Beginners' guide

From ArchWiki
Revision as of 02:01, 8 September 2015 by Pid1 (talk | contribs) (UEFI/GPT: These two should match.)
Jump to: navigation, search

ro:Ghidul începătorilor zh-cn:Beginners' guide zh-tw:Beginners' guide

This document will guide you through the process of installing Arch Linux using the Arch Install Scripts. Before installing, you are advised to skim over the FAQ.

The community-maintained ArchWiki is the primary resource that should be consulted if issues arise. The IRC channel (irc:// and the forums are also excellent resources if an answer cannot be found elsewhere. In accordance with the Arch Way, you are encouraged to type man command to read the man page of any command you are unfamiliar with.


Arch Linux should run on any i686 compatible machine with a minimum of 256 MB RAM. A basic installation with all packages from the base group should take less than 800 MB of disk space.

The installation media and their GnuPG signatures can be acquired from the Download page. The single ISO image supports both 32bit and 64bit systems; this guide assumes you use the latest available version.

It is highly recommended to verify the image signature before use, especially when downloading from an HTTP mirror, as these are run by volunteers who could serve malicious images. On a system with GnuPG installed, do this by downloading the PGP signature (under Checksums) to the ISO directory, and run:

$ gpg --verify archlinux-<version>-dual.iso.sig

If the public key is not found, import it with gpg --recv-keys key-id.

Alternatively, run from an existing Arch Linux installation:

$ pacman-key -v archlinux-<version>-dual.iso.sig

Now, choose one of the methods from Category:Getting and installing Arch to #Boot the installation medium on the target machine(s). As the installation process retrieves packages from a remote repository, these methods require an internet connection; See Offline installation of packages when none is available.

Boot the installation medium

Point the current boot device to the media containing the Arch installation media. This is typically achieved by pressing a key during the POST phase, as indicated on the splash screen. Refer to your motherboard's manual for details.

When the Arch menu appears, select Boot Arch Linux and press Enter to enter the installation environment. See README.bootparams for a list of boot parameters. If the live system fails to launch, see Boot problems for possible workarounds.

You will be logged in as the root user and presented with a Zsh shell prompt. Zsh provides advanced tab completion and other features as part of the grml config. For modifying or creating configuration files, nano and vim are suggested.

UEFI mode

In case you have a UEFI motherboard with UEFI mode enabled, the CD/USB will automatically launch Arch Linux via systemd-boot.

To verify you are booted in UEFI mode, run:

# efivar -l

Should efivar not list the UEFI variables properly, check if all requirements are met.

Keyboard layout

The default keymap for the virtual console is set to us. To change the layout for the installation process (only), run:

# loadkeys layout

where layout is a keymap file from the /usr/share/kbd/keymaps/ directory tree. Use localectl list-keymaps to list all available choices.

See Console fonts if certain characters appear as white squares or other symbols.

Establish an internet connection

The dhcpcd daemon is enabled on boot, and will attempt to start a wired connection. The ELinks browser is included with the installation media, and can be used to access captive portal login forms.

Should checking the connection with ping fail, configure the network as explained below. See Category:Network configuration for other methods, such as a modem connection.

To preserve settings, copy modified configuration files to the new system before configuring the base system.


Use netctl's wifi-menu to list available networks and make a connection:

# wifi-menu

If your computer has more than one Wi-Fi device, list the wireless devices with iw dev. Names are likely to start with wl, for example wlp3s0.

Relay the name to wifi-menu:

# wifi-menu wlp3s0

For networks which require both a username and password, see WPA2 Enterprise#netctl.

See Wireless for more information.

Wired (Static IP)

Identify the device names with ip link. Ethernet names are likely to start with en, for example enp0s25.

Configure a static profile for dhcpcd in /etc/dhcpcd.conf. See man dhcpcd.conf for more examples.

# nano /etc/dhcpcd.conf
interface enp0s25
static ip_address=
static routers=
static domain_name_servers=

Restart dhcpcd.service:

# systemctl restart dhcpcd.service

See Static IP address for more information.

