Difference between revisions of "Beginners' guide"

From ArchWiki
Jump to navigation Jump to search
m (→‎Partition table types: Minor rewording)
m (→‎Create new partition table: Removing unneeded zap all command)
Line 395: Line 395:
 
{{Warning|If dual-booting with an existing installation of Windows on a UEFI/GPT system, do not erase the partition table. Doing so will destroy all existing data on the device, including the UEFI partition with the Windows {{ic|.efi}} file required to boot it.}}
 
{{Warning|If dual-booting with an existing installation of Windows on a UEFI/GPT system, do not erase the partition table. Doing so will destroy all existing data on the device, including the UEFI partition with the Windows {{ic|.efi}} file required to boot it.}}
  
If you want to start from scratch — and do not intend to keep existing partitions on a device — its existing partition table can be completely erased. This simplifies the creation of new partitions and avoids problems if converting disks from MBR to GPT or vice versa. Use the following command:
+
If you want to start from scratch — and do not intend to keep existing partitions on a device — its existing partition table can be completely erased. This simplifies the creation of new partitions and avoids problems if converting disks from MBR to GPT or vice versa. Open each device whose partition table must be (re)created with:
 
 
# sgdisk --zap-all /dev/sd''x''
 
 
 
Now open each device whose partition table must be (re)created with:
 
  
 
  # parted /dev/sd''x''
 
  # parted /dev/sd''x''
  
{{Warning|MBR is designed specifically for use with BIOS systems, and GPT is designed for UEFI. It is not recommended for less experienced users to break this convention as both have features and/or limitations that may be incompatible with your hardware (e.g. MBR cannot cope with devices larger than 2 TiB). If for any reason you do not wish to follow this convention, see http://mjg59.dreamwidth.org/8035.html and http://rodsbooks.com/gdisk/bios.html for more information and possible workarounds.}}
+
{{Warning|MBR is designed specifically for use with BIOS systems, and GPT is designed for (U)EFI. It is not recommended for less experienced users to break this convention as both have features and/or limitations that may be incompatible with your hardware (e.g. MBR cannot cope with devices larger than 2 TiB). If for any reason you do not wish to follow this convention, see http://mjg59.dreamwidth.org/8035.html and http://rodsbooks.com/gdisk/bios.html for more information and possible workarounds.}}
  
 
To then create a new MBR/msdos partition table for BIOS systems, use the following command:
 
To then create a new MBR/msdos partition table for BIOS systems, use the following command:

Revision as of 16:02, 3 January 2015

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://irc.freenode.net/#archlinux) 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.

System requirements

Arch Linux should run on any i686 compatible machine with a minimum of 64 MB RAM. A basic installation with all packages from the base group should take less than 800 MB of disk space. If you are working with limited space, this can be trimmed down considerably, but you will have to know what you are doing.

Prepare the latest installation medium

The latest release of the installation media can be obtained from the Download page. Note that the single ISO image supports both 32 and 64-bit architectures. It is highly recommended to always use the latest ISO image.

Tip: The archboot ISO images can take several steps explained in this guide interactively. See Archboot for details.
  • Install images are signed and it is highly recommended to verify their signature before use. Download the .sig file from the download page (or one of the mirrors listed there) to the same directory as the .iso file. On Arch Linux, use pacman-key -v iso-file.sig as root; in other environments make use, still as root, of gpg2 directly with gpg2 --verify iso-file.sig. The file integrity checksums md5 and sha1 are also provided
    Note: The gpg2 verification will fail if you have not downloaded the public key corresponding to the RSA key ID. See http://sparewotw.wordpress.com/2012/10/31/how-to-verify-signature-using-sig-file/ for details
Note: The quality of optical drives and the discs themselves varies greatly. Generally, using a slow burn speed is recommended for reliable burns. If you are experiencing unexpected behaviour from the disc, try burning at the lowest speed supported by your burner

Installing over the network

Instead of writing the boot media to a disc or USB stick, you may alternatively boot the ISO image over the network. This works well when you already have a server set up. Please see the PXE article for more information, and then continue to #Boot the installation medium.

Install from an existing Linux system

Alternatively, it is possible to install from an already running Linux system. See Install from Existing Linux. This can be useful particularly if you plan to install Arch via VNC or SSH remotely. Users seeking to perform the Arch Linux installation remotely via an SSH connection should read Install from SSH for additional tips.

