SYSTEMD.SPECIAL(7)

📦
version 262.1 · 16 versions
, in arch rolling

✨ systemd.special ✨ — special systemd units, the VIP list 💅

okay bestie, so this is the manual page for systemd.special — the little celebrity roster of units that systemd treats extra special. Some of them have deep internal meaning and you literally cannot rename them (systemd will cry), and others are just vibes: standard meanings that every system should have on deck. 🌸

🌈 SYNOPSIS 🌈

like, the whole guest list, no cap:

basic.target, bluetooth.target, cryptsetup-pre.target,
cryptsetup.target, veritysetup-pre.target, veritysetup.target,
ctrl-alt-del.target, blockdev@.target, boot-complete.target,
default.target, emergency.target, exit.target, factory-reset.target,
factory-reset-now.target, final.target, first-boot-complete.target,
getty.target, getty-pre.target, graphical.target, halt.target,
hibernate.target, hybrid-sleep.target, suspend-then-hibernate.target,
initrd.target, initrd-fs.target, initrd-root-device.target,
initrd-root-fs.target, initrd-usr-fs.target, integritysetup-pre.target,
integritysetup.target, kbrequest.target, kexec.target,
local-fs-pre.target, local-fs.target, machines.target,
multi-user.target, network-online.target, network-pre.target,
network.target, nss-lookup.target, nss-user-lookup.target, paths.target,
poweroff.target, printer.target, reboot.target,
remote-cryptsetup.target, remote-integritysetup.target,
remote-veritysetup.target, remote-fs-pre.target, remote-fs.target,
rescue.target, rpcbind.target, shutdown.target, sigpwr.target,
sleep.target, slices.target, smartcard.target, sockets.target,
soft-reboot.target, sound.target, ssh-access.target,
storage-target-mode.target, suspend.target, swap.target, sysinit.target,
system-update.target, system-update-pre.target, time-set.target,
time-sync.target, timers.target, tpm2.target, umount.target,
usb-gadget.target, -.slice, capsule.slice, machine.slice, system.slice,
user.slice, -.mount, dbus.service, dbus.socket, display-manager.service,
init.scope, syslog.socket, system-update-cleanup.service

💖 DESCRIPTION 💖

so here's the tea 🫖 — a handful of units get treated specially by systemd. Some of them have real internal semantics baked in and absolutely cannot be renamed (sorry, they're famous, they can't just rebrand). The rest are just units with a standard, agreed-upon meaning that should honestly be present on all systems, like, every single one. 👑

👑 UNITS MANAGED BY THE SYSTEM SERVICE MANAGER 👑

🎀 Special System Units 🎀

-.mount

the root mount point, bestie — aka the mount unit for the / path. this one is unconditionally active the entire time the system is up, because that mount point is literally where the basic userspace is running from. it's the main character's house. 🏠

added in version 235.

basic.target

a special target unit covering basic boot-up. iconic behavior. ✨

systemd automatically adds a dependency of the type After= for this target unit to all services (except for those with DefaultDependencies=no — they opted out, respect it).

usually, this should pull-in all local mount points plus /var/, /tmp/ and /var/tmp/, swap devices, sockets, timers, path units and other basic initialization necessary for general purpose daemons. the mentioned mount points are special cased to allow them to be remote, iykyk.

this target usually does not pull in any non-target units directly, but rather does so indirectly via other early boot targets. it's instead meant as a synchronization point for late boot services. refer to bootup(7) for details on the targets involved. 📚

boot-complete.target

this target is intended as a generic synchronization point for services that shall determine or act on whether the boot process completed successfully. it's giving "did we make it? did we slay?"

order units that are required to succeed for a boot process to be considered successful before this unit, and add a Requires= dependency from the target unit to them. then order units that shall only run when the boot process is considered successful after the target unit, and pull in the target from it, also with Requires=. note that by default this target unit is not part of the initial boot transaction — it's supposed to be pulled in only if required by units that want to run only on successful boots. so it's like, invite-only. 🙈

see systemd-boot-check-no-failures.service(8) for a service that implements a generic system health check and orders itself before boot-complete.target.

see systemd-bless-boot.service(8) for a service that propagates boot success information to the boot loader, and orders itself after boot-complete.target.

added in version 240.

ctrl-alt-del.target

systemd starts this target whenever Control+Alt+Del is pressed on the console. the keyboard smash of destiny. ⌨️

usually, this should be aliased (symlinked) to reboot.target.

cryptsetup.target

a target that pulls in setup services for all encrypted block devices. 🔐 locked and loaded.

veritysetup.target

a target that pulls in setup services for all verity integrity protected block devices. keeping it honest fr fr. 🛡️

added in version 248.