Update the system clock

Use systemd-timesyncd to ensure that your system clock is accurate. To start it:

# timedatectl set-ntp true

To check the service status, use timedatectl status. See Time for more information.

Prepare the storage devices

  • In general, partitioning or formatting will make existing data inaccessible and subject to being overwritten, i.e. destroyed, by subsequent operations. For this reason, all data that needs to be preserved must be backed up before proceeding.
  • If dual-booting with an existing installation of Windows on a UEFI/GPT system, avoid reformatting the UEFI partition, as this includes the Windows .efi file required to boot it. Furthermore, Arch must follow the same firmware boot mode and partitioning combination as Windows. See Windows and Arch dual boot#Important information.

In this step, the storage devices that will be used by the new system will be prepared. Read Partitioning for a more general overview.

Users intending to create stacked block devices for LVM, disk encryption or RAID, should keep those instructions into consideration when preparing the partitions. If intending to install to a USB flash key, see Installing Arch Linux on a USB key.

Identify the devices

The first step is to identify the devices where the new system will be installed. The following command will show all the available devices:

# lsblk

This will list all devices connected to your system along with their partition schemes, including that used to host and boot live Arch installation media (e.g. a USB drive). Not all devices listed will therefore be viable or appropriate mediums for installation. Results ending in rom, loop or airoot can be ignored.

Devices (e.g. hard disks) will be listed as sdx, where x is a lower-case letter starting from a for the first device (sda), b for the second device (sdb), and so on. Existing partitions on those devices will be listed as sdxY, where Y is a number starting from 1 for the first partition, 2 for the second, and so on. In the example below, only one device is available (sda), and that device uses only one partition (sda1):

sda               8:0    0    80G  0 disk
└─sda1            8:1    0    80G  0 part

The sdxY convention will be used in the examples provided below for partition tables, partitions, and file systems. As they are just examples, it is important to ensure that any necessary changes to device names, partition numbers, and/or partition sizes (etc.) are made. Do not just blindly copy and paste the commands.

If the existing partition scheme needs not be changed, skip to #File systems and swap, otherwise continue reading the following section.

Partition table types

If you are installing alongside an existing installation (i.e. dual-booting), a partition table will already be in use. If the devices are not partitioned, or the current partitions table or scheme needs to be changed, you will first have to determine the partition tables (one for each device) in use or to be used.

There are two types of partition table:

Any existing partition table can be identified with the following command for each device:

# parted /dev/sdx print

Partitioning tools

Warning: Using a partitioning tool that is incompatible with your partition table type will likely result in the destruction of that table, along with any existing partitions/data.

For each device to be partitioned, a proper tool must be chosen according to the partition table to be used. Several partitioning tools are provided by the Arch installation medium, including:

  • parted: MBR and GPT
  • fdisk, cfdisk, sfdisk: MBR and GPT
  • gdisk, cgdisk, sgdisk: GPT

Devices may also be partitioned before booting the installation media, for example through tools such as GParted (also provided as a live CD).

Using parted in interactive mode

All the examples provided below make use of parted, as it can be used for both BIOS/MBR and UEFI/GPT. It will be launched in interactive mode, which simplifies the partitioning process and reduces unnecessary repetition by automatically applying all partitioning commands to the specified device.

In order to start operating on a device, execute:

# parted /dev/sdx

You will notice that the command-line prompt changes from a hash (#) to (parted): this also means that the new prompt is not a command to be manually entered when running the commands in the examples.

To see a list of the available commands, enter:

(parted) help

When finished, or if wishing to implement a partition table or scheme for another device, exit from parted with:

(parted) quit

After exiting, the command-line prompt will change back to #.

Create new partition table

You need to (re)create the partition table of a device when it has never been partitioned before, or when you want to change the type of its partition table. Recreating the partition table of a device is also useful when the partition scheme needs to be restructured from scratch.

