First off, I'm a bit of a Linux newb, but have a good amount of Windows experience. Here's the specs on the system:

Tyan Thunder S2885
Dual Opteron 248
4GB Corsair PC2700 Reg. ECC
4x160GB WD Hard Drives
Adapctec 2410SA Raid 5 setup
Radeon 9600XT

I had to install the software on a USB DVD drive, because the drive in the system is SATA and isn't detected. The USB drive is detected fine, as is the RAID setup. I format and put Grub on the boot sector, because I believe putting it on the MBR is bad with a RAID 5 setup. The system proceeds to hang everytime starting the graphical Anaconda installer. Try the text installer, and no problems. Figure I'll leave it because once the software is fully installed, it probably won't have problems with the graphics anymore.

After the full install and reboot, I get to Grub just fine, and pick the SMP kernel. As predicted, now that the drivers are installed, the graphical install comes up without a problem, for me to accept the license agreement, set the date/time, etc. I click "Yes", and then click "Next" to bring up the time/date setup. The system hangs everytime it tries to bring up the time/date setup. Found bad CMOS battery from factory. Re-install to be safe, and try again, same problem.

Thinking there might be more hardware problem than just the battery, I try Suse 9.1 Pro install. Graphical YaST installer works fine, and indeed gets past Grub and everything else, to work just fine the entire time. Can't use ATI drivers, cause they only work with 2.4 kernel, but meh, it proves that Linux can make it onto the machine.

I try RedHat again, try MBR instead of boot sector, same problem. Install again, back on boot sector, and go to rescue mode. Find that the entire file system is read-only, and even as root, can't be changed to read/write. Is that normal if you haven't made it past the full install? One of the guys here got it to work on his AMD 64 pc at home, so it seems like the distro is okay. What are we doing wrong???

Also, even though Suse ran, we really can't use it. This is going to be the head of a cluster, and apparently the whole cluster community always uses RedHat. So that others can reproduce our results, we really really really want to use RedHat. Any and all help will be appreciated.

It sounds like a mainboard or RAM issue, the question is whether it's hardware or software.

I would consider running MemTest to eliminate the possibility that the RAM is faulty. Then I would check Google, and Red Hat's HCL and Bugzilla to see if there are known issues with that model of board or chipset. As a last resort I would make sure that the BIOS was the latest version. After that, don't take the chance - replace it.

Different distributions use different driver sets and hardware discovery routines, so where the hardware is flaky you may find that an install will succeed with one distro and fail with another. Since you've already found one fault with the board I think that you do need to treat it as suspect.

Thanks for the reply! Checked RedHat's HCL, and it's 100% compatible, nothing in Bugzilla that is like it, that I could find. It is the newest BIOS, but I'll try replacing it anyways, like you said. Also contacting RH tech support right now. I'll try to the memtest, but how long does that take?!?! I did it on a box with only 768mb, starting it when I left on Friday, and it was still running when I came in Monday morning. Is 4gb going to take a few weeks? Although, it's not like it's going anywhere without the os installed...

I'm basically willing to do anything to not have to pry the mobo out of there. It was tight enough of a fit in the first place, and with thermistors on the chips and everything, it'll be a monster PITA. I really don't care if the BIOS can't keep the time and date if we unplug it. If I can get it up and running, without any random reboots or anything, I'll take it, broken time/date and all. I'm almost starting to wish that I had just bought a box pre-built from Monarch.....almost.

MemTest cycles continously through it's tests until you exit. I've haven't used it in a while but I think it took about an afternoon to go through all of the tests.

The risk is that you have a fault with the board, and such faults get worse over time. So it's more convenient to swap the hardware out now than after you've put it into service and hit problems six months on. The clock chip isn't really relevent except that it means the board came out the factory with at least one fault (so there may be more), and that you can get the board replaced with a fresh one now without the supplier arguing.