virt-dib - Run diskimage-builder elements
virt-dib -B DIB-LIB [options] elements...
Virt-dib is a tool for using the elements of
diskimage-builder to build a new disk image, generate new ramdisks, etc.
Virt-dib is intended as safe replacement for
diskimage-builder and its
ramdisk-image-create mode, see "COMPARISON WITH DISKIMAGE-BUILDER" for a quick comparison with usage of
diskimage-builder is part of the TripleO OpenStack project: https://wiki.openstack.org/wiki/TripleO.
virt-dib \ -B /path/to/diskimage-builder/lib \ -p /path/to/diskimage-builder/elements \ --envvar DIB_RELEASE=jessie \ --name debian-jessie \ debian vm
This builds a Debian Jessie (8.x) disk image, suitable for running as virtual machine, saved as debian-jessie.qcow2.
virt-dib \ -B /path/to/diskimage-builder/lib \ -p /path/to/diskimage-builder/elements \ --ramdisk \ --name ramdisk \ ubuntu deploy-ironic
This builds a ramdisk for the Ironic OpenStack component based on the Ubuntu distribution.
Set the path to the library directory of
diskimage-builder. This is usually the lib subdirectory in the sources and when installed, and /usr/share/diskimage-builder/lib when installed in /usr.
This parameter is mandatory, as virt-dib needs to provide it for the elements (as some of them might use scripts in it). Virt-dib itself does not make use of the library directory.
Use the specified architecture for the output image. The default value is the same as the host running virt-dib.
Right now this option does nothing more than setting the
ARCH environment variable for the elements, and it's up to them to produce an image for the requested architecture.
Generate checksum files for the generated image. The supported checksums are MD5, and SHA256.
Use ANSI colour sequences to colourize messages. This is the default when the output is a tty. If the output of the program is redirected to a file, ANSI colour sequences are disabled unless you use this option.
Set the debug level to
LEVEL, which is a non-negative integer number. The default is
This debug level is different than what -x and -v set, and it increases the debugging information printed out. Specifically, this sets the
DIB_DEBUG_TRACE, and any value >
0 enables tracing in the scripts executed.
Set the repository and tag for docker.
This is used only when the formats include
docker, and it is required in that case.
Add the specified disk to be used as helper drive where to cache files of the elements, like disk images, distribution packages, etc.
See "HELPER DRIVE".
Specify the format of the helper drive. If this flag is not given then it is auto-detected from the drive itself.
If working with untrusted raw-format guest disk images, you should ensure the format is always specified.
This option is used only if --drive is specified.
See "HELPER DRIVE".
Add a new path with elements. Paths are used in the same order as the -p parameters appear, so a path specified first is looked first, and so on.
Obviously, it is recommended to add the path to the own elements of
diskimage-builder, as most of the other elements will rely on them.
Install additional packages in the image being built.
This relies on the
install-packages binary provided by the package management elements.
This option can be specified multiple times, each time with multiple packages separated by comma.
Carry or set an environment variable for the elements.
See "ENVIRONMENT VARIABLES" below for more information on the interaction and usage of environment variables.
This option can be used in two ways:
Carry the environment variable
VARIABLE. If it is not set, nothing is exported to the elements.
Set the environment variable
VARIABLE with value
VALUE for the elements, regardless whether an environment variable with the same name exists.
This can be useful to pass environment variable without exporting them in the environment where virt-dib runs.
Ignore the specified element.
Ignore any element script named
SCRIPT, whichever element it is in.
This can be useful in case some script does not run well with virt-dib, for example when they really need
Set the list of output formats, separating them with comma.
Supported formats are:
Import the image to docker, running docker import. The target for the image must be specified using --docker-target.
Please note this operation usually requires the docker service to be enabled, otherwise it will fail. Furthermore, docker is run using sudo(8), so make sure the user has the permissions to run at least docker.
qcow2(enabled by default)
QEMU's qcow2. This output format requires the
Raw disk format.
An squashfs filesystem, compressed with XZ. This output format requires the
squashfs feature; see also "AVAILABILITY" in guestfs(3).
An uncompressed tarball.
A tarball compressed with gzip.
Virtual Hard Disk disk image. This output format requires the
Please note that the version of
vhd-util tool needs to be patched to support the
convert subcommand, and to be bootable. The patch is available here: https://github.com/emonty/vhd-util/blob/master/debian/patches/citrix.
Set the filesystem type to use for the root filesystem. The default is
See also "guestfs_filesystem_available" in guestfs(3).
Set the path in the host where cache the resources used by the elements of the
extra-data.d phase. The default is ~/.cache/image-create.
Please note that most of the resources fetched in phases other than
extra-data.d will be cached in the helper drive specified with --drive; see also "HELPER DRIVE".
