Image Description Elements

Note

This document provides a reference for the elements and attributes of the KIWI NG XML document in version 9.24.44

<image>

The toplevel of any KIWI NG image description

<image schemaversion="7.4" name="{exc_image_base_name}">
    <!-- descendants -->
</image>

The image definition starts with an image tag and requires the schema format at version 7.4. The attribute name specifies the name of the image which is also used for the filenames created by KIWI. Because we don’t want spaces in filenames the name attribute must not have any spaces in its name.

The following optional attributes can be inserted in the image tag:

displayname

Allows setup of the boot menu title for the selected boot loader. So you can have suse-SLED-foo as the image name but a different name as the boot display name. Spaces are not allowed in the display name because it causes problems for some boot loaders and kiwi did not take the effort to separate the ones which can display them correctly from the ones which can’t

id

sets an identification number which appears as file /etc/ImageID within the image.

<include>

Optional include of XML file content from file

<image schemaversion="7.4" name="{exc_image_base_name}">
    <include from="file://description.xml"/>
</image>

with file description.xml as follows:

<image>
    <description type="system">
        <author>name</author>
        <contact>contact</contact>
        <specification>text</specification>
    </description>
</image>

This will replace the include statement with the contents of description.xml. The validation of the result happens after the inclusion of all include references. The value for the from attribute is interpreted as an URI, as of now only local URI types are supported as well as the this:// resource locator which translates into the path to the KIWI image description.

Note

The include information must be embedded into an <image> root node. Only the inner elements of the root node will be included. The processing of XML data via XSLT always requires a root node which is the reason why this is required to be specified for include files as well.

Note

Nesting of include statements in other include files is not supported. This will lead to unresolved include statements in the final document and will cause the runtime checker to complain about it.

Note

The include is implemented via a XSLT stylesheet and therefore expects an XML document. Other markup formats are not supported as include reference.

<description>

Provide an image identity.

<description type="system">
  <author>name</author>
  <contact>contact</contact>
  <specification>text</specification>
</description>

The mandatory description section contains information about the creator of this image description. The attribute type could be either of the value system which indicates this is a system image description or at value boot for custom kiwi boot image descriptions.

The following optional sub sections can be inserted below the description tag:

license

Specifies the license name which applies to this image description.

<preferences>

Setup image type and layout.

<preferences arch="arch">
  <version>1.2.3</version>
  <packagemanager name="zypper"/>
  <type image="tbz"/>
</preferences>

The mandatory preferences section contains information about the supported image type(s), the used package manager, the version of this image, and further optional elements. The preferences section can be configured to apply only for a certain architecture. In this case specify the arch attribute with a value as it is reported by uname -m

<preferences><version>

The mandatory image version must be a three-part version number of the format: Major.Minor.Release. In case of changes to the image description the following rules should apply:

  • For smaller image modifications that do not add or remove any new packages, only the release number is incremented. The XML description file(config.xml) remains unchanged.

  • For image changes that involve the addition or removal of packages the minor number is incremented and the release number is reset.

  • For image changes that changes the behavior or geometry of the image file the major number is incremented.

<preferences><packagemanager>

The mandatory packagemanager element specifies which package manager should be used to handle software packages. The packagemanager setup is connected to the distribution used to build the image. The following table shows which package manager is connected to which distributor:

Distributor

Package Manager

SUSE

zypper

RedHat

dnf

Debian Based

apt

Arch Linux

pacman

In general the specification of one preferences section is sufficient. However, it’s possible to specify multiple preferences sections and distinguish between the sections via the profiles attribute.

In combination with the above the preferences element supports the following optional elements:

<preferences><rpm-locale-filtering>

locale-filtering can be set to “true” or “false”. If set to “true” it sets the install_lang macro for RPM based installations to the RPM configured locale list. This results in language specific files to become filtered out by rpm if they don’t match the configured list.

<preferences>
  <rpm-locale-filtering>true</rpm-locale-filtering>
</preferences>

Note

It depends on the individual package design if the install_lang macro contents apply to the package or not.

<preferences><rpm-check-signatures>

Specifies whether package signatures should be checked or not

<preferences>
  <rpm-check-signatures>true</rpm-check-signatures>
</preferences>

<preferences><rpm-excludedocs>

Specifies whether files marked as documentation should be skipped during installation

<preferences>
  <rpm-excludedocs>true</rpm-excludedocs>
</preferences>

<preferences><keytable>

Specifies the name of the console keymap to use. The value corresponds to a map file in /usr/share/kbd/keymaps/xkb.

<preferences>
  <keytable>us</keytable>
</preferences>

<preferences><timezone>

Specifies the time zone. Available time zones are located in the /usr/share/zoneinfo directory. Specify the attribute value relative to /usr/share/zoneinfo. For example, specify Europe/Berlin for /usr/share/zoneinfo/Europe/Berlin.

<preferences>
  <timezone>Europe/Berlin</timezone>
</preferences>

<preferences><locale>

Specifies the name of the UTF-8 locale to use, which defines the contents of the RC_LANG system environment variable used in the image and to run the custom scripts specified as part of the KIWI NG image description. Please note only UTF-8 locales are supported here which also means that the encoding must not be part of the locale information. This means you need to specify the locale using the 4-digit name like the following example: en_US or en_US,de_DE

<preferences>
  <locale>en_US</locale>
</preferences>

<preferences><bootsplash-theme>

Specifies the name of the plymouth bootsplash theme to use

<preferences>
  <bootsplash-theme>bgrt</bootsplash-theme>
</preferences>

<preferences><bootloader-theme>

Specifies the name of the bootloader theme to use if that used bootloader has theme support.

