Add static network configuration option and make <Back> work.

[virt-p2v.git] / virt-p2v.pod

=head1 NAME

virt-p2v - P2V (physical to virtual machine) migration tool

=head1 SUMMARY

virt-p2v

=head1 DESCRIPTION

virt-p2v is a live CD for migrating physical machines to virtual
machine guests.

In the simplest mode of operation, you take a pre-built live CD ISO
from the main website (L<http://et.redhat.com/~rjones/virt-p2v/>) and
burn it to a CD-R.  Then insert the CD-R into the physical machine
which must be migrated, reboot, and follow the on-screen instructions.
See B<STANDARD USAGE> section below.

You may also build a customized live CD.  Typically this will contain
things like server details specific to your organization, so that the
live CD can run mostly or completely automatically.  See B<BUILDING A
CUSTOM LIVE CD> section below.

In both cases, files and disk images are transferred from the physical
machine over the network to the virtualization host machine over ssh.
Therefore C<sshd> must be running on the virtualization host, and must
be accessible to that host.  See B<SERVER REQUIREMENTS> section below.

The C<virt-p2v> script must only be run from the live CD.  It isn't
designed to run outside this environment and could do Bad Things to
your machine if you try it.  The script contains some checks to try to
stop you from doing this.

Virt-p2v does not modify the physical machine, its disks,
configuration etc.

=head1 STANDARD USAGE

After booting the live CD-R, you are presented with a series of
questions.  This section explains each question.

=over 4

=item Remote host

Enter the name or IP address of the virtualization host.  This is the
host running Xen (or any other virtualization system supported by
libvirt, eg. QEMU).  This host should be accessible on the network and
running an SSH daemon (C<sshd>).

=item Remote port

This is the port name or number of the SSH server on the remote host.
The default is C<22> which is the standard SSH port.

=item Remote directory

Enter the directory on the remote host where disk image(s) and
configuration file(s) must reside.

Note that if the remote host is running SELinux then you may not be
able to start a Xen guest unless its disk image(s) are located in the
default directory, C</var/lib/xen/images>.

=item Remote username

Enter the remote SSH username to use to log in to the remote host.

If you use the default username of C<root> then you should ensure that
remote root logins are enabled on the remote host
(ie. C<PermitRootLogin yes> in C</etc/ssh/sshd_config>).

=item Network configuration

Choose the way that the live CD configures network access.  The
current options are:

=over 4

=item Automatic configuration

In this mode, the live CD attempts to reuse the network configuration
from the physical machine's root filesystem.  You should probably try
this method even though occasionally it does not work.

=item Ask for fixed IP address and gateway

In this mode the live CD will ask you for a fixed IP address and
gateway address, and will configure your chosen interface with these.

=item Configure from the shell

In this mode you will be dropped into a command shell and you will
need to issue the correct sequence of C</sbin/ifconfig> commands in
order to configure the network interface.

A typical sequence of commands which should bring up the network
interface would be:

 /sbin/ifconfig eth0 AA.BB.CC.DD
 /sbin/route add default gw GG.HH.II.JJ eth0

where C<AA.BB.CC.DD> is the IP address and C<GG.HH.II.JJ> is the
gateway.

=item QEMU user network

This option configures the network for use inside a QEMU user network
(L<http://fabrice.bellard.free.fr/qemu/qemu-doc.html#SEC30>).  It
should only be used by developers.

=back

=item Devices

This question lists out all local block devices (hard disk drives and
similar) and asks you to choose which will be sent to the remote host.
You must send at least one block device.

=item Root device

This question lists out possible root filesystems and asks you to
choose the right one.  Choose the filesystem which would normally be
mounted as C</> on the system.

Virt-p2v performs some autodetection and is in most cases able to work
out which filesystems are possible root filesystems.  It displays what
it thinks is on each filesystem, but leaves it up to the user to make
a final decision.

The root filesystem is critical because it contains C</etc/fstab>.
This is used during P2V both to determine how other filesystems are
normally mounted on the machine, and because this file and others
under C</etc> may need to be modified during P2V conversion.

If the machine has more than one root filesystem (typically because
the machine is dual-booted with another operating system), then you
must choose only one of them to perform the P2V conversion on.

