Maintenance and recovery

Run commands as your regular user unless a section explicitly uses the live ISO or chroot.

Routine maintenance

Read Arch Linux news before upgrading, then perform a full system upgrade:

sudo pacman -Syu

If you use AUR packages, also review and apply their updates with your helper, rebuilding packages affected by library changes. Restart affected applications and services; reboot after kernel updates to use the new kernel.

Review .pacnew and .pacsave files using pacdiff from pacman-contrib. Merge changes deliberately rather than overwriting your configuration:

sudo pacdiff

pacdiff defaults to vimdiff; install vim or set DIFFPROG to an installed diff editor you know how to use.

Inspect orphaned and foreign packages periodically:

pacman -Qdt
pacman -Qm

Review each candidate before removing it. An orphan may still be useful to you. See System maintenance and package cache cleanup.

Memory and swap

Keep the default swappiness until you have measured a problem. Swappiness controls the relative cost of swapping versus filesystem paging; it is not a percentage of RAM usage at which swapping starts. See the kernel documentation.

For compressed swap in RAM, install zram-generator and create /etc/systemd/zram-generator.conf:

sudo pacman -S zram-generator
sudo nano /etc/systemd/zram-generator.conf

A simple configuration is:

[zram0]
zram-size = min(ram / 2, 4096)

Reboot to activate it, then check zramctl and swapon --show. This allocates a compressed swap device up to half of RAM, capped at 4 GiB. Zram cannot store a hibernation image; hibernation needs disk-backed swap and resume configuration. See Zram.

If the desktop becomes unresponsive under memory pressure, consider earlyoom after checking which OOM policy is already active. Do not stack multiple OOM daemons without understanding their interaction. To choose earlyoom:

sudo pacman -S earlyoom
sudo systemctl enable --now earlyoom.service

It can terminate applications with unsaved work; review its configuration for your workload.

Backups

Personal files

Back up personal files to another device or remote storage and test restoring them. Déjà Dup is one option:

sudo pacman -S deja-dup

Include documents, saves, credentials and application data you need. .config and .local/share often contain important state; do not exclude them wholesale. Caches can usually be excluded. A synchronized folder is not by itself a versioned backup.

Encryption of your system disk does not encrypt a backup stored elsewhere. Configure encryption on the backup destination when it contains private data.

System snapshots

Timeshift can create rsync snapshots of the ext4 system used here:

sudo pacman -S timeshift cronie
sudo systemctl enable --now cronie.service

Configure the destination and schedule in Timeshift. Its Btrfs mode requires a supported subvolume layout; it is not applicable to this ext4 installation. Timeshift excludes user data by default and is not a substitute for a separate personal backup.

Upstream supports GRUB2 for bootloader restoration. With systemd-boot, plan to restore the ESP and boot entries separately using the recovery procedure below. Keep a separate backup of /boot, /etc and, for encryption, the LUKS header.

Package and service lists

Record explicitly installed repository and foreign packages separately:

pacman -Qqen > packages-official.txt
pacman -Qqem > packages-foreign.txt
systemctl list-unit-files --state=enabled > enabled-services.txt
systemctl --user list-unit-files --state=enabled > enabled-user-services.txt

Save these with your backups. After reinstalling and fully upgrading, restore available official packages with:

sudo pacman -S --needed - < packages-official.txt

Review foreign packages individually and rebuild them from trusted sources. Package names and availability may have changed. Re-enable needed services deliberately, distinguishing system services from user services.

Recover from the installation ISO

Boot a current ISO in the same firmware mode, connect to the internet and identify the devices with lsblk -f. Run the following as root in the live ISO, substituting your actual partitions.

For the unencrypted example:

mount /dev/nvme0n1p2 /mnt

For the encrypted example, use these commands instead:

cryptsetup open /dev/nvme0n1p2 cryptroot
vgchange -ay main
mount /dev/mapper/main-root /mnt

On UEFI, mount the ESP before entering the system. Also mount any separate filesystems needed for the repair:

mount --mkdir /dev/nvme0n1p1 /mnt/boot
arch-chroot /mnt

On BIOS, skip the ESP mount and run arch-chroot /mnt directly. Inspect the error and configuration before changing packages. If an upgrade was interrupted, complete a full pacman -Syu; reinstall affected packages if necessary.

After fixing initramfs configuration or reinstalling kernel packages:

mkinitcpio -P

For GRUB, regenerate /boot/grub/grub.cfg with grub-mkconfig -o /boot/grub/grub.cfg. If its boot files need reinstalling, repeat the appropriate GRUB installation command first.

For systemd-boot, use bootctl --esp-path=/boot install if the boot manager needs reinstalling. Restore or recreate the loader entries and verify their kernel paths and root/encryption options. If the ESP was lost, reinstall the kernel packages while it is mounted, then regenerate the initramfs; reinstalling the boot manager alone does not restore kernels.

Exit the chroot, unmount with umount -R /mnt, then reboot. Keep the ISO available until you have verified a successful boot.