AIF Configuration File

From ArchWiki
Jump to navigation Jump to search

This template has only maintenance purposes. For linking to local translations please use interlanguage links, see Help:i18n#Interlanguage links.

Local languages: Català – Dansk – English – Español – Esperanto – Hrvatski – Indonesia – Italiano – Lietuviškai – Magyar – Nederlands – Norsk Bokmål – Polski – Português – Slovenský – Česky – Ελληνικά – Български – Русский – Српски – Українська – עברית – العربية – ไทย – 日本語 – 正體中文 – 简体中文 – 한국어

External languages (all articles in these languages should be moved to the external wiki): Deutsch – Français – Română – Suomi – Svenska – Tiếng Việt – Türkçe – فارسی


The Arch Installation Framework is the backbone to the Arch Linux distribution installer. The AIF executes all of the routines used to install Arch. More information about AIF proper can be found in the official installation guide.

The AIF configuration file contains variables that define how Arch will be installed. This article describes these variables, and how to make use of the configuration file.

The AIF configuration file

The aif command can be executed with a configuration file which defines how Arch is to be installed. This is a similar concept to Red Hat's kickstart and Debian's preseed. But the similarities end with the concept, thanks to the Arch KISS philosophy the pain involved in automated installs is swept away. AIF is written entirely in bash, and the AIF configuration file is just that, bash. All of the variables and function definitions in the AIF configuration file are sourced shortly after the install begins and applied to the install.

With this in mind, the AIF configuration file is simply a representation of all of that data which can be defined in the global scope of the AIF. This means that available variables can be found in the AIF source code.

Optional variables

The following variables are available in the AIF configuration file:


The source variable will be either CD or net, and will define whether the packages used are locally available or found over the network/Internet.

for local files:


or for a source over the network/Internet



The local directory containing Arch Linux packages, used if SOURCE=cd



The network location used to sync packages over the network. What is important about this variable is that it is the same line that is placed in /etc/pacman.d/mirrorlist. So it is wise to use the $repo and $arch variables:





The groups to install, like base




Packages to exclude from the install

TARGET_PACKAGES_EXCLUDE='reiserfsprogs nano'


Packages to include in the install

TARGET_PACKAGES='openssh vim'

Mandatory variables

These variables are the hard ones, and honestly the primary reason for this wiki entry, because they, like the hard drives they configure, can be complicated


The GRUB_DEVICE variable is a simple one, it is the device that will hold the grub mbr. This is almost always the primary hard drive.



The partitions line is used to load up sfdisk for disk carving. It is a space delimited list which begins with the device to carve up and then a set of colon delimited arguments which define the device sizes, type and options.

PARTITIONS='/dev/sda 100:ext2:+ 512:swap *:ext4'

Here the disk to be carved up is /dev/sda, 3 partitions will be created, /dev/sda1 will be 100 Megabytes, Linux type, and the "+" sets the grub bootable flag. /dev/sda2 is 512 Megabytes and the swap type, /dev/sda3 will take up the rest of the space, and have the type Linux.

Entry syntax:


      FILESYSTEM_TYPE := { raw | ext2 | ext3 | ext4 | xfs | jfs | vfat | brtfs | swap | lvm-pv | lvm-vg | lvm-lv }

If you want to skip partitioning, override one of the worker functions to do nothing:

 worker_prepare_disks () {


The BLOCKDATA variable is used to determine which partitions are formatted and how they are to be mounted. While static partitions are simple and straightforward, LVM and crypt setups on the other hand can get somewhat messy.

Since AIF translates components of the variables into arguments for commands, understanding what the commands are and how the arguments are passed will greatly assist in understanding how to use the variables.

