mylex scsi raid with ibm ultrastar project

[xazax]

Im having trouble configuring linux/unix on my computers.
Harddware:
cpu = Dual PII 333mhz
ram = 1gb
video= voodo 16mb ram
raid = mylex dac960pl 16mb dram
-bios version 4.xx
-firmware 3.52--xx
-target drives 1 and 2 on channel 0
-raid level 0
-write policy "write back"
-system drives initialized
-system drives recognized as one system drive -
-@ 70001mb
-bios set to 8gb partition not 2gb
drives= 2 ibm ultrastar 36.x gb harddrives
Oses Tested:
Linux
Redhat 7.1
SuSE 7.1
SlackWare 7.1
Slackware 8.0
Mandrake 8.1
Unix
FreeBSD 4.4
FreeBSD 4.5
Description of problem:
RedHat Linux 7.1:
System goes through P.O.S.T recoginizing all of my hardware.
Mylex Bios recognizes ibm harddrives as one system drive.
CMOS set to boot to IDE cdrom, installation of linux begins.
I'm skipping over to the "expert" mode since that was the first sign of progress.
In expert mode I was able to run fdisk and create my partition table.
I allocated anywhere from:

100 - 400mb /boot partition
2047mb <swap> partition
remainder space / partition

The files systems get formatted, no problem (sometimes), reads package information, transfers the install image to the harddrives and the installation takes off.
Here's where the problems begin to vary. Either the comptuter locks up completely to where I cannot move my mouse or numlock check my keyboard anywhere from 10 to 35 minutes, or the systems gives me an error like:
oops
cpu0
kernel trap:
unable to handle virtual address ax00000123 etc.
I wont get too much into FreeBSD because the effect is very similar in different terminology but the error is in that category.
Slackwares are not able to even get fdisk to open the drives at all.
DOS that I skipped to mention did install to the same harddrives.
I am lost and stranded. Does any one know a solution to this problem?
I really hope that it is caused by some user ignorant error(me) and not defective harddware. I have two identical server machines like this(the only difference is the color of the cpu heatsinks).
Additional notes:
I did specify smp support during my installation.
The dac960pl driver was loaded from linux as opposed to when the old firmware was on there. There has been a little progress from when I first started this project.
Please Help!!! S.O.S
Thanks
xazax

 

[bjverzal]

You should post this in the Linux forum.

 

[lej]

Hi, from your info, it looks as if you have set a 2047MB Swapp partition. If I'm not mistaken, you should keep the swapp partition to less than 128MB.

Try setting the swap partition to less than 128MB and give that a try.

Good luck

 

[aixat]

Hello,
check with manufacture of scsi controller if they have tested your setup. if they have not tested your setup then you are out of luck.

 

[lmbedore]

To read the dac960xx RAID cards on slackware, you'll need to load the RAID.s kernel. This is possible off the CD (hit F2 and F3 for details at boot-time), or from floppy if you have the patience.

It sounds like a hardware problem that won't be resolved from any particular distribution, though. I'd run a RAM testing utility of some sort on the machine to ensure that the RAM is good. I'd also try swapping video cards. It may also be worth doing a text-based install on redhat to avoid loading/stressing the video card anyways.

This problem could also be due to something as simple as RAM timing/clocking. If your BIOS allows you to, try setting all the RAM timings (CAS, CAS/RAS, etc) to the slowest possible.

 

[xazax]

ThanX alot for everyones help.

I have tried Windows 2000, and Windows NT 4 just to check whether there was any incompatibilities with UNIX/Linux.


I have tried different combinations of hardware minimizing the amount of memory and even with a single CPU.

I have never, ever experienced such a head ache with any other hardware.

I have 11 other computers that run just fine.

I have determined that those motherboards are retarded and am going to upgrade.

BTW, I decided this long ago. Im going to find a way to remove this post so it does not waste anyone elses time.

Thanks again, people.