There is a good chance that it has to due with your grub.conf configuation.  What I would do is get both the SCSI and PERC installed and seeing file systems when booting from SCSI.  Then copy the contents of your root volume from the SCSI file system to the RAID file system.  Do the same for all other file systems that are on the SCSI device.  Determine the new root volume device name, ie /dev/sdb2.  Then add a grub.conf stanza entry to match that new device name and reboot.  I would keep a copy of the old grub stanza just to ensure you have a bootable system.

Once your system boots, you can remove the old SCSI controller, and update your grub.conf kernel SCSI device again if needed. 

Dear Eric,

Thanks a lot for your reply.

I have tried the way you said, but didn't success. Do you think it is possibly because of something protection from the server such as ESCD(Extended System Configuration Data Area)?? Because when i tried to cloning the entire Linux system in another PC, i have no problem with it. All the problem appear when i cloning the system in DELL PowerEdge 8450.

Which one is better, Cloning Drive (Norton Ghost, Acronis True Image) or manual copy the content?

May let me tell you the procedure i have done, so you can get clearly :

1. Cloning the original disk to new drive, with the same capacity.(to make sure the original disk is safe. i don't want to take any risk). The clone success, but when i start the new drive that has cloned, the GRUB cannot appear. So i get the rescue disk to start the system and overwrite the GRUB loader. The system start normally.

FYI, i didn't find this kind of error when i simulate on another system (non-branded server).

2. Install MegaRAID PERC3/DC to the server. Linux can detect the driver automatically.

3. Make a RAID-1 drive from 2 new HDD, and the copy the Drive(on step 2) to the new RAID partition. I got message "GRUB READ ERROR". I also didn't find this error when i simulate on another server.

For this error, i also have rescue disk and edit grub.conf to start for RAID disk. But didn't success. Some times i got message "kernel panic"

So do you think what should i do?

Thanks a lot

I still believe the problem is with grub.  Either by cloning or manually coping the data should result in successful migration to the RAID controller.  The trick is the procedure of doing so.

If you can successfully boot the system using a rescue floppy disk and recognize the boot and root partitions, then you can probably bring up the system quickly.

I would first start by recreating the MBR using grub-install.  This will remake the stage 1 and stage 2 boot loaders that Grub will use on the new disk.

I would then ensure that the kernel stanza has the correct root partition declared.  If you are using a Label in grub, I would change it to reflect the actual SCSI device the root partition is on.  i.e.  /dev/sda2

Lastly I would double check the modules.conf to ensure the RAID controller module is listed as the first SCSI controller, and then remake the initrd.

Since you are able to boot using a floppy disk, the problem most likely is grub recognizing the boot device properly.

1. ensure bios is enabled

2. then post what you have in fdisk /l, /proc/scsi/scsi, and modules.conf

dear Eric,

i've always tried the way as u told me. But it really didn't works in PowerEdge8450. After reboot, system only displaying "GRUB READ ERROR".

Then i try another way, by removing GRUB from MBR, and then reinstall GRUB in SDA1 sector 0. And... it works. But then i wanna know, can you tell me the difference for the two way? Does it have any pros and cons to install GRUB in MBR or sector 0? Do you prefer in MBR or sector 0?

Other question, can you give me suggestion where to find the PERC3/DC driver for RedHat 7.2 kernel 2.4.7-10smp and FreeBSD 4.11?   i need it for other PowerEdge 8450 server.

Thanks a lot.