Installing on a virtual machine

Installing on a virtual machine is a good way to become familiar with Arch Linux and its installation procedure without leaving your current operating system and repartitioning the storage drive. It will also let you keep this Beginners' Guide open in your browser throughout the installation. Some users may find it beneficial to have an independent Arch Linux system on a virtual drive, for testing purposes.

Examples of virtualization software are VirtualBox, VMware, QEMU, Xen, Parallels.

The exact procedure for preparing a virtual machine depends on the software, but will generally follow these steps:

  1. Create the virtual disk image that will host the operating system.
  2. Properly configure the virtual machine parameters.
  3. Boot the downloaded ISO image with a virtual CD drive.
  4. Continue with Boot the installation medium.

The following articles may be helpful:

Boot the installation medium

Most modern systems allow you to select the boot device during the POST phase, usually by pressing the F12 key while the BIOS splash screen is visible. Select the device which contains the Arch ISO. Alternatively, you may need to change the boot order in your computer's BIOS.

To do this, press a key (usually Delete, F1, F2, F11 or F12) during the POST phase. This will take you into the BIOS settings screen where you can set the order in which the system searches for devices to boot from. Set the device which contains the Arch ISO as the first device from which boot is attempted. Select "Save & Exit" (or your BIOS's equivalent) and the computer should then complete its normal boot process.

When the Arch menu appears, select "Boot Arch Linux" and press Enter to enter the live environment where you will run the actual installation (if booting from a UEFI boot disk, the option may look more like "Arch Linux archiso x86_64 UEFI").

Testing if you are booted into UEFI mode

In case you have a UEFI motherboard and UEFI Boot mode is enabled (and is preferred over BIOS/Legacy mode), the CD/USB will automatically launch Arch Linux via Gummiboot and you will get the following menu (white letters on black background), with the first item highlighted:

Arch Linux archiso x86_64 UEFI USB
UEFI Shell x86_64 v1
UEFI Shell x86_64 v2
EFI Default Loader

If you do not remember which menu you had at boot time, or if you want to make sure you booted into UEFI mode, run:

# efivar -l

If efivar lists the UEFI variables properly, then you have booted in UEFI mode. If not check whether all the requirements listed in Unified Extensible Firmware Interface are met.

Troubleshooting boot problems

Change the language

You are now presented with a shell prompt, automatically logged in as root. Your shell is Zsh; this will provide you advanced Tab completion, and other features as part of the grml config. For editing text files, the console editor nano is suggested. If you are not familiar with it, see nano#Usage. If you have (or plan on having) a dual boot setup with Windows, see Windows and Arch Dual Boot.

Tip: These are optional for the majority of users. Useful only if you plan on writing in your own language in any of the configuration files, if you use diacritical marks in the Wi-Fi password, or if you would like to receive system messages (e.g. possible errors) in your own language. Changes here only affect the installation process.

By default, the keyboard layout is set to us. If you have a non-US keyboard layout, run:

# loadkeys layout

...where layout can be fr, uk, dvorak, be-latin1, etc. See this wikipedia article for a 2-letter country code list. Use the command localectl list-keymaps to list all available keymaps.

If some glyphs of your language's alphabet (e.g. accented and non Latin letters) show up as white squares or as other symbols, you may want to change the console font with one from /usr/share/kbd/consolefonts/. For example:

# setfont lat9w-16

You can run the showconsolefont command to display the full contents of the loaded font. Note that the font name is case-sensitive, so type it exactly as you see it. See Fonts#Console fonts for more information.

By default, the language is set to English (US). If you would like to change the language for the install process (German, in this example), remove the # in front of the locale you want from /etc/locale.gen, along with English (US). Please choose the UTF-8 entries:

# nano /etc/locale.gen
en_US.UTF-8 UTF-8
de_DE.UTF-8 UTF-8
# locale-gen
# export LANG=de_DE.UTF-8

Establish an internet connection

Warning: As of v197, udev no longer assigns network interface names according to the wlanX and ethX naming scheme. If you are coming from a different distribution or are reinstalling Arch and not aware of the new interface naming style, please do not assume that your wireless interface is named wlan0, or that your wired interface is named eth0. You can use the command ip link to discover the names of your interfaces.