Open each device whose partition table must be (re)created with:

# parted /dev/sdx

To then create a new MBR/msdos partition table for BIOS systems, use the following command:

(parted) mklabel msdos

To create a new GPT partition table for UEFI systems instead, use:

(parted) mklabel gpt

Partition schemes

You can decide the number and size of the partitions the devices should be split into, and which directories will be used to mount the partitions in the installed system (also known as mount points). The mapping from partitions to directories is the partition scheme, which must comply with the following requirements:

  • At least a partition for the / (root) directory must be created.
  • When using a UEFI motherboard, one EFI System Partition must be created

In the examples below it is assumed that a new and contiguous partitioning scheme is applied to a single device. Some optional partitions will also be created for the /boot and /home directories: see also Arch filesystem hierarchy for an explanation of the purpose of the various directories; if separate partitions for directories like /boot or /home are not created, these will simply be contained in the / partition. Also the creation of an optional partiton for swap space will be illustrated.

If not already open in a parted interactive session, open each device to be partitioned with:

# parted /dev/sdx

The following command will be used to create partitions:

(parted) mkpart part-type fs-type start end
  • part-type is one of primary, extended or logical, and is meaningful only for MBR partition tables.
  • fs-type is an identifier chosen among those listed by entering help mkpart as the closest match to the file system that you will use in #File systems and swap. The mkpart command does not actually create the file system: the fs-type parameter will simply be used by parted to set a 1-byte code that is used by boot loaders to "preview" what kind of data is found in the partition, and act accordingly if necessary. See also Wikipedia:Disk partitioning#PC partition types.
Tip: Most Linux native file systems map to the same partition code (0x83), so it is perfectly safe to e.g. use ext2 for an ext4-formatted partition.
  • start is the beginning of the partition from the start of the device. It consists of a number followed by a unit, for example 1M means start at 1MiB.
  • end is the end of the partition from the start of the device (not from the start value). It has the same syntax as start, for example 100% means end at the end of the device (use all the remaining space).
Warning: It is important that the partitions do not overlap each other: if you do not want to leave unused space in the device, make sure that each partition starts where the previous one ends.
Note: parted may issue a warning like:
Warning: The resulting partition is not properly aligned for best performance.
In this case, read Partitioning#Partition alignment and follow GNU Parted#Alignment to fix it.

The following command will be used to flag the partition that contains the /boot directory as bootable:

(parted) set partition boot on
  • partition is the number of the partition to be flagged (see the output of the print command).

UEFI/GPT examples

In every instance, a special bootable EFI System Partition is required.

If creating a new EFI System Partition, use the following commands (a size of 512MiB is suggested):

(parted) mkpart ESP fat32 1MiB 513MiB
(parted) set 1 boot on

The remaining partition scheme is entirely up to you. For one other partition using 100% of remaining space:

(parted) mkpart primary ext4 513MiB 100%

For separate / (20GiB) and /home (all remaining space) partitions:

(parted) mkpart primary ext4 513MiB 20.5GiB
(parted) mkpart primary ext4 20.5GiB 100%

And for separate / (20GiB), swap (4GiB), and /home (all remaining space) partitions:

(parted) mkpart primary ext4 513MiB 20.5GiB
(parted) mkpart primary linux-swap 20.5GiB 24.5GiB
(parted) mkpart primary ext4 24.5GiB 100%

BIOS/MBR examples

For a minimum single primary partition using all available disk space, the following command would be used:

(parted) mkpart primary ext4 1MiB 100%
(parted) set 1 boot on

In the following instance, a 20GiB / partition will be created, followed by a /home partition using all the remaining space:

(parted) mkpart primary ext4 1MiB 20GiB
(parted) set 1 boot on
(parted) mkpart primary ext4 20GiB 100%

In the final example below, separate /boot (100MiB), / (20GiB), swap (4GiB), and /home (all remaining space) partitions will be created:

(parted) mkpart primary ext4 1MiB 100MiB
(parted) set 1 boot on
(parted) mkpart primary ext4 100MiB 20GiB
(parted) mkpart primary linux-swap 20GiB 24GiB
(parted) mkpart primary ext4 24GiB 100%

File systems and swap

Once the partitions have been created, each must be formatted with an appropriate file system, except for swap partitions. All available partitions on the intended installation device can be listed with the following command:

# lsblk /dev/sdx

With the exceptions noted below, it is recommended to use the ext4 file system:

