Bootsector DATA
 

When I look at the BS data that I've copied with the dd command with less, I can see LILO and GRUB in there- I don't think this is right; I don't think the bootloader is correctly writing itself (or overwriting itself) to the bootsector. If it's not being written correctly, it's not going to be extracted correctly which would explain my problem.

The SuSE 9 install disc can boot Linux, what commands is it using to do so? Any way to check?

Same Problem, different distro
 

MrOuija,

I have the same problem with Fedora Core 1. Due to conflicts with my ATI Radeon 9800 Pro, I had to figure out a way to get the system booting into init 3 mode (w/o X). Obviously, GRUB was the bootloader of choice because the LILO package wasn't even installed at all.

I used the rescue image from the original boot CD to boot into rescue mode, chroot'ed to make /mnt/sysimage the new root folder; then I looked for the LILO package on all three CDs and for sure found it on the last one, installed LILO package; finally I made a copy of a file called lilo.conf.anaconda, which was already prepared by the installation script on the fly through the installation process.

Now, here's what: I have a SATA Raid Volume that hold my Windows XP, and I have a regular ATA drive that is connected to the IDE port on the motherboard. The XP partition is NTFS only, and the IDE drive has the following partitions:

/dev/hda1 -> /boot (100megs, to stay below the 1024 cyl. limit)
/dev/hda2 -> / (30+gigs)
/dev/hda3 -> swap (1 gig)
/dev/hda4 -> extended partition
/dev/hda5 -> fat32 (~8 gigs)

Right. So, on my IDE drive I don't have a primary DOS partition. Okay? Because that would allow the XP installation to put boot.ini on to the IDE drive. That means that ntldr and boot.ini are, in fact, on the SATA Raid Volume in the NTFS partition.

This post made it clear that dd .... bs=512 count=1 is needed, so I extracted the LILO bootloader from the first bootsector of /dev/hda1. With the same lilo.conf that I used to write the loader to the first boot sector, I also created a boot sector on a floppy disk by typing:

lilo -b /dev/fd0

I'd say that at this point, both floppy disk and /dev/hda1 have a valid and exactly the same boot sector on it using LILO. By the way, the directive boot= in the lilo.conf determines the device/partition that holds the boot sector, so changing that directive to point to the floppy disk is wrong in my eyes. I also found that the directive 'linear' breaks the process of putting the bootsector on the drive, so I changed that to 'lba32' as well.

I put the floppy in the drive and the BIOS boots from the floppy; then a prompt appears, I hit enter, and voila: there's my Fedora (booting the kernel image off the /dev/hda1 partition, which is pretty fast).

Then, I use the file that I already copied to my C:\ drive (remember its NTFS on a SATA Raid), I choose "Linux" and the screen goes blank and it says:

LI (blinking cursor)

Long story short, my suggestion to MrOuija is to use a floppy disk ;-))) Anybody reading this post that knows what's going on with ntldr and that 512 bytes long file with the boot sector?

Thanks for any suggestions... ;-)

P.S.: Sorry for the long post and for sounding a bit fuzzy. It's kinda hard to explain what happened on your screen/PC to somebody else. Right ? ;-)

Try the bootpart program I mentioned in the previous post, it works better than extracting the raw bootsector with dd.

I have tried it
 

I've tried just about everything- with no success. Methinks that when the system tries to boot the .lin file from the RAID array- it loses the other half of striped data located on the second physical drive and takes a yit- In other words- Only half of LILO/GRUB/WTF is loading- causing errors, because the system has no clue how to read a RAID partition. This is why I tried moving the bootsector to a FAT32 partition on the Linux drive, again, with no luck. I think Im just going to be booting off the CD for a while :(

Does grub even recognise the drive, ie when u're in the grub console(using a grub bootdisk, no linux) does it recogise your harddisk configuration(does root (hd1,0) etc work). Tried installing grub into the mbr?

Can't
 

The MBR is striped across 2 hard drives- It can't write correctly to it, and thus- other information will be overwritten. Tried this once already.

Re: Can't
 

Really? Are your raid stripes smaller than 512 bytes?

If grub is able to recognise the raid volume, you can try install grub into the mbr of the linux drive, set in bios to boot from it then add appropriate entries to menu.lst to load windows.

No
 

They aren't, not now that I think about it. They're 64k.

Hello
I have a similar problem. My setup is more common (no raid):
One hard disk 80 GB partitioned in:
hda1: 75 Gb FAT32
hda2: 4 Gb ext3 (Debian)
hda3: Swap

I have been using this setup for two years, booting with NTLDR on hda1 and adding the boot sector of hda2 (bootsect.lin) into boot.ini, and everything was ok. I use grub instead of lilo, i hate lilo :rolleyes:

Well, yesterday I resized hda1 to make a new10 gb partition, and linux fdisk changed the volume labels in this way:
hda1: 65 Gb FAT32
hda2: 10 Gb unformatted
hda3: 4 Gb ext3 (Debian)
hda4: swap.

I changed fstab, kernel 'root=' parameter and extracted the new bootsector with dd, but now I can't boot into linux. It hangs with no error, only a blinking cursor. If I chage ative partition to hda3, it doesn't boot either because hda3 it's over 2048 hard disk cylinder and 2 gb limit.

So now I can only boot using loadlin.

I can't understand why doesn't work now if it have been right for two years, doing the same process. I hope someone see the mystery.

Thanks.