Difference between revisions of "Beginners' guide"

From ArchWiki
Jump to navigation Jump to search
(Partition schemes: "help mkpart" is more up to date than the manual (suggested by User:Lahwaacz in Talk:Beginners'_guide#Linked to 'parted' Manual doesn't list ext3 or ext4 for fs-type))
Line 396: Line 396:
  
 
* {{ic|''part-type''}} is one of {{ic|primary}}, {{ic|extended}} or {{ic|logical}}, and is meaningful only for MBR partition tables.
 
* {{ic|''part-type''}} is one of {{ic|primary}}, {{ic|extended}} or {{ic|logical}}, and is meaningful only for MBR partition tables.
* {{ic|''fs-type''}} is an identifier chosen among those listed in the [http://www.gnu.org/software/parted/manual/parted.html#mkpart manual] as the closest match to the file system that you will use in [[#Create filesystems]]. The ''mkpart'' command does not actually create the file system: the {{ic|''fs-type''}} parameter will simply be used by ''parted'' to set a 1-byte code that is used by boot loaders to "preview" what kind of data is found in the partition, and act accordingly if necessary. See also [[Wikipedia:Disk partitioning#PC partition types]].
+
* {{ic|''fs-type''}} is an identifier chosen among those listed by entering {{ic|help mkpart}} as the closest match to the file system that you will use in [[#Create filesystems]]. The ''mkpart'' command does not actually create the file system: the {{ic|''fs-type''}} parameter will simply be used by ''parted'' to set a 1-byte code that is used by boot loaders to "preview" what kind of data is found in the partition, and act accordingly if necessary. See also [[Wikipedia:Disk partitioning#PC partition types]].
 
: {{Tip|Most [[Wikipedia:File_system#Linux|Linux native file systems]] map to the same partition code ([[Wikipedia:Partition type#PID_83h|0x83]]), so it is perfectly safe to e.g. use {{ic|ext2}} for an ''ext4''-formatted partition.}}
 
: {{Tip|Most [[Wikipedia:File_system#Linux|Linux native file systems]] map to the same partition code ([[Wikipedia:Partition type#PID_83h|0x83]]), so it is perfectly safe to e.g. use {{ic|ext2}} for an ''ext4''-formatted partition.}}
 
* {{ic|''start''}} is the beginning of the partition from the start of the device. It consists of a number followed by a [http://www.gnu.org/software/parted/manual/parted.html#unit unit], for example {{ic|1M}} means start at 1MiB.
 
* {{ic|''start''}} is the beginning of the partition from the start of the device. It consists of a number followed by a [http://www.gnu.org/software/parted/manual/parted.html#unit unit], for example {{ic|1M}} means start at 1MiB.

Revision as of 13:20, 10 April 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.

Minimum system requirements

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

Tip: Compared to the regular ISO images, the archboot images can take several steps explained in this guide interactively. See Archboot for details.

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

It is highly recommended to verify the image signature before use, especially when downloading from an HTTP mirror, as these are run by volunteers who could theoretically serve malicious images. [1] On a system with GnuPG installed, do this by downloading the PGP signature (under Checksums) to the ISO directory, and run:

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

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

Alternatively, run from an existing Arch Linux installation:

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

Now choose one of the methods from the table below to #Boot the installation medium on the target machine(s). As the installation process retrieves packages from a remote repository, these methods require an internet connection; see Offline installation of packages when none is available.

Method Articles Conditions
Write the image on flash media or optical disc, then boot from it.
  • Installation on one, or a few machines at most
  • Obtain a directly bootable system
Mount the image on a server machine and have clients boot it over the network.
  • Client-server model
  • Wired (1Gbit+) network connection
Mount the image in a running Linux system and install Arch from a chroot environment.
  • Replace an existing system with reduced downtime
  • Install on the local machine, or a remote one via VNC or SSH
Set up a virtual machine and install Arch as a guest system.
  • Operating system compatible with virtualization software
  • Obtain an isolated system for learning, testing or debugging

Boot the installation medium

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

When the Arch menu appears, select "Boot Arch Linux" and press Enter to enter the live environment where you will perform the actual installation.

You will be presented with a Zsh shell prompt, logged in as the root user. Zsh provides advanced tab completion and other features as part of the grml config. For editing text files, the console editor nano is suggested.

Booting into UEFI mode

Warning: While the choice to install in EFI mode is forward looking, early vendor UEFI implementations carried more bugs than their BIOS counterparts. It is advised to do a search relating your particular mainboard model before proceeding.

