Difference between revisions of "Beginners' guide"

From ArchWiki
Jump to navigation Jump to search
m (update link(s) (avoid redirect if the titles differ only in capitalization) (testing https://github.com/lahwaacz/wiki-scripts/blob/master/update-links-avoid-redirect.py))
(→‎Example: Advise to use plain lsblk first, since it includes the SIZE column, which lsblk -f does not. I think we can get away without mentioning the -o <columnlist> option, because "lsblk" / "lsblk -f" combination includes all the important columns.)
Line 358: Line 358:
The Arch Linux install media includes the following partitioning tools: {{ic|fdisk}}, {{ic|gdisk}}, {{ic|cfdisk}}, {{ic|cgdisk}} and {{ic|parted}}.
The Arch Linux install media includes the following partitioning tools: {{ic|fdisk}}, {{ic|gdisk}}, {{ic|cfdisk}}, {{ic|cgdisk}} and {{ic|parted}}.
{{Tip|Use the {{ic|lsblk -f}} or {{ic|lsblk -o NAME,FSTYPE,SIZE,LABEL}} command to list the hard disks attached to your system, along with the sizes of their existing partitions. This will help you to be confident you are partitioning the right disk.}}
{{Tip|Use the {{ic|lsblk}} command to list the hard disks attached to your system, along with the sizes of their existing partitions. This will help you to be confident you are partitioning the right disk. {{ic|lsblk -f}} will show additional information about Labels, UUIDs and filesystem types.}}
The example system will contain a 15 GB root partition, and a [[Partitioning#/home|home]] partition for the remaining space. Choose either MBR or GPT, as described above. Do not choose both!
The example system will contain a 15 GB root partition, and a [[Partitioning#/home|home]] partition for the remaining space. Choose either MBR or GPT, as described above. Do not choose both!

Revision as of 09:10, 4 September 2014

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.



Note: If you wish to install from an existing GNU/Linux distribution, please 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.

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. Dowload 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
  • Burn the ISO image on a CD or DVD with your preferred software. On Arch, that is covered in Optical disc drive#Burning
    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
  • Or you can write the ISO image to a USB stick. For detailed instructions, see USB flash installation media

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.

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

  • If you are using an Intel video chipset and the screen goes blank during the boot process, the problem is likely an issue with Kernel mode setting. A possible workaround may be achieved by rebooting and pressing Tab over the entry that you are trying to boot (i686 or x86_64). At the end of the string type nomodeset and press Enter. Alternatively, try video=SVIDEO-1:d which, if it works, will not disable kernel mode setting. You can also try i915.modeset=0. See the Intel article for more information.
  • If the screen does not go blank and the boot process gets stuck while trying to load the kernel, press Tab while hovering over the menu entry, type acpi=off at the end of the string and press Enter.


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#nano usage. If you have (or plan on having) a dual boot setup with Windows, see Windows and Arch Dual Boot.

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. 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.
Note: Since the ISO released on 2014.04 (but maybe even on previous ones) there seems to be a problem in getting an IP address with DHCP if you are using the family of routers "FritzBox!". At this time models 7390[1] and 7112[2] seem to have this issue, but other models may be affected. The issue seems to be between the dhcpcd client and the FritzBox! routers and the way they assign IP addresses. The solution to the problem seems to be as follows: in your FritzBox! settings, manually delete the entry related to the IP address that identifies your machine. Also, disable the option "Assign always the same IP address to this machine". Now restart the DHCP process or simply reboot your computer and you should get an IP address as usual. If it does not work, try also to reboot your FritzBox!. Once your computer gets the IP address, you can re-enable the previously disabled option.

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 ( 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, 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 drive.


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:

# 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.
  • 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 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 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.


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

First, identify the name of your wireless interface:

# iw dev
        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".

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 network configuration for more detailed information.

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

# wifi-menu wlp3s0

You should now have a working network connection. If you do not, try #Without wifi-menu or check 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.

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.

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 (e.g. "Linksys etc...") 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 passphrase >> /etc/wpa_supplicant.conf
# ip link set interface up
# wpa_supplicant -B -D nl80211 -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.

If wpa_supplicant complains about an unsupported driver at step 4, just leave out the -D nl80211 parameter:

# wpa_supplicant -B -c /etc/wpa_supplicant.conf -i interface

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 drive

Warning: Partitioning can destroy data. You are strongly cautioned and advised to backup any critical data before proceeding.
Note: If you are installing to a USB flash key, see Installing Arch Linux on a USB key.
Tip: If you want to create any stacked block devices for LVM, disk encryption or RAID, do it now.

Choose a partition table type

Note: If Arch and Windows are dual-booting from same disk, then Arch should follow the same firmware boot mode and partitioning combination used by the installed Windows in the disk. Otherwise Windows will fail to boot. See Windows and Arch Dual Boot#Important information for details.

