Suse 9.3 quirks

[time]

Others may care to add other examples to this thread.

I installed Suse 9.3 on an nForce4 (Athlon 64) PC with virgin drives. Everything was fine. I experimented with software RAID, including breaking it, and somewhere along the line, Linux started having a 'kernel panic' while booting.

I didn't think much of it, until I attempted to reinstall Suse 9.3 and ended up with the same error. Specifically, it couldn't mount sda2, the root partition in this case.

To cut a long story short, I deleted the partitions and used the Hitachi Drive Fitness Test tool to initialize the boot sector. I also cleared the PC CMOS. Still, as soon as the installation came to rebooting, Suse had a kernel panic.

Eventually, I discovered I could fix the problem by restricting the swap partition (sda1) to less than the 1GB default, eg 1000MB instead.

If anyone can shed any light, that would be great, otherwise I'll just hope that this info might help someone one day.

 

[Buck]

I deleted the partitions and used the Hitachi Drive Fitness Test tool to initialize the boot sector.

Initialize the boot sector? Do you mean write zeros? Besides erasing the boot sector, you also need to erase the BPB at sector 74. The best thing is to write zeros to the first few hundred sectors of the drive.

 

[time]

The best thing is to write zeros to the first few hundred sectors of the drive.

What tool would you suggest?

 

[Buck]

There DFT v4.05 utility has the option: Erase Disk. It looks like you ran the Erase Boot Sector option.

 

[Mercutio]

A very tiny amount of research suggests that it's a bug in 9.3.

I had some difficulty with mounting pre-existing hardware RAID on my 3ware cards under 9.3 as well. When I pulled my existing drives and tried new ones in a plain old software array, the array wouldn't format (boo). I compiled a new kernel and things started working for me with both the new drives and the old ones. I don't know whether there's something about SuSE 9.3 that doesn't like RAID or just a coincidence, but I certainly did have problems.

I've found SuSE to be great for "desktop" linux, but right now I'm still trying to decide if I'd rather putz around upgrading my file servers to SuSE or if I should just install Centos.

 

[Buck]

After playing with Suse 9.3 for a bit, I think I'll set it up as a file server in my house. Plus, I might start up Apache and use it for web development. But as far as a desktop environment goes, and the Windows applications that must be run, it doesn't cut the mustard.

I am impressed with Yast and Suse's online update capability.

 

[time]

After more experimentation with different swap partition sizes in different places, I realized it only failed in the original place at the original size. I also used Hitachi DFT to zero about 5% of the disk, but then experienced a different problem.

How's this for a possible explanation:

Linux software RAID uses superblocks at the end of each device. Given that it's partition-based, I assume there's probably one at the end of each partition. Simply deleting the partition wouldn't erase the superblock, but truncating the partition would effectively hide it.

Similarly, erasing only the first few percent of a disk wouldn't touch the superblock at the end of the last partition - you'd have to zero the entire disk.

In other words, perhaps YAST is trying to be too clever by half and recognizes a RAID member even when RAID isn't currently selected?

 

[Handruin]

I'm noticing an odd quirk with suse 9.3. The HD LED stays lit on my system shortly after the boot process begins. This was never the case with SuSe 9.2. The problem is fixed on reboot, but as soon as suse begins loading, it stays on all the time.

 

[i]

If some fancy flash screen appears during boot up, can you clear it? (Maybe try pressing ESC?) What is loading when your HDD LED goes solid?

Not that there're are particularly technical suggestions, but it seems you are not alone. And that's Mandrake ... so perhaps this is a kernel issue with your motherboard? The person in the other discussion had a Intel D915PBL motherboard, with a 3.4GHz P4.

Also, it seems one of the other people in the discussion actually found their LED signals inverted (i.e., off = activity, on = idle) from time to time. Is it that specific in your case? Or just on all the time, regardless of activity?

 

[Handruin]

I'm running this on my dell sc420. My situation is not due to constant HD activity, the system is not using many resources. The light glows solid and never flickers, even under activity. My LED is not reversed like the forum members described.

I noticed the LED glow solid right after the first SATA probing message appears on the screen. I can try to take a picture of it if you're interested.

The only time the LED functions normal, is after POST, and right after the boot menu loads. Shortly after suse begins the boot, the LED glows solid from this point forward.

 

[i]

I noticed the LED glow solid right after the first SATA probing message appears on the screen. I can try to take a picture of it if you're interested.

No, that's ok, I understand what you mean. I thought that was the same thing the original poster in the other discussion was experiencing, but maybe I misread.

Anyway ... I wish I had some suggestions. I might try some creative searches tomorrow.

Just as a tangent, I have a PCMCIA network card for my old laptop. The last time I tried the famous "Tom's Root Boot Disk" floppy-based Linux distro, I found the network card drivers caused the activity light to operate in an inverse fashion. On meant no activity, and off meant activity. It was very strange to see, because for years under Win98 and then some old Redhat version it'd been working in the expected and normal fashion.

Aside from the strange LED issue, the card worked just fine under that floppy distro though...