dbus.service

a special unit for the D-Bus bus daemon. as soon as this service is fully started up, systemd will connect to it and register its service. it's the group chat everyone joins. 💬

dbus.socket

a special unit for the D-Bus system bus socket. all units with Type=dbus automatically gain a dependency on this unit. automatic, no effort, slay. ✨

default.target

the default unit systemd starts at bootup. usually, this should be aliased (symlinked) to multi-user.target or graphical.target. see bootup(7) for more discussion.

the default unit systemd starts at bootup can be overridden with the systemd.unit= kernel command line option, or more conveniently, with the short names like single, rescue, 1, 3, 5, ...; see systemd(1).

if default.target is not present on disk and no unit was requested on the kernel command line, the default set at compile time will be used (graphical.target).

if the default unit or one of the fallbacks cannot be started, systemd starts rescue.target. the backup plan always shows up. 🫶

for typical unit files please set "WantedBy=" to a regular target (like multi-user.target or graphical.target), instead of default.target, since such a service will also be run on special boots like on system update, emergency boot... important!!

display-manager.service

the display manager service. usually, this should be aliased (symlinked) to gdm.service or a similar display manager service. your login screen's whole identity. 💅

emergency.target

a special target unit that starts an emergency shell on the main console. this target does not pull in other services or mounts. it's the most minimal version of starting the system in order to acquire an interactive shell — the only processes running are usually just the system manager (PID 1) and the shell process. bare minimum, no extras.

this unit may be used by specifying emergency on the kernel command line; it is also used when a file system check on a required file system fails and boot-up cannot continue. compare with rescue.target, which serves a similar purpose, but also starts the most basic services and mounts all file systems. so rescue.target is the slightly-more-dressed-up sibling.

in many ways booting into emergency.target is similar to the effect of booting with "init=/bin/sh" on the kernel command line, except that emergency mode provides you with the full system and service manager, and allows starting individual units in order to continue the boot process in steps. glow-up fr. ✨

note that depending on how emergency.target is reached, the root file system might be mounted read-only or read-write (no remounting is done specially for this target). for example, the system may boot with root mounted read-only when ro is used on the kernel command line and remain this way for emergency.target, or the system may transition to emergency.target after the system has been partially booted and disks have already been remounted read-write. so like, don't assume. 🙈

exit.target

a special service unit for shutting down the system or user service manager. it is equivalent to poweroff.target on non-container systems, and also works in containers. versatile queen. 👑

systemd will start this unit when it receives the SIGTERM or SIGINT signal when running as user service daemon.

normally, this (indirectly) pulls in shutdown.target, which in turn should be conflicted by all units that want to be scheduled for shutdown when the service manager starts to exit.

added in version 186.

factory-reset.target

a special target to request a factory reset operation. this will typically persistently store a request flag for the next boot and then reboot in order to reset the system to factory state. it's giving "delete the whole account and start over." 🧼

see Factory Reset[1] for more information.

added in version 250.

factory-reset-now.target

a special target that is started on boots that shall execute a factory reset. it may be used to pull in additional services that shall be invoked during a factory reset operation. it also acts as an ordering barrier: once the target is reached, the factory reset state is marked as "completed". 🎀

see Factory Reset[1] for more information.

added in version 258.

final.target

a special target unit that is used during the shutdown logic and may be used to pull in late services after all normal services are already terminated and all mounts unmounted. the after-party, basically. 🌙

getty.target

a special target unit that pulls in statically configured local TTY getty instances. all the terminals, lined up. 🖥️

graphical.target

a special target unit for setting up a graphical login screen. this pulls in multi-user.target.

units that are needed for graphical logins shall add Wants= dependencies for their unit to this unit (or multi-user.target) during installation. this is best configured via WantedBy=graphical.target in the unit's [Install] section. do it the right way bestie. ✨

hibernate.target

a special target unit for hibernating the system. this pulls in sleep.target. 😴

hybrid-sleep.target

a special target unit for hibernating and suspending the system at the same time. this pulls in sleep.target. why choose one when you can have both. 💜

added in version 196.

suspend-then-hibernate.target

a special target unit for suspending the system for a period of time, waking it and putting it into hibernate. this pulls in sleep.target. a little nap first, then the big sleep. 🌸

added in version 239.

halt.target

a special target unit for shutting down and halting the system. note that this target is distinct from poweroff.target in that it generally really just halts the system rather than powering it down. so like, stopping vs. fully leaving. 🛑

applications wanting to halt the system should not start this unit directly, but should instead execute systemctl halt (possibly with the --no-block option) or call systemd(1)'s org.freedesktop.systemd1.Manager.Halt D-Bus method directly. don't do it the wrong way!!

imports.target

a target unit that pulls in all disk image download jobs to execute on system boot. this is used by systemd-import-generator(8). downloading on boot, no cap. 📥

added in version 258.

init.scope

this scope unit is where the system and service manager (PID 1) itself resides. it is active as long as the system is running. it's the origin story. 🌟

added in version 235.

