Unit state

A unit may be:

• active: aka started, bound, plugged in, ...
• inactive: aka stopped, unbound, unplugged, ...
• failed: like inactive, but an error of some kind occurred
• activating: changing from inactive to active
• deactivating: changing from active to inactive
• maintenance: inactive + maintenance operation in progress
• reloading: active + reloading the configuration
• refreshing: active + new mountpoint is being activated inside unit namespace

Unit type

• service: units deescribing how to start and control daemons;
• socket: units defining local IPC and network sockets;
• device: units exposing kernel devices to systemd;
• mount: units encoding information about filesystem mount points;
• automount: units providing automount capabalities;
• swap: basically like mount ounits;
• target: units for syncronization points and to group units;
• path: units used to activate services when filesystem ojects change;
• timer: units used to activate a unit with a timer.
• slice: units to group units for resource management;
• scope: units for managing services not started by themself

Unit Options

Units metadata options

• Description=
• Documentation=

Dependencies options

• Wants=: Weak requirements for this unit. Units listed in Wants= option will start together with this unit. If for any reason they fail to start, this unit should start anyway.
• Requires=: units started together with this unit. If one of them fails to start, this unit won't start. If one of them is stopped, this unit will stop. If one of them is restarted, this unit will restart.
• Requisite=: unit won't start if these are not running
• BindsTo=
• PartOf=
• Upholds=
• Conflicts=
• Before/After=: Configure ordering dependencies between units. This unit activation is delayed until all units listed in After= are started. Activation of units listed in the Before= option is delayed until this unit is started.

Unit condition and assert options

Before a unit is started, Condition and Assert options a checked. These are checked the precise moment the unit would start, meaning the ordering dependencies (After/Before options) is respected.

If a Condition options is not met, the unit will be skipped. If an Assert option is not met, the unit will be aborted with an error message. In either case, the unit won't change its state (for example, from inactive to failed

• <Condition/Assert>Architecture=
• <Condition/Assert>Firmware=
• <Condition/Assert>Virtualization=
• <Condition/Assert>Host=
• <Condition/Assert>KernelCommandLine=
• <Condition/Assert>KernelVersion=
• <Condition/Assert>Credentials=
• <Condition/Assert>Environment=
• <Condition/Assert>Security=
• <Condition/Assert>Capability= = <Condition/Assert>ACPower=
• <Condition/Assert>NeeedsUpdate
• <Condition/Assert>FirstBoot
• <Condition/Assert>PathExists-
• <Condition/Assert>PathExistsGlob=
• <Condition/Assert>PathIsDirectory=
• <Condition/Assert>PathIsSymbolicLink=
• <Condition/Assert>PathIsMountPoint=
• <Condition/Assert>PathIsReadWrite=
• <Condition/Assert>PathIsEncrypted=
• <Condition/Assert>DirectoryIsNotEmpty=
• <Condition/Assert>FileIsNotEmpty=
• <Condition/Assert>FileIsExecutable=
• <Condition/Assert>User=
• <Condition/Assert>Group=
• <Condition/Assert>ControlGroupController=
• <Condition/Assert>Memory=
• <Condition/Assert>CPUs=
• <Condition/Assert>CPUFeature=
• <Condition/Assert>OSRelease=
• <Condition/Assert>MemoryPressue=
• <Condition/Assert>CPUPressue=
• <Condition/Assert>IOPressue=