Re: [Xen-users] Re: How to Backup and Restore MBR within Logical Volumes?

[Mr. Teo En Ming (Zhang Enming)]

On Thu, Nov 12, 2009 at 6:37 PM, Fajar A. Nugraha  wrote:
> On Thu, Nov 12, 2009 at 5:10 PM, Mr. Teo En Ming (Zhang Enming)
>  wrote:
> > Hi,
> >
> > I don't think I will be using the LVM snapshot method. I will simply
>
> ensure
> > that my domU is not running,
>
> Ah, so you can live with the down time. It makes things a lot simpler then.
> > then I proceed to do the losetup and kpartx
> > procedure.
>
> Like others mentioned, you don't need losetup for LVs. It will only
> make things slower.
> > Subsequently I will backup the MBR of my domU by dd-ing
> > /dev/loopX and clone all the filesystems of the domU using fsarchiver.
> >
> > I would like to use fsarchiver because it can support LZMA compression.
>
> LZMA
> > can compress better than bzip2 and decompress 3x faster than bzip2.
> > partimage only supports gzip and bzip2.
>
> If your concern is the size and speed of backup, here's one final note
> from me: you might want to look at zfs. Possible use scenarios :
> use opensolaris as dom0. Probably the most supported option when it
>
> comes to xen + zfs, but you need to be familiar with how opensolaris
> works.
> use external storage server running solaris/opensolaris as iscsi
>
> SAN. Think of it as poor-man's Netapp.
> use zfs-fuse inside Linux dom0, and store domU storage as files
> use zfs-fuse on domU, and do backup from domU instead of dom0.
>
> With zfs, you'd get zfs snapshot (which is like instaneous in-place
> backup), incremental block-level send (can greatly reduce backup
> storage size if your data doesn't change often), optional transparent
> compression for live data (like what ntfs has).
>
> --
> Fajar

I am not familiar with OpenSolaris but that may change with time.

[Xen-users] Re: How to Backup and Restore MBR within Logical Volumes?

[Mr. Teo En Ming (Zhang Enming)]

On Thu, Nov 12, 2009 at 8:53 PM, Mr. Teo En Ming (Zhang Enming) <
[e-mail]> wrote:
> On Thu, Nov 12, 2009 at 6:37 PM, Fajar A. Nugraha  wrote:
> > On Thu, Nov 12, 2009 at 5:10 PM, Mr. Teo En Ming (Zhang Enming)
> >  wrote:
> > > Hi,
> > >
> > > I don't think I will be using the LVM snapshot method. I will simply
> >
> > ensure
> > > that my domU is not running,
> >
> > Ah, so you can live with the down time. It makes things a lot simpler
> > then.
> > > then I proceed to do the losetup and kpartx
> > > procedure.
> >
> > Like others mentioned, you don't need losetup for LVs. It will only
> > make things slower.
> > > Subsequently I will backup the MBR of my domU by dd-ing
> > > /dev/loopX and clone all the filesystems of the domU using fsarchiver.
> > >
> > > I would like to use fsarchiver because it can support LZMA compression.
> >
> > LZMA
> > > can compress better than bzip2 and decompress 3x faster than bzip2.
> > > partimage only supports gzip and bzip2.
> >
> > If your concern is the size and speed of backup, here's one final note
> > from me: you might want to look at zfs. Possible use scenarios :
> > use opensolaris as dom0. Probably the most supported option when it
> >
> > comes to xen + zfs, but you need to be familiar with how opensolaris
> > works.
> > use external storage server running solaris/opensolaris as iscsi
> >
> > SAN. Think of it as poor-man's Netapp.
> > use zfs-fuse inside Linux dom0, and store domU storage as files
> > use zfs-fuse on domU, and do backup from domU instead of dom0.
> >
> > With zfs, you'd get zfs snapshot (which is like instaneous in-place
> > backup), incremental block-level send (can greatly reduce backup
> > storage size if your data doesn't change often), optional transparent
> > compression for live data (like what ntfs has).
> >
> > --
> > Fajar
>
> I am not familiar with OpenSolaris but that may change with time.
>
> --
> Mr. Teo En Ming (Zhang Enming) Dip(Mechatronics) BEng(Hons)(Mechanical
> Engineering)
> Alma Maters:
> (1) Singapore Polytechnic
> (2) National University of Singapore
> My Primary Blog: http://teo-en-ming-aka-zhang-enming.blogspot.com
> My Secondary Blog: http://enmingteo.wordpress.com
> My Youtube videos: http://www.youtube.com/user/enmingteo
> Email: [e-mail]
> Mobile Phone (Starhub Prepaid): [phone]
> Street: Bedok Reservoir Road
> Country: Singapore

Hi,

I have just finished writing my backup/cloning script. Could anyone help me
verify whether my backup script will work?