🎀 initrd.target

This is the default target in the initrd, basically the twin of default.target in the main system. Its whole job is to mount the real root and transition over to it. See bootup(7) for more discussion. 💖

Added in version 245.

🦋 initrd-fs.target

systemd-fstab-generator(8) automatically adds dependencies of type Before= to sysroot-usr.mount and all mount points found in /etc/fstab that have the x-initrd.mount mount option set and do not have the noauto mount option set. It's also indirectly ordered after sysroot.mount. So the moment this target is reached, the /sysroot/ hierarchy is fully set up — prepped and ready for the transition to the host OS. ✨

Added in version 199.

🌸 initrd-root-device.target

A special initrd target unit that gets reached when the root filesystem device is available, but before it has been mounted. systemd-fstab-generator(8) and systemd-gpt-auto-generator(8) automatically set up the appropriate dependencies to make this happen. 🫶

Added in version 230.

💅 initrd-root-fs.target

systemd-fstab-generator(8) automatically adds dependencies of type Before= to the sysroot.mount unit, which is generated from the kernel command line's root= setting (or equivalent). 🎀

Added in version 199.

⭐ initrd-usr-fs.target

systemd-fstab-generator(8) automatically adds dependencies of type Before= to the sysusr-usr.mount unit, which is generated from the kernel command line's usr= switch. Services can order themselves after this target unit so they run once the /sysusr/ hierarchy becomes available — this is for systems that come up initially without a root file system, but with an initialized /usr/ that they need to access before setting up the root file system they'll ultimately switch to. On systems where usr= is not used, this target is ordered after sysroot.mount and thus is mostly equivalent to initrd-root-fs.target. In effect, on any system, once this target is reached the file system backing /usr/ is mounted — though possibly at two different locations, either below the /sysusr/ or the /sysroot/ hierarchies. no cap. 💜

Added in version 249.

🌈 kbrequest.target

systemd starts this target whenever Alt+ArrowUp is pressed on the console. ⚠️ Heads up bestie: any user with physical access to the machine will be able to do this, with no authentication, so this should be used carefully.

fr fr, think before you enable this one. 💀

🔥 kexec.target

A special target unit for shutting down and rebooting the system via kexec.

Applications that want to reboot the system should not start this unit directly — instead they should execute systemctl kexec (possibly with the --no-block option) or call systemd-logind(8)'s org.freedesktop.login1.Manager.RebootWithFlags() D-Bus method directly. ✨

See systemd-kexec.service(8) for further details of the operation this target pulls in.

🧚 local-fs.target

systemd-fstab-generator(8) automatically adds dependencies of type Before= to all mount units that refer to local mount points for this target unit. On top of that, it adds dependencies of type Wants= to this target unit for those mounts listed in /etc/fstab that have the auto mount option set. 🎀

👑 machines.target

A standard target unit for starting all the containers and other virtual machines. See systemd-nspawn@.service for an example. 🦋

Added in version 233.

💖 multi-user.target

A special target unit for setting up a multi-user system (non-graphical). This is pulled in by graphical.target.

Units that are needed for a multi-user system shall add Wants= dependencies for their unit to this unit during installation. This is best configured via WantedBy=multi-user.target in the unit's [Install] section. slay. ✨

🌐 network-online.target

Units that strictly require a configured network connection should pull in network-online.target (via a Wants= type dependency) and order themselves after it. This target unit is intended to pull in a service that delays further execution until the network is sufficiently set up. What precisely this requires is left to the implementation of the network managing service. 🤷♀️

Note the distinction between this unit and network.target, bestie — it matters!

This unit is an active unit (i.e. pulled in by the consumer rather than the provider of this functionality) and it pulls in a service which possibly adds substantial delays to further execution. In contrast, network.target is a passive unit (i.e. pulled in by the provider of the functionality, rather than the consumer) that usually does not delay execution much. Usually, network.target is part of the boot of most systems, while network-online.target is not — except when at least one unit requires it. Also see Running Services After the Network Is Up[2] for more information. 📖

All mount units for remote network file systems automatically pull in this unit, and order themselves after it. Note that networking daemons that simply provide functionality to other hosts (as opposed to consume functionality of other hosts) generally do not need to pull this in. 🙈

Note that this unit is only useful during the original system start-up logic. After the system has completed booting up, it will not track the online state of the system anymore. Because of this it cannot be used as a network connection monitor concept — it is purely a one-time system start-up concept. periodt. 💅

Added in version 200.

🍓 paths.target

A special target unit that sets up all path units (see systemd.path(5) for details) that shall be active after boot.

It is recommended that path units installed by applications get pulled in via Wants= dependencies from this unit. This is best configured via a WantedBy=paths.target in the path unit's [Install] section. 🌸

Added in version 199.

🔌 poweroff.target

A special target unit for shutting down and powering off the system.