<preferences>
  <bootloader-theme>openSUSE</bootloader-theme>
</preferences>

Along with the version and the packagemanager at least one image type element must be specified to indicate which image type should be build.

<preferences><release-version>

Specifies the distribution global release version as consumed by package managers. Currently the release version is not set or set to 0 for package managers which requires a value to operate. With the optional release-version section, users have an opportunity to specify a custom value which is passed along the package manager to define the distribution release.

Note

The release version information is currently used in dnf and microdnf package managers only. It might happen that it gets applied to the other package manager backends as well. This will happen on demand though.

<preferences><type>

At least one type element must be configured. It is possible to specify multiple type elements in a preferences block. To set a given type description as the default image use the boolean attribute primary and set its value to true:

<preferences>
  <type image="typename" primary="true"/>
</preferences>

The image type to be created is determined by the value of the image attribute. The following list describes the supported types and possible values of the image attribute:

image=”tbz”

A simple tar archive image. The tbz type packs the contents of the image root tree into a xz compressed tarball.

image=”btrfs|ext2|ext3|ext4|squashfs|xfs”

A filesystem image. The image root tree data is packed into a filesystem image of the given type. An image of that type can be loop mounted and accessed according to the capabiities of the selected filesystem.

image=”iso”

An iso image which can be dumped on a CD/DVD or USB stick and boots off from this media without interfering with other system storage components. A useful pocket system for testing and demo and debugging purposes.

image=”oem”

An image representing an expandable system disk. This means after deployment the system can resize itself to the new disk geometry. The resize operation is configurable as part of the image description and an installation image for CD/DVD, USB stick and Network deployment can be created in addition. For use in cloud frameworks like Amazon EC2, Google Compute Engine or Microsoft Azure this disk type also supports the common virtual disk formats.

image=”docker”

An archive image suitable for the docker container engine. The image can be loaded via the docker load command and works within the scope of the container engine

image=”oci”

An archive image that builds a container matching the OCI (Open Container Interface) standard. The container should be able to run with any oci compliant container engine.

image=”appx”

An archive image suitable for the Windows Subsystem For Linux container engine. The image can be loaded From a Windows System that has support for WSL activated.

image=”kis”

An optional root filesystem image associated with a kernel and initrd. The use case for this component image type is highly customizable. Many different deployment strategies are possible.

For completion of a type description, there could be several other optional attributes and child elements. The type element supports a plethora of optional attributes, some of these are only relevant for certain build types and will be covered in extra chapters that describes the individual image types more detailed. Certain attributes are however useful for nearly all build types and will be covered next:

bootpartition=”true|false”:

Boolean parameter notifying KIWI NG whether an extra boot partition should be used or not (the default depends on the current layout). This will override KIWI NG’s default layout.

bootpartsize=”nonNegativeInteger”:

For images with a separate boot partition this attribute specifies the size in MB. If not set the boot partition size is set to 200 MB

efipartsize=”nonNegativeInteger”:

For images with an EFI fat partition this attribute specifies the size in MB. If not set the EFI partition size is set to 20 MB

efiparttable=”msdos|gpt”:

For images with an EFI firmware specifies the partition table type to use. If not set defaults to the GPT partition table type

dosparttable_extended_layout=”true|false”:

For oem disk images, specifies to make use of logical partitions inside of an extended one. If set to true and if the msdos table type is active, this will cause the fourth partition to be an extended partition and all following partitions will be placed as logical partitions inside of that extended partition. This setting is useful if more than 4 primary partitions needs to be created in an msdos table

btrfs_quota_groups=”true|false”:

Boolean parameter to activate filesystem quotas if the filesystem is btrfs. By default quotas are inactive.

btrfs_root_is_snapshot=”true|false”:

Boolean parameter that tells KIWI NG to install the system into a btrfs snapshot. The snapshot layout is compatible with snapper. By default snapshots are turned off.

btrfs_root_is_readonly_snapshot=”true|false”:

Boolean parameter notifying KIWI NG that the btrfs root filesystem snapshot has to made read-only. if this option is set to true, the root filesystem snapshot it will be turned into read-only mode, once all data has been placed to it. The option is only effective if btrfs_root_is_snapshot is also set to true. By default the root filesystem snapshot is writable.

bootstrap_package=”package_name”:

For use with the apt packagemanager only. Specifies the name of a bootstrap package which provides a bootstrap tarball in /var/lib/bootstrap/PACKAGE_NAME.ARCH.tar.xz. The tarball will be unpacked and used as the bootstrap rootfs to begin with. This allows for an alternative bootstrap method preventing the use of debootstrap. For further details see Circumvent debootstrap.

compressed=”true|false”:

Specifies whether the image output file should be compressed or not. This option is only used for filesystem only images or for the pxe or cpio types.

editbootconfig=”file_path”:

Specifies the path to a script which is called right before the bootloader is installed. The script runs relative to the directory which contains the image structure.

editbootinstall=”file_path”:

Specifies the path to a script which is called right after the bootloader is installed. The script runs relative to the directory which contains the image structure.

filesystem=”btrfs|ext2|ext3|ext4|squashfs|xfs”:

The root filesystem

firmware=”efi|uefi”:

Specifies the boot firmware of the appliance, supported options are: bios, ec2, efi, uefi, ofw and opal. This attribute is used to differentiate the image according to the firmware which boots up the system. It mostly impacts the disk layout and the partition table type. By default bios is used on x86, ofw on PowerPC and efi on ARM.

force_mbr=”true|false”:

Boolean parameter to force the usage of a MBR partition table even if the system would default to GPT. This is occasionally required on ARM systems that use a EFI partition layout but which must not be stored in a GPT. Note that forcing a MBR partition table incurs limitations with respect to the number of available partitions and their sizes.

fsmountoptions=”option_string”:

Specifies the filesystem mount options which are passed via the -o flag to mount and are included in /etc/fstab.

fscreateoptions=”option_string”:

Specifies the filesystem options used to create the filesystem. In KIWI NG the filesystem utility to create a filesystem is called without any custom options. The default options are filesystem specific and are provided along with the package that provides the filesystem utility. For the Linux ext[234] filesystem, the default options can be found in the /etc/mke2fs.conf file. Other filesystems provides this differently and documents information about options and their defaults in the respective manual page, e.g man mke2fs. With the fscreateoptions attribute it’s possible to directly influence how the filesystem will be created. The options provided as a string are passed to the command that creates the filesystem without any further validation by KIWI NG. For example, to turn off the journal on creation of an ext4 filesystem the following option would be required:

<preferences>
  <type fscreateoptions="-O ^has_journal"/>
</preferences>
kernelcmdline=”string”:

Additional kernel parameters passed to the kernel by the bootloader.

root_clone=”number”

For oem disk images, this attribute allows to create number clone(s) of the root partition, with number >= 1. A clone partition is content wise an exact byte for byte copy of the origin root partition. However, to avoid conflicts at boot time the UUID of any cloned partition will be made unique. In the sequence of partitions, the clone(s) will always be created first followed by the partition considered the origin. The origin partition is the one that will be referenced and used by the system. Also see Partition Clones

boot_clone=”number”

Same as root_clone but applied to the boot partition if present

luks=”passphrase|file:///path/to/keyfile”:

Supplying a value will trigger the encryption of the partition serving the root filesystem using the LUKS extension. The supplied value represents either the passphrase string or the location of a key file if specified as file://... resource. When using a key file it is in the responsibility of the user how this key file is actually being used. By default any distribution will just open an interactive dialog asking for the credentials at boot time !

luks_version=”luks|luks1|luks2”:

Specify which LUKS version should be used. If not set and by default luks is used. The interpretation of the default depends on the distribution and could result in either ‘luks1’ or ‘luks2’. The specification of the LUKS version allows using a different set of luksformat options. To investigate the differences between the two please consult the cryptsetup manual page.

target_blocksize=”number”:

Specifies the image blocksize in bytes which has to match the logical blocksize of the target storage device. By default 512 Bytes is used, which works on many disks. You can obtain the blocksize from the SSZ column in the output of the following command:

blockdev --report $DEVICE
target_removable=”true|false”:

Indicate if the target disk for oem images is deployed to a removable device e.g a USB stick or not. This only affects the EFI setup if requested and in the end avoids the creation of a custom boot menu entry in the firmware of the target machine. By default the target disk is expected to be non-removable

selinux_policy.attribute=”targeted|mls|minimum”:

The selinux_policy attribute sets the SELinux policy to use. targeted policy is the default policy. Only change this option if you want to use the mls or minimum policy.

spare_part=”number”:

Request a spare partition right before the root partition of the requested size. The attribute takes a size value and allows a unit in MB or GB, e.g 200M. If no unit is given the value is considered to be mbytes. A spare partition can only be configured for the disk image type oem

spare_part_mountpoint=”dir_path”:

Specify mount point for spare partition in the system. Can only be configured for the disk image type oem

spare_part_fs=”btrfs|ext2|ext3|ext4|xfs”:

Specify filesystem for spare partition in the system. Can only be configured for the disk image type oem

spare_part_fs_attributes=”attribute_list”:

Specify filesystem attributes for the spare partition. Attributes can be specified as comma separated list. Currently the attributes no-copy-on-write and synchronous-updates are available. Can only be configured for the disk image type oem

spare_part_is_last=”true|false”:

Specify if the spare partition should be the last one in the partition table. Can only be configured for the oem type with oem-resize switched off. By default the root partition is the last one and the spare partition lives before it. With this attribute that setup can be toggled. However, if the root partition is no longer the last one the oem repart/resize code can no longer work because the spare part would block it. Because of that moving the spare part at the end of the disk is only applied if oem-resize is switched off. There is a runtime check in the KIWI NG code to check this condition

devicepersistency=”by-uuid|by-label”:

Specifies which method to use for persistent device names. This will affect all files written by kiwi that includes device references for example etc/fstab or the root= parameter in the kernel commandline. By default by-uuid is used

squashfscompression=”uncompressed|gzip|lzo|lz4|xz|zstd”:

Specifies the compression type for mksquashfs

standalone_integrity=”true|false”:

For the oem type only, specifies to create a standalone dm_integrity layer on top of the root filesystem

integrity_keyfile=”filepath”:

For the oem type only and in combination with the standalone_integrity attribute, protects access to the integrity map using the given keyfile.

integrity_metadata_key_description=”string”:

For the oem type only and in combination with the embed_integrity_metadata attribute, specifies a custom description of an integrity key as it is expected to be present in the kernel keyring. The information is placed in the integrity metadata block. If not specified kiwi creates a key argument string instead which is based on the given integrity_keyfile filename. The format of this key argument is:

:BASENAME_OF_integrity_keyfile_WITHOUT_FILE_EXTENSION
embed_integrity_metadata=”true|false”:

For the oem type only, and in combination with the standalone_integrity attribute, specifies to write a binary block at the end of the partition serving the root filesystem, containing information to create the dm_integrity device map in the following format:

|header|0xFF|dm_integrity_meta|0xFF|0x0|
header:

Is a string of the following information separated by spaces

  • version: currently set to 1

  • fstype: name of filesystem attribute

  • access: either ro or rw depending on the filesystem capabilities

  • integrity: fixed identifier value

dm_integrity_meta:

Is a string of the following information separated by spaces

  • provided_data_sectors: number of data sectors

  • sector_size: sector size in byte, defaults to 512

  • parameter_count: number of parameters needed to construct the integrity device map. After the parameter_count a list of space separated parameters follows and the parameter_count specifies the quantity of these parameters

  • parameters: The first element of the parameter list contains information about the used hash algorithm which is not part of the superblock and provided according to the parameters passed along when KIWI NG calls integritysetup. As of now this defaults to:

    • internal_hash:sha256

    All subsequent parameters are taken from the flags field of the dm-integrity superblock. see the dm-integrity documentation on the web for possible flag values.

verity_blocks=”number|all”:

For the oem type only, specifies to create a dm verity hash from the number of given blocks (or all) placed at the end of the root filesystem For later verification of the device, the credentials information produced by veritysetup from the cryptsetup tools are needed. This data as of now is only printed as debugging information to the build log file. A concept to persistently store the verification metadata as part of the partition(s) will be a next step.

embed_verity_metadata=”true|false”:

For the oem type only, and in combination with the verity_blocks attribute, specifies to write a binary block at the end of the partition serving the root filesystem, containing information for dm_verity verification in the following format:

|header|0xFF|dm_verity_credentials|0xFF|0x0|
header:

Is a string of the following information separated by spaces

  • version: currently set to 1

  • fstype: name of filesystem attribute

  • access: either ro or rw depending on the filesystem capabilities

  • verity: fixed identifier value

dm_verity_credentials:

Is a string of the following information separated by spaces

  • hash_type: hash type name as returned by veritysetup

  • data_blksize: data blocksize as returned by veritysetup

  • hash_blksize: hash blocksize as returned by veritysetup

  • data_blocks: number of data blocks as returned by veritysetup

  • hash_start_block: hash start block as required by the kernel to construct the device map

  • algorithm: hash algorithm as returned by veritysetup

  • root_hash: root hash as returned by veritysetup

  • salt: salt hash as returned by veritysetup

overlayroot=”true|false”:

For the oem type only, specifies to use an overlayfs based root filesystem consisting out of a squashfs compressed read-only root filesystem combined with a write-partition or tmpfs. The optional kernel boot parameter rd.root.overlay.readonly can be used to point the write area into a tmpfs instead of the existing persistent write-partition. In this mode all written data is temporary until reboot of the system. The kernel boot parameter rd.root.overlay.size can be used to configure the size for the tmpfs that is used for the overlayfs mount process if rd.root.overlay.readonly is requested. That size basically configures the amount of space available for writing new data during the runtime of the system. The default value is set to 50% which means one half of the available RAM space can be used for writing new data. By default the persistent write-partition is used. The size of that partition can be influenced via the optional <size> element in the <type> section or via the optional <oem-resize> element in the <oemconfig> section of the XML description. Setting a fixed <size> value will set the size of the image disk to that value and results in an image file of that size. The available space for the write partition is that size reduced by the size the squashfs read-only system needs. If the <oem-resize> element is set to true an eventually given <size> element will not have any effect because the write partition will be resized on first boot to the available disk space.

overlayroot_write_partition=”true|false”:

For the oem type only, allows to specify if the extra read-write partition in an overlayroot setup should be created or not. By default the partition is created and the kiwi-overlay dracut module also expect it to be present. However, the overlayroot feature can also be used without dracut (initrd_system="none") and under certain circumstances it is handy to configure if the partition table should contain the read-write partition or not.

overlayroot_readonly_partsize=”mbsize”:

Specifies the size in MB of the partition which stores the squashfs compressed read-only root filesystem in an overlayroot setup. If not specified kiwi calculates the needed size by a preliminary creation of the squashfs compressed file. However this is only accurate if no changes to the root filesystem data happens after this calculation, which cannot be guaranteed as there is at least one optional script hook which is allowed and applied after the calculation. In addition the pre-calculation requires some time in the build process. If the value can be provided beforehand this also speeds up the build process significantly

bootfilesystem=”btrfs|ext2|ext3|ext4|xfs|fat32|fat16”:

If an extra boot partition is required this attribute specify which filesystem should be used for it. The type of the selected bootloader might overwrite this setting if there is no alternative possible though.

flags=”overlay|dmsquash”:

For the iso image type specifies the live iso technology and dracut module to use. If set to overlay the kiwi-live dracut module will be used to support a live iso system based on squashfs+overlayfs. If set to dmsquash the dracut standard dmsquash-live module will be used to support a live iso system based on the capabilities of the upstream dracut module.

