I have a system with two promise ultra 133 tx2's in it. For what ever reason it will not boot when I hook a hard drive up to these. I did the RHEL 4 install with the cards in the system, but no hard drives hooked up to them. Now when I reboot and plug a hard drive into one of them it gets to the grub screen then just stops. It doesn't seem froozen, since I can press the caps lock and numlock buttons and the lights go on and off respectivly, but it doesn't do anything. The screen says :

Booting 'Red Hat Enterprise Linux WS (2.6.9-22.EL)'

root (hd0,0)
Filesystem type is ext2fs, partition type 0x83
kernel /vmlinuz-2.6.9-22.EL ro root=LABEL=/1 rhgb quiet
[Linux-bzImage, setup=0x1400, size=0x1644e8]
initrd /initrd-2.6.9-22.EL.img
[Linux-initrd @ 0x37f77000, 0x78694 bytes]


And then it just sits after that.

When I boot with no drives attached to the two controller cards it works fine. And whe I do lspci it dectects both cards correctly. What else do I need to do in order to get these to work??

Provided that you removed all hard drives from your on-board IDE controller, I would have expected what you did to work, since you are using the LABEL= method to identify the partitions.

One thing that I have noticed when using more than one Promise Ultra100/133 TX2 card in a system is that which card will get the lower device names (i.e., hde, hdf, hdg, hdh) is not always obvious from one motherboard to another.

One theory about your problem would be that you may have placed what you think is (hd0) on the “second” Promise card and that it is really (hd1), (hd2), etc., assuming you have hard drives on both cards.

Now for some diagnostics. Connect your drives to the cards and then boot into rescue mode using your installation media (boot: linux rescue). Don’t search for or mount the installation (although you may want to do that later to fix any problems).

From the prompt, run “fdisk -l” and/or “fdisk -l /dev/hd[e-l]” to verify that the drives are recognized and to see which drive letters they are identified by.

You may find that you need to reorder your drives on the cards to move what you think is (hd0) to the correct position.

If the order seems to be correct, then boot into rescue mode again and search for/mount the installation (i.e., chroot) and change all LABEL= references in /boot/grub/grub.conf and /etc/fstab to the device names. For example, root=LABEL=/1 might be changed to something like root=/dev/hde2, based on what you see using fdisk.

Be sure to back up both files before you modify them (i.e., “cp grub.conf grub.conf.backup").

Wait, so I can't use the onboard ide controller card and the ultra 133tx2 at the same time? I was under the impression that this was possible.

also, about the LABEL=, what exactly is the purpose of having that. I don't remember doing anything different but I see that it is in the fstab:
LABEL=/1 / ext3 defaults 1 1
LABEL=/boot /boot ext3 defaults 1 2
etc...

how does it know which device (sda1,hdb3, etc) I want these partitions to be?

Sure, you can use as many of the onboard controller and the Promise card IDE channels as you want. But when you relocate some of the drives to the Promise card and leave some on the motherboard controller, the BIOS drive order probably changes.

The physical drive that was (hd0) might become (hd3), for example. Grub gets “confused” when that happens, because it will be looking for your stuff on the wrong drive. So you need to manually modify /boot/grub/grub.conf to reflect the new order.

You probably need to go through and reconfigure/reinstall grub, too.

The installer created ext2 labels for the partitions (e.g., /1) and has used those labels instead of the actual device names. Personally, I prefer using the device names.

If you revert to using device names, then you need to modify /boot/grub/grub.conf and /etc/fstab and replace the LABEL= designations with the actual device names.

You need to rephrase the question, because I don’t understand what you want to know.

But the answer will probably be “Connect the drives to the Promise cards and run “fdisk -l” from rescue mode.”

Redhat likes to use labels instead device nodes because it is easier for SCSI users to add new hard drives with out messing up the setup. Setting labels for IDE devices is optional. Labels are set during formatting.

IDE devices range from /dev/hda to /dev/hdz so you can use as many controllers as you want.

You will have to run mkinitrd if Redhat did not include the module to load up your controllers in the initrd file.

just a thought- this computer has two sata drive hooked up to it as well. RHEL4 is installed on those two sata drives, is it possible that they are conflicting with the drives I try to plug into the controller card?

I tried plugging the two ide drives that are normally plugged into the motherboard into the controller card, and it gave the same results, doesn't this mean that the problem is something else? Since its trying to boot from the sata drives, it really shouldn't matter as far as booting goes where the ide drives are plugged into. To me this seems like its something else, but I'm not sure.

Depending on the SATA controller that you are using. The module (driver) can use either IDE node layout (/dev/hda to /dev/hdz) or SCSI node layout (/dev/sda to /dev/sdz). If it is SCSI, you have to use labels for each partition and have permanent device for the /boot partition that stores grub's config. IDE controllers are pernament even if you add new hard drives. The controller on the motherboard always comes first unless there is a BIOS option to change this behavior.

At the grub screen try going into console mode. Then type root (hd<tab>. <tab> means hit the tab key and not actually typing the word <tab>. You will get a list of devices. After you select a device, you can get a list on the partition. You may have to repeat root (hd<tab> to find the correct device. By typing root (hd0,<tab> will print a list of partitions for the first hard drive. The grub console can help you view files, but I suggest that you do not write to any files.

Take out quiet at the kernel line. It does help to find out what the kernel is doing. It is probably probing other IDE or SATA devices that are not yet connected. Usually it is IDE. On IDE channels that are not connected include hdX=noprobe. X designates the device such as hda=noprobe for the primary master.

OK, so far I've tried pretty much all of the suggestions in this thread so far and its still a no-go. One thing I did notice today: I unplugged the two sata drives (one of those hat the /boot partition) and had an ide cdrom plugged into the onboard controller and an ide drive plugged into one of the pci cards, and tried going into rescue mode in linux. This froze the computer. Doesn't this seem like something else is going on besides it looking for the wrong partition? I think I'm going to try installing the updated bios to one of the pci cards to see if that makes a difference.

edit- at this point I think this is just a compatability problem between the motherboard and these pci cards. I put one of the pci cards in a different computer and hooked an ide drive up to it and its working like it should.

well apparently it is just a software thing. I connected a drive to the pci card and loaded up knoppix and did an fdisk -l. And there it listed /dev/hde as the drive that was connected. I'm really at a loss now. I tried adding hde=noprobe to the grub line and that didn't help. Any other suggestions?