Applications wanting to power off the system should not start this unit directly — they should instead execute systemctl poweroff (possibly with the --no-block option) or call systemd-logind(8)'s org.freedesktop.login1.Manager.PowerOff D-Bus method directly. ✨

🔄 reboot.target

A special target unit for shutting down and rebooting the system.

Applications wanting to reboot the system should not start this unit directly — they should instead execute systemctl reboot (possibly with the --no-block option) or call systemd-logind(8)'s org.freedesktop.login1.Manager.Reboot() D-Bus method directly. 💫

See systemd-reboot.service(8) for further details of the operation this target pulls in.

🔐 remote-cryptsetup.target

Similar to cryptsetup.target, but for encrypted devices which are accessed over the network. It is used for crypttab(5) entries marked with _netdev. 🦋

Added in version 235.

🛡️ remote-integritysetup.target

Similar to integritysetup.target, but for integrity protected devices which are accessed over the network. It is used for integritytab(5) entries marked with _netdev. ⭐

Added in version 258.

✅ remote-veritysetup.target

Similar to veritysetup.target, but for verity protected devices which are accessed over the network. It is used for veritytab(5) entries marked with _netdev. 💜

Added in version 248.

🌍 remote-fs.target

Similar to local-fs.target, but for remote mount points. (you already know the vibe from the local one 💅)

🚑 rescue.target

A special target unit that pulls in the base system (including system mounts) and spawns a rescue shell. Isolate to this target in order to administer the system in single-user mode — with all file systems mounted but with no services running, except for the most basic. 🫶

Compare with emergency.target, which is much more reduced and does not provide the file systems or most basic services. Compare with multi-user.target — this target could be seen as single-user.target.

Use the "systemd.unit=rescue.target" kernel command line option to boot into this mode. A short alias for this kernel command line option is "1", for compatibility with SysV. 🎀

🛑 shutdown.target

A special target unit that terminates the services on system shutdown.

Services that shall be terminated on system shutdown shall add Conflicts= and Before= dependencies to this unit for their service unit, which is implicitly done when DefaultDependencies=yes is set (the default). ✨

⚡ sigpwr.target

A special target that is started when systemd receives the SIGPWR process signal, which is normally sent by the kernel or UPS daemons when power fails. 🔋

😴 sleep.target

A special target unit that is pulled in by suspend.target, hibernate.target, suspend-then-hibernate.target, and hybrid-sleep.target and may be used to hook units into the sleep state logic. 💤

In order to hook external programs before the actual system sleep operation, place their command line in a service unit file's ExecStart= line (use Type=oneshot), and ensure the unit is pulled in by sleep.target and ordered before it. In order to hook program code after the actual system sleep operation (i.e. to be run after the system woke up again), place the command in ExecStop= instead, and make sure to enable StopWhenUnneeded= and RemainAfterExit=. Both approaches can be combined into one unit file in order to run programs both before and after the sleep operation. 🌙

Example 1. Combined Example

[Unit]
DefaultDependencies=no
StopWhenUnneeded=yes
Before=sleep.target

[Service]
Type=oneshot
RemainAfterExit=yes
ExecStart=/usr/bin/some-before-command
ExecStop=/usr/bin/some-after-command

[Install]
WantedBy=sleep.target

🍰 slices.target

A special target unit that sets up all slice units (see systemd.slice(5) for details) that shall always be active after boot. By default, the generic system.slice slice unit as well as the root slice unit -.slice are pulled in and ordered before this unit (see below). 💖

ok so literally one last thing about slices.target bestie, then we're moving on, promise 💅

Adding slice units to slices.target is generally not necessary. Like, when some unit that uses Slice= gets started, the slice you named gets started automatically — the system already knows. Adding WantedBy=slices.target lines to the [Install] section should only be done for units that need to be always active. And if you do that? Care needs to be taken, fr, to avoid creating a loop through the automatic dependencies on "parent" slices. 🌀

Added in version 229.

  • sockets.target — a special target unit that sets up all socket units (peep systemd.socket(5) for the deets) that shall be active after boot. ✨

Services that can be socket-activated shall add Wants= dependencies to this unit for their socket unit during installation. The bestie-approved way to configure this is a WantedBy=sockets.target in the socket unit's [Install] section. 🫶

  • soft-reboot.target — a special target unit for shutting down and rebooting the userspace of the system, while leaving the kernel running. The kernel is fine. The kernel is thriving. 🌈

Apps that wanna reboot the system should not start this unit directly, no cap. Instead they should execute systemctl soft-reboot (possibly with the --no-block option) or call systemd-logind(8)'s org.freedesktop.login1.Manager.RebootWithFlags() D-Bus method directly. 👑

See systemd-soft-reboot.service(8) for further details of the operation this target pulls in.

Added in version 254.