Specify the default installation type. Defaults to
package to use package based installations by default.
This option is used to make the output more machine friendly when being parsed by other programs. See "MACHINE READABLE OUTPUT" below.
Change the amount of memory allocated to the appliance. Increase this if you find that the virt-dib execution runs out of memory.
The default can be found with this command:
Add the specified options to mkfs(1), to be able to fine-tune the root filesystem creation; the options are passed to the driver of mfks(1), and not to mfks(1) itself. Note that --fs-type is used to to change the filesystem type.
You should use --mkfs-options at most once. To pass multiple options, separate them with space, eg:
virt-dib ... --mkfs-options '-O someopt -I foo'
Enable or disable network access from the guest during the installation.
Enabled is the default. Use --no-network to disable access.
The network only allows outgoing connections and has other minor limitations. See "NETWORK" in virt-rescue(1).
This does not affect whether the guest can access the network once it has been booted, because that is controlled by your hypervisor or cloud environment and has nothing to do with virt-dib.
If you use --no-network, then the environment variable
DIB_OFFLINE is set to
1, signaling the elements that they should use only cached resources when available. Note also that, unlike with
diskimage-builder where elements may still be able to access to the network even with
DIB_OFFLINE=, under virt-dib network will not be accessible at all.
Set the name of the output image file. The default is
According to the chosen name, there will be the following in the current directory:
For each output format, a file named after the output image with the extension depending on the format; for example: $NAME.qcow2, $NAME.raw, etc.
Not applicable in ramdisk mode, see "RAMDISK BUILDING".
A directory containing any files created by the elements, for example dib-manifests directory (created by the
manifests element), ramdisks and kernels in ramdisk mode, and so on.
When --checksum is specified, there will be files for each supported checksum type; for example: $NAME.ext.md5, $NAME.ext.sha256, etc.
Not applicable in ramdisk mode, see "RAMDISK BUILDING".
Don't delete the output files on failure to build. You can use this to debug failures to run scripts.
The default is to delete the output files if virt-dib fails (or, for example, some script that it runs fails).
Specify a different Python interpreter to use. Parts of
diskimage-builder are implemented in Python, and thus an interpreter is needed.
PYTHON can either be an executable filename (e.g. python2, which is then searched in
$PATH), or a full path (e.g. /usr/bin/python2). If not specified, the default value is python.
Don't print ordinary progress messages.
Pass --qemu-img-options option(s) to the qemu-img(1) command to fine-tune the output format. Options available depend on the output format (see --formats) and the installed version of the qemu-img program.
You should use --qemu-img-options at most once. To pass multiple options, separate them with commas, eg:
virt-dib ... --qemu-img-options cluster_size=512,preallocation=metadata ...
Set the ramdisk building mode.
See "RAMDISK BUILDING".
Set the name for the additional element added in ramdisk building mode. The default is
See "RAMDISK BUILDING".
Set the label for the root filesystem in the created image.
Please note that some filesystems have different restrictions on the length of their labels; for example, on
ext2/3/4 filesystems labels cannot be longer than 16 characters, while on
xfs they have at most 12 characters.
The default depends on the actual filesystem for the root partition (see --fs-type): on
cloudimg-rootfs on any other filesystem.
Select the size of the output disk, where the size can be specified using common names such as
32G (32 gigabytes) etc. The default size is
To specify size in bytes, the number must be followed by the lowercase letter b, eg:
See also virt-resize(1) for resizing partitions of an existing disk image.
Skip the inclusion of the
Enable N ≥ 2 virtual CPUs for scripts to use.
Do not compress resulting qcow2 images. The default is to compress them.
Enable debugging messages.
Display version number and exit.
Enable tracing of libguestfs API calls.
diskimage-builder, the environment of the host is not inherited in the appliance when running most of the elements (i.e. all except the ones in the
To set environment for the elements being run, it is necessary to tell virt-dib to use them, with the option --envvar. Such option allows to selectively export environment variables when running the elements, and it is the preferred way to pass environment variables to the elements.
To recap: if you want the environment variable
MYVAR (and its content) to be available to the elements, you can do either
export MYVAR # whichever is its value virt-dib ... --envvar MYVAR ...
virt-dib ... --envvar MYVAR=value_of_it ...
Virt-dib runs most of the element in its own appliance, and thus not on the host. Because of this, there is no possibility for elements to cache resources directly on the host.
To solve this issue, virt-dib allows the usage of an helper drive where to store cached resources, like disk images, distribution packages, etc. While this means that there is a smaller space available for caching, at least it allows to limit the space on the host for caches, without assuming that elements will do that by themselves.