The dhcpcd network daemon starts automatically during boot and it will attempt to start a wired connection. Try to ping a server to see if a connection was established. For example, Google's webservers:

# ping -c 3 www.google.com
PING www.l.google.com (74.125.132.105) 56(84) bytes of data.
64 bytes from wb-in-f105.1e100.net (74.125.132.105): icmp_req=1 ttl=50 time=17.0 ms
64 bytes from wb-in-f105.1e100.net (74.125.132.105): icmp_req=2 ttl=50 time=18.2 ms
64 bytes from wb-in-f105.1e100.net (74.125.132.105): icmp_req=3 ttl=50 time=16.6 ms

--- www.l.google.com ping statistics ---
3 packets transmitted, 3 received, 0% packet loss, time 2003ms
rtt min/avg/max/mdev = 16.660/17.320/18.254/0.678 ms

If you get a ping: unknown host error, first check if there is an issue with your cable or wireless signal strength. If not, you will need to set up the network manually, as explained below. Once a connection is established move on to #Prepare the storage devices.

Wired

Follow this procedure if you need to set up a wired connection via a static IP address.

First, disable the dhcpcd service which was started automatically at boot:

Note: That service may have another name dhcpcd@your_ethernet_interface.service, for example dhcpcd@enp0s25.service. Just press Tab two times after systemctl stop dhcpcd to choose the proper one
# systemctl stop dhcpcd.service

Identify the name of your ethernet interface.

# ip link
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN mode DEFAULT
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
2: enp2s0f0: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN mode DEFAULT qlen 1000
    link/ether 00:11:25:31:69:20 brd ff:ff:ff:ff:ff:ff
3: wlp3s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP mode DORMANT qlen 1000
    link/ether 01:02:03:04:05:06 brd ff:ff:ff:ff:ff:ff

In this example, the ethernet interface is enp2s0f0. If you are unsure, your ethernet interface is likely to start with the letter "e", and unlikely to be "lo" or start with the letter "w".

You also need to know these settings:

  • Static IP address.
  • Subnet mask in CIDR notation, for example /24 is the CIDR notation of 255.255.255.0 netmask.
  • Gateway's IP address.
  • Name servers' (DNS) IP addresses.
  • Domain name (unless you are on a local LAN, in which case you can make it up).

Activate the connected ethernet interface (e.g. enp2s0f0):

# ip link set enp2s0f0 up

Add the address:

# ip addr add ip_address/mask_bits dev interface_name

For example:

# ip addr add 192.168.1.2/24 dev enp2s0f0

For more options, run man ip.

Add your gateway like this, substituting your own gateway's IP address:

# ip route add default via ip_address

For example:

# ip route add default via 192.168.1.1

Edit resolv.conf, substituting your name servers' IP addresses and your local domain name:

# nano /etc/resolv.conf
nameserver 61.23.173.5
nameserver 61.95.849.8
search example.com
Note: Currently, you may include a maximum of three nameserver lines. In order to overcome this limitation, you can use a locally caching nameserver like dnsmasq.

You should now have a working network connection. If you do not, check the detailed Network configuration page.

Wireless

Warning: Wireless chipset firmware packages (for cards which require them) are pre-installed under /usr/lib/firmware in the live environment (on CD/USB stick) but must be explicitly installed to your actual system to provide wireless functionality after you reboot into it! Package installation is covered later in this guide. Ensure installation of both your wireless module and firmware before rebooting! See Wireless network configuration if you are unsure about the requirement of corresponding firmware installation for your particular chipset.

Run this command if you need wireless connectivity (Wi-Fi) during the installation process:

# wifi-menu

It should bring you a menu of wifi networks if you have only one wifi device in the computer (mostly the case in laptops).

Follow this procedure if that didn't work:

First, identify the name of your wireless interface:

# iw dev
phy#0
        Interface wlp3s0
                ifindex 3
                wdev 0x1
                addr 00:11:22:33:44:55
                type managed

In this example, wlp3s0 is the available wireless interface. If you are unsure, your wireless interface is likely to start with the letter "w", and unlikely to be "lo" or start with the letter "e".

Now use netctl's wifi-menu to connect to a network:

# wifi-menu wlp3s0

See the sample configuration in WPA2 Enterprise#netctl for networks that require both a username and password.

You should now have a working wireless network connection. If you do not or even failed to identify the wireless interface, see #Without wifi-menu below or the detailed Wireless network configuration page.

