FC3 cannot boot on master if slave present
I am going to try and be as specific as possible (that means this will be a lengthy post) but this is a fascinating and rather severe (read annoying) issue that I spent a great deal of time diagnosing without a resolution.
I have been running Fedora Core 3 on two hard drives (identical Seagate ST380011A). What I mean by this is that each hard drive was set up exactly the same: grub, fedora core 3, files, everything, such that I could swap drives should either one fail. I do not have a Raid controller. During the installation of FC3 on each drive, only the drive being installed to was present--the operating system on each drive did not "know" about the other. Each hard drive was partitioned identically, and as appears below:
Disk /dev/hda: 80.0 GB, 80026361856 bytes
255 heads, 63 sectors/track, 9729 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Device Boot Start End Blocks Id System
/dev/hda1 * 1 13 104391 83 Linux
/dev/hda2 14 796 6289447+ 83 Linux
/dev/hda3 797 9598 70702065 83 Linux
/dev/hda4 9599 9729 1052257+ 5 Extended
/dev/hda5 9599 9729 1052226 82 Linux swap
For over a week I was able to run one drive as the master and one as the slave using cable select. The bios was set to automatically detect the hard drives via the cable select method. I had no trouble booting the machine with both drives present. Yesterday evening, after having used the computer all day, I rebooted only to find the computer would not boot. The problem was that during the boot process errors appeard that the operating system was halting because the drive was Read Only. I tried booting from the second hard drive, and encountered exactly the same error.
This is what I did.
I set the Bios to stop using cable select, specified all drive settings for master and slave, and set the jumpers to master and slave, then rebooted, trying this with each drive while the other remained present. I encountered the same Read Only Hard Drive error. When I removed one of the hard drives and booted from the other, everything was fine. Even when I returned the bios to autodetecting hard drives and using cable select everything was fine--as long as a second hard drive was not present. I have complete control over the problem: it does not exist when one drive is present, and it exists when both drives are present--and it does not matter which drive is master or slave, or whether cable select is active or I force master and slave upon the hard drives.
I decided to do a test with one of the hard drives and reinstall FC3 on it--a fresh install, surely, there would be no problems. I set that drive as master, and the other as slave, and booted. The OS reported that drive as Read Only and halted. I tried booting the second drive using the new OS drive as slave, and received the same Read Only errors and halted. I tried using cable select and setting the jumpers in every configuration and received the same problem.
The only thing I have not yet tried is formatting one of the drives and then booting from the master and using the blank drive as a slave.
I think I have a hard drive problem, but I won't be able to verify this until I purchase a replacement (I'm looking at obtaining a pair of Maxtor 120GB hdds within a month). The reason I think this is a hard drive problem is because everything was fine until last night. There had been no os changes, no bios changes--and the bios settings are fine--I even did a reset on them via removing the bios battery then setting up once again. It seems unlikely that the data on two hard drives would become corrupt at the same time. Certainly possible, but highly unlikely. And even then, it doesn't matter since I have reinstalled the OS on both hard drives and continue to have no problems provided only one hard drive is present in the machine at a given time (and again, it matters not which hard drive it is).
I understand that complex systems do approach the concept (as we know it) of personality. Is this simply sibling rivalry?
I would appreciate knowing if anyone has any thoughts on the matter, or has heard of similar situations and possible solutions to the problem.
Thank you very much if you have read even this far.
Ariadne
|