Difference between revisions of "Beginners' guide"

From ArchWiki
Jump to navigation Jump to search
(→‎EFISTUB: refindx64.efi -> refind_x64.efi)
m (→‎Mount the partitions: add a note about mounting a lot of things to the same place.)
Line 332: Line 332:
  # mkdir /mnt/home
  # mkdir /mnt/home
  # mount /dev/sda2 /mnt/home
  # mount /dev/sda2 /mnt/home
{{Note| Do not mount more than one partition to a directory. Having two /dev/sdxY partitions mounted to /mnt is bad. You have to manually create the /mnt/boot or /mnt/home directories to mount.}}
In case you have a UEFI motherboard, mount the UEFI partition:
In case you have a UEFI motherboard, mount the UEFI partition:

Revision as of 09:09, 25 January 2013

ro:Ghidul începătorilor/Instalare zh-CN:Beginners' Guide/Installation

Tip: This is part of a multi-page article for The Beginners' Guide. Click here if you would rather read the guide in its entirety.


You are now presented with a shell prompt, automatically logged in as root.

Change the language

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.

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, be-latin1, etc. See here for a comprehensive list.

The font should also be changed, because most languages use more glyphs than the 26 letter English alphabet. Otherwise some foreign characters may show up as white squares or as other symbols. Note that the name is case-sensitive, so please type it exactly as you see it:

# setfont Lat2-Terminus16

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 entry.

Use Template:Keypress to exit, and when prompted to save changes, press Template:Keypress and Template:Keypress to use the same filename.

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

Remember, Template:Keypress activates and deactivates the keymap.

Establish an internet connection

Warning: udev no longer assigns network interface names according to the wlanX and ethX naming scheme. If you're 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 "ip" utility to discover the names of your interfaces.

From systemd-197's release and onward, udev now assigns predictable, stable network interface names that deviate from the legacy incremental naming scheme (wlan0, wlan1, etc.). These interface names are guaranteed to be persistent across reboots, which solves the problem of the lack of predictability of network interface name assignment. For more information about why this was necessary, read http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames .

The dhcpcd network daemon is started automatically at boot and it will attempt to start a wired connection, if available. Try pinging a website to see if it was successful. And since Google is always on...

# ping -c 3 www.google.com
PING www.l.google.com ( 56(84) bytes of data.
64 bytes from wb-in-f105.1e100.net ( icmp_req=1 ttl=50 time=17.0 ms
64 bytes from wb-in-f105.1e100.net ( icmp_req=2 ttl=50 time=18.2 ms
64 bytes from wb-in-f105.1e100.net ( 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, you will need to set up the network manually, as explained below.

Otherwise, move on to Prepare the storage drive.


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

First, 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 case, the ethernet interface is enp2s0f0. If you're unsure, your ethernet interface is likely to start with the letter "e", and unlikely to be "lo" or start with the letter "w". You can also use iwconfig and see which interfaces are not wireless:

# iwconfig
enp2s0f0  no wireless extensions.
wlp3s0    IEEE 802.11bgn  ESSID:"NETGEAR97"  
          Mode:Managed  Frequency:2.427 GHz  Access Point: 2C:B0:5D:9C:72:BF   
          Bit Rate=65 Mb/s   Tx-Power=16 dBm   
          Retry  long limit:7   RTS thr:off   Fragment thr:off
          Power Management:on
          Link Quality=61/70  Signal level=-49 dBm  
          Rx invalid nwid:0  Rx invalid crypt:0  Rx invalid frag:0
          Tx excessive retries:0  Invalid misc:430   Missed beacon:0
lo        no wireless extensions.

In this example, neither enp2s0f0 nor the loopback device have wireless extensions, meaning enp2s0f0 is our ethernet interface.

You also need to know these settings:

  • Static IP address.
  • Subnet mask.
  • Gateway's IP address.
  • Name servers' (DNS) IP addresses.
  • Domain name (unless you're 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>/<subnetmask> dev <interface>

For example:

# ip addr add 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

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

# nano /etc/resolv.conf
nameserver 61.95.849.8
search example.com
Note: Currently, you may include a maximum of 3 nameserver lines.

