fixing LILO from a rescue shell

[Vdubjunkie]

Ok, so I finally broke down and decided to upgrade my server from 8.2 to 9.0. Everything installed fine, that is until I had a nice brain fart and agreed to let it try to boot from /dev/hda. My /boot (as well as all other partitions of any use to the OS) are on /dev/hdd.

 

No problem, right? I'll just use the rescue disk to get to lilo.conf and adjust it, then run /sbin/lilo. Well, since the rescue disk mounts everything to /mnt I can edit /mnt/etc/lilo.conf, but when I run /mnt/sbin/lilo it complains because nothing is where it is supposed to be. It doesn't know to prepend /mnt to everything.

 

Next I read to drop right to a shell in the rescue disk and type "chroot /mnt" which is supposed to mount everything to / as it should be. Well, that failed too :mad: Then I read on and saw where one guy said after upgrading that his /boot/initrd-<kernel version> file didn't get created. Guess what, neither did mine. In fact the old one got wasted too. So, the symlink "initrd" points to nothing, and I should use mkinitrd to create the new initrd-<kernel version>, right? Sure, except that the kernel version folder isn't where it expects it to be. I backed up mkinitrd and edited it so that everything I could find in there referenced /mnt..., but it still failed looking for things under their native location.

 

HEEEELP!!!! I don't know what to do. I've even tried to re-upgrade hoping LILO would get fixed, but no luck.

 

[bvc]

In rescue mode, did you first choose 'mount your partitions under mount'? Then do chroot /mnt, IF that is where your lilo stuff is (/etc/lilo.conf and /sbin/lilo).

 

What do you mean..."agreed to let it try to boot from /dev/hda"? let what?

 

You should also be able to boot without the initrd.img....just take it out of the kernel boot command if you can successfully rin lilo. This is exactly why I use grub...much easier to edit the boot command and there's no need to run anything after editing its config file. It's all rt there to see and edit to your liking :wink: If you can get everything mounted properly you can re-run mkinitrd.

[Vdubjunkie]

Actually, no I didn't mount first prior to trying chroot /mnt. The article I read didn't specify that. When I get home in the morning I will try that to see if it works.

 

Oh yeah, I guess I was kind of vague. Right toward the end of the install procedure the install process asks you what device to boot from and offers /dev/hda as the default. Funny thing there was that after I edited the lilo.conf file and reran the upgrade it had /dev/hdd as the default.

 

I did not know initrd.img wasn't necessary. However if I can successfully run lilo, I would imagine I would be able to run mkinitrd as well, right? That is good to know though, thanks. Perhaps I should do some hardcore boot process edumacation.

 

This is what I love about problems in Linux. You figure out so much in the process of "fixing" it. :P

[bvc]

after I edited the lilo.conf file and reran the upgrade it had /dev/hdd as the default

because the 'Upgrade' upgraded a previously found install and modified accordingly.....(I guess :P )

 

I would imagine I would be able to run mkinitrd as well, right?

Yes

 

This is what I love about problems in Linux. You figure out so much in the process of "fixing" it.

yea, I've had several friends, that always called me for help or wanted me to come over to fix stuff. After a while I just told them they needed to TRY and fix it themselves.....and if they screwed up, fix that, and if they screwed that up....and so on. There's no better way to learn than to break it and HAVE to fix it, IMO. I used parted to try and 'rescue' my partition table....it didn't work... so the option 'mklabel' sounded applicable....hehehe, I lost everything with that one command (thought you might get a laugh out of that one :wink: )

[Steve Scrimpshire]

This one time....at band camp....er, I mean in Windows...I had this program that could password-protect files/directories (maybe it was PowerDesk) and so, I decided that it would be a good idea to password-protect the whole C: drive using it. Everything was ok until I rebooted, then I got:

 

Non-system disk or disk error

Remove disk and hit any key to continue.

 

And there was no disk in any drive. I had no rescue disk and no backups. Fun fun fun.

[Vdubjunkie]

ok. that was cool, i was able to run mkinitrd (which seemed to be caught in an endless loop with an error so i killed it) and lilo. The problem is this. Lilo complained that it could not find /dev/hdd. Truly, /dev/hdd didn't exist anymore. Neither did /mnt/dev/hdd for that matter. Aaaargh!!!

What do I do now?

[Steve Scrimpshire]

You will have to edit /etc/lilo.conf and change or remove any of the entries that refer to /dev/hdd. then run lilo -v

 

vim /etc/lilo.conf

[hit 'i']

[do your editing]

[hit esc, then :wq and hit enter]

lilo -v

[Vdubjunkie]

You will have to edit /etc/lilo.conf and change or remove any of the entries that refer to /dev/hdd.

 

Well, as it turns out I had a disk in the machine which used to run NT4. I haven't booted to NT4 in.... lemme see now... a REALLY long time, so I simply pulled a new install on that drive. Cool thing was it found and installed my scsi raid array. Anyway, what I am wondering is, since I do still have the ability to try to fix the original install, by removing all reference to the partition it is supposed to boot to, how is it going to know where to boot?

 

I do afterall prefer to figure out how I could have fixed this problem instead of pulling a reinstall as I did.

[iphitus]

Actually thae same problem with lilo happened to me yesterday. I deited Lilo.conf appropriately and tried lilo -v. Wouldn't work complained.

Then i just went into the rescue menu told it to restore Lilo and it did with the edited lilo.conf

 

Try editing /mnt/etc/lilo.conf

Then just select "Restore lilo" or waht it may be in the rescue GUI

 

Worked for me

James