format=”gce|ova|qcow2|vagrant|vmdk|vdi|vhd|vhdx|vhd-fixed”:

For disk image type oem, specifies the format of the virtual disk such that it can run on the desired target virtualization platform.

formatoptions=”string”:

Specifies additional format options passed on to qemu-img formatoptions is a comma separated list of format specific options in a name=value format like qemu-img expects it. kiwi will take the information and pass it as parameter to the -o option in the qemu-img call

fsmountoptions=”string”:

Specifies the filesystem mount options which also ends up in fstab The string given here is passed as value to the -o option of mount

fscreateoptions=”string”:

Specifies options to use at creation time of the filesystem

force_mbr=”true|false”:

Force use of MBR (msdos table) partition table even if the use of the GPT would be the natural choice. On e.g some arm systems an EFI partition layout is required but must not be stored in a GPT. For those rare cases this attribute allows to force the use of the msdos table including all its restrictions in max partition size and amount of partitions

gpt_hybrid_mbr=”true|false”:

For GPT disk types only: Create a hybrid GPT/MBR partition table

hybridpersistent=”true|false”:

For the live ISO type, triggers the creation of a partition for a COW file to keep data persistent over a reboot

hybridpersistent_filesystem=”ext4|xfs”:

For the live ISO type, set the filesystem to use for persistent writing if a hybrid image is used as disk on e.g a USB Stick. By default the ext4 filesystem is used.

initrd_system=”kiwi|dracut|none”:

Specify which initrd builder to use, default is set to dracut. If set to none the image is build without an initrd. Depending on the image type this can lead to a non bootable system as its now a kernel responsibility if the given root device can be mounted or not.

metadata_path=”dir_path”:

Specifies a path to additional metadata required for the selected image type or its tools used to create that image type.

Note

Currently this is only effective for the appx container image type.

installboot=”failsafe-install|harddisk|install”:

Specifies the bootloader default boot entry for the initial boot of a KIWI NG install image.

Note

This value is only evaluated for grub

install_continue_on_timeout=”true|false”:

Specifies the boot timeout handling for the KIWI NG install image. If set to “true” the configured timeout or its default value applies. If set to “false” no timeout applies in the boot menu of the install image.

installprovidefailsafe=”true|false”:

Specifies if the bootloader menu should provide an failsafe entry with special kernel parameters or not

installiso=”true|false”

Specifies if an install iso image should be created. This attribute is only available for the oem type. The generated ISO image is an hybrid ISO which can be used as disk on e.g a USB stick or as ISO.

installpxe=”true|false”:

Specifies if a tarball that contains all data for a pxe network installation should be created. This attribute is only available for the oem type.

mediacheck=”true|false”:

For ISO images, specifies if the bootloader menu should provide an mediacheck entry to verify ISO integrity or not. Disabled by default and only available for the x86 arch family.

mdraid=”mirroring|striping”:

Setup software raid in degraded mode with one disk Thus only mirroring and striping is possible

primary=”true|false”:

Specifies this type to be the primary type. If no type option is given on the commandline, KIWI NG will build this type

ramonly=”true|false”:

For all images that are configured to use the overlay filesystem this setting forces any COW(Copy-On-Write) action to happen in RAM.

rootfs_label=”string”:

Label name to set for the root filesystem. By default ROOT is used

volid=”string”:

For the ISO type only, specifies the volume ID (volume name or label) to be written into the master block. There is space for 32 characters.

vhdfixedtag=”GUID_string”:

For the VHD disk format, specifies the GUID

derived_from=”string”:

For container images, specifies the image URI of the container image. The image created by KIWI NG will use the specified container as the base root to work on.

ensure_empty_tmpdirs=”true|false”:

For OCI container images, specifies whether to ensure /run and /tmp directories are empty in the container image created by Kiwi. Default is true.

publisher=”string”:

For ISO images, specifies the publisher name of the ISO.

The following sections shows the supported child elements of the type element including references to their usage in a detailed type setup:

<preferences><type><luksformat>

The luksformat element is used to specify additional luks options passed on to the cryptsetup luksFormat call. The element requires the attribute luks to be set in the <type> section referring to luksformat. Several custom settings related to the LUKS and LUKS2 format features can be setup. For example the setup of the dm_integrity feature:

<luksformat>
  <option name="--cipher" value="aes-gcm-random"/>
  <option name="--integrity" value="aead"/>
</luksformat>

<preferences><type><bootloader>

The bootloader element is used to select the bootloader. At the moment, grub2, isolinux, zipl and grub2_s390x_emu (a combination of zipl and a userspace GRUB2) are supported. The special custom entry allows to skip the bootloader configuration and installation and leaves this up to the user, which can be done by using the editbootinstall and editbootconfig custom scripts.

In addition to the mandatory name attribute, the following optional attributes are supported:

console=”console|gfxterm|serial”:

Specifies the bootloader console. The attribute is available for the grub and isolinux bootloader types. By default, a graphics console setup is used.

grub_template=”filename”:

Specifies a custom grub bootloader template file which will be used instead of the one provided with Kiwi. A static bootloader template to create the grub config file is only used in Kiwi if the native method via the grub mkconfig toolchain does not work properly. As of today, this is only the case for live and install ISO images. Thus, this setting only affects the oem and iso image types.

