Re: [Xen-users] lomount + lvm

[John Haxby <john.haxby@xxxxxxxxxx>]

Brian Krusic wrote:
> Forgot, my conf files disk line of interest looks like this;
>
> disk = [ "tap:aio:/var/lib/xen/images/foo.img,xvda,w" ]
First make sure your guest isn't running unless you want to trash its 
file systems.
losetup -f /var/lib/xen/images/foo.img
losetup -a
# Make a note of which device corresponds to /var/lib/images/foo.img,
# 'll call it /dev/loopN but it's probably /dev/loop0
kpartx -va /dev/loopN

You'll get two new entries in /dev/mapper now: /dev/mapper/loopNp1 and 
/dev/mapper/loopNp2.   loopNp1 is /boot (asume you have got a standard 
layout).   loopNp2 is a volume group.   You can just mount
/dev/loopNp1 to poke around the /boot file system.

Now

vgscan

This is where you might come unstuck.   The default volume group for Red 
Hat and similar is "VolGroup00".  If your dom0 is using LVM and so is 
the guest then you'll have do VolGroup00's and that's bad.   The best 
thing to do now is to boot a rescue image in a different domU and rename 
the guest's volume group.  You'll need to undo the kpartx and losetup 
(see below first) and when you've all finished then you'll need to 
either fix up the guest's /boot/initrd*.img, /etc/fstab and 
/boot/grub/grub.conf to hold the new name or you'll have to rename it 
back again in the rescue guest.
Anyway, assuming you don't get a clash:

vgchange -ay VolGroup00

The guest's file systems are now in /dev/VolGroup00 and you can mount 
them as normal.
To undo everything:

1.  umount any file systyems you mounted
2.  vgchange -an VolGroup00
3.  kpartx -d /dev/loopN
4. losetup -d /dev/loopN

And next time you build a system, change the name of its volume group so 
you don't wind up with two systems with the same volume group name!   
And I wish Red Hat had listened to me years ago when I said that 
"VolGroup00" was a really poor idea.
jch



_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users