In case you have a UEFI motherboard with UEFI mode enabled, the CD/USB will automatically launch Arch Linux via Gummiboot and present the following menu:

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

To verify you are booted in UEFI mode, run:

# efivar -l

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

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.

Keyboard layout

Note: Changes here only affect the installation process.

The default keyboard layout is set to US, the locale to en_US.UTF-8. To change the keyboard layout, run:

# loadkeys layout

where layout is a two-letter country code. Use localectl list-keymaps to list all available keymaps.

If certain special characters appear as white squares or other symboles, you may wish to change the console font. See Fonts#Previewing_and_testing for details.

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 of the live system and 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. 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.

Tip:
  • The elinks browser is available in the live system: it can be useful for example to authenticate in RADIUS-protected networks.
  • The system you are going to install in this guide makes no pre-assumptions regarding network access. For an easy start after the first boot, it may be helpful to stick to the method that got you connected with the live medium and copy relevant configuration to the new system before you #Chroot and configure the base system later.

Static IP

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

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".

See Network_configuration#Static_IP_address for required settings. Configure a static profile for dhcpcd in /etc/dhcpcd.conf with your settings, for example:

interface enp2s0f0
static ip_address=192.168.0.10/24
static routers=192.168.0.1
static domain_name_servers=192.168.0.1 8.8.8.8

Restart dhcpcd.service:

# systemctl restart dhcpcd.service

You should now have a working network connection. If you do not, see 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.

Use netctl's wifi-menu to connect to a wireless network:

# wifi-menu

This should bring you a menu of wifi networks if your computer has only one Wi-Fi device (mostly the case in laptops).

If your computer has more than one Wi-Fi device, you need to choose one and pass its interface name to wifi-menu. First, identify the name of the needed interface:

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

This example shows wlp3s0 as the only available wireless interface, for simplicity. If you are unsure, wireless interfaces are likely to start with the letter "w", and unlikely to be "lo" or start with the letter "e".

Now try wifi-menu again by passing it the interface name:

# 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 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 UEFI 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

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

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.
  • 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). [2] If for any reason you do not wish to follow this convention, see [3] and [4] for more information and possible workarounds.

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

# parted /dev/sdx

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

(parted) mklabel msdos

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

(parted) mklabel gpt

Partition schemes

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

  • At least a partition for the / (root) directory must be created.
  • Depending on the motherboard's firmware interface, the chosen #Partition table types, and in some cases also the chosen boot loader, the following additional partitions must be created:

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

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

# parted /dev/sdx

The following command will be used to create partitions:

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

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

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

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

(parted) mkpart primary ext3 513MiB 100%

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

(parted) mkpart primary ext3 513MiB 20.5GiB
(parted) mkpart primary ext3 20.5GiB 100%

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

(parted) mkpart primary ext3 513MiB 20.5GiB
(parted) mkpart primary linux-swap 20.5GiB 24.5GiB
(parted) mkpart primary ext3 24.5GiB 100%

BIOS/MBR examples

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

(parted) mkpart primary ext3 1MiB 100%
(parted) set 1 boot on

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

(parted) mkpart primary ext3 1MiB 20GiB
(parted) set 1 boot on
(parted) mkpart primary ext3 20GiB 100%

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

(parted) mkpart primary ext3 1MiB 100MiB
(parted) set 1 boot on
(parted) mkpart primary ext3 100MiB 20GiB
(parted) mkpart primary linux-swap 20GiB 24GiB
(parted) mkpart primary ext3 24GiB 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.
  • 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

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. Should you change your mirror list at a later stage, refresh all package lists with pacman -Syyu. See Mirrors for more information.

Install the base system

The 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.

See Pacman#Troubleshooting and Pacman-key#Troubleshooting in case of errors.

Generate an fstab

UUIDs are used because they have certain advantages (see fstab#Identifying filesystems). If you prefer 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. See fstab#Field definitions for syntax information.

Chroot and configure the base system

Next, chroot into your newly installed system:

# arch-chroot /mnt /bin/bash

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.

Warning: 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 hiccup in your fresh install.

Locale

Locales define which language the system uses and other regional considerations, such as currency denomination, numerology and character sets. Possible values are listed in /etc/locale.gen, with the active locale defined in locale.conf files.

All entries in locale.gen are commented out (preceded by #) by default. Uncomment en_US.UTF-8 UTF-8, as well as other needed localisations. UTF-8 is highly recommended over other options.