Currently this disk is either required to have a single partition on it, or the first partition on it will be used. A disk with the latter configuration can be easily created with guestfish(1) like the following:
guestfish -N filename.img=fs:ext4:10G exit
The above will create a disk image called filename.img, 10G big, with a single partition of type ext4; see "PREPARED DISK IMAGES" in guestfish(1).
It is recommended for it to be ≥ 10G or even more, as elements will cache disk images, distribution packages, etc. As with any disk image, the helper disk can be easily resized using virt-resize(1) if more space in it is needed.
The drive can be accessed like any other disk image, for example using other tools of libguestfs such as guestfish(1):
guestfish -a filename.img -m /dev/sda1
If no helper drive is specified with --drive, all the resources cached during a virt-dib run will be discarded.
Inside the helper drive, it is possible to find the following resources:
This directory is set as
HOME environment variable during the build. It contains mostly the image cache (saved as /home/.cache/image-create), and whichever other resource is cached in the home directory of the user running the various tools.
These are the logs of the elements being run within the libguestfs appliance, which means all the phases except
Virt-dib can emulate also
ramdisk-image-create, which is a secondary operation mode of
diskimage-builder. Instead of being a different tool name, virt-dib provides easy access to this mode using the --ramdisk switch.
In this mode:
there is an additional ramdisk element added (see --ramdisk-element)
no image is produced (so --formats is ignored)
$NAME.d (see --name) will contain initrd, kernel, etc
Virt-dib uses the standard temporary directory used by libguestfs, see "ENVIRONMENT VARIABLES" in guestfs(3).
By default this location is /tmp (default value for
TMPDIR), which on some systems may be on a tmpfs filesystem, and thus defaulting to a maximum size of half of physical RAM. If virt-dib exceeds this, it may hang or exit early with an error. The solution is to point
TMPDIR to a permanent location used as temporary location, for example:
mkdir local-tmp env TMPDIR=$PWD/local-tmp virt-dib ... rm -rf local-tmp
Because of virt-dib runs most of the elements in its own appliance, all the tools and libraries used by elements running outside the guest (typically
cleanup.d) need to be present in the appliance as well. In case they are not, scripts will fail typically with a
command not found error.
For tools and libraries packaged by the distribution, the easy solution is to tell libguestfs to include additional packages in the appliance. This is doable by e.g. creating a new file with the additional packages:
# echo wget > /usr/lib64/guestfs/supermin.d/dib-my-extra
The actual path to the supermin.d directory depends on the distribution; additional files can list more packages, each in its own line. For more details, see supermin(1).
Virt-dib is intended as safe replacement for
diskimage-builder and its
ramdisk-image-create mode; the user-notable differences consist in:
the command line arguments; some of the arguments are the same as available in
diskimage-builder, while some have different names:
disk-image-create virt-dib ----------------- -------- -a ARCH --arch ARCH --image-size SIZE --size SIZE --max-online-resize SIZE doable using --mkfs-options -n --skip-base -o IMAGENAME --name IMAGENAME -p PACKAGE(S) --extra-packages PACKAGE(S) -t FORMAT(S) --formats FORMAT(S) -x --debug 1 -x -x --debug 2 -x -x [-x ...] --debug 3/4/etc
the location of non-image output files (like ramdisks and kernels)
the way some of the cached resources are saved: using an helper drive, not directly on the disk where virt-dib is run
the need to specify a target size for the output disk, as opposed to
diskimage-builder calculating an optimal one
the handling of environment variables, see "ENVIRONMENT VARIABLES".
Furthermore, other than the libguestfs own environment variables (see "ENVIRONMENT VARIABLES" in guestfs(3)), virt-dib does not read any other environment variable: this means that all the options and behaviour changes are specified solely using command line arguments
extra tools needed on some out-of-chroot phases need to be available in the appliance, see "EXTRA DEPENDENCIES".
Elements themselves should notice no difference in they way they are run; behaviour differences may due to wrong assumptions in elements, or not correct virt-dib emulation.
Known issues at the moment:
The --machine-readable option can be used to make the output more machine friendly, which is useful when calling virt-dib from other programs, GUIs etc.
Use the option on its own to query the capabilities of the virt-dib binary. Typical output looks like this:
$ virt-dib --machine-readable virt-dib output:qcow2 output:tar output:raw output:vhd
A list of features is printed, one per line, and the program exits with status 0.
output: features refer to the output formats (--formats command line option) supported by this binary.
Virt-dib has been tested with
diskimage-builder (and its elements) ≥ 0.1.43; from time to time also with
Previous versions may work, but it is not guaranteed.
This program returns 0 if successful, or non-zero if there was an error.
guestfs(3), guestfish(1), virt-resize(1), http://libguestfs.org/.
Pino Toscano (
ptoscano at redhat dot com)
Copyright (C) 2015 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.