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


Follow this procedure if you need wireless connectivity (Wi-Fi) during the installation process.

If you're coming from another distribution, or if this is your first time installing Arch Linux since the deprecation of the old interface naming scheme, you might be surprised to learn that the first wireless interface is not named "wlan0". In fact, none of the interfaces are automatically prefixed with "wlan" any longer. Don't panic; simply execute iwconfig to discover the name of your wireless interface.

The wireless drivers and utilities are now available to you in the live environment of the installation media. A good knowledge of your wireless hardware will be of key importance to successful configuration. Note that the following quick-start procedure executed at this point in the installation will initialize your wireless hardware for use in the live environment of the installation media. These steps (or some other form of wireless management) must be repeated from the actual installed system after booting into it.

Also note that these steps are optional if wireless connectivity is unnecessary at this point in the installation; wireless functionality may always be established later.

Note: The following examples use wlp3s0 for the interface and linksys for the ESSID. Remember to change these values according to your setup.

The basic procedure will be:

  • Identify the wireless interface:
# lspci | grep -i net

Or, if using a USB adapter:

# lsusb
  • Ensure udev has loaded the driver, and that the driver has created a usable wireless kernel interface with iwconfig:
Note: If you do not see output similar to this, then your wireless driver has not been loaded. If this is the case, you must load the driver yourself. Please see Wireless Setup for more detailed information.
# iwconfig
enp2s0f0  no wireless extensions.
wlp3s0    IEEE 802.11bgn  ESSID:"NETGEAR97"  
          Mode:Managed  Frequency:2.427 GHz  Access Point: 2C:B0:5D:9C:72:BF   
          Bit Rate=65 Mb/s   Tx-Power=16 dBm   
          Retry  long limit:7   RTS thr:off   Fragment thr:off
          Power Management:on
          Link Quality=61/70  Signal level=-49 dBm  
          Rx invalid nwid:0  Rx invalid crypt:0  Rx invalid frag:0
          Tx excessive retries:0  Invalid misc:430   Missed beacon:0
lo        no wireless extensions.

In this example, wlp3s0 is the available wireless interface.

  • Bring the interface up with:
# ip link set wlp3s0 up

A small percentage of wireless chipsets also require firmware, in addition to a corresponding driver. If the wireless chipset requires firmware, you are likely to receive this error when bringing the interface up:

# ip link set wlp3s0 up
SIOCSIFFLAGS: No such file or directory

If unsure, invoke dmesg to query the kernel log for a firmware request from the wireless chipset.

Example output from 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.

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 Setup if you are unsure about the requirement of corresponding firmware installation for your particular chipset.

Next, use netcfg's wifi-menu to connect to a network:

# wifi-menu wlp3s0
Warning: At the moment, netcfg's wifi-menu, when executed without arguments, will look for "wlan0". Execute wifi-menu with your interface as the argument in order to use it.

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

xDSL (PPPoE), analog modem or ISDN

If you have a router in bridge mode, run:

# pppoe-setup

To use these settings and connect to your ISP, run:

# pppoe-start

You may also need to adjust your resolv.conf:

# echo nameserver > /etc/resolv.conf

If you have a dial-up or ISDN connection, 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 drive

Warning: Partitioning can destroy data. You are strongly cautioned and advised to backup any critical data before proceeding.

Absolute beginners are encouraged to use a graphical partitioning tool. GParted is a good example, and is provided as a "live" CD. It is also included on live CDs of most Linux distributions such as Ubuntu and Linux Mint. A drive should first be partitioned and the partitions should be formatted with a file system before rebooting.

It's possible to set up a swap file at any point after installation, so there is no need to decide on swap size now. See Swap for details if you wish to set up a swap partition now (but note that it's much easier to resize a file than a partition).

If you have already done so, proceed to Mount the partitions.

Otherwise, see the following example.


The Arch Linux install media includes the following partitioning tools:

This example uses cfdisk, but it can easily be followed using cgdisk, which will allow for GPT instead of MBR partitioning.