You have to choose between GUID Partition Table (GPT) and Master Boot Record (MBR), see also Partitioning#Choosing between GPT and MBR.

Partitioning tool

Absolute beginners are encouraged to use a graphical partitioning tool. GParted is a good example, and is provided as a live CD. A drive should first be partitioned and afterwards the partitions should be formatted with a file system.

While gparted may be easier to use, if you just want to create a few partitions on a new disk you can get the job done quickly by just using one of the fdisk variants which are included on the install medium. In the next section short usage instructions for both gdisk and fdisk follow.

Erase partition table

If you want to start from scratch, and do not intend to keep existing partitions, erase the partition table with the following command. This simplifies creating new partitions and avoids problems with converting disks from MBR to GPT and vice versa.

# sgdisk --zap-all /dev/sda

Partition scheme

You can decide into how many partitions the disk should be split, and for which directory each partition should be used in the system. The mapping from partitions to directories (frequently called 'mount points') is the Partition scheme. The simplest, and not a bad choice, is to make just one huge / partition. Another popular choice is to have a / and a /home partition.

Additional required partitions:

  • If you have a UEFI motherboard, you will need to create an extra EFI System Partition.
  • 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 an extra BIOS Boot Partition of size 1 or 2 MiB and EF02 type code. Syslinux does not need one.
  • If you have a requirement for a Disk encryption of the system itself, this must be reflected in your partition scheme. It is unproblematic to add encrypted folders, containers or home directories after the system is installed.

See Swap for details if you wish to set up a swap partition or swap file. A swap file is easier to resize than a partition and can be created at any point after installation, but cannot be used with a Btrfs filesystem.

If you have already created your partitions, proceed to #Create filesystems. Otherwise, see the following example.


The Arch Linux install media includes the following partitioning tools: fdisk, gdisk, cfdisk, cgdisk and parted.

Tip: Use the lsblk command to list the hard disks attached to your system, along with the sizes of their existing partitions. This will help you to be confident you are partitioning the right disk. lsblk -f will show additional information about Labels, UUIDs and filesystem types.

The example system will contain a 15 GB root partition, and a home partition for the remaining space. Choose either MBR or GPT, as described above. Do not choose both!

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

Using cgdisk to create GPT partitions

Launch cgdisk with:

# cgdisk /dev/sda
Tip: If cgdisk cannot change your disk to GPT, parted can.


  • Choose New (or press N) – Enter for the first sector (2048) – type in 15GEnter for the default hex code (8300) – Enter for a blank partition name.


  • Press the down arrow a couple of times to move to the larger free space area.
  • Choose New (or press N) – Enter for the first sector – Enter to use the rest of the drive (or you could type in the desired size; for example 30G) – Enter for the default hex code (8300) – Enter for a blank partition name.

Here is what it should look like:

Part. #     Size        Partition Type            Partition Name
            1007.0 KiB  free space
   1        15.0 GiB    Linux filesystem
   2        123.45 GiB  Linux filesystem

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 Q) to exit without saving changes and then restart cgdisk.

If you are satisfied, choose Write (or press Shift+W) to finalize and to write the partition table to the drive. Type yes and choose Quit (or press Q) to exit without making any more changes.

Using fdisk to create MBR partitions
Note: There is also cfdisk, which is similar in UI to cgdisk, but it currently does not automatically align the first partition properly. That is why the classic fdisk tool is used here.

Launch fdisk with:

# fdisk /dev/sda

Create the partition table:

  • Command (m for help): type o and press Enter

Then create the first partition:

  1. Command (m for help): type n and press Enter
  2. Partition type: Select (default p): press Enter
  3. Partition number (1-4, default 1): press Enter
  4. First sector (2048-209715199, default 2048): press Enter
  5. Last sector, +sectors or +size{K,M,G,T,P} (2048-209715199....., default 209715199): type +15G and press Enter

Then create a second partition:

  1. Command (m for help): type n and press Enter
  2. Partition type: Select (default p): press Enter
  3. Partition number (1-4, default 2): press Enter
  4. First sector (31459328-209715199, default 31459328): press Enter
  5. Last sector, +sectors or +size{K,M,G,T,P} (31459328-209715199....., default 209715199): press Enter

Now preview the new partition table:

  • Command (m for help): type p and press Enter
Disk /dev/sda: 107.4 GB, 107374182400 bytes, 209715200 sectors
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x5698d902

   Device Boot     Start         End     Blocks   Id  System
/dev/sda1           2048    31459327   15728640   83   Linux
/dev/sda2       31459328   209715199   89127936   83   Linux

Then write the changes to disk:

  • Command (m for help): type w and press Enter

If everything went well fdisk will now quit with the following message:

The partition table has been altered!

Calling ioctl() to re-read partition table.
Syncing disks. 

In case this does not work because fdisk encountered an error, you can use the q command to exit.

