opensuse 11.0 - Wierd things

BACKGROUND

I have 1 ide hard drive — sda and 1 sata hard drive — sdb
Sda is formated with seperate partitions for each of boot,local,opt,usr,var,and home
Sdb has 3 partitions sdb2 and sdb3 contain data. sdb1 is Linux (40 gb)

Using the RC1 dvd,
First I tried to do an upgrade install on sdb1 - it appeared to work got a bootable system; I was satisfied; so I then tried to do an upgrade install on sda; This resulted in a system that would not boot (kernel panic - see earlier thread). Not only would sda not boot, but now sdb1 would not boot.
I then downloaded 11.0 GM - same thing happened, finally got my head out of my backside and did an md5sum check - dvd was broken; downloaded image was also broken.

What this post is all about
new download, new image, new DVD, (successful media check) ---- fresh install on sdb1. told it to install on sdb1; made sure the boot configuration showed sdb1 as the boot default;

After the install when the system booted for the first time - it booted sda! and menu.lst had NO entry to boot sdb1. Also all menu.lst entries were pointed to the “pae” kernel NOT the “default” kernel

Am I smoking some bad stuff or is there really a bug in the opensuse 11.0 GM DVD?

Just took a look at the boot folder — the only images there are the pae images — no default kernel images

Shoulda checked bugzilla — that would have explained some of the things I’ve experienced.

(bugzilla shows the bug closed with RC1, but it appears to be still there with the GM dvd)

garyg the first wrote:

> Just took a look at the boot folder — the only images there are the
> pae images — no default kernel images

I have met this problem as well ! It seems to come about because the
BIOS supplies details about the IDE subsystem to the install program
but because of the naming of the drives in Linux is done using the SCSI
notation the SATA drives are considered first.

I got round this by disconnecting my SCSI drives. Do the install and
setting up. Then connect the other drives, which are then recognised
and numbered correctly.

HTH.


Best Regards:
Baron.

VOn Sun, 29 Jun 2008 20:56:53 GMT
Baron <baron.nospam@linuxmaniac.nospam.net> wrote:

>garyg the first wrote:
>
>> Just took a look at the boot folder — the only images there are the
>> pae images — no default kernel images
>
>I have met this problem as well ! It seems to come about because the
>BIOS supplies details about the IDE subsystem to the install program
>but because of the naming of the drives in Linux is done using the SCSI
>notation the SATA drives are considered first.
>
>I got round this by disconnecting my SCSI drives. Do the install and
>setting up. Then connect the other drives, which are then recognised
>and numbered correctly.
>

I noticed this today on my 11 install and thought I’d have problems,
but I didn’t. It was confusing at first when I got to the partitioning
(I always do the expert partitioning) because it was pointing to sdb
instead of sda. In my 10.3 install, sda is a 160GB IDE drive with the
OS only on it and sdb is a 232GB SATA with /home and various data
directories. These were reversed and I almost stopped the install, but
figured, WTH, I’ll see what happens.

Install went ok, and when I tried rebooting back into 10.3, it still
worked ok so I let it be. :slight_smile:

One interesting difference – when I first installed 10.3 (from 10.1)
it saw my LVM setup on the 232GB drive and picked up the correct mount
points. The 11 install saw my LVM setup but I had to manually set the
mount points. No problem, just an observation.


Kevin Nathan (Arizona, USA)
Linux is not a destination, it’s a journey – enjoy the trip!

Linux 2.6.22.18-0.2-default
9:28pm up 0:06, 21 users, load average: 1.71, 2.27, 1.24

Kevin Nathan wrote:

> VOn Sun, 29 Jun 2008 20:56:53 GMT
> Baron <baron.nospam@linuxmaniac.nospam.net> wrote:
>
>>garyg the first wrote:
>>
>>> Just took a look at the boot folder — the only images there are
>>> the pae images — no default kernel images
>>
>>I have met this problem as well ! It seems to come about because the
>>BIOS supplies details about the IDE subsystem to the install program
>>but because of the naming of the drives in Linux is done using the
>>SCSI notation the SATA drives are considered first.
>>
>>I got round this by disconnecting my SCSI drives. Do the install and
>>setting up. Then connect the other drives, which are then recognised
>>and numbered correctly.
>>
>
> I noticed this today on my 11 install and thought I’d have problems,
> but I didn’t. It was confusing at first when I got to the partitioning
> (I always do the expert partitioning) because it was pointing to sdb
> instead of sda. In my 10.3 install, sda is a 160GB IDE drive with the
> OS only on it and sdb is a 232GB SATA with /home and various data
> directories. These were reversed and I almost stopped the install, but
> figured, WTH, I’ll see what happens.
>
> Install went ok, and when I tried rebooting back into 10.3, it still
> worked ok so I let it be. :slight_smile:

In my case it took a couple of goes before the penny dropped. Once I
realised what was going on unplugging the SCSI cable got things moving
in the right direction.

In some ways it very bad, because a newby with two drives of different
interfaces is going to end up with a trashed system and one that won’t
boot from where the install was done. Made worse by the new install
actually working until the machine is re-booted.

> One interesting difference – when I first installed 10.3 (from 10.1)
> it saw my LVM setup on the 232GB drive and picked up the correct mount
> points. The 11 install saw my LVM setup but I had to manually set the
> mount points. No problem, just an observation.

I seem to recall another post saying something like this. I wonder if
its worth doing a bug file on these issues ?


Best Regards:
Baron.

On Mon, 30 Jun 2008 16:36:29 GMT
Baron <baron.nospam@linuxmaniac.nospam.net> wrote:

>Kevin Nathan wrote:
>
>> VOn Sun, 29 Jun 2008 20:56:53 GMT
>> Baron <baron.nospam@linuxmaniac.nospam.net> wrote:
>>
>In my case it took a couple of goes before the penny dropped. Once I
>realised what was going on unplugging the SCSI cable got things moving
>in the right direction.
>
>In some ways it very bad, because a newby with two drives of different
>interfaces is going to end up with a trashed system and one that won’t
>boot from where the install was done. Made worse by the new install
>actually working until the machine is re-booted.
>

I think what saved me was that my 10.3 has a separate /boot partition
and when I select 10.3 from the 11 boot screen, it runs my 10.3 grub
menu.

>> One interesting difference – when I first installed 10.3 (from 10.1)
>> it saw my LVM setup on the 232GB drive and picked up the correct
>> mount points. The 11 install saw my LVM setup but I had to manually
>> set the mount points. No problem, just an observation.
>
>I seem to recall another post saying something like this. I wonder if
>its worth doing a bug file on these issues ?
>

I don’t know enough about LVM to know if this is really a bug. After I
thought about it a while, I remember I installed 10.3 with my
10.1 /home directory being available. The install of 11 was completely
clean and won’t get my /home dir until I’ve verified everything runs as
I need it to run . . .


Kevin Nathan (Arizona, USA)
Linux is not a destination, it’s a journey – enjoy the trip!

Linux 2.6.22.18-0.2-default
9:49pm up 1 day 0:26, 21 users, load average: 0.97, 0.69, 0.98

garyg the first wrote:

>
> Just took a look at the boot folder — the only images there are the
> pae images — no default kernel images
>
>

Can’t speak to your original issue, but PAE is the new “default” for openSUSE 11, when applicable. It was decided to use PAE for PAE-capable systems, whether or not they have 4GB+, due to the additional benefits. The old “-default” kernel is still available as an option, though.

Cheers,
KV

Hello KV
Thanks for the info.
But ---- what is ‘PAE’?:o

here ya go PAE

Andy

here ya go PAE

Andy

It’s there for a reason: the default kernel is the PAE kernel on the DVD (it was true with RC1, and is still true with GM; that doesn’t change until after a later update). Amazingly, I don’t mind it, even though my CPU (Pentium 4 Northwood-C) doesn’t support PAE; in fact, I’ve stayed with the latest PAE kernel (though a non-PAE version of the same kernel is available).

PAE support in this kernel is modular; in short, it won’t load unless it detects the CPU supports it. In addition to the modular PAE support, this same kernel also has modular SMP/multicore support (this kernel replaced the old -bigsmp option for HT and multicore systems), and I haven’t taken any support of processing-cycle or other performance lumps despite the non-presence of PAE in my processor.

Thank you, Andy.

Gary