Notes regarding UEFI boot:
  • If you have a UEFI motherboard, you will need to create an extra UEFI System partition.
  • It is recommended to always use GPT for UEFI boot, as some UEFI firmwares do not allow UEFI-MBR boot.
Notes regarding GPT partitioning:
  • If you are not dual booting with Windows, then it is advisable to use GPT instead of MBR. Read GPT for a list of advantages.
  • If you have a BIOS motherboard (or plan on booting in BIOS compatibility mode) and you want to setup GRUB on a GPT-partitioned drive, you will need to create a 2 MiB "BIOS Boot Partition". Syslinux doesn't need one.
  • Some BIOS systems may have issues with GPT. See http://mjg59.dreamwidth.org/8035.html and http://rodsbooks.com/gdisk/bios.html for more info and possible workarounds.
Note: If you are installing to a USB flash key, see Installing Arch Linux on a USB key.
# cfdisk /dev/sda

The example system will contain a 15 GB root partition, and a home partition for the remaining space.

It should be emphasized that partitioning is a personal choice and that this example is only for illustrative purposes. See Partitioning.



Here's how it should look like:

Name    Flags     Part Type    FS Type          [Label]       Size (MB)
sda1    Boot       Primary     Linux                             15360
sda2               Primary     Linux                             133000*

Double check and make sure that you are happy with the partition sizes as well as the partition table layout before continuing.

If you would like to start over, you can simply select Quit (or press Template:Keypress) to exit without saving changes and then restart cfdisk.

If you are satisfied, choose Write (or press Template:Keypress) to finalize and to write the partition table to the drive. Type "yes" and choose Quit (or press Template:Keypress) to exit cfdisk without making any more changes.

Simply partitioning is not enough; the partitions also need a filesystem. To format the partitions with an ext4 filesystem:

Warning: Double check and triple check that it's actually /dev/sda1 and /dev/sda2 that you want to format.
# mkfs.ext4 /dev/sda1
# mkfs.ext4 /dev/sda2

If you have made a partition dedicated to swap (code 82), don't forget to format and activate it with:

# mkswap /dev/sdaX
# swapon /dev/sdaX

Mount the partitions

Each partition is identified with a number suffix. For example, sda1 specifies the first partition of the first drive, while sda designates the entire drive.

To display the current partition layout:

# lsblk /dev/sda

Pay attention, because the mounting order is important.

First, mount the root partition on /mnt. Following the example above (yours may be different), it would be:

# mount /dev/sda1 /mnt

Then mount the /home partition and any other separate partition (/boot, /var, etc), if you have any:

# mkdir /mnt/home
# mount /dev/sda2 /mnt/home
Note: Do not mount more than one partition to a directory. Having two /dev/sdxY partitions mounted to /mnt is bad. You have to manually create the /mnt/boot or /mnt/home directories to mount.

In case you have a UEFI motherboard, mount the UEFI partition:

# mkdir /mnt/boot/efi
# mount /dev/sdaX /mnt/boot/efi

Select a mirror

Before installing, 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's worth getting it right.

# nano /etc/pacman.d/mirrorlist
## Arch Linux repository mirrorlist
## Sorted by mirror score from mirror status page
## Generated on 2012-MM-DD

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

If you want, you can make it the only mirror available by getting rid of everything else (using Template:Keypress), but it's usually a good idea to have a few more, in case the first one goes offline.

  • 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.
  • Whenever in the future you change your list of mirrors, always remember to force pacman to refresh all package lists with pacman -Syy. This is considered to be good practice and will avoid possible headaches. See Mirrors for more information.
  • If you're 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 and base-devel groups without prompting.

# pacstrap -i /mnt base base-devel
Note: If pacman fails to verify your packages, check the system time with cal. If the system date is invalid (e.g. it shows year 2010), signing keys will be considered expired (or invalid), signature checks on packages will fail and installation will be interrupted. Make sure to correct the system time, either by doing so manually or with the ntp client, and retry running the pacstrap command. Refer to Time page for more information on correcting system time.
  • base: Software packages from the [core] repo to provide the minimal base environment.
  • base-devel: Extra tools from [core] such as make, and automake. Most beginners should choose to install it, as it will likely be needed to expand the system. The base-devel group will be required to install software from the Arch User Repository.