Without wifi-menu

Bring the interface up with:

# ip link set wlp3s0 up

To verify that the interface is up, inspect the output of the following command:

# ip link show wlp3s0
3: wlp3s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state DOWN mode DORMANT group default qlen 1000
    link/ether 00:11:22:33:44:55 brd ff:ff:ff:ff:ff:ff

The UP in <BROADCAST,MULTICAST,UP,LOWER_UP> is what indicates the interface is up, not the later state DOWN.

Most wireless chipsets require firmware in addition to a corresponding driver. The kernel tries to identify and load both automatically. If you get output like SIOCSIFFLAGS: No such file or directory, this means you will need to manually load the firmware. If unsure, invoke dmesg to query the kernel log for a firmware request from the wireless chipset. For example, if you have an Intel chipset which requires and has requested firmware from the kernel at boot:

# dmesg | grep firmware
firmware: requesting iwlwifi-5000-1.ucode

If there is no output, it may be concluded that the system's wireless chipset does not require firmware.

Next, scan for available networks using iw dev wlp3s0 scan | grep SSID, then connect to a network with:

# wpa_supplicant -B -i wlp3s0 -c <(wpa_passphrase "ssid" "psk")

You need to replace ssid with the name of your network and psk with your wireless password, leaving the quotes around the network name and password.

Finally, you have to give your interface an IP address. This can be set manually or using dhcp:

# dhcpcd wlp3s0

If that does not work, issue the following commands:

# echo 'ctrl_interface=DIR=/run/wpa_supplicant' > /etc/wpa_supplicant.conf
# wpa_passphrase "ssid" "psk" >> /etc/wpa_supplicant.conf
# ip link set interface up
# wpa_supplicant -B -D nl80211,wext -c /etc/wpa_supplicant.conf -i interface
# dhcpcd -A interface

Setting the interface up at step 3 may not be needed, but does no harm in any case.

Analog modem, ISDN, or PPPoE DSL

For xDSL, dial-up, and ISDN connections, see Direct Modem Connection.

Behind a proxy server

If you are behind a proxy server, you will need to export the http_proxy and ftp_proxy environment variables. See Proxy settings for more information.

Prepare the storage devices

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

Warning: Partitioning will destroy existing data. Before proceeding, you must backup all data that needs to be preserved.
Tip:

Identify the devices

The first step is 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 — the live Arch installation media (e.g. a USB drive). Not all devices listed will therefore be viable or appropriate mediums for installation. To filter out inappropriate results, the command can optionally be amended as follows:

# lsblk | grep -v "rom\|loop\|airoot"

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):

NAME            MAJ:MIN RM   SIZE RO TYPE MOUNTPOINT
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 #Create filesystems, 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.

Warning: If Arch and Windows are dual-booting from same device, then Arch must follow the same firmware boot mode and partitioning combination already used, or Windows will fail to boot. See Windows and Arch Dual Boot#Important information for more details.

There are two types of partition table:

  • MBR: Intended for BIOS systems (also referred to as "msdos")
  • GPT: Intended for (U)EFI systems

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

# parted /dev/sdx print

Partitioning tools

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
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.
Tip: The devices may also be partitioned before booting the Arch installation media, possibly using alternative live systems with other partitioning tools. For example beginners might find it easier to use a graphical partitioning tool such as GParted, which is also provided as a live CD and works with both MBR and GPT partition tables.

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

Warning: If dual-booting with an existing installation of Windows on a UEFI/GPT system, do not erase the partition table. Doing so will destroy all existing data on the device, including the UEFI partition with the Windows .efi file required to boot it.

If you want to start from scratch — and do not intend to keep existing partitions on a device — its existing partition table can be completely erased. This simplifies the creation of new partitions and avoids problems if converting disks from MBR to GPT or vice versa. Open each device whose partition table must be (re)created with:

# parted /dev/sdx
Warning: MBR is designed specifically for use with BIOS systems, and GPT is designed for (U)EFI. It is not recommended for less experienced users to break this convention as both have features and/or limitations that may be incompatible with your hardware (e.g. MBR cannot cope with devices larger than 2 TiB). If for any reason you do not wish to follow this convention, see http://mjg59.dreamwidth.org/8035.html and http://rodsbooks.com/gdisk/bios.html for more information and possible workarounds.

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:

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 one of the supported file systems listed in the manual. The partition will be properly formatted in #Create filesystems.
  • 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.
Ignore/Cancel?
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.