Entry syntax:

  FS_STRING      := { FS_DESCRIPTION | no_fs }


     RECREATE    := { yes | no }                     # Re-format?
     PARAMS      := { PARAMETERS | no_params }       # Formating utility parameters
     MOUNT_PATH  := { MOUNTPOINT | no_mountpoint }   # Partition mountpoint
     MOUNT_OPTS  := { MOUNTOPTS  | no_opts }         # Mounting options (see: man mount)
     MOUNT_LOCATION := { target | runtime | no }     # Where to mount partition:
                                                     #    target  - on system being installed
                                                     #    runtime - on host system
                                                     #    no      - do not mount

  TYPE        := { raw | ext2 | ext3 | ext4 | xfs | jfs |
                   vfat | brtfs | swap | lvm-pv | lvm-vg | lvm-lv }
  LABEL_NAME  := { LABEL | no_label }

   PARAMS/MOUNT_OPTS - spaces need to be replaced with underscores (_)
Simple file systems

Simple file systems can be defined on top of partitions derived from the PARTITIONS variable.

A simple example:

BLOCKDATA='/dev/vda1 raw no_label ext2;yes;/boot;target;no_opts;no_label;no_params
/dev/vda2 raw no_label swap;yes;no_mountpoint;target;no_opts;no_label;no_params
/dev/vda3 raw no_label ext4;yes;/;target;no_opts;no_label;no_params'

In this scenario, /dev/vda1 is a raw block device with no label, it is to be formatted with the ext2 file system, yes verifies that it will be formatted, /boot is the mount point, target <Not sure what target means yet>, no_opts is where file system mount options are placed, no_label can be replaced with a file system label. The parameters section is used by more advanced partitioning mechanisms.

Advanced file systems

This example will create a /boot partition and a LVM volume group called *system* with 5 partitions:

/boot     => /dev/sda1              (ext3)
/         => /dev/system/root [LVM] (xfs)
/home     => /dev/system/home [LVM] (xfs)
/tmp      => /dev/system/tmp  [LVM] (xfs)
/var      => /dev/system/var  [LVM] (xfs)
/usr      => /dev/system/usr  [LVM] (xfs)

The entry has extra whitespaces in order to increase readability (all whitespaces get removed at the end).

BLOCKDATA=$(echo '
/dev/sda1               raw    no_label ext3;    yes; /boot        ; target; no_opts; no_label; no_params
/dev/sda2               raw    no_label lvm-pv;  yes; no_mountpoint; target; no_opts; no_label; no_params
/dev/sda2+              lvm-pv no_label lvm-vg;  yes; no_mountpoint; target; no_opts; system  ; /dev/sda2

/dev/mapper/system      lvm-vg system   lvm-lv;  yes; no_mountpoint; target; no_opts; home    ; 1025M
/dev/mapper/system      lvm-vg system   lvm-lv;  yes; no_mountpoint; target; no_opts; root    ; 1024M
/dev/mapper/system      lvm-vg system   lvm-lv;  yes; no_mountpoint; target; no_opts; tmp     ; 1024M
/dev/mapper/system      lvm-vg system   lvm-lv;  yes; no_mountpoint; target; no_opts; var     ; 1024M
/dev/mapper/system      lvm-vg system   lvm-lv;  yes; no_mountpoint; target; no_opts; usr     ; 2048M

/dev/mapper/system-home lvm-lv home     xfs;     yes; /home        ; target; no_opts; no_label; no_params
/dev/mapper/system-root lvm-lv root     xfs;     yes; /            ; target; no_opts; no_label; no_params
/dev/mapper/system-tmp  lvm-lv tmp      xfs;     yes; /tmp         ; target; no_opts; no_label; no_params
/dev/mapper/system-var  lvm-lv var      xfs;     yes; /var         ; target; no_opts; no_label; no_params
/dev/mapper/system-usr  lvm-lv usr      xfs;     yes; /usr         ; target; no_opts; no_label; no_params'| sed -r 's/ *; */;/g' | tr -s ' ')

Using the AIF configuration file


If you want to set the current keymap:

# aif -p partial-keymap

and network settings:

# aif -p partial-configure-network

(both settings are also automatically written to the installed system)

# aif -p automatic -c <file>


If the AIF is installed, examples can be found in /usr/share/aif/examples/. Otherwise, see them on the project page.