=item Hypervisor

This question asks you to choose the hypervisor / virtualization
system in use on the remote host.

If you select I<Xen>, I<QEMU> or I<KVM> then virt-p2v will produce a
configuration file which is customized for the selected system.  If
you select I<Other> then virt-p2v will produce a generic configuration
file which will probably require hand-modification to work.

See also L<http://libvirt.org/format.html>.

=item Machine architecture

This question asks you to choose the machine architecture.  Virt-p2v
can normally detect this, so you should leave it as I<Auto-detect>.

=item Memory

This question asks you to choose the amount of memory (RAM) in
megabytes assigned to the virtual machine.

If the entry is left blank, then virt-p2v will try to autodetect how
much RAM is present in the physical machine and use that, and this is
probably a good choice for most simple migrations.

=item Virtual CPUs

This question asks you to choose the number of virtual CPUs assigned
to the virtual machine.  Choosing C<1> causes the virtual machine to
be uniprocessor, and choosing some number greater than 1 causes the
virtual machine to be SMP.

If the entry is left blank, then virt-p2v will try to autodetect how
many CPU cores are present in the physical machine and use that, and
this is probably a good choice for most simple migrations.

=item MAC address

Here you should enter a MAC address for the virtual machine's emulated
network card.  MAC addresses are written as C<aa:bb:cc:dd:ee:ff> where
C<aa>, C<bb> etc are hexadecimal octets.

Leaving it blank will cause virt-p2v to choose a random MAC address
within the C<00:16:3e:..> space reserved for Xen guests.  These MAC
addresses are not tested for uniqueness so there is a very small
chance that they could coincide, which would leave a guest unable to
access the virtual network.

=item Verify and proceed

In this step you are asked to verify the settings above.  If any are
incorrect, use the I<Back> button to navigate back to the setting.  If
all settings are correct, use the I<OK> button to begin the P2V
conversion.

=item Network autoconfiguration

If you selected network autoconfiguration above then virt-p2v tries to
autoconfigure the network and ping the remote host.  It then asks
I<Did automatic network configuration work?>

You should answer C<y> here if it worked.

Answering C<n> will drop you into a command shell.

You can also switch to another virtual console if you need to perform
additional tests.  See section B<GETTING A SHELL> below.

=item SSH connection

Unless you have set up an SSH key, or the SSH server on the remote
host allows passwordless logins, then for each file that has to be
transferred to the remote host you will need to confirm the identity
of the remote host and/or enter a password.

To understand more about this, please see the L<ssh(1)> manual page.

=back

=head2 BOOTING P2V GUEST ON VIRTUALIZATION HOST

Once the P2V conversion has been completed, and assuming it was
successful, you will find a configuration file and one or more disk
images on the remote host.

The files will be located in the directory selected, usually
C</var/lib/xen/images>.  The names of the files are made up of:

C<p2v-I<hostname>-I<YYYYMMDDHHMM>.conf> or
C<p2v-I<hostname>-I<YYYYMMDDHHMM>-hdI<X>.img>

To simply start up the guest, use the following commands as root:

 virsh define p2v-foo-2008MMDDHHMM.conf
 virsh start foo

For QEMU/KVM do:

 virsh -c qemu:///system define p2v-foo-2008MMDDHHMM.conf
 virsh -c qemu:///system start foo

For other hypervisors you will need to edit the configuration file and
read L<http://libvirt.org/uri.html>.

=head1 GETTING A SHELL

During all stages of P2V questions and conversion you can get a root
shell on the physical machine.  Use I<ALT> I<F2> keys to switch to the
second virtual console, then log in as I<root> with no password.

=head2 LOG FILE

Virt-p2v writes a detailed log file to C</tmp/virt-p2v.log>.  (Note
that this C</tmp> directory is a ramdisk on the live CD, not the same
as the C</tmp> directory of the physical machine, and more importantly
it disappears when the machine is rebooted).

If you are reporting a bug, please always supply this file.

=head1 SERVER REQUIREMENTS

The virtualization host (remote host) must be running an SSH daemon
(C<sshd>), accessible from the physical machine which is being
migrated.

Previous versions of virt-p2v could use a special virt-p2v server.
However this capability has been removed since there was practically
no benefit.

