systemd-system.conf, system.conf.d, systemd-user.conf, user.conf.d — System and session service manager configuration files
When run as a system instance, systemd interprets the configuration file
system.conf and the files in
system.conf.d directories; when
run as a user instance, it interprets the configuration file
user.conf (either in
the home directory of the user, or if not found, under
/etc/systemd/) and the files
user.conf.d directories. These configuration files contain a few settings
controlling basic manager operations.
See systemd.syntax(7) for a general description of the syntax.
The default configuration is set during compilation, so configuration is only needed when it is
necessary to deviate from those defaults. The main configuration file is either in
/etc/systemd/ and contains commented out
entries showing the defaults as a guide to the administrator. Local overrides can be created by creating
drop-ins, as described below. The main configuration file can also be edited for this purpose (or a copy
/etc/ if it's shipped in
/usr/) however using drop-ins for
local configuration is recommended over modifications to the main configuration file.
In addition to the "main" configuration file, drop-in configuration snippets are read from
/etc/systemd/*.conf.d/. Those drop-ins have higher precedence and override the
main configuration file. Files in the
*.conf.d/ configuration subdirectories are
sorted by their filename in lexicographic order, regardless of in which of the subdirectories they
reside. When multiple files specify the same option, for options which accept just a single value, the
entry in the file sorted last takes precedence, and for options which accept a list of values, entries
are collected as they occur in the sorted files.
When packages need to customize the configuration, they can install drop-ins under
/usr/. Files in
/etc/ are reserved for the local administrator,
who may use this logic to override the configuration files installed by vendor packages. Drop-ins have to
be used to override package drop-ins, since the main configuration file has lower precedence. It is
recommended to prefix all filenames in those subdirectories with a two-digit number and a dash, to
simplify the ordering of the files. This also defined a concept of drop-in priority to allow
distributions to ship drop-ins within a specific range lower than the range used by users. This should
lower the risk of package drop-ins overriding accidentally drop-ins defined by users.
To disable a configuration file supplied by the vendor, the recommended way is to place a symlink
/dev/null in the configuration directory in
/etc/, with the
same filename as the vendor configuration file.
All options are configured in the [Manager] section:
Configures various parameters of basic manager operation. These options may be overridden by the respective process and kernel command line arguments. See systemd(1) for details.
Defines what action will be performed
if user presses Ctrl-Alt-Delete more than 7 times in 2s.
Can be set to "
or disabled with "
none". Defaults to
Configures the CPU affinity for the service manager as well as the default CPU
affinity for all forked off processes. Takes a list of CPU indices or ranges separated by either
whitespace or commas. CPU ranges are specified by the lower and upper CPU indices separated by a
dash. This option may be specified more than once, in which case the specified CPU affinity masks are
merged. If the empty string is assigned, the mask is reset, all assignments prior to this will have
no effect. Individual services may override the CPU affinity for their processes with the
CPUAffinity= setting in unit files, see
Configures the NUMA memory policy for the service manager and the default NUMA memory policy
for all forked off processes. Individual services may override the default policy with the
NUMAPolicy= setting in unit files, see
Configures the NUMA node mask that will be associated with the selected NUMA policy. Note that
local NUMA policies don't require explicit NUMA node mask and
value of the option can be empty. Similarly to
NUMAPolicy=, value can be overridden
by individual services in unit files, see
Configure the hardware watchdog at runtime and at reboot. Takes a timeout value in
seconds (or in other time units if suffixed with "
w"), or the special strings
off" or "
default". If set to "
0") the watchdog logic is disabled: no watchdog device is opened,
configured, or pinged. If set to the special string "
default" the watchdog is opened
and pinged in regular intervals, but the timeout is not changed from the default. If set to any other
time value the watchdog timeout is configured to the specified value (or a value close to it,
depending on hardware capabilities).
RuntimeWatchdogSec= is set to a non-zero value, the watchdog hardware
/dev/watchdog0 or the path specified with
the kernel option
systemd.watchdog-device=) will be programmed to automatically
reboot the system if it is not contacted within the specified timeout interval. The system manager
will ensure to contact it at least once in half the specified timeout interval. This feature requires
a hardware watchdog device to be present, as it is commonly the case in embedded and server
systems. Not all hardware watchdogs allow configuration of all possible reboot timeout values, in
which case the closest available timeout is picked.
RebootWatchdogSec= may be used to configure the hardware watchdog when the
system is asked to reboot. It works as a safety net to ensure that the reboot takes place even if a
clean reboot attempt times out. Note that the
RebootWatchdogSec= timeout applies
only to the second phase of the reboot, i.e. after all regular services are already terminated, and
after the system and service manager process (PID 1) got replaced by the
systemd-shutdown binary, see system
details. During the first phase of the shutdown operation the system and service manager remains
running and hence
RuntimeWatchdogSec= is still honoured. In order to define a
timeout on this first phase of system shutdown, configure
JobTimeoutAction= in the [Unit] section of the
shutdown.target unit. By default
to 0 (off), and
RebootWatchdogSec= to 10min.