The template file should contain a Template string and can use the following variables:

Variable

Description

search_params

parameters needed for grub’s search command to locate the root volume

default_boot

number of the default menu item to boot

kernel_file

the name of the kernel file

initrd_file

the name of the initial ramdisk file

boot_options

kernel command line options for booting normally

failsafe_boot_options

kernel command line options for booting in failsafe mode

gfxmode

the resolution to use for the bootloader; passed to grub’s gfxmode command

theme

the name of a graphical theme to use

boot_timeout

the boot menu timeout, set by the timeout attribute

boot_timeout_style

the boot timeout style, set by the timeout_style attribute

serial_line_setup

directives used to initialize the serial port, set by the serial_line attribute

title

a title for the image: this will be the <image> tag’s displayname attribute or its name attribute if displayname is not set; see: <image>

bootpath

the bootloader lookup path

boot_directory_name

the name of the grub directory

efi_image_name

architecture-specific EFI boot binary name

terminal_setup

the bootloader console mode, set by the console attribute
serial_line=”string”:

Specifies the bootloader serial line setup. The setup is effective if the bootloader console is set to use the serial line. The attribute is available for the grub bootloader only.

timeout=”number”:

Specifies the boot timeout in seconds prior to launching the default boot option. By default, the timeout is set to 10 seconds. It makes sense to set this value to 0 for images intended to be started non-interactively (e.g. virtual machines).

timeout_style=”countdown|hidden”:

Specifies the boot timeout style to control the way in which the timeout interacts with displaying the menu. If set, the display of the bootloader menu is delayed after the timeout expired. In countdown mode, an indication of the remaining time is displayed. The attribute is available for the grub loader only.

targettype=”CDL|LDL|FBA|SCSI”:

Specifies the device type of the disk zipl should boot. On zFCP devices, use SCSI; on DASD devices, use CDL or LDL; on emulated DASD devices, use FBA. The attribute is available for the zipl loader only.

<preferences><type><containerconfig>

Used to describe the container configuration metadata in docker or wsl image types. For details see: Build a Container Image and: Build a WSL Container Image

<preferences><type><vagrantconfig>

Used to describe vagrant configuration metadata in a disk image that is being used as a vagrant box. For details see: Image Description for Vagrant

<preferences><type><systemdisk>

Used to describe the volumes of the disk area which contains the root filesystem. Volumes are either a feature of the used filesystem or LVM is used for this purpose. For details see: Custom Disk Volumes

Note

When both <partitions> and <systemdisk> are used, <partitions> are evaluated first and mount points defined in <partitions> cannot be redefined as <systemdisk> volumes. The two types define a complete disk setup, so there cannot be any overlapping volumes or mount points. As a result, whatever is written in <partitions> cannot be expressed in the same way in <volumes>.

<preferences><type><partitions>

Used to describe the geometry of the disk on the level of the partition table. For details see: Custom Disk Partitions

<preferences><type><oemconfig>

Used to customize the deployment process in an oem disk image. For details see: OEM Customization

<preferences><type><size>

Used to customize the size of the resulting disk image in an oem image. For details see: Modifying the Size of the Image

<preferences><type><machine>

Used to customize the virtual machine configuration which describes the components of an emulated hardware. For details see: Customizing the Virtual Machine

<preferences><type><installmedia>

Used to customize the installation media images created for oem images deployment. For details see: Installation Media Customization

<repository>

Setup software sources for the image.

<repository>
  <source path="uri"/>
</repository>

The mandatory repository element specifies the location and type of a repository to be used by the package manager as a package installation source. KIWI NG supports apt, dnf, pacman and zypper as package managers, specified with the packagemanager element. The repository element has the following optional attributes:

alias=”name”

Specifies an alternative name for the configured repository. If the attribute is not specified KIWI NG will generate a random alias name for the repository.

components=”name”

Used for Debian (apt) based repositories only. Specifies the component name that should be used from the repository. By default the main component is used

distribution=”name”

Used for Debian (apt) based repositories only. Specifies the distribution name to be used on call of debootstrap

imageonly=”true|false”

Specifies whether or not this repository should be configured in the resulting image without using it at build time. By default the value is set to false

repository_gpgcheck=”true|false”

Specifies whether or not this specific repository is configured to to run repository signature validation. If not set, no value is appended into the repository configuration file. If set the relevant key information needs to be provided on the KIWI NG commandline using the --signing-key option or via the <signing> element as part of the <repository> setting in the image description.

customize=”/path/to/custom_script”

Custom script hook which is invoked with the repo file as parameter for each file created by KIWI NG.

Note

If the script is provided as relative path it will be searched in the image description directory

imageinclude=”true|false”

Specifies whether the given repository should be configured as a repository in the image or not. The default behavior is that repositories used to build an image are not configured as a repository inside the image. This feature allows you to change the behavior by setting the value to true.