This will give you a basic Arch system. Other packages can be installed later using pacman.

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.

Note: If you encounter errors running genfstab or later in the install process, do not run genfstab again; just edit the fstab file.
# genfstab -U -p /mnt  | sed 's/rw,relatime,data=ordered/defaults,relatime/' >> /mnt/etc/fstab
# nano /mnt/etc/fstab
Warning: The fstab file should always be checked after generating it. If you made an EFI system partition earlier, then genfstab has incorrectly added options to your EFI system partition. This will in fact prevent your computer from booting from that drive, so you need to remove all options for the EFI partition except for noatime. For the other partitions that use it, be sure to replace "codepage=cp437" with "codepage=437" or else when you next reboot, any mounts with this option will fail and systemd will halt and drop into recovery mode. This should be fixed by linux 3.8

A few considerations:

  • Only the root (/) partition needs 1 for the last field. Everything else should have either 2 or 0 (see fstab#Field definitions).

Chroot and configure the base system

Next, we chroot into our newly installed system:

# arch-chroot /mnt
Note: Use arch-chroot /mnt /bin/bash to chroot into a bash 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.


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.

There are two files that need editing: locale.gen and locale.conf.

  • The locale.gen file is empty by default (everything is commented out) and you need to remove the # in front of the line(s) you want. You may uncomment more lines than just English (US), as long as you choose their UTF-8 encoding:
# nano /etc/locale.gen
en_US.UTF-8 UTF-8
de_DE.UTF-8 UTF-8
# locale-gen

This will run on every glibc upgrade, generating all the locales specified in /etc/locale.gen.

  • The locale.conf file doesn't exist by default. Setting only LANG should be enough. It will act as the default value for all other variables.
# echo LANG=en_US.UTF-8 > /etc/locale.conf
# export LANG=en_US.UTF-8
Note: If you set some other language than English at the beginning of the install, the above commands would be something like:
# echo LANG=de_DE.UTF-8 > /etc/locale.conf
# export LANG=de_DE.UTF-8

To use other LC_* variables, first run locale to see the available options. An advanced example can be found here.

Warning: Using the LC_ALL variable is strongly discouraged because it overrides everything.

Console font and keymap

If you set a keymap at the beginning of the install process, load it now, as well, because the environment has changed. For example:

# loadkeys de-latin1
# setfont Lat2-Terminus16

To make them available after reboot, edit vconsole.conf:

# nano /etc/vconsole.conf
  • KEYMAP – Please note that this setting is only valid for your TTYs, not any graphical window managers or Xorg.
  • FONT – Available alternate console fonts reside in /usr/share/kbd/consolefonts/. The default (blank) is safe, but some foreign characters may show up as white squares or as other symbols. It's recommended that you change it to Lat2-Terminus16, because according to /usr/share/kbd/consolefonts/README.Lat2-Terminus16, it claims to support "about 110 language sets".
  • Possible option FONT_MAP – Defines the console map to load at boot. Read man setfont. Removing it or leaving it blank is safe.

See Console fonts and man vconsole.conf 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 <Zone>, 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 zone file /usr/share/zoneinfo/<Zone>/<SubZone> using this command:

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


# ln -s /usr/share/zoneinfo/Europe/Minsk /etc/localtime

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

To synchronize your "UTC" time over the internet, see NTPd.

  • 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

If you have (or planning on having) a dual boot setup with Windows:

  • Recommended: Set both Arch Linux and Windows to use UTC. A quick registry fix is needed. Also, be sure to prevent Windows from synchronizing the time on-line, because the hardware clock will default back to localtime.
  • Not recommended: Set Arch Linux to localtime and disable any time-related services, like NTPd . This will let Windows take care of hardware clock corrections and you will need to remember to boot into Windows at least two times a year (in Spring and Autumn) when DST kicks in. So please don't ask on the forums why the clock is one hour behind or ahead if you usually go for days or weeks without booting into Windows.