  • storage-target-mode.target — a special target unit that you can boot into which selects the "Storage Target Mode" for the OS. In this mode all local storage disks are exposed to external systems as block devices. It invokes systemd-storagetm.service(8), which exposes all local disks as NVMe-TCP devices so they can be reached over the network. It might as well invoke other services too, ones that make local disks available via other mechanisms. 💾✨

Added in version 255.

  • suspend.target — a special target unit for suspending the system. It pulls in sleep.target. 😴
  • swap.target — similar to local-fs.target, but for swap partitions and swap files. Same energy, different filesystem-adjacent situation. 🔄
  • sysinit.target — systemd automatically adds dependencies of the types Requires= and After= for this target unit to all services, except for those with DefaultDependencies=no. So yeah, everybody's connected to it unless they opted out. 🕸️

This target pulls in the services required for system initialization. System services pulled in by this target should declare DefaultDependencies=no and specify all their dependencies manually, including access to anything more than a read only root filesystem. For details on the dependencies of this target, refer to bootup(7). 📖

  • syslog.socket — the socket unit syslog implementations should listen on. All userspace log messages will be made available on this socket. For more information about syslog integration, please consult the Syslog Interface[3] document. 📨
  • system-update.target, system-update-pre.target, system-update-cleanup.service — a special target unit that is used for offline system updates. systemd-system-update-generator(8) will redirect the boot process to this target if /system-update or /etc/system-update exists. For more information see systemd.offline-updates(7). 🔧

Updates should happen before system-update.target is reached, and the services which implement them should cause the machine to reboot. The main units executing the update should order themselves after system-update-pre.target but not pull it in. Services which want to run during system updates only, but before the actual system update is executed, should order themselves before this unit and pull it in. As a safety measure, if this does not happen, and /system-update or /etc/system-update still exists after system-update.target is reached, system-update-cleanup.service will remove the symlinks and reboot the machine. 💀

not the cleanup service doing a lil emergency reboot on you, but okay, that's the safety net, bestie.

Added in version 186.

  • timers.target — a special target unit that sets up all timer units (see systemd.timer(5) for details) that shall be active after boot. ⏰

It is recommended that timer units installed by applications get pulled in via Wants= dependencies from this unit. This is best configured via WantedBy=timers.target in the timer unit's [Install] section.

Added in version 199.

  • umount.target — a special target unit that unmounts all mount and automount points on system shutdown. Bye bye, mounts. 👋

Mounts that shall be unmounted on system shutdown shall add Conflicts dependencies to this unit for their mount unit, which is implicitly done when DefaultDependencies=yes is set (the default).

✨ Special System Units for Devices ✨

Some target units are automatically pulled in as devices of certain kinds show up in the system. These may be used to automatically activate various services based on the specific type of the available hardware. It's giving plug and play but make it systemd. 🔌

  • bluetooth.target — this target is started automatically as soon as a Bluetooth controller is plugged in or becomes available at boot. 🦷💙

This may be used to pull in Bluetooth management daemons dynamically when Bluetooth hardware is found.

  • printer.target — this target is started automatically as soon as a printer is plugged in or becomes available at boot. 🖨️

This may be used to pull in printer management daemons dynamically when printer hardware is found.

  • smartcard.target — this target is started automatically as soon as a smartcard controller is plugged in or becomes available at boot. 💳

This may be used to pull in smartcard management daemons dynamically when smartcard hardware is found.

  • sound.target — this target is started automatically as soon as a sound card is plugged in or becomes available at boot. 🔊

This may be used to pull in audio management daemons dynamically when audio hardware is found.

  • usb-gadget.target — this target is started automatically as soon as a USB Device Controller becomes available at boot. 🔌

This may be used to pull in usb gadget dynamically when UDC hardware is found.

Added in version 242.

  • tpm2.target — this target is started automatically if a TPM2 device is discovered, either by the OS or by the firmware. It acts as a synchronization point for services that require TPM2 device access. The target unit is enqueued by systemd-tpm2-generator(8) if it detects that the firmware has discovered a TPM2 device but the OS kernel has not activated a driver for it yet. It is also pulled in whenever systemd-udevd.service(8) discovers a TPM2 device. The target unit is ordered after the /dev/tpmrm0 device node, so that it only becomes active once the TPM2 device is actually accessible. Early boot programs that intend to access the TPM2 device should hence order themselves after this target unit, but not pull it in. 🔐

Added in version 256.