Warning: If dual-booting with an existing installation of Windows on a UEFI/GPT system, the existing UEFI partition must not be recreated. Doing so will destroy the Windows .efi file required to boot it.

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

(parted) mkpart ESP fat32 1M 513M
(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 ext3 513M 100%

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

(parted) mkpart primary ext3 513M 20.5G
(parted) mkpart primary ext3 20.5G 100%

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

(parted) mkpart primary ext3 513M 20.5G
(parted) mkpart primary linux-swap 20.5G 24.5G
(parted) mkpart primary ext3 24.5G 100%

BIOS/MBR examples

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

(parted) mkpart primary ext3 1M 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 ext3 1M 20G
(parted) set 1 boot on
(parted) mkpart primary ext3 20G 100%

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

(parted) mkpart primary ext3 1M 100M
(parted) set 1 boot on
(parted) mkpart primary ext3 100M 20G
(parted) mkpart primary linux-swap 20G 24G
(parted) mkpart primary ext3 24G 100%

Create filesystems

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
Warning: If dual-booting with an existing installation of Windows on a UEFI/GPT system, do not re-format the UEFI partition. Doing so will destroy all existing data on that partition, including the Windows .efi file required to boot it.
Note:
  • If a new UEFI system partition has been created on a UEFI/GPT system, it must be formatted with a fat32 or vfat32 file system. Failure to do so will result in an unbootable installation:
# mkfs.vfat -F32 /dev/sdxY
  • If you plan to use GRUB on a BIOS/GPT system, please note that the BIOS Boot Partition has nothing to do with the /boot mountpoint. It will be used by GRUB directly. Do not create a filesystem on it, and do not mount it.

Activate swap

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

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

Mount the partitions

Note: Swap partitions must not be mounted here.

The / (root) partition must be mounted first: this is because any directories such as /boot or /home that have separate partitions will have to be created in the root file system. The /mnt directory of the live system will be used to mount the root partition, and consequently all the other partitions will stem from there. If the root partition's name is sdxR, do:

# mount /dev/sdxR /mnt

Once the / partition has been mounted, any remaining partitions may be mounted in any order. The general procedure is to first create the mount point, and then mount the partition to it. If using a separate /boot partition:

# mkdir -p /mnt/boot
# mount /dev/sdxB /mnt/boot
Note: Using /boot is recommended also for mounting the EFI System Partition on UEFI/GPT system. See EFISTUB and related articles for alternatives.

If using a separate /home partition:

# mkdir -p /mnt/home	
# mount /dev/sdxH /mnt/home

Once all the remaining partitions, if any, have been mounted, the devices are ready to install Arch.

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 = http://mirror.example.xyz/archlinux/$repo/os/$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.

Tip:
  • Use the Mirrorlist Generator to get an updated list for your country. HTTP mirrors are faster than FTP, because of something called keepalive. With FTP, pacman has to send out a signal each time it downloads a package, resulting in a brief pause. For other ways to generate a mirror list, see Sorting mirrors and Reflector.
  • Arch Linux MirrorStatus reports various aspects about the mirrors such as network problems with mirrors, data collection problems, the last time mirrors have been synced, etc.
Note:
  • Whenever in the future you change your mirrorlist, refresh all package lists with pacman -Syyu, to ensure that the package lists are updated consistently. See Mirrors for more information.
  • If you are using an older installation medium, your mirrorlist might be outdated, which might lead to problems when updating Arch Linux (see FS#22510). Therefore it is advised to obtain the latest mirror information as described above.
  • Some issues have been reported in the Arch Linux forums regarding network problems that prevent pacman from updating/synchronizing repositories (see [1] and [2]). When installing Arch Linux natively, these issues have been resolved by replacing the default pacman file downloader with an alternative (see Improve pacman performance for more details). When installing Arch Linux as a guest OS in VirtualBox, this issue has also been addressed by using "Host interface" instead of "NAT" in the machine properties.

Install the base system

The base system is installed using the pacstrap script. The -i switch can be omitted if you wish to install every package from the base group without prompting. To build packages from the AUR or with ABS, you will also need the base-devel group.

# pacstrap -i /mnt base base-devel

Other packages can be installed later using pacman.