Note

Scope of repository uri’s

The repository is configured in the image according to the source path as specified with the path attribute of the source element. Therefore, if the path is not a fully qualified URL, you may need to adjust the repository file in the image to accommodate the expected location. It is recommended that you use the alias attribute in combination with the imageinclude attribute to avoid having unpredictable random names assigned to the repository you wish to include in the image.

password=”string”

Specifies a password for the given repository. The password attribute must be used in combination with the username attribute. Dependent on the repository location this information may not be used.

username=”name”

Specifies a user name for the given repository. The username attribute must be used in combination with the password attribute. Dependent on the repository location this information may not be used.

prefer-license=”true|false”

The repository providing this attribute will be used primarily to install the license tarball if found on that repository. If no repository with a preferred license attribute exists, the search happens over all repositories. It’s not guaranteed in that case that the search order follows the repository order like they are written into the XML description.

priority=”number”

Specifies the repository priority for this given repository. Priority values are treated differently by different package managers. Repository priorities allow the package management system to disambiguate packages that may be contained in more than one of the configured repositories. The zypper package manager for example prefers packages from a repository with a lower priority over packages from a repository with higher priority values. The value 99 means “no priority is set”. For other package managers please refer to the individual documentation about repository priorities.

sourcetype=”baseurl|metalink|mirrorlist”

Specifies the source type of the repository path. Depending on if the source path is a simple url or a pointer to a metadata file or mirror list, the configured package manager needs to be setup appropriately. By default the source is expected to be a simple repository baseurl.

use_for_bootstrap=”true|false”

Used for Debian (apt) based repositories only. It specifies whether this repository should be the one used for bootstrapping or not. It is set to ‘false’ by default. Only a single repository is allowed to be used for bootstrapping, if no repository is set for the bootstrap the last one in the description XML is used.

<repository><source>

The location of a repository is specified by the path attribute of the mandatory source child element:

<repository alias="kiwi">
  <source path="obs://Virtualization:Appliances:Builder/openSUSE_Leap_15.3"/>
</repository>

The location specification may include the %arch macro which will expand to the architecture of the image building host. The value for the path attribute may begin with any of the following location indicators:

  • dir:///local/path An absolute path to a directory accessible through the local file system.

  • ftp://<ftp://> A ftp protocol based network location.

  • http://<http://> A http protocol based network location.

  • https://<https://> A https protocol based network location.

    Note

    https repositories

    When specifying a https location for a repository it is generally necessary to include the openssl certificates and a cracklib word dictionary as package entries in the bootstrap section of the image configuration. The names of the packages to include are individual to the used distribution. On SUSE systems as one example this would be openssl-certs and cracklib-dict-full

  • iso://<iso://> An absolute path to an .iso file accessible via the local file system. KIWI NG will loop mount the the .iso file to a temporary directory with a generated name. The generated path is provided to the specified package manager as a directory based repository location.

  • obs://Open:Build:Service:Project:Name A reference to a project in the Open Build Service (OBS). KIWI NG translates the given project path into a remote url at which the given project hosts the packages.

  • obsrepositories:/ A placeholder for the Open Build Service (OBS) to indicate that all repositories are taken from the project configuration in OBS.

A repository <source> element can optionally contain one ore more signing keys for the packages from this repository like shown in the following example:

<repository alias="kiwi">
  <source path="obs://Virtualization:Appliances:Builder/openSUSE_Leap_15.3">
    <signing key="/path/to/sign_key_a"/>
    <signing key="/path/to/sign_key_b"/>
  </source>
</repository>

All signing keys from all repositories will be collected and incorporated into the keyring as used by the selected package manager.

<packages>

Setup software components to be installed in the image.

<packages type="type"/>

The mandatory packages element specifies the setup of a packages group for the given type. The value of the type attribute specifies at which state in the build process the packages group gets handled, supported values are as follows:

type=”bootstrap”

Bootstrap packages, list of packages to be installed first into a new (empty) root tree. The packages list the required components to support a chroot environment from which further software components can be installed

type=”image”

Image packages, list of packages to be installed as part of a chroot operation inside of the new root tree.

type=”uninstall|delete”

Packages to be uninstalled or deleted. For further details see Uninstall System Packages

type=”image_type_name”

Packages to be installed for the given image type name. For example if set to type=”iso”, the packages in this group will only be installed if the iso image type is build.

The packages element must contain at least one child element of the following list to provide specific configuration information for the specified packages group:

<packages><package>

<packages type="image"/>
  <package name="name" arch="arch"/>
</packages>

The package element installs the given package name. The optional arch attribute can be used to limit the installation of the package to the host architecture from which KIWI NG is called. The arch attribute is also available in all of the following elements.

<packages><namedCollection>

<packages type="image" patternType="onlyRequired">
  <namedCollection name="base"/>
</packages>

The namedCollection element is used to install a number of packages grouped together under a name. This is a feature of the individual distribution and used in the implementation of the KIWI NG package manager backend. At the moment collections are only supported for SUSE and RedHat based distributions. The optional patternType attribute is used to control the behavior of the dependency resolution of the package collection. onlyRequired installs only the collection and its required packages. plusRecommended installs the collection, any of its required packages and any recommended packages.

Note

Collections on SUSE

On SUSE based distributions collections are called patterns and are just simple packages. To get the names of the patterns such that they can be used in a namedCollection type the following command: $ zypper patterns. If for some reason the collection name cannot be used it is also possible to add the name of the package that provides the collection as part of a package element. To get the names of the pattern packages type the following command: $ zypper search patterns. By convention all packages that starts with the name “patterns-” are representing a pattern package.

Note

Collections on RedHat

On RedHat based distributions collections are called groups and are extra metadata. To get the names of these groups type the following command: $ dnf group list. Please note that group names are allowed to contain whitespace characters.

<packages><collectionModule>

<packages type="bootstrap">
    <collectionModule name="module" stream="stream" enable="true|false"/>
</packages>

In CentOS Stream >= 8 and Red Hat Enterprise Linux >= 8, there are Application Streams that are offered in the form of modules (using Fedora Modularity technology). To build images that use this content KIWI NG offers to enable/disable modules when using the dnf or microdnf package manager backend. Modules are setup prior the bootstrap phase and its setup persists as part of the image.