🧚 Special Passive System Units 🧚

A number of special system targets are defined that can be used to properly order boot-up of optional services. These targets are generally not part of the initial boot transaction, unless they are explicitly pulled in by one of the implementing services. Note specifically that these passive target units are generally not pulled in by the consumer of a service, but by the provider of the service. So like, here's the vibe:

  • a consuming service should order itself after these targets (as appropriate), but not pull it in.
  • a providing service should order itself before these targets (as appropriate) and pull it in (via a Wants= type dependency).
🚨 Important bestie alert: these passive units cannot be started manually, i.e. systemctl start time-sync.target will fail with an error. They can only be pulled in by dependency. This is enforced since they exist for ordering purposes only and thus are not useful as only unit within a transaction. Periodt. 💅
  • blockdev@.target — this template unit is used to order mount units and other consumers of block devices after services that synthesize these block devices. In particular, this is intended to be used with storage services (such as systemd-cryptsetup@.service(8)/systemd-veritysetup@.service(8)) that allocate and manage a virtual block device. Storage services are ordered before an instance of blockdev@.target, and the consumer units after it. The ordering is particularly relevant during shutdown, as it ensures that the mount is deactivated first and the service backing the mount later. The blockdev@.target instance should be pulled in via a Wants= dependency of the storage daemon and thus generally not be part of any transaction unless a storage daemon is used. The instance name for instances of this template unit must be a properly escaped block device node path, e.g. blockdev@dev-mapper-foobar.target for the storage device /dev/mapper/foobar. 🗂️

Added in version 245.

  • cryptsetup-pre.target — this passive target unit may be pulled in by services that want to run before any encrypted block device is set up. All encrypted block devices are set up after this target has been reached. Since the shutdown order is implicitly the reverse start-up order between units, this target is particularly useful to ensure that a service is shut down only after all encrypted block devices are fully stopped. 🔒

Added in version 215.

  • veritysetup-pre.target — this passive target unit may be pulled in by services that want to run before any verity integrity protected block device is set up. All verity integrity protected block devices are set up after this target has been reached. Since the shutdown order is implicitly the reverse start-up order between units, this target is particularly useful to ensure that a service is shut down only after all verity integrity protected block devices are fully stopped. 🛡️

Added in version 248.

  • first-boot-complete.target — this passive target is intended as a synchronization point for units that need to run once during the first boot. Only after all units ordered before this target have finished will the machine-id(5) be committed to disk, marking the first boot as completed. If the boot is aborted at any time before that, the next boot will re-run any units with ConditionFirstBoot=yes. 🎀

Added in version 247.

  • getty-pre.target — a special passive target unit. Users of this target are expected to pull it in the boot transaction via a dependency (e.g. Wants=). Order your unit before this unit if you want to make use of the console just before getty is started. 🖥️

Added in version 235.

  • imports-pre.target — a passive unit that's ordered before literally every disk image download job that runs at system boot. systemd-import-generator(8) is the one using it, bestie. ✨

Added in version 258.

  • local-fs-pre.target — this target gets automatically ordered before all your local mount points that are marked with auto (scroll up, it's explained above 👆). Wanna run some units before all the local mounts happen? This is the moment. 💅
  • network.target — okay so this one is supposed to mean "network functionality is available" but ngl the definition is only very weakly defined, like what does that even mean. That said, at minimum these two things should hold:
    • At start-up, any configured synthetic network devices (so not the physical ones that need hardware to show up and get probed, but the virtual ones like bridge devices and similar that get created programmatically) that don't depend on any underlying hardware should be allocated by the time this target is reached. These interfaces do not have to have finished their IP level configuration yet by the time network.target is reached. no cap.
    • At shutdown, a unit ordered after network.target gets stopped before the network — to whatever level it's been set up by then — gets shut down. So when you're writing service files that need network access on shutdown, order them after this target, but don't pull it in. Also see Running Services After the Network Is Up[2] for more. 🫶

And I gotta emphasize this fr fr: at start-up there is no guarantee that hardware-based devices have even shown up by the time this target is reached, or that they've got complete IP configuration. For that vibe, use network-online.target as described above. 👑

  • network-pre.target — a passive target unit that services can pull in if they wanna run before any network is set up, like for setting up a firewall, for example. All network management software orders itself after this target, but does not pull it in. Also see Running Services After the Network Is Up[2] for more. 💖

Added in version 214.

  • nss-lookup.target — use this as the synchronization point for all host/network name service lookups. Note that this is independent of UNIX user/group name lookups — for those, use nss-user-lookup.target. All services where having full host/network name resolution available is essential should be ordered after this target, but should not pull it in. 🌸
  • nss-user-lookup.target — the synchronization point for all regular UNIX user/group name service lookups. Note that this is independent of host/network name lookups — those get nss-lookup.target. All services where having the full user/group database available is essential should be ordered after this target, but should not pull it in. And all services that provide parts of the user/group database should be ordered before this target, and pull it in. ⚠️ Note that this unit only matters for regular users and groups — system users and groups have to be resolvable during earliest boot already, so they don't need any special ordering against this target. ✨
  • remote-fs-pre.target — this target unit is automatically ordered before all mount point units (see above) and cryptsetup/veritysetup devices marked with _netdev. You can use it to run certain units before remote encrypted devices and mounts get established. Heads up tho: this unit is generally not part of the initial transaction, unless the unit that wants to be ordered before all remote mounts pulls it in via a Wants= type dependency. And if the unit wants to get pulled in by the first remote mount that shows up, it should use network-online.target (see above) instead. 💫
  • rpcbind.target — the portmapper/rpcbind pulls in this target and orders itself before it, just to announce its availability. iconic behaviour honestly.
  • ssh-access.target — service and socket units that provide remote SSH secure shell access to the local system should pull in this unit and order themselves before it. It's meant to act as a milestone indicating if and when SSH access into the system is available. It should only become active when an SSH port is bound for remote clients (so if SSH is being used as a local privilege escalation mechanism, it should not involve this target unit) — and that's regardless of the protocol choices, i.e. regardless of whether IPv4, IPv6 or AF_VSOCK is used. 💅

