8.04 -> 10.04 2nd machine kernel panic - not syncing VFS

[rsteinmetz70112]

This is the second machine I have upgraded from 8.04 -> 10.04 Both machines have identical hardware both machines have the same files system layout.

The Root file system is a Logical Volume of software Raid /boot is reiserfs

When I boot I get the following message;


kernel panic - not syncing VFS: unable to mount rootfs on unknown - block (0,0)

When I look a the Grub menu it has the following entries;


root (hd0,0)
kernel /vmlinuz-2.6.32-23-generic root=/dev/mapper/romulus-root ro quiet splash

There is no initrd line and I do not get an initramfs shell when it fails, like I did with the first machine.

The Grub menu for the other machine is different. It points root (hd0,1) and has an initrd entry. If I add those manually I get a file not found error.

I can boot a Live CD and using the noraid option assemble the arrays and recover the volume groups.

Any ideas on how to proceed from here?

[rsteinmetz70112]

One Part Solved.

The upgrade did not add the initrd line to the GRUB menu.lst file.

Now the machine behaves like my other one did, It fails to assemble the arrays and recover the volume groups so it drops into the initramfs shell.

[rsteinmetz70112]

Another Piece of the picture. The arrays are not assembled.

After I drop into the initramfs shell if I run

(initramfs) mdadm -A --scan
(initramfs) exit
Things then go as planned. It appears that the initrd script is looking for the volume groups before they are available because the arrays they exist on have not been assembled.

The quick fix seems to be to add

mdadm -A --scan
to the file

/usr/share/initramfs-tools/scripts/local-top
and run

# update-initramfs -u
Does anyone know why this isn't a good idea? Can someone please point me to an explanation of the order the initial script is assembled in?

There seems to be a bug in the upgrade process used to assemble the initial Grub menu and in the initial initramfs when Logical Volumes exist on top of RAID Arrays.