Note:
  • If pacstrap hangs with error: failed retrieving file 'core.db' from mirror... : Connection time-out, yet your mirrors are configured correctly, try setting a different name server.
  • If in the middle of the installation of base packages you get a request to import a PGP key, agree to download the key to proceed. This is likely to happen if the Arch ISO you are using is out of date. If you are unable to add the PGP key successfully, try upgrading the package archlinux-keyring as follows: pacman -S archlinux-keyring

See Pacman#Troubleshooting and Pacman-key#Troubleshooting for more information.

Generate an fstab

Generate an fstab file with the following command. UUIDs will be used because they have certain advantages (see fstab#Identifying filesystems). If you would prefer to use labels instead, replace the -U option with -L:

# genfstab -U -p /mnt >> /mnt/etc/fstab
# nano /mnt/etc/fstab
Warning: The fstab file should always be checked after generating it. If you encounter errors running genfstab or later in the install process, do not run genfstab again; just edit the fstab file.

A few considerations:

  • The last field determines the order in which partitions are checked at start up: use 1 for the (non-Btrfs) root partition, which should be checked first; 2 for all other partitions you want checked at start up; and 0 means 'do not check' (see fstab#Field definitions).
  • All Btrfs partitions should have 0 for this field. Normally, you will also want your swap partition to have 0.

Chroot and configure the base system

Next, chroot into your newly installed system:

# arch-chroot /mnt /bin/bash
Tip: Leaving out /bin/bash will chroot into the sh shell.

At this stage of the installation, you will configure the primary configuration files of your Arch Linux base system. These can either be created if they do not exist, or edited if you wish to change the defaults.

Closely following and understanding these steps is of key importance to ensure a properly configured system.

Note: Do not assume that the tools you used from the ISO are automatically installed. For example, if you used wifi-menu to gain network access during the installation and want to continue so after the first boot, you will have to install dialog to use it. The following section specifies such cases, do follow it closely to avoid a hick-up in your fresh install.

Locale

Locales are used by glibc and other locale-aware programs or libraries for rendering text, correctly displaying regional monetary values, time and date formats, alphabetic idiosyncrasies, and other locale-specific standards. These values are defined in locale.gen and locale.conf.

The locale.gen file has all entries commented out by default. To uncomment a line remove the # in front. UTF-8 is highly recommended over ISO-8859. Uncomment en_US.UTF-8 UTF-8 as well as other localisations users might need:

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

Generate the locale(s) specified in /etc/locale.gen:

# locale-gen

Create the /etc/locale.conf file substituting your chosen locale:

Tip: Setting en_US.UTF-8 as the system-wide locale allows to keep system logs in English for easier troubleshooting. Users can override this setting for their environment as described in Locale#Setting the locale.
# echo LANG=en_US.UTF-8 > /etc/locale.conf
Note:
  • The locale specified in the LANG variable must be uncommented in /etc/locale.gen.
  • The locale.conf file does not exist by default. Setting only LANG should be enough as it will act as the default value for all other variables.

Export substituting your chosen locale:

# export LANG=en_US.UTF-8
Tip: To use other locales for other LC_* variables, run locale to see the available options and add them to locale.conf. It is not recommended to set the LC_ALL variable. See Locale for details.

Console font and keymap

If you changed the default console keymap and font in #Change the language, you will have to edit /etc/vconsole.conf accordingly (create it if it does not exist) to make those changes persist in the installed system, for example:

# nano /etc/vconsole.conf
KEYMAP=de-latin1
FONT=lat9w-16
Warning: If you set KEYMAP to a different value than the one you initially set with loadkeys, and then you #Set the root password, you may have problems logging into the new system after rebooting, because some keys may be mapped differently between the two layouts.

Note that these settings are only valid for your virtual consoles, not in Xorg. See Fonts#Console fonts for more information.

Time zone

Available time zones and subzones can be found in the /usr/share/zoneinfo/Zone/SubZone directories.

To view the available zones, check the directory /usr/share/zoneinfo/:

# ls /usr/share/zoneinfo/

Similarly, you can check the contents of directories belonging to a subzone:

# ls /usr/share/zoneinfo/Europe

Create a symbolic link /etc/localtime to your subzone file /usr/share/zoneinfo/Zone/SubZone using this command:

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

Example:

# ln -s /usr/share/zoneinfo/Europe/Minsk /etc/localtime
Note: If you get ln: failed to create symbolic link '/etc/localtime': File exists, check the existing file with ls -l /etc/localtime and add the -f option to the ln command to overwrite it.