Create filesystems

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 is actually /dev/sda1 and /dev/sda2 that you want to format. You can use lsblk to help with this.
# mkfs.ext4 /dev/sda1
# mkfs.ext4 /dev/sda2

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

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

For UEFI, you should format the EFI System Partition (for example /dev/sdXY) with:

# mkfs.fat -F32 /dev/sdXY

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 -f
Note: Do not mount more than one partition to the same directory. And 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

In case you have a UEFI motherboard, mount the EFI System Partition to /boot. Whilst other mountpoints are viable, using /boot is recommended as explained in the EFISTUB article.

# mkdir /mnt/boot
# mount /dev/sdXY /mnt/boot

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 2012-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.

  • 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 mirrorlist, refresh all package lists with pacman -Syy, 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 [3] and [4]). 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. You may also want to include base-devel, as you will need these packages should you want to build packages from the AUR or using the ABS:

# pacstrap -i /mnt base base-devel
  • 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 pacman fails to verify your packages, stop the process with Ctrl+C and check the system time with cal. If the system date is invalid (e.g. it shows the 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, using the command ntpd -qg, and retry running the pacstrap command. Refer to Time page for more information on correcting system time.
  • If pacman complains that error: failed to commit transaction (invalid or corrupted package), run the following command:
# pacman-key --init && pacman-key --populate archlinux

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:

# 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
Note: Leave out /bin/bash to 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.


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 has everything commented out by default. To uncomment a line remove the # in the front. Using UTF-8 is highly recommended over ISO-8859:

# 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
Note: This will also run with every update of glibc.

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

# echo LANG=en_US.UTF-8 > /etc/locale.conf
  • 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#Setting the locale system-wide 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
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


# 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.


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

Add the same hostname to /etc/hosts:

# nano /etc/hosts
# /etc/hosts: static lookup table for host names

#<ip-address>	<hostname.domain.org>	<hostname>	localhost.localdomain	localhost	myhostname
::1		localhost.localdomain	localhost

# End of file

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.

  • 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.


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

Notice the /24 in Address which is the CIDR notation of a netmask.

Enable above created profile to start it at every boot:

# netctl enable my_network
Using systemd-networkd

See systemd-networkd.


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: This must be done after your reboot when you are 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 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
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

Set the root password with:

# passwd

Install and configure a bootloader

For BIOS motherboards

For BIOS systems, several boot loaders are available, see Boot loaders for a complete list. Choose one as per your convenience. Here, two of the possibilities are given as examples:

  • 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 in the syslinux article.
  • 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.

If you opted for a GUID partition table (GPT) for your hard drive earlier, you need to install the gptfdisk package now for the installation of syslinux to work:

# pacman -S gptfdisk

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

# pacman -S syslinux
# syslinux-install_update -iam

After installing Syslinux, 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 /dev/sda1).

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

If adding UUID rather than partition number the syntax is APPEND root=UUID=partition_uuid rw.

Do the same for the fallback entry.

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


Install the grub package and then run grub-install to install the bootloader:

# pacman -S grub
# grub-install --target=i386-pc --recheck /dev/sda
  • Change /dev/sda to reflect the drive you installed Arch on. Do not append a partition number (do not use sdaX).
  • For GPT-partitioned drives on BIOS motherboards, you also need a "BIOS Boot Partition". See GPT-specific instructions in the GRUB page.
  • 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.

While using a manually created grub.cfg is absolutely fine, it is 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
Note: It is possible that multiple redundant menu entries will be generated. See GRUB#Redundant_menu_entries.

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. Here, two of the possibilities are given as examples:

  • gummiboot is a minimal UEFI Boot Manager which provides a menu for EFISTUB kernels and other UEFI applications.
  • GRUB is a more complete bootloader, useful if you run into problems with Gummiboot.

No matter which one you choose, first install the dosfstools package, so you can manipulate your EFI System Partition after installation:

# pacman -S dosfstools
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.

Install the gummiboot package and run gummiboot install to install the bootloader to the EFI System Partition:

# pacman -S gummiboot
# gummiboot install
Warning: Gummiboot and the Linux Kernel will not automatically update if your EFI System Partition is not mounted at /boot.

You will need to manually create a configuration file to add an entry for Arch Linux to the gummiboot manager. Create /boot/loader/entries/arch.conf and add the following contents, replacing /dev/sdaX with your root partition, usually /dev/sda2:

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

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


Install the grub and efibootmgr packages and run grub-install to install the bootloader:

# pacman -S grub efibootmgr
# grub-install --target=x86_64-efi --efi-directory=/boot --bootloader-id=arch_grub --recheck

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

Tip: To automatically search for other operating systems on your computer, install os-prober before running the next command. However os-prober is not known to properly detect UEFI OSes.
# 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

Reboot the computer:

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


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 General recommendations#System administration and General recommendations#Package management.

See the rest of the General recommendations article for 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.