# mkfs.ext4 /dev/sdxY

If a new UEFI system partition has been created on a UEFI/GPT system, it must be formatted with a fat32 file system:

# mkfs.fat -F32 /dev/sdxY

If a swap partition has been created, it must be set up and activated with:

# mkswap /dev/sdxY
# swapon /dev/sdxY

Mount the root partition to the /mnt directory of the live system:

# mount /dev/sdxY /mnt

Remaining partitions (except swap) may be mounted in any order, after creating the respective mount points. For example, when using a /boot partition:

# mkdir -p /mnt/boot
# mount /dev/sdxZ /mnt/boot

/boot is also recommended for mounting the EFI System Partition on a UEFI/GPT system. See EFISTUB and related articles for alternatives.

Select a mirror

You may want to edit the mirrorlist file and place your preferred mirror first. A copy of this file will be installed on your new system by pacstrap as well, so it is worth getting it right.

# nano /etc/pacman.d/mirrorlist
## Arch Linux repository mirrorlist
## Sorted by mirror score from mirror status page
## Generated on YYYY-MM-DD

Server =$repo/os/$arch

If you want, you can make it the only mirror available by deleting all other lines, but it is usually a good idea to have a few more, in case the first one goes offline. Should you change your mirror list at a later stage, refresh all package lists with pacman -Syyu. See Mirrors for more information.

Install the base system

The pacstrap script installs the base system. To build packages from the AUR or with ABS, the base-devel group is also required.

# pacstrap -i /mnt base base-devel

The -i switch ensures prompting before package installation. Other packages can later be installed with pacman.

See Pacman and Pacman-key in case of errors.

Generate an fstab

UUIDs are used because they have certain advantages (see Identifying filesystems). If you prefer labels instead, replace the -U option with -L:

# genfstab -U /mnt > /mnt/etc/fstab

The fstab file should always be checked after generating it, and edited in case of errors. See Field definitions for syntax information.

# cat /mnt/etc/fstab

Configure the base system

Change root into the new system:

# arch-chroot /mnt /bin/bash

The close following and understanding of these steps is key to a properly configured system. Tools from the live installation, such as dialog and wpa_supplicant, are not automatically installed. The following sections specify such cases.


The Locale defines which language the system uses, and other regional considerations such as currency denomination, numerology, and character sets.

Possible values are listed in /etc/locale.gen. Uncomment en_US.UTF-8 UTF-8, as well as other needed localisations. UTF-8 is highly recommended over other options.

# nano /etc/locale.gen
#en_SG ISO-8859-1
en_US.UTF-8 UTF-8
#en_US ISO-8859-1

Before locales can be enabled, they must be generated:

# locale-gen

Create /etc/locale.conf, where LANG refers to the first column of an uncommented entry in /etc/locale.gen:

# echo LANG=en_US.UTF-8 > /etc/locale.conf

If you changed the keymap in #Keyboard layout, create the /etc/vconsole.conf file accordingly. KEYMAP must match the value initially set with loadkeys, to ensure correct entry of the Root password on reboot. For example:

# nano /etc/vconsole.conf


Time zones are available in /usr/share/zoneinfo/Zone/SubZone. To list them, run:

$ ls -l /usr/share/zoneinfo

To set the default zone, create a symbolic link /etc/localtime to the respective subzone file:

# ln -sf /usr/share/zoneinfo/Zone/SubZone /etc/localtime

It is highly recommended to set the Time standard to UTC, and adjust the Time skew:

# hwclock --systohc --utc

If other operating systems are installed on the machine, they must be configured accordingly.


As mkinitcpio was run on installation of linux with pacstrap, most users can use the defaults provided in mkinitcpio.conf.

For special configurations, set the correct hooks in /etc/mkinitcpio.conf and re-generate the initramfs image:

# mkinitcpio -p linux

Boot loader

See Boot loaders for available choices and configurations. If you have an Intel CPU, microcode updates must be configured after installing the boot loader.