Hardware clock

Set the hardware clock mode uniformly between your operating systems. Otherwise, they may overwrite the hardware clock and cause time shifts.

You can generate /etc/adjtime automatically by using one of the following commands:

  • UTC (recommended):
    Note: Using UTC for the hardware clock does not mean that software will display time in UTC.
    # hwclock --systohc --utc
  • localtime (discouraged; used by default in Windows):
    Warning: Using localtime may lead to several known and unfixable bugs. However, there are no plans to drop support for localtime.
    # hwclock --systohc --localtime

Kernel modules

Tip: This is just an example, you do not need to set it. All needed modules are automatically loaded by udev, so you will rarely need to add something here. Only add modules that you know are missing.

For kernel modules to load during boot, place a *.conf file in /etc/modules-load.d/, with a name based on the program that uses them:

# nano /etc/modules-load.d/virtio-net.conf
# Load 'virtio-net.ko' at boot.

virtio-net

If there are more modules to load per *.conf, the module names can be separated by newlines. A good example are the VirtualBox Guest Additions.

Empty lines and lines starting with # or ; are ignored.

Hostname

Set the hostname to your liking (e.g. arch):

# echo myhostname > /etc/hostname
Note: You no longer need to specify the hostname in /etc/hosts, because systemd will provide host name resolution via the myhostname nss module which is enabled by default in /etc/nsswitch.conf.

Configure the network

You need to configure the network again, but this time for your newly installed environment. The procedure and prerequisites are very similar to the one described above, except we are going to make it persistent and automatically run at boot.

As a first step, identify the network interface name you want to configure the connection for with ip link.

Note:
  • For more in-depth information on network configration, visit Network configuration and Wireless network configuration.
  • If you would like to use the old interface naming scheme (ie. eth* and wlan*) you can accomplish this by creating an empty file at /etc/udev/rules.d/80-net-setup-link.rules which will mask the file of the same name located under /usr/lib/udev/rules.d.

Wired

Dynamic IP
Using dhcpcd

If you only use a single fixed wired network connection, you do not need a network management service and can simply enable the dhcpcd service for the interface:

# systemctl enable dhcpcd@interface_name.service
Using netctl

Copy a sample profile from /etc/netctl/examples to /etc/netctl:

# cd /etc/netctl
# cp examples/ethernet-dhcp my_network

Edit the profile as needed (update Interface from eth0 to the interface name of the system.

# nano my_network

Enable the my_network profile:

# netctl enable my_network
Note: You will get the message "Running in chroot, ignoring request.". This can be ignored for now.
Using netctl-ifplugd
Warning: You cannot use this method in conjunction with explicitly enabling profiles, such as netctl enable profile.

Alternatively, you can use netctl-ifplugd, which gracefully handles dynamic connections to new networks.

Install ifplugd, which is required for netctl-ifplugd:

# pacman -S ifplugd

Then enable for interface that you want:

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

Copy a sample profile from /etc/netctl/examples to /etc/netctl:

# cd /etc/netctl
# cp examples/ethernet-static my_network

Edit the profile as needed (modify Interface, Address, Gateway and DNS):

# nano my_network

For the Address take care to include the correct netmask (the /24 in the sample profile equates to a netmask of 255.255.255.0) or the profile will fail to start. See also CIDR notation.

Enable above created profile to start it at every boot:

# netctl enable my_network
Using systemd-networkd

See systemd-networkd.

Wireless

Note: If your wireless adapter requires a firmware (as described in the above Establish an internet connection section and also in the article Wireless network configuration#Device driver), install the package containing your firmware. Most of the time, the linux-firmware package will contain the needed firmware. Though for some devices, the required firmware might be in its own package. For example:
# pacman -S zd1211-firmware
See Wireless network configuration#Installing driver/firmware for more info.

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 interface_name (where interface_name is the interface of your wireless chipset).

# wifi-menu interface_name
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

Copy a network profile from /etc/netctl/examples to /etc/netctl:

# cd /etc/netctl
# cp examples/wireless-wpa my-network

Edit the profile as needed (modify Interface, ESSID and Key):

# nano my-network

Enable above created profile to start it at every boot:

# netctl enable my-network
Connect automatically to known networks
Warning: You cannot use this method in conjunction with explicitly enabling profiles, such as 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.

Analog modem, ISDN or PPPoE DSL

For xDSL, dial-up and ISDN connections, see Direct Modem Connection.

