Working with Multiple LVM Hard Disks under Fedora Core Linux

by Marion Bates <mbates at whoopis.com> and Josh Pyke <josh dot pyke at dartmouth.edu>

Scenario: You have a hard disk from one Linux computer, and you want the data off of it. So you attach it to another Linux machine. But, all hell breaks loose because both hard disks have the same Fedora default Logical Volume Management (LVM) configuration, same volume names, etc. so you can't mount them both at the same time. What to do?

This howto documents one possible method which worked for us, but may not necessarily be the best. For example purposes, it assumes that, to begin with, both host's LVM groups are named "VolGroup00", and that our approach is to rename the host's boot disk, leaving the secondary attached drive (the one we want to recover) untouched.

With the secondary, to-be-recovered drive NOT attached, boot the Fedora Core installer disk into rescue mode ("linux rescue"). When offered the chance to auto-detect and mount drives, select "skip."

At the root shell, type:
```
lvm
```
and you will be in the Logical Volume Management shell. Then:
```
vgscan
```
The output will look something like this:
```
  Reading all physical volumes.  This may take a while...
  Found volume group "VolGroup00" using metadata type lvm2
```
Now:
```
vgrename VolGroup00 VolGroup99
```
We have now renamed the boot device's volume group to something other than VolGroup00. Exit lvm.

Make a mountpoint with the same hierarchy as your newly-named group, and mount the logical volume containing your / partition:
```
mkdir -p /mnt/VolGroup99/LogVol00 
mount -t ext3 /dev/VolGroup99/LogVol00 /mnt/VolGroup99/LogVol00 
```

Now, note that if your initial install created a separate /boot partition (which it probably did, for Fedora default install), then you will need to mount that as well, in the old-fashioned way:
```
mkdir /mnt/hda1
mount /dev/hda1 /mnt/hda1
```
Note that this assumes the usual, a single IDE hard disk. If you're unsure of which partition contains /boot, run
```
fdisk /dev/hda
p (for print)
```
The first, and smallest, partition with the asterisk in the Boot column is the one. Note the device name and mount that as noted above. E.g. if your disk is SCSI, it will probably be /dev/sda1.

Now make the requisite edits to /etc/fstab and /etc/grub.conf. In fstab, find the lines:

/dev/VolGroup00/LogVol00 /                       ext3    defaults        1 1
...
/dev/VolGroup00/LogVol01 swap                    swap    defaults        0 0

and change them to:

/dev/VolGroup00/LogVol00 /                       ext3    defaults        1 1
...
/dev/VolGroup99/LogVol01 swap                    swap    defaults        0 0

NOTE: "Joe" writes: "in Step 5, I think you meant:

       ...
       and change them to:
       /dev/VolGroup99 ...
                    ^^ this was 00

He is probably right...I no longer remember! Govern yourself accordingly. :)

Then in /boot/grub/grub.conf, fine the line:

kernel /vmlinuz-2.6.19-1.2911.fc6 ro root=/dev/VolGroup00/LogVol00 rhgb quiet acpi=off

and change it to

kernel /vmlinuz-2.6.19-1.2911.fc6 ro root=/dev/VolGroup99/LogVol00 rhgb quiet acpi=off

Of course, your vmlinuz version and other kernel arguments may differ.

Delete or rename the lvm config's cache file:
```
rm /etc/lvm/.cache
```

Now, fix initrd. This is the ugliest part of the process, but it looks more complicated than it is. Basically, the initrd image (init RAM disk) that Fedora creates at the time of initial install, contains hard-coded references to your logical volume group(s). So you need to edit those, and it's multiple steps because the initrd is a disk image in gzipped archive format. Soooo....thanks to someguy's howto [url]:
```
cd /mnt/hda1
mkdir newinit
cd newinit
gunzip -c ../initrd-2.6.9-1.724_FC3.img | cpio -idmv
          # or whatever initrd version file you have
```
Once that finishes, you've got an unpacked version of the initrd. Edit the "init" file, which is a nash script (NOT a normal bash shell script, so don't get cute with bash stuff). Find all instances of "VolGroup00" and change them to "VolGroup99". Save and exit.
Then, repackage up a new image. First rename the original one out of the way to save it so you can restore it later if need be. Then, still within the "newinit" directory, do:
```
 find . | cpio --quiet -c -o > ../newinitrd
 cd ..     
 		# You are now in /boot, aka /mnt/hda1
 gzip -9 < newinitrd > initrd-2.6.9-1.724_FC3.img 
 		# or whatever it was called before...EXACTLY
```

Type
```
sync
unmount /mnt/hda1
unmount /mnt/VolGroup99/LogVol00
sync
```
Exit the shell to reboot. If all goes well...your system should boot up normally.

Now (whew!), you can finally (try to) mount that secondary hard disk. Attach it (reboot if necessary) and do:

lvm (to enter lvm shell)
vgscan

	Reading all physical volumes.  This may take a while...
	Found volume group "VolGroup99" using metadata type lvm2
	Found volume group "VolGroup00" using metadata type lvm2

Remember, 99 is your newly-renamed, main, boot drive. VolGroup00 is the secondary, to-be-recovered drive.

Still in the lvm shell, type:

vgchange --available y

	2 logical volume(s) in volume group "VolGroup99" now active
	2 logical volume(s) in volume group "VolGroup00" now active

Exit lvm and do:

mkdir -p /mnt/VolGroup00/LogVol00
mount -t ext3 /dev/VolGroup00/LogVol00 /mnt/VolGroup00/LogVol00

and your secondary disk's contents should now be available under that mountpoint.

NOTE:

References:

Josh Pyke.

Deron Meranda's post on the Fedora list, re: editing initrd.

TLDP's LVM Howto.