Install the grub package. To search for other operating systems, also install os-prober:

# pacman -S grub os-prober

Install the bootloader to the drive Arch was installed to:

# grub-install --recheck /dev/sda

Generate grub.cfg:

# grub-mkconfig -o /boot/grub/grub.cfg

See GRUB for more information.


Here, the installation drive is assumed to be GPT-partioned, and have the EFI System Partition (gdisk type EF00, formatted with FAT32) mounted at /boot.

bootctl is part of systemd, and as such part of the base installation.

# bootctl install

Create a boot entry in /boot/loader/entries/arch.conf, replacing /dev/sda2 with the root partition:

# nano /boot/loader/entries/arch.conf
title          Arch Linux
linux          /vmlinuz-linux
initrd         /initramfs-linux.img
options        root=/dev/sda2 rw

Modify /boot/loader/loader.conf to select the default entry (without .conf) suffix:

# nano /boot/loader/loader.conf
timeout 3
default arch

See systemd-boot for more information.

Network configuration

Configure the network for the newly installed environment. The procedure is similar to #Establish an internet connection, except made persistent for subsequent boots. Select one daemon to handle the network.


Set the hostname to your liking:

# echo myhostname > /etc/hostname

Add the same hostname to /etc/hosts:

# nano /etc/hosts
#<ip-address> <> <hostname> localhost.localdomain localhost myhostname
::1   localhost.localdomain localhost myhostname


dhcpcd is the default method in the install medium, and part of the base installation. When only requiring a single wired connection, enable the dhcpcd service:

# systemctl enable dhcpcd@interface.service

Where interface is an ethernet device name. If static IP settings are required, adjust the profile configuration as described in Static IP.

See Configure the IP address for more information.


For wireless, dhcpcd and systemd-networkd require a separate configuration of the connection in the wireless backend, wpa_supplicant, first. If you anticipate to connect the machine to different wireless networks over time, a tool which provides its own connection management may be easier to handle. Aside from netctl introduced below, Wireless network configuration#Automatic setup lists other choices.

Warning: Wireless chipset firmware packages (for cards which require them) are pre-installed under /usr/lib/firmware in the live environment, but must be installed separately on the new system. See Wireless network configuration#Installing driver/firmware for details.

Install iw and wpa_supplicant, which you will need to connect to a network:

# pacman -S iw wpa_supplicant
Adding wireless networks
Using wifi-menu

Install dialog, which is required for wifi-menu:

# pacman -S dialog

After finishing the rest of this installation and rebooting, you can connect to the network with:

# wifi-menu device_name

Where device_name is the interface of your wireless chipset.

Warning: Do not use wifi-menu now. Instead, wait until you have finished this guide and have rebooted. It will not work now because a process spawned by this command will conflict with the one you have running outside of the chroot. Alternatively, you could just configure a network profile manually using the following templates, so that you do not have to worry about using wifi-menu at all.
Using manual netctl profiles

See Netctl for general information, and Netctl#Example profiles for how to source and configure an example profile.

Connect automatically to known networks
Warning: This method cannot be used with explicitely enabled profiles, i.e. through netctl enable profile.

Install wpa_actiond, which is required for netctl-auto:

# pacman -S wpa_actiond

Enable the netctl-auto service, which will connect to known networks and gracefully handle roaming and disconnects:

# systemctl enable netctl-auto@interface_name.service
Tip: netctl also provides netctl-ifplugd, which can be used to handle wired profiles in conjunction with netctl-auto.

Unmount the partitions and reboot

Set the root password with:

# passwd

Exit from the chroot environment:

# exit

Partitions will be unmounted automatically by systemd on shutdown. You may however unmount manually as a safety measure:

# umount -R /mnt

If the partition is "busy", you can find the cause with fuser. Reboot the computer.

# reboot

Remove the installation media, or you may boot back into it. You can log into your new installation as root, using the password you specified with passwd.


Your new Arch Linux base system is now a functional GNU/Linux environment ready to be built into whatever you wish or require for your purposes. You are now strongly advised to read the General recommendations article, especially the first two sections. Its other sections provide links to post-installation tutorials like setting up a graphical user interface, sound or a touchpad.

For a list of applications that may be of interest, see List of applications.