Create an initial ramdisk environment

Tip: Most users can skip this step and use the defaults provided in mkinitcpio.conf. The initramfs image (from the /boot folder) has already been generated based on this file when the linux package (the Linux kernel) was installed earlier with pacstrap.

Here you need to set the right hooks if the root is on a USB drive, if you use RAID, LVM, if using a multi-device Btrfs volumes as root, or if /usr is on a separate partition.

Edit /etc/mkinitcpio.conf as needed and re-generate the initramfs image with:

# mkinitcpio -p linux

Set the root password

Set the root password with:

# passwd

Install and configure a bootloader

If you have an Intel CPU (on AMD it is automatic) you might want to configure your bootloader to setup microcode updates.

For BIOS motherboards

For BIOS systems, several boot loaders are available, see Boot loaders for a complete list. Choose one as per your convenience. Possible choices include:

  • Syslinux#Installation is (currently) limited to loading only files from the partition where it was installed. Its configuration file is considered to be easier to understand. An example configuration can be found in Syslinux#Examples.
  • GRUB is more feature-rich and supports more complex scenarios. Its configuration file(s) is more similar to 'sh' scripting language, which may be difficult for beginners to manually write. It is recommended that they automatically generate one.

Here, installation with GRUB and MBR is demonstrated. Install the grub package and then run grub-install to install the bootloader:

# pacman -S grub
# grub-install --target=i386-pc --recheck /dev/sda
Note:
  • Change /dev/sda to reflect the drive you installed Arch on. Do not append a partition number (do not use sdaX).
  • A sample /boot/grub/grub.cfg gets installed as part of the grub package, and subsequent grub-* commands may not over-write it. Ensure that your intended changes are in grub.cfg, rather than in grub.cfg.new or some such file.

Automatically generate grub.cfg:

Tip: To automatically search for other operating systems on your computer, install os-prober (pacman -S os-prober) before running the next command.
# grub-mkconfig -o /boot/grub/grub.cfg

For more information on configuring and using GRUB, see GRUB.

For UEFI motherboards

For UEFI systems, several boot loaders are available, see Boot loaders for a complete list. Choose one as per your convenience. Possible choices include:

  • gummiboot is a minimal UEFI Boot Manager which provides a menu for EFISTUB kernels and other UEFI applications. This is recommended for beginners, especially those wishing to dual-boot with other installed operating systems such as Windows 8.
  • GRUB is a more complete bootloader, useful if you run into problems with Gummiboot.

Here, installation with gummiboot is demonstrated. First install dosfstools to manipulate the EFI System Partition post-installation, and efibootmgr to create bootable .efi stubs (used by bootmanager installation scripts):

# pacman -S dosfstools efibootmgr
Note: For UEFI boot, the drive needs to be GPT-partitioned and an EFI System Partition (512 MiB or larger, gdisk type EF00, formatted with FAT32) must be present. In the following examples, this partition is assumed to be mounted at /boot. If you have followed this guide from the beginning, you have already done all of these.
Note: It is strongly recommended to have the EFI System Partition mounted at /boot as this is required to automatically update Gummiboot.

Install the gummiboot package and run the automated installation script, replacing $esp with the location of your EFI System Partiton, usually /boot:

# pacman -S gummiboot
# gummiboot --path=$esp install

Gummiboot will automatically be detected by firmware that requires that the bootable bootx64.efi stub be placed in $esp/EFI/boot, and will in turn automatically detect the presence of any other installed operating systems using .efi stubs. However, it will still be necessary to manually create a configuration file for Gummiboot.

First, create $esp/loader/entries/arch.conf and add the following, replacing /dev/sdaX with your root partition (e.g. /dev/sda1):

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

Second, create $esp/loader/loader.conf and add the following, replacing the timeout value (in seconds) with your own choice:

# nano $esp/loader/loader.conf
default  arch
timeout  5

See gummiboot for more information.

Unmount the partitions and reboot

Exit from the chroot environment:

# exit
Note: While partitions are unmounted automatically by systemd on shutdown, you may do so manually with umount -R /mnt as a safety measure. If the partition is "busy", you can find the cause with fuser.

Reboot the computer:

# reboot
Tip: Be sure to remove the installation media, otherwise you will boot back into it. You can log into your machine as the user "root" and the password that you specified with the passwd command. The default password is "root".

Post-installation

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.