virt-v2v-output-openstack - Using virt-v2v to convert guests to OpenStack
virt-v2v [-i* options] -o openstack
-oo server-id=SERVER
[-oo guest-id=GUEST]
[-oo verify-server-certificate=false]
[-oo os-username=admin] [-oo os-*=*]
virt-v2v [-i* options] -o glance
This page documents how to use virt-v2v(1) to convert guests to run on OpenStack. There are two output modes you can select, but only -o openstack should be used normally.
Full description: "OUTPUT TO OPENSTACK"
This is the modern method for uploading to OpenStack via the REST API. Guests can be directly converted into Cinder volumes.
Full description: "OUTPUT TO GLANCE"
This is the old method for uploading to Glance. Unfortunately Glance is not well suited to storing converted guests (since virt-v2v deals with "pets" not templated "cattle"), so this method is not recommended unless you really know what you are doing.
To output to OpenStack, use the -o openstack option.
When virt-v2v is converting to OpenStack, it is unusual in that virt-v2v must be running inside a virtual machine running on top of the OpenStack overcloud. This virtual machine is called the "conversion appliance". Note this virtual machine is unrelated to the guest which is being converted.
The reason for this is because to create Cinder volumes that will contain the guest data (for the converted guest) we must attach those Cinder volumes to an OpenStack virtual machine.
The openstack
command must be installed in the appliance. We use it for communicating with OpenStack.
When virt-v2v is running in the conversion appliance, you must supply the name or UUID of the conversion appliance to virt-v2v, eg:
$ openstack server list
+--------------------------------------+-----------+--------+
| ID | Name | Status |
+--------------------------------------+-----------+--------+
| bbb0147a-44b9-4d19-9a9d-10ca9a984744 | test1 | ACTIVE |
+--------------------------------------+-----------+--------+
# virt-v2v [...] \
-o openstack -oo server-id=bbb0147a-44b9-4d19-9a9d-10ca9a984744
or:
# virt-v2v [...] -o openstack -oo server-id=test1
You can run many parallel conversions inside a single conversion appliance if you want, subject to having enough resources available. However OpenStack itself imposes a limit that you should be aware of: OpenStack cannot attach more than around 25 disks [the exact number varies with configuration] to a single appliance, and that limits the number of guests which can be converted in parallel, because each guest's disk must be attached to the appliance while being copied.
Converting to OpenStack requires access to the tenant (non-admin) API endpoints. You will need to either set up your $OS_*
environment variables or use output options on the virt-v2v command line to authenticate with OpenStack.
For example:
export OS_USERNAME=admin
or:
virt-v2v [...] -o openstack -oo os-username=admin
are equivalent, and have the same effect as using --os-username on the command line of OpenStack tools.
Normally there is a file called overcloudrc or keystonerc_admin which you can simply source
to set everything up.
If you need to copy overcloudrc to another machine, check if it references the OS_CLOUD
environment variable. If so, you may have to copy the clouds.yaml file from /etc/openstack/ or $HOME/.config/openstack/ to the other machine as well. See the section "CLOUD CONFIGURATION" in the openstack CLI manual.
Because virt-v2v must access Cinder volumes which are presented as /dev devices to the conversion appliance, virt-v2v must usually run as root in -o openstack mode.
If you use sudo
to start virt-v2v and you are using environment variables for authentication, remember to use the sudo -E
option to preserve the environment.
virt-v2v [...] -o openstack -oo guest-id=123-456-7890
You may optionally specify -oo guest-id=... on the command line. The ID (which can be any string) is saved on each Cinder volume in the virt_v2v_guest_id
volume property.
This can be used to find disks associated with a guest, or to associate which disks are related to which guests when converting many guests.
Using -oo verify-server-certificate=false you can tell the openstack client to ignore the server certificate when connecting to the OpenStack API endpoints. This has the same effect as passing the --insecure option to the openstack
command.
The final command to convert the guest, running as root, will be:
# virt-v2v [-i options ...] \
-o openstack -oo server-id=NAME|UUID [-oo guest-id=ID]
If you include authentication options on the command line then:
# virt-v2v [-i options ...] \
-o openstack -oo server-id=NAME|UUID -oo os-username=admin [etc]
Guests are converted as Cinder volume(s) (one volume per disk in the original guest). To boot them use the openstack server create --volume
option:
$ openstack volume list
+--------------------------------------+---------------+-----------+
| ID | Name | Status |
+--------------------------------------+---------------+-----------+
| c4d06d15-22ef-462e-9eff-ab54ab285a1f | fedora-27-sda | available |
+--------------------------------------+---------------+-----------+
$ openstack server create \
--flavor x1.small \
--volume c4d06d15-22ef-462e-9eff-ab54ab285a1f \
myguest
$ openstack console url show myguest
To specify the Cinder volume type, use -os. If not specified then no Cinder volume type is used.
The following options are not supported with OpenStack: -oa, -of.
Note this is a legacy option. In most cases you should use "OUTPUT TO OPENSTACK" instead.
To output to OpenStack Glance, use the -o glance option.
This runs the glance(1) CLI program which must be installed on the virt-v2v conversion host. For authentication to work, you will need to set OS_*
environment variables. See "OpenStack: Authentication" above.
Virt-v2v adds metadata for the guest to Glance, describing such things as the guest operating system and what drivers it requires. The command glance image-show
will display the metadata as "Property" fields such as os_type
and hw_disk_bus
.
Glance image upload doesn't appear to correctly handle sparseness. For this reason, using qcow2 will be faster and use less space on the Glance server. Use the virt-v2v -of qcow2 option.
If the guest has a single disk, then the name of the disk in Glance will be the name of the guest. You can control this using the -on option.
Glance doesn't have a concept of associating multiple disks with a single guest, and Nova doesn't allow you to boot a guest from multiple Glance disks either. If the guest has multiple disks, then the first (assumed to be the system disk) will have the name of the guest, and the second and subsequent data disks will be called guestname-disk2
, guestname-disk3
etc. It may be best to leave the system disk in Glance, and import the data disks to Cinder.
virt-v2v(1), https://docs.openstack.org/python-openstackclient/latest/cli/man/openstack.html, glance(1).
Richard W.M. Jones
Copyright (C) 2009-2020 Red Hat Inc.
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., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
To get a list of bugs against libguestfs, use this link: https://bugzilla.redhat.com/buglist.cgi?component=libguestfs&product=Virtualization+Tools
To report a new bug against libguestfs, use this link: https://bugzilla.redhat.com/enter_bug.cgi?component=libguestfs&product=Virtualization+Tools
When reporting a bug, please supply:
The version of libguestfs.
Where you got libguestfs (eg. which Linux distro, compiled from source, etc)
Describe the bug accurately and give a way to reproduce it.
Run libguestfs-test-tool(1) and paste the complete, unedited output into the bug report.