<SCRIPT>

#!/bin/sh

# Script to Backup/Clone Xen Host/Dom0 and all DomUs which are using Logical
Volumes as Virtual Hard Disks

# Written by:

# Mr. Teo En Ming (Zhang Enming) Dip(Mechatronics) BEng(Hons)(Mechanical
Engineering)
# Alma Maters:
# (1) Singapore Polytechnic
# (2) National University of Singapore
# Primary Blog: http://teo-en-ming-aka-zhang-enming.blogspot.com
# Secondary Blog: http://enmingteo.wordpress.com
# Youtube Videos: http://www.youtube.com/user/enmingteo
# Email: [e-mail]
# MSN: [e-mail]
# Mobile Phone (Starhub Prepaid): [phone]
# Street: Bedok Reservoir Road
# Country: Singapore

# First written: 13 November 2009 Friday 12:38 A.M. Singapore time
# Last updated: 13 November 2009 Friday 12:38 A.M. Singapore time

# REFERNCE: "Geek Sheet: Bare-metal backup and recovery", May 7th, 2008,
Jason Perlow
# URL: http://blogs.zdnet.com/BTL/?p=8759

# Usage Instructions:

# Boot up your computer/server with System Rescue CD version 1.3.2 for
i386/amd64.
# Then execute this backup script. You need to adapt this script to work for
your environment.
# Download System Rescue CD from http://www.sysresccd.org/Main_Page

###############################################################################################################
# Declare Variables
###############################################################################################################

HARDDISK=/dev/sda
DEST=/media/hitachi/test

###############################################################################################################
# Cloning Xen Host/Dom0
###############################################################################################################

# Backup MBR

dd if=$HARDDISK of=$DEST/f11-xen-dom0-sda.mbr bs=512 count=1

# Activate all logical volumes in all volume groups

vgchange -ay

# Backup UUIDs of PVs

pvdisplay > $DEST/f11-xen-dom0-pvdisplay.txt

# Backup LVM Metadata

# Backup the configuration of dedicated volume group for dom0

vgcfgbackup -d -v vg_fedora11_host -f $DEST/vg_fedora11_host.vgcfg.backup

# Backup the configuration of dedicated volume group for domUs

vgcfgbackup -d -v virtualmachines -f $DEST/virtualmachines.vgcfg.backup

# Backup /boot partition and all logical volumes of dedicated volume group
for host/dom0

fsarchiver savefs -v -z 9 -j 2 $DEST/f11-xen-dom0-filesystems.fsa /dev/sda1
/dev/vg_fedora11_host/lv_home \
/dev/vg_fedora11_host/lv_root /dev/vg_fedora11_host/lv_var

###############################################################################################################
# Cloning Xen-based Virtual Machines/DomUs/Guest Operating Systems/VMs
###############################################################################################################

# The following 8 virtual machines have physical volumes, so it is not
possible to image using partimage or
# fsarchiver directly. The only convenient way is to dd the logical volume
for the VM, ie. sector by sector
# copy. Other methods would be very tedious.

# VM 1: Fedora 11 x86_64 PV domU

dd if=/dev/virtualmachines/f11-pv-hpc-node0001 | gzip --fast >
$DEST/f11-pv-hpc-node0001.img

# VM 2: Fedora 11 x86_64 PV domU

dd if=/dev/virtualmachines/f11-pv-hpc-node0002 | gzip --fast >
$DEST/f11-pv-hpc-node0002.img

# VM 3: Fedora 11 x86_64 PV domU

dd if=/dev/virtualmachines/f11-pv-hpc-node0003 | gzip --fast >
$DEST/f11-pv-hpc-node0003.img

# VM 4: Fedora 11 x86_64 PV domU

dd if=/dev/virtualmachines/f11-pv-hpc-node0004 | gzip --fast >
$DEST/f11-pv-hpc-node0004.img

# VM 5: Fedora 11 x86_64 PV domU

dd if=/dev/virtualmachines/f11-pv-hpc-node0005 | gzip --fast >
$DEST/f11-pv-hpc-node0005.img

# VM 6: Fedora 11 x86_64 PV domU

dd if=/dev/virtualmachines/f11-pv-hpc-node0006 | gzip --fast >
$DEST/f11-pv-hpc-node0006.img

# VM 7: Fedora 11 x86_64 PV domU

dd if=/dev/virtualmachines/f11-pv-hpc-node0007 | gzip --fast >
$DEST/f11-pv-hpc-node0007.img

# VM 8: Fedora 11 x86_64 PV domU

dd if=/dev/virtualmachines/f11-pv-hpc-node0008 | gzip --fast >
$DEST/f11-pv-hpc-node0008.img