Added in version 256.

  • time-set.target — services responsible for setting the system clock (CLOCK_REALTIME) from a local source (like a maintained timestamp file or an imprecise real-time clock) should pull in this target and order themselves before it. And services where you just want approximate, roughly monotonic time should be ordered after this unit, but not pull it in. ⏰

This target does not give you the accuracy guarantees of time-sync.target (see below), but it also doesn't depend on remote clock sources being reachable — meaning it typically doesn't get delayed by network problems and similar drama. Use this target for services where approximate clock accuracy and rough monotonicity is what you want, but activation shall not be delayed waiting on possibly unreliable network communication. 🦋

The service manager automatically adds dependencies of type After= for this target unit to all timer units with at least one OnCalendar= directive. ✨

systemd-timesyncd.service(8) is a simple daemon that pulls in this target and orders itself before it. Besides implementing the SNTP network protocol, it maintains a timestamp file on disk whose modification time gets regularly updated. At service start-up the local system clock is set from that modification time, making sure it increases roughly monotonically. 🕰️

⚠️ Note that ordering a unit after time-set.target only has an effect if there's actually a service ordered before it that delays it until the clock is adjusted for rough monotonicity. Otherwise this target might get reached before the clock is adjusted to be roughly monotonic. So enable systemd-timesyncd.service(8), or an alternative NTP implementation, to delay the target. 💖

Added in version 242.

  • time-sync.target — services that indicate completed synchronization of the system clock (CLOCK_REALTIME) to a remote source should pull in this target and order themselves before it. And services where accurate time is essential should be ordered after this unit, but not pull it in. ⏰✨

The service manager automatically adds dependencies of type After= for this target unit to all timer units with at least one OnCalendar= directive. 💅

This target gives you stricter clock accuracy guarantees than time-set.target (see above), but it likely requires network communication and thus introduces unpredictable delays. Services that need clock accuracy and can tolerate network communication delays should use this target. Services that just need a less accurate clock — approximate and roughly monotonic behaviour — should use time-set.target instead. 🌈

⚠️ Note that ordering a unit after time-sync.target only has an effect if there's actually a service ordered before it that delays it until clock synchronization is reached. Otherwise this target might get reached before the clock is synchronized to any remote accurate reference clock. When using systemd-timesyncd.service(8), enable systemd-time-wait-sync.service(8) to delay the target; or use an equivalent service for other NTP implementations. 😭

Table 1. Comparison 💖

  • time-set.target:
    • "quick" to reach
    • typically uses local clock sources, boot process not affected by availability of external resources
    • reliable, because local
    • typically guarantees an approximate and roughly monotonic clock only
    • implemented by systemd-timesyncd.service
  • time-sync.target:
    • "slow" to reach
    • typically uses remote clock sources, inserts dependencies on remote resources into boot process
    • unreliable, because typically network involved
    • typically guarantees an accurate clock
    • implemented by systemd-time-wait-sync.service

🍰 Special Slice Units 🍰

There are four ".slice" units that form the basis of the hierarchy for assigning resources to services, users, and virtual machines or containers. Wanna know all the slice unit deets? That's systemd.slice(5), bestie. 📚

  • -.slice — the root slice, aka the root of the whole slice hierarchy. It usually doesn't contain units directly, but you can use it to set defaults for the entire tree. 🌳

Added in version 206.

  • machine.slice — bestie, by default every virtual machine and container registered with systemd-machined lives here. This one gets pulled in by systemd-machined.service.

Added in version 206.

  • capsule.slice — by default, all capsules encapsulated in capsule@.service hang out in this slice.

Added in version 255.

  • system.slice — by default, every system service that systemd starts is found right here.

Added in version 206.

  • user.slice — by default, all user processes and services started on behalf of the user — including the per-user systemd instance — live in this slice. It's pulled in by systemd-logind.service.

Added in version 206.