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

Before locales can be enabled, they must be generated:

# locale-gen

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

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

Export the chosen locale:

# export LANG=en_US.UTF-8
Note:
  • Choosing en_US.UTF-8 as the system locale allows to keep system logs in English for easier troubleshooting. Users may override this setting for their session as described in Locale#Setting the locale.
  • LANG acts as the default value for the locale-related LC_* variables. To use other locales for these 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 #Keyboard layout, create /etc/vconsole.conf to make those changes persist in the installed system. It is important KEYMAP matches the value initially set with loadkeys, to ensure correct entry of the root password on reboot.

# nano /etc/vconsole.conf
KEYMAP=de-latin1
FONT=lat9w-16

These settings only apply to virtual consoles, not 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, and listed with the ls command. Create a symbolic link /etc/localtime to your subzone file /usr/share/zoneinfo/Zone/SubZone:

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

You may use tab completion to show available zones and subzones. Example:

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

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

If you have multiple operating systems installed in the same machine, they will all derive the current time from the same hardware clock, which must be set to either UTC or localtime. For this reason you must make sure that all the operating systems see the hardware clock as providing time in the same chosen standard, otherwise some of them will perform the time zone adjustement for the system clock, while others will not.

In particular, it is strongly recommended to set the hardware clock to UTC, in order to avoid conflicts between the installed operating systems. For example, if the hardware clock was set to localtime, more than one operating system may adjust it after a DST change, thus resulting in an overcorrection; more problems may arise when travelling between different time zones and using one of the operating systems to reset the system/hardware clock.

To set the hardware clock to UTC in Linux, run:

# hwclock --systohc --utc

The hwclock command also generates the /etc/adjtime file.

Note: Using UTC for the hardware clock does not mean that software will display time in UTC. However, the system setup/BIOS interface will instead: this should be neither surprising nor treated as a bug.
Warning: Windows systems use localtime by default. Using localtime on Arch systems may lead to several known and unfixable bugs, but there are no plans to drop support for localtime. It is, though, recommended to set Windows to use UTC instead, and prevent it from synchronising time. See Time#UTC in Windows.

Kernel modules

Needed kernel modules are automatically loaded by udev, so you will rarely need to load modules manually. See Kernel modules for details.

Hostname

Set the hostname to your liking:

# echo myhostname > /etc/hostname

Add the same hostname to /etc/hosts:

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

Configure the network

You need to configure the network again, but this time for your newly installed environment. The procedure and prerequisites are 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 configuration, visit Network configuration and Wireless network configuration.
  • If you would like to use the old interface naming scheme (i.e. ethX and wlanX) 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.

Now select a daemon to handle the configuration and operation. Several are listed below; only select one of them for the new system.

Wired

Using dhcpcd

A simple option for adapter configuration is to use the DHCP Client Daemon, the method used by default with the install medium. See Dhcpcd#Running.

Users requiring only single wired network connection can simply enable the dhcpcd service for the interface:

# systemctl enable dhcpcd@interface_name.service

If static IP settings are required, adjust the profile configuration as described in #Static IP.

Using systemd-networkd

The Arch default init system, systemd includes built-in support for managing adapters using both DHCP and static IP setups. Configuration is simple. See Systemd-networkd#Required_services_and_setup.

Using netctl

Another option is netctl which is a CLI-based tool used to configure and manage network connections via user-created profiles. Create a profile as shown in netctl#Example profiles, then enable it as described in netctl#Basic method.

Wireless

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

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.

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: This method cannot be used with explicitely enabled profiles, i.e. through netctl enable profile.

Install wpa_actiond, which is required for netctl-auto:

# pacman -S wpa_actiond

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

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

Analog modem, ISDN or PPPoE DSL

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

Create an initial ramdisk environment

Note: As mkinitcpio was run on installation of linux with pacstrap, most users can skip this step and use the defaults provided in mkinitcpio.conf.

Set the correct hooks in /etc/mkinitcpio.conf and re-generate the initramfs image:

# mkinitcpio -p linux

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. 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; to have GRUB search for other installed operating systems, install os-prober in addition:

# pacman -S grub os-prober

Install the bootloader to the drive Arch was installed to (do not append a partition number, or /dev/sdaX):

# grub-install --target=i386-pc --recheck /dev/sda

Automatically generate grub.cfg:

# grub-mkconfig -o /boot/grub/grub.cfg
Note: A sample /boot/grub/grub.cfg is included with the grub package, and subsequent grub-* commands may not over-write it. Ensure your intended changes are in grub.cfg, rather than in grub.cfg.new or similar file.

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 a UEFI boot entry (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.
  • 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 (most likely /dev/sda2 if /dev/sda1 is the ESP):

# 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

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

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.