# VM 9: FreeBSD 8.0 RC2 UNIX amd64 HVM domU

dd if=/dev/virtualmachines/freebsd | gzip --fast > $DEST/freebsd.img

# VM 10: OpenSolaris 2009.06 UNIX amd64 PV domU

dd if=/dev/virtualmachines/opensolaris | gzip --fast > $DEST/opensolaris.img

# VM 11: Rocks 5.1 x86_64 HPC Compute Cluster HVM domU

kpartx -av /dev/virtualmachines/rocks0001

dd if=/dev/virtualmachines/rocks0001 of=$DEST/rocks0001.mbr bs=512 count=1

partimage -d -M -b -z1 save /dev/mapper/virtualmachines-rocks0001p1
$DEST/virtualmachines-rocks0001p1.img

partimage -d -M -b -z1 save /dev/mapper/virtualmachines-rocks0001p2
$DEST/virtualmachines-rocks0001p2.img

partimage -d -M -b -z1 save /dev/mapper/virtualmachines-rocks0001p5
$DEST/virtualmachines-rocks0001p5.img

kpartx -dv /dev/virtualmachines/rocks0001

# VM 12: Rocks 5.1 x86_64 HPC Compute Cluster HVM domU

kpartx -av /dev/virtualmachines/rocks0002

dd if=/dev/virtualmachines/rocks0002 of=$DEST/rocks0002.mbr bs=512 count=1

partimage -d -M -b -z1 save /dev/mapper/virtualmachines-rocks0002p1
$DEST/virtualmachines-rocks0002p1.img

partimage -d -M -b -z1 save /dev/mapper/virtualmachines-rocks0002p2
$DEST/virtualmachines-rocks0002p2.img

partimage -d -M -b -z1 save /dev/mapper/virtualmachines-rocks0002p5
$DEST/virtualmachines-rocks0002p5.img

kpartx -dv /dev/virtualmachines/rocks0002

# VM 13: Rocks 5.1 x86_64 HPC Compute Cluster HVM domU

kpartx -av /dev/virtualmachines/rocks0003

dd if=/dev/virtualmachines/rocks0003 of=$DEST/rocks0003.mbr bs=512 count=1

partimage -d -M -b -z1 save /dev/mapper/virtualmachines-rocks0003p1
$DEST/virtualmachines-rocks0003p1.img

partimage -d -M -b -z1 save /dev/mapper/virtualmachines-rocks0003p2
$DEST/virtualmachines-rocks0003p2.img

partimage -d -M -b -z1 save /dev/mapper/virtualmachines-rocks0003p5
$DEST/virtualmachines-rocks0003p5.img

kpartx -dv /dev/virtualmachines/rocks0003

# VM 14: Rocks 5.1 x86_64 HPC Compute Cluster HVM domU

kpartx -av /dev/virtualmachines/rocks0004

dd if=/dev/virtualmachines/rocks0004 of=$DEST/rocks0004.mbr bs=512 count=1

partimage -d -M -b -z1 save /dev/mapper/virtualmachines-rocks0004p1
$DEST/virtualmachines-rocks0004p1.img

partimage -d -M -b -z1 save /dev/mapper/virtualmachines-rocks0004p2
$DEST/virtualmachines-rocks0004p2.img

partimage -d -M -b -z1 save /dev/mapper/virtualmachines-rocks0004p5
$DEST/virtualmachines-rocks0004p5.img

kpartx -dv /dev/virtualmachines/rocks0004

# VM 15: Rocks 5.1 x86_64 HPC Compute Cluster HVM domU

kpartx -av /dev/virtualmachines/rocks0005

dd if=/dev/virtualmachines/rocks0005 of=$DEST/rocks0005.mbr bs=512 count=1

partimage -d -M -b -z1 save /dev/mapper/virtualmachines-rocks0005p1
$DEST/virtualmachines-rocks0005p1.img

partimage -d -M -b -z1 save /dev/mapper/virtualmachines-rocks0005p2
$DEST/virtualmachines-rocks0005p2.img

partimage -d -M -b -z1 save /dev/mapper/virtualmachines-rocks0005p5
$DEST/virtualmachines-rocks0005p5.img

kpartx -dv /dev/virtualmachines/rocks0005

# VM 16: Slackware64 13.0 amd64 HVM domU

kpartx -av /dev/virtualmachines/slackware64

dd if=/dev/virtualmachines/slackware64 of=$DEST/slackware64.mbr bs=512
count=1

partimage -d -M -b -z1 save /dev/mapper/virtualmachines-slackware64p1
$DEST/virtualmachines-slackware64p1.img

partimage -d -M -b -z1 save /dev/mapper/virtualmachines-slackware64p2
$DEST/virtualmachines-slackware64p2.img

kpartx -dv /dev/virtualmachines/slackware64

# VM 17: Ubuntu 9.10 Karmic Koala Linux HVM domU

kpartx -av /dev/virtualmachines/ubuntu910

dd if=/dev/virtualmachines/ubuntu910 of=$DEST/ubuntu910.mbr bs=512 count=1

partimage -d -M -b -z1 save /dev/mapper/virtualmachines-ubuntu910p1
$DEST/virtualmachines-ubuntu910p1.img

kpartx -dv /dev/virtualmachines/ubuntu910

# VM 18: Windows XP Home Edition SP3 32-bit HVM domU with VGA passthrough
(requires Intel VT-d)

kpartx -av /dev/virtualmachines/winxphome32

dd if=/dev/virtualmachines/winxphome32 of=$DEST/winxphome32.mbr bs=512
count=1

partimage -d -M -b -z1 save /dev/mapper/virtualmachines-winxphome32p1
$DEST/virtualmachines-winxphome32p1.img

kpartx -dv /dev/virtualmachines/winxphome32

# EOF

</SCRIPT>

Thank you!