ati framebuffer is broken :confused: lol

Going to stick with 2.6.35.7 as it works.

17

As an aside, installing the RPM via failsafe mode at init 3 worked with out issues, this lends weight to the problem being the:

hwinfo --framebuffer

command

edit, however, now the ati framebuffer is broken :confused: lol

Going to stick with 2.6.35.7 as it works.
cbowlby, so you got kernel 2.6.35.7 to work. How did you install this new kernel?

Thank You,

18

On 2010-10-13 02:36, cbowlby wrote:
>
> As an aside, installing the RPM via failsafe mode at init 3 worked with
> out issues,

I thought so.

> this lends weight to the problem being the:
>
> hwinfo --framebuffer

Interesting. I don’t know why they have to use that command, this is new to me.

> command
>
> edit, however, now the ati framebuffer is broken :confused: lol

Ouch! Bad luck.

Well, you can also report that error on another bugzilla :slight_smile:

Tell us in this thread the outcome of that bugzilla, please. It may take months, anyway, but it is
the only method to really solve those problems.

–
Cheers / Saludos,

Carlos E. R.
(from 11.2 x86_64 “Emerald” at Telcontar)

19

On 2010-10-13 04:06, jdmcdaniel3 wrote:

>> How did you install this new kernel?

Using the rpm command. CLI.

–
Cheers / Saludos,

Carlos E. R.
(from 11.2 x86_64 “Emerald” at Telcontar)

20

actually, to get 2.6.35.7 installed I had to drop to failsafe mode, set init 3 as the run level, and then I was able to compile and then install from the source. That sakc script made things easier, however I was able to get it done both with that script and manually.