👑 UNITS MANAGED BY THE USER SERVICE MANAGER 👑

✨ Special User Units ✨

So when systemd is running as a user instance, these special units are the ones available to you:

  • default.target — this is THE main target of the user service manager, started by default the moment the service manager gets invoked. All the various services that make up a normal user session should be pulled into this target. In that sense, default.target is giving very much multi-user.target in the system instance energy — but plot twist: it's a real unit, not an alias. 💅

Added in version 242.

  • capsule@.target — the main target of capsule service managers, started by default, instantiated with the capsule name. This is how you can define different sets of units that get started for different capsules via generic unit definitions. For the tea on capsules, see capsule@.service(5).

Added in version 255.

On top of those, these units are also available, and they have definitions similar to their system counterparts: exit.target, shutdown.target, sockets.target, timers.target, paths.target, bluetooth.target, printer.target, smartcard.target, sound.target.

💫 Special Passive User Units 💫

  • graphical-session.target — this target is active whenever any graphical session is running. Its whole job is to stop user services that only apply to a graphical (X, Wayland, etc.) session when that session is terminated. Those services should have "PartOf=graphical-session.target" in their [Unit] section. And a target for a particular session (e.g. gnome-session.target) starts and stops "graphical-session.target" with "BindsTo=graphical-session.target". 🦋

Which services actually get started by a session target is decided by the "Wants=" and "Requires=" dependencies. For services that can be enabled independently, you should use symlinks in ".wants/" and ".requires/" — see systemd.unit(5). Those symlinks should either ship inside packages, or get added dynamically after installation, for example using "systemctl add-wants", see systemctl(1).

Example 2. Nautilus as part of a GNOME session

"gnome-session.target" pulls in Nautilus as a top-level service:

[Unit]
Description=User systemd services for GNOME graphical session
Wants=nautilus.service
BindsTo=graphical-session.target

And "nautilus.service" gets stopped when the session stops:

[Unit]
Description=Render the desktop icons with Nautilus
PartOf=graphical-session.target

[Service]
...

Added in version 234.

  • graphical-session-pre.target — this target holds services that set up the environment or the global configuration of a graphical session. Think SSH/GPG agents (they need to export an environment variable into all desktop processes) or migrating obsolete d-conf keys after an OS upgrade (that has to happen before any process that might use them even starts). This target must be started before you start a graphical session like gnome-session.target. ✨

Added in version 234.

  • xdg-desktop-autostart.target — the XDG specification defines a way to autostart applications using XDG desktop files, and systemd ships systemd-xdg-autostart-generator(8) for the XDG desktop files in the autostart directories. Desktop Environments can opt in to use this service by adding a Wants= dependency on xdg-desktop-autostart.target.

Added in version 246.

🎀 Special User Slice Units 🎀

There are four ".slice" units that form the basis of the user hierarchy for assigning resources to user applications and services. See systemd.slice(5) for the details on slice units, and the documentation about Desktop Environments[4] for even more info.

  • -.slice — the root slice, aka the root of the user's slice hierarchy. It usually doesn't contain units directly, but it can be used to set defaults for the whole tree. 🌳

Added in version 247.

  • app.slice — by default, all user services and applications managed by systemd are found in this slice. Every interactively launched application like web browsers and text editors, plus all non-critical services, should be placed into this slice.

Added in version 247.

  • session.slice — all essential services and applications required for the session should use this slice. These are the services that either can't be restarted easily, or where latency issues might mess with the interactivity of the system and applications. This includes the display server, screen readers, and other services such as DBus or XDG portals. Such services should be configured to be part of this slice by adding Slice=session.slice to their unit files. 💜

Added in version 247.

  • background.slice — all services running low-priority background tasks should use this slice. That way, resources can be preferentially assigned to the other slices. Examples include non-interactive tasks like file indexing or backup operations, where latency is simply not important.

Added in version 247.

🌈 SEE ALSO 🌈

systemd(1), systemd.unit(5), systemd.service(5), systemd.socket(5), systemd.target(5), systemd.slice(5), bootup(7), systemd-fstab-generator(8), user@.service(5)

📝 NOTES 📝

  1. Factory Reset https://systemd.io/FACTORY_RESET
  2. Running Services After the Network Is Up https://systemd.io/NETWORK_ONLINE
  3. Syslog Interface https://systemd.io/SYSLOG
  4. Desktop Environments https://systemd.io/DESKTOP_ENVIRONMENTS

and that's the whole page, bestie!! 🫶 you read every last unit, every slice, every "Added in version" like the icon you are 👑💖

This one came from systemd 262.1, and the page itself is SYSTEMD.SPECIAL(7) — go forth and slay your unit files, no cap. 💅✨🎀

📦 where this page comes from
  • arch rolling: systemd 262.1-1 · LGPL-2.1-or-later AND CC0-1.0 AND GPL-2.0-or-later AND MIT-0
  • 💖 girlified by deepseek-ai/DeepSeek-V4.1-Flash on : a language model's rewrite of the original, under its terms