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.


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.


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

# echo myhostname > /etc/hostname
Note: You no longer need to edit /etc/hosts. The nss-myhostname package will provide host name resolution, and is installed on all systems by default.

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.

Note: For more in-depth information on network configration, visit Configuring Network and Wireless Setup.


Dynamic IP

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. Where <interface> is your wired interface:

# systemctl enable dhcpcd@<interface>.service
Warning: systemd now renames network interfaces. Moreover, dhcpcd@.service always creates symlink named dhcpcd@eth0.service, no matter what you supply for <interface>. The command above won't work.

Alternatively, you can use netcfg's net-auto-wired, which gracefully handles dynamic connections to new networks:

Install ifplugd, which is required for net-auto-wired:

# pacman -S ifplugd

Set up the dhcp profile and enable the net-auto-wired service:

# cd /etc/network.d
# ln -s examples/ethernet-dhcp .
# systemctl enable net-auto-wired.service
Static IP

Install ifplugd, which is required for net-auto-wired:

# pacman -S ifplugd

Copy a sample profile from /etc/network.d/examples to /etc/network.d:

# cd /etc/network.d
# cp examples/ethernet-static .

Edit the profile as needed:

# nano ethernet-static

Enable the net-auto-wired service:

# systemctl enable net-auto-wired.service


You will need to install additional programs to be able to configure and manage wireless network profiles for netcfg.

NetworkManager and Wicd are other popular alternatives.

  • Install the required packages:
# pacman -S wireless_tools wpa_supplicant wpa_actiond dialog

If your wireless adapter requires a firmware (as described in the above Establish an internet connection section and also here), install the package containing your firmware. For example:

# pacman -S zd1211-firmware
  • After finishing the rest of this installation and rebooting, you can connect to the network with wifi-menu <interface> (where <interface> is the interface of your wireless chipset), which will generate a profile file in /etc/network.d named after the SSID. There are also templates available in /etc/network.d/examples/ for manual configuration.
# wifi-menu <interface>
Warning: If you're using wifi-menu, this must be done *after* your reboot when you're no longer chrooted. The 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 templates previously mentioned so that you don't have to worry about using wifi-menu at all.
  • Enable the net-auto-wireless service, which will connect to known networks and gracefully handle roaming and disconnects:
# systemctl enable net-auto-wireless.service
Note: Netcfg also provides net-auto-wired, which can be used in conjunction with net-auto-wireless.
  • Make sure that the correct wireless interface (usually wlan0) is set in /etc/conf.d/netcfg:
# nano /etc/conf.d/netcfg

It is also possible to define a list of network profiles that should be automatically connected, using the AUTO_PROFILES variable in /etc/conf.d/netcfg. If AUTO_PROFILES is not set, all known wireless networks will be tried.

xDSL (PPPoE), analog modem or ISDN

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

Configure pacman

Pacman is the Arch Linux package manager. It is highly recommended to study and learn how to use it. Read man pacman, have a look at the pacman article, or check out the Pacman Rosetta article for a comparison to other popular package managers.

For repository selections and pacman options, edit pacman.conf:

# nano /etc/pacman.conf

Most people will want to use [core], [extra] and [community].

If you installed Arch Linux x86_64, it's recommended that you enable the [multilib] repository, as well (to be able to run both 32 bit and 64 bit applications):

Note: When choosing repos, be sure to uncomment both the [repo_name] header lines, as well as the lines below. Failure to do so will result in the selected repository being omitted! This is a very common error. A correct example for the multilib repository is found below.
SigLevel = PackageRequired
Include = /etc/pacman.d/mirrorlist

You will then need to update the package list by running pacman with the -Sy switch. Failing to do so will generate "warning: database file for 'multilib' does not exist" error when next using pacman.

See Official Repositories for more information, including details about the purpose of each repository.

For software unavailable directly through pacman, see Arch User Repository.

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, or if /usr is on a separate partition.