=head1 BUILDING A CUSTOM LIVE CD

To build a custom live CD you must download the source for virt-p2v
from L<http://et.redhat.com/~rjones/virt-p2v/> or from the Mercurial
source repository (see website for details).

Please read the C<README> file to find the dependencies which are all
in Fedora &gt; 8 or EPEL &gt; 5.

The steps to creating a custom live CD are:

=over 4

=item 1. Edit C<virt-p2v> and adjust defaults

Find the section "TO MAKE A CUSTOM virt-p2v SCRIPT ..." which is near
to the top of this file.  Edit the defaults in this section as
detailed below.

=item 2. C<virt-p2v --test> to verify your changes

This command should not print anything at all.  If it prints any
message, then you will need to fix the error by going back to the
first step.

=item 3. C<make build> or C<make update> to build a custom live CD

C<make build> will create a complete ISO from scratch.  C<make update>
can be used to build a "quick" developer ISO by updating an existing
ISO image.  See section B<ISO ATTACHMENTS> below for more details.

=item 4. Burn the ISO to a CD-R and test

=back

=head2 EDITING DEFAULTS IN THE C<virt-p2v> SCRIPT

For each default, setting it to C<None> will ask the user.  All of the
defaults are set to C<None> in the standard, uncustomized virt-p2v
script, and so the standard script asks all the questions.

You may edit C<virt-p2v> and change the defaults, in which case the
user will not be questioned.  In this way you can make the script
partially or fully automated.

I<Note about OCaml code:>  C<None> and C<Some foo> are similar to the
concept of a NULL pointer versus non-NULL pointer in other languages.
This a variant type defined as:

type E<945> option = None | Some of E<945>

=over 4

=item C<greeting>

If this is C<true> then we wait for a keypress after boot and at a
couple of other stages.  If set to C<false> then we try not to wait
for any keypresses (so more automated live CDs are possible).

=item C<remote_host>

Set this to C<Some "hostname"> or C<Some "IP-address"> to
provide the name of the remote host.

=item C<remote_port>

Set this to C<Some port> (eg. C<Some 22>) to provide the port number
of the remote host's SSH daemon.

=item C<remote_directory>

Set this to C<Some "path"> (eg. C<Some "/var/lib/xen/images">) to
provide the directory where we update P2V converted images and
configuration files.

=item C<remote_username>

Set this to C<Some "username"> (eg. C<Some "root">) to provide the SSH
username to use on the remote system.

=item C<devices_to_send>

Set this to a list of block devices to send to the remote system.
For example, C<Some ["sda"; "sdb"]>.

=item C<root_filesystem>

Set this to the name of the root filesystem.

For a disk partition (eg. C</dev/sda1>), use:

 Some (Part ("sda", "1"))

For a logical volume (eg. C</dev/VolGroup00/LogVol00>), use:

 Some (LV ("VolGroup00", "LogVol00"))

=item C<network>

Set this to the choice for network setup.  Use one of:

=over 4

=item C<Some Auto>

for auto-configuration

=item C<Some Static>

to specify the interface, address, netmask and gateway statically

=item C<Some Shell>

to launch a shell

=item C<Some QEMUUserNet>

to use a QEMU user network (developers only)

=back

=item C<static_network_config>

This setting only applies if C<network> is set to C<Some Static>,
in which case you should set this to the static network settings,
a tuple of (interface, address, netmask, gateway, nameserver):

 Some ("eth0", "192.168.2.5", "255.255.255.0", "192.168.2.1", "192.168.2.1")

=item C<hypervisor>

Set this to the choice of hypervisor or virtualization system.  The
choices are: C<Some Xen>, C<Some QEMU> or C<Some KVM>.

=item C<architecture>

Set this to the architecture.  The choices are:
C<Some I386> (i386 and up, 32 bit),
C<Some X86_64> (AMD and Intel x86-64, 64 bit),
C<Some IA64> (Intel IA64),
C<Some PPC> (PowerPC, 32 bit),
C<Some PPC64> (PowerPC, 64 bit),
C<Some SPARC> (Sun SPARC, 32 bit),
C<Some SPARC64> (Sun SPARC, 64 bit),
C<OtherArch "foo"> (a hypothetical architecture called I<foo>), or
C<UnknownArch> to auto-detect the architecture.