There are the following constraints when adding collectionModule elements:

  • collectionModule elements can only be specified as part of the <packages type="bootstrap"> section. This is because the setup of modules must be done once and as early as possible in the process of installing the image root tree.

  • Disabling a module can only be done as a whole and therefore the stream attribute is not allowed for disabling modules. For enabling modules the stream` attribute is optional

  • The enable attribute is mandatory because it should be an explicit setting if a module is effectively used or not.

<packages><archive>

<packages type="image"/>
  <archive name="name" target_dir="some/path"/>
</packages>

The archive element takes the name attribute and looks up the given name as file on the system. If specified relative KIWI NG looks up the name in the image description directory. The archive is installed using the tar program. Thus the file name is expected to be a tar archive. The compression of the archive is detected automatically by the tar program. The optional target_dir attribute can be used to specify a target directory to unpack the archive.

<packages><ignore>

<packages type="image"/>
  <ignore name="name"/>
</packages>

The ignore element instructs the used package manager to ignore the given package name at installation time. Please note whether or not the package can be ignored is up to the package manager. Packages that are hard required by other packages in the install procedure cannot be ignored and the package manager will simply ignore the request.

<packages><product>

<packages type="image">
  <product name="name"/>
</packages>

The product element instructs the used package manager to install the given product. What installation of a product means is up to the package manager and also distribution specific. This feature currently only works on SUSE based distributions

<users>

Setup image users.

<users>
  <user
    name="user"
    groups="group_list"
    home="dir"
    id="number"
    password="text"
    pwdformat="encrypted|plain"
    realname="name"
    shell="path"
  />
</users>

The optional users element contains the user setup KIWI NG should create in the system. At least one user child element must be specified as part of the users element. Multiple user elements may be specified.

Each user element represents a specific user that is added or modified. The following attributes are mandatory:

name=”name”:

the UNIX username

password=”string”

The password for this user account. It can be provided either in cleartext form or encrypted. An encrypted password can be created using openssl as follows:

$ openssl passwd -1 -salt xyz PASSWORD

It is also possible to specify the password as a non encrypted string by using the pwdformat attribute and setting it’s value to plain. KIWI NG will then encrypt the password prior to the user being added to the system.

Warning

plain text passwords

We do not recommend plain passwords as they will be readable in the image configuration in plain text

All specified users and groups will be created if they do not already exist. The defined users will be part of the group(s) specified with the groups attribute or belong to the default group as configured in the system. If specified the first entry in the groups list is used as the login group.

Additionally, the following optional attributes can be specified:

home=”path”:

The path to the user’s home directory

groups=”group_a,group_b,group_c:id”:

A comma separated list of UNIX groups. The first element of the list is used as the user’s primary group. The remaining elements are appended to the user’s supplementary groups. When no groups are assigned then the system’s default primary group will be used. If a group should be of a specific group id, it can be appended to the name separated by a colon.

Note

Group ID’s can only be set for groups that does not yet exist at the time when KIWI NG creates them. A check is made if the desired group is already present and if it exists the user will become a member of that group but any given group ID from the KIWI NG configuration will not be taken into account. Usually all standard system groups are affected by this behavior because they are provided by the OS itself. Thus it’s by intention not possible to overwrite the group ID of an existing group.

id=”number”:

The numeric user id of this account.

pwdformat=”plain|encrypted”:

The format in which password is provided. The default if not specified is encrypted.

<profiles>

Manage image namespace(s).

<profiles>
  <profile name="name" description="text"/>
</profiles>

The optional profiles section lets you maintain one image description while allowing for variation of other sections that are included. A separate profile element must be specified for each variation. The profile child element, which has name and description attributes, specifies an alias name used to mark sections as belonging to a profile, and a short description explaining what this profile does.

For example to mark a set of packages as belonging to a profile, simply annotate them with the profiles attribute as shown below:

<packages type="image" profiles="profile_name">
  <package name="name"/>
</packages>

It is also possible to mark sections as belonging to multiple profiles by separating the names in the profiles attribute with a comma:

<packages type="image" profiles="profile_A,profile_B">
  <package name="name"/>
</packages>

If a section tag does not have a profiles attribute, it is globally present in the configuration. If global sections and profiled sections contains the same sub-sections, the profiled sections will overwrite the global sections in the order of the provided profiles. For a better overview of the result configuration when profiles are used we recommend to put data that applies in any case to non profiled (global) sections and only extend those global sections with profiled data. For example:

<preferences>
  <version>1.2.3</version>
  <packagemanager name="zypper"/>
</preferences>

<preferences profiles="oem_qcow_format">
  <type image="oem" filesystem="ext4" format="qcow2"/>
</preferences>

<preferences profiles="oem_vmdk_format">
  <type image="oem" filesystem="ext4" format="vmdk"/>
</preferences>

The above example configures two version of the same oem type. One builds a disk in qcow2 format the other builds a disk in vmdk format. The global preferences section without a profile assigned will be used in any case and defines those preferences settings that are common to any build process. A user can select both profiles at a time but that will result in building the disk format that is specified last because one is overwriting the other.

Use of one or more profile(s) during image generation is triggered by the use of the --profile command line argument. multiple profiles can be selected by passing this option multiple times.