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

# mkinitcpio -p linux
Note: Arch VPS installations on QEMU (e.g. when using virt-manager) may need virtio modules in mkinitcpio.conf to be able to boot.
# nano /etc/mkinitcpio.conf
MODULES="virtio virtio_blk virtio_pci virtio_net"

Set the root password and add a regular user

Set the root password with:

# passwd
Warning: Linux is a multi-user operating system. You should not perform everyday tasks using the root account. It is considered a very poor practice and could be extremely dangerous. The root account should only be used for administrative tasks.

Then add a normal user account. The user archie is just an example.

# useradd -m -g users -G wheel -s /bin/bash archie
# passwd archie

If you wish to start over, use userdel. The -r option will remove the user's home directory and its content, along with the user's settings (the so-called "dot" files).

# userdel -r archie

For more information, read Users and Groups.

Install and configure a bootloader

For BIOS motherboards

For BIOS systems, there are three bootloaders - Syslinux, GRUB, and LILO. Choose the bootloader as per your convenience. Below only Syslinux and GRUB are explained.

  • Syslinux 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 here.
  • GRUB is more feature-rich and supports more complex scenarios. Its configuration file(s) is more similar to a scripting language, which may be difficult for beginners to manually write. It is recommended that they automatically generate one.
Note: Some BIOS systems may have issues with GPT. See http://mjg59.dreamwidth.org/8035.html and http://rodsbooks.com/gdisk/bios.html for more info and possible workarounds.

Install the syslinux package and then use the syslinux-install_update script to automatically install the files (-i), mark the partition active by setting the boot flag (-a), and install the MBR boot code (-m):

Note: If you have partitioned the drive as GPT, install gptfdisk package, as well (pacman -S gptfdisk), because it contains sgdisk, which will be used to set the GPT-specific boot flag.
# pacman -S syslinux
# syslinux-install_update -i -a -m

Configure syslinux.cfg to point to the right root partition. This step is vital. If it points to the wrong partition, Arch Linux will not boot. Change /dev/sda3 to reflect your root partition (if you partitioned your drive as in the example, your root partition is sda1). Do the same for the fallback entry.

# nano /boot/syslinux/syslinux.cfg
LABEL arch
        APPEND root=/dev/sda3 ro

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


Install the grub-bios package and then run grub-install:

Note: Change /dev/sda to reflect the drive you installed Arch on. Do not append a partition number (do not use sdaX).
Note: For GPT-partitioned drives on BIOS motherboards, GRUB needs a 2 MiB "BIOS Boot Partition".
# pacman -S grub-bios
# grub-install --target=i386-pc --recheck /dev/sda
# cp /usr/share/locale/en\@quot/LC_MESSAGES/grub.mo /boot/grub/locale/en.mo

While using a manually created grub.cfg is absolutely fine, it's recommended that beginners automatically generate one:

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 boot, the drive needs to be GPT-partitioned, and a UEFI System Partition (512 MiB or higher, FAT32, type EF00) must be present and mounted on /boot/efi. If you have followed this guide from the beginning, you've already done all of these.

While there are other UEFI bootloaders available, using EFISTUB is recommended. Below are instructions for setting up EFISTUB and GRUB.

Note: Syslinux does not yet support UEFI.

The Linux kernel can act as its own bootloader using EFISTUB. This is the UEFI boot method recommended by developers and simpler compared to grub-efi-x86_64. The below steps set up rEFInd (a fork of rEFIt) to provide a menu for EFISTUB kernels, as well as for booting other UEFI bootloaders. You can also use gummiboot instead of rEFInd. Both rEFInd and gummiboot can detect Windows UEFI bootloader in case of dual-boot.

1. Boot in UEFI mode and load efivars kernel module before chrooting:

# modprobe efivars      # before chrooting

2. Mount the UEFISYS partition at /mnt/boot/efi, chroot and copy the kernel and initramfs files as described below.

  • Create /boot/efi/EFI/arch/ directory.
  • Copy /boot/vmlinuz-linux to /boot/efi/EFI/arch/vmlinuz-arch.efi. The .efi file extension is very important as some UEFI firmwares refuse to launch a file without this extension. Important: Remember that the file is called vmlinuz, but not vmlinux.
  • Copy /boot/initramfs-linux.img to /boot/efi/EFI/arch/initramfs-arch.img.
  • Copy /boot/initramfs-linux-fallback.img to /boot/efi/EFI/arch/initramfs-arch-fallback.img.