=item C<memory>

Set this to the size of memory in megabytes, eg. C<Some 256>.  If you
set this to C<Some 0> then virt-p2v will try to autodetect the amount
of RAM installed on the physical machine.

=item C<vcpus>

Set this to the number of virtual CPUs, eg. C<Some 1>.  If you set
this to C<Some 0> then virt-p2v will try to autodetect the number of
CPU cores on the physical machine.

=item C<mac_address>

Set this to the MAC address for the virtual network card, eg. C<Some
"aa:bb:cc:dd:ee:ff">.  If you set this to C<Some ""> then virt-p2v
will choose a random MAC address within the C<00:16:3e:..> space
reserved for Xen guests.  These MAC addresses are not tested for
uniqueness so there is a very small chance that they could coincide,
which would leave a guest unable to access the virtual network.

=back

=head2 ISO ATTACHMENTS

Rebuilding a custom ISO is time-consuming.  You can make a "quick"
developer ISO by updating an existing ISO image with a new custom
C<virt-p2v> script.  This is useful for testing purposes.

From the source directory, assuming that you have downloaded or
built an existing C<virt-p2v-*.iso>, you can just do:

 make update

or the equivalent manual command:

 ./iso-attach virt-p2v-VERSION.iso virt-p2v

=head1 BOOTING FROM A USB KEY INSTEAD OF A CD

If you wish to boot from a USB keydrive, use the livecd-iso-to-disk
tool:

 livecd-iso-to-disk virt-p2v-$VERSION.iso /dev/sdX1

(Replace /dev/sdX1 with the actual USB device).

In my experience I also had to set up a suitable MBR:

 cat /usr/lib/syslinux/mbr.bin > /dev/sdX

=head1 TESTING AN ISO UNDER QEMU OR KVM

If you have a virtual guest running under QEMU or KVM then
you can test the P2V conversion process on the guest.

(Technically this is a V2V -- virtual to virtual -- conversion).

From the source directory do:

 make boot HDA=qemuimage.img

where C<qemuimage.img> is the name of the QEMU/KVM image.

You can also supply an C<HDB> parameter to specify a second disk.

=head1 MAILING LIST

Please direct questions to the et-mgmt-tools mailing list
L<http://www.redhat.com/mailman/listinfo/et-mgmt-tools>
<et-mgmt-tools @ redhat . com>

=head1 SEE ALSO

L<virsh(1)>,
L<http://www.libvirt.org/ocaml/>,
L<http://www.libvirt.org/>,
L<http://et.redhat.com/~rjones/>,
L<http://caml.inria.fr/>

=head1 AUTHORS

Richard W.M. Jones <rjones @ redhat . com>

=head1 COPYRIGHT

(C) Copyright 2007-2008 Red Hat Inc., Richard W.M. Jones
http://libvirt.org/

This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 2 of the License, or
(at your option) any later version.

This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
GNU General Public License for more details.

You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.

=head1 REPORTING BUGS

Bugs can be viewed on the Red Hat Bugzilla page:
L<https://bugzilla.redhat.com/>.

If you find a bug in virt-p2v, please follow these steps to report it:

=over 4

=item 1. Check for existing bug reports

Go to L<https://bugzilla.redhat.com/> and search for similar bugs.
Someone may already have reported the same bug, and they may even
have fixed it.

=item 2. Capture debug and error messages

At the point where you get the error or unexpected behaviour,
go to the second virtual console (I<ALT> I<F2>) and look at
the logfile C</tmp/virt-p2v.log>.  Please make sure that
this file is attached to your bug report.

=item 3. Get version of virt-p2v

The version is in the name of the ISO.  If you have built a custom
virt-p2v ISO, please describe any changes that you have made.

=item 4. Submit a bug report.

Go to L<https://bugzilla.redhat.com/> and enter a new bug.
Please describe the problem in as much detail as possible.

Remember to include the version number (step 3) and to
attach the log file (step 2).

=item 5. Assign the bug to rjones @ redhat.com

Assign or reassign the bug to B<rjones @ redhat.com> (without the
spaces).  You can also send me an email with the bug number if you
want a faster response.

=back