Every time the kernel and initramfs files are updated in /boot, they need to be updated in /boot/efi/EFI/arch. This can be automated either using systemd or using incron (for non-systemd setups).

3. In this guide you set up a bootloader GUI called rEFInd. Alternative bootloaders can be found on the page UEFI Bootloaders#Booting EFISTUB. For the recommended rEFInd bootloader install the following packages:

# pacman -S refind-efi efibootmgr

4. Install rEFInd to the UEFISYS partition (summarized from UEFI Bootloaders#Using rEFInd):

# mkdir -p /boot/efi/EFI/refind
# cp /usr/lib/refind/refind_x64.efi /boot/efi/EFI/refind/refind_x64.efi
# cp /usr/lib/refind/config/refind.conf /boot/efi/EFI/refind/refind.conf
# cp -r /usr/share/refind/icons /boot/efi/EFI/refind/icons

5. Create a refind_linux.conf file with the kernel parameters to be used by rEFInd:

# nano /boot/efi/EFI/arch/refind_linux.conf
"Boot to X"          "root=/dev/sdaX ro rootfstype=ext4 systemd.unit=graphical.target"
"Boot to console"    "root=/dev/sdaX ro rootfstype=ext4 systemd.unit=multi-user.target"
Note: refind_linux.conf is copied in the directory /boot/efi/EFI/arch/ where the initramfs and the kernel have been copied to in step 2.
Note: In refind_linux.conf, sdaX refers to your root file system, not your boot partition, if you created them separately.

6. Add rEFInd to UEFI boot menu using efibootmgr.

Warning: Using efibootmgr on Apple Macs may brick the firmware and may need reflash of the motherboard ROM. For Macs, use mactel-bootAUR, or "bless" from within Mac OS X.
# efibootmgr -c -g -d /dev/sdX -p Y -w -L "rEFInd" -l '\EFI\refind\refind_x64.efi'
Note: In the above command, X and Y denote the drive and partition of the UEFISYS partition. For example, in /dev/sdc5, X is "c" and Y is "5".

7. (Optional) As a fallback, in case efibootmgr created boot entry does not work, copy refind_x64.efi to /boot/efi/EFI/boot/boot_x64.efi as follows:

# cp -r /boot/efi/EFI/refind/* /boot/efi/EFI/boot/
# mv /boot/efi/EFI/boot/refind_x64.efi /boot/efi/EFI/boot/boot_x64.efi
Note: In case you have a system with 32-bit EFI, like pre-2008 Macs, install grub-efi-i386 instead, and use --target=i386-efi.
# pacman -S grub-efi-x86_64 efibootmgr
# grub-install --target=x86_64-efi --efi-directory=/boot/efi --bootloader-id=arch_grub --recheck
# cp /usr/share/locale/en\@quot/LC_MESSAGES/grub.mo /boot/grub/locale/en.mo

The next command creates a menu entry for GRUB in the UEFI boot menu. However, as of grub-efi-x86_64 version 2.00, grub-install tries to create a menu entry, so running efibootmgr may not be necessary. See UEFI#efibootmgr for more info.

# efibootmgr -c -g -d /dev/sdX -p Y -w -L "Arch Linux (GRUB)" -l '\EFI\arch_grub\grubx64.efi'

Next, while using a manually created grub.cfg is absolutely fine, it's recommended that beginners automatically generate one:

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.

Unmount the partitions and reboot

Exit from the chroot environment:

# exit

Since the partitions are mounted under /mnt, we use the following command to unmount them:

# umount /mnt/{boot,home,}

Reboot the computer:

# reboot
Tip: Be sure to remove the installation media, otherwise you will boot back into it.

Template:Beginners' Guide navigation