KExecWatchdogSec= may be used to additionally enable the watchdog when kexec
is being executed rather than when rebooting. Note that if the kernel does not reset the watchdog on
kexec (depending on the specific hardware and/or driver), in this case the watchdog might not get
disabled after kexec succeeds and thus the system might get rebooted, unless
RuntimeWatchdogSec= is also enabled at the same time. For this reason it is
recommended to enable
KExecWatchdogSec= only if
RuntimeWatchdogSec= is also enabled.
These settings have no effect if a hardware watchdog is not available.
Configure the hardware watchdog device pre-timeout value.
Takes a timeout value in seconds (or in other time units similar to
RuntimeWatchdogSec=). A watchdog pre-timeout is a
notification generated by the watchdog before the watchdog reset might
occur in the event the watchdog has not been serviced. This notification
is handled by the kernel and can be configured to take an action (i.e.
generate a kernel panic) using
Not all watchdog hardware or drivers support generating a pre-timeout and
depending on the state of the system, the kernel may be unable to take the
configured action before the watchdog reboot. The watchdog will be configured
to generate the pre-timeout event at the amount of time specified by
RuntimeWatchdogPreSec= before the runtime watchdog timeout
RuntimeWatchdogSec=). For example, if the we have
RuntimeWatchdogPreSec=10, then the pre-timeout event
will occur if the watchdog has not pinged for 20s (10s before the
watchdog would fire). By default,
defaults to 0 (off). The value set for
must be smaller than the timeout value for
This setting has no effect if a hardware watchdog is not available or the
hardware watchdog does not support a pre-timeout and will be ignored by the
kernel if the setting is greater than the actual watchdog timeout.
Configure the action taken by the hardware watchdog device
when the pre-timeout expires. The default action for the pre-timeout event
depends on the kernel configuration, but it is usually to log a kernel
message. For a list of valid actions available for a given watchdog device,
check the content of the
file. Typically, available governor types are
Availability, names and functionality might vary depending on the specific device driver
in use. If the
pretimeout_available_governors sysfs file is empty,
the governor might be built as a kernel module and might need to be manually loaded
pretimeout_noop.ko), or the watchdog device might not support
Configure the hardware watchdog device that the
runtime and shutdown watchdog timers will open and use. Defaults
/dev/watchdog0. This setting has no
effect if a hardware watchdog is not available.
Controls which capabilities to include in the
capability bounding set for PID 1 and its children. See
for details. Takes a whitespace-separated list of capability
names as read by
Capabilities listed will be included in the bounding set, all
others are removed. If the list of capabilities is prefixed
with ~, all but the listed capabilities will be included, the
effect of the assignment inverted. Note that this option also
affects the respective capabilities in the effective,
permitted and inheritable capability sets. The capability
bounding set may also be individually configured for units
for units, but note that capabilities dropped for PID 1 cannot
be regained in individual units, they are lost for
Takes a boolean argument. If true, ensures that PID 1 and all its children can never gain new privileges through execve(2) (e.g. via setuid or setgid bits, or filesystem capabilities). Defaults to false. General purpose distributions commonly rely on executables with setuid or setgid bits and will thus not function properly with this option enabled. Individual units cannot disable this option. Also see No New Privileges Flag.
Takes a space-separated list of architecture
identifiers. Selects from which architectures system calls may
be invoked on this system. This may be used as an effective
way to disable invocation of non-native binaries system-wide,
for example to prohibit execution of 32-bit x86 binaries on
64-bit x86-64 systems. This option operates system-wide, and
acts similar to the
SystemCallArchitectures= setting of unit
for details. This setting defaults to the empty list, in which
case no filtering of system calls based on architecture is
applied. Known architecture identifiers are
arm" and the special
native". The latter implicitly
maps to the native architecture of the system (or more
specifically, the architecture the system manager was compiled
for). Set this setting to "
prohibit execution of any non-native binaries. When a binary
executes a system call of an architecture that is not listed
in this setting, it will be immediately terminated with the
Sets the timer slack in nanoseconds for PID 1,
which is inherited by all executed processes, unless
overridden individually, for example with the
TimerSlackNSec= setting in service units
(for details see
The timer slack controls the accuracy of wake-ups triggered by
system timers. See
for more information. Note that in contrast to most other time
span definitions this parameter takes an integer value in
nano-seconds if no unit is specified. The usual time units are
combined as the value. If
name, the system manager will use unit
names in status messages (e.g. "
systemd-journald.service"), instead of the longer
and more informative descriptions set with
Description= (e.g. "
Logging Service"). If
combined, the system manager will use both unit names
and descriptions in status messages (e.g. "
systemd-journald.service - Journal Logging
details about unit names and
Sets the default accuracy of timer units. This
controls the global default for the
AccuracySec= setting of timer units, see
AccuracySec= set in individual
units override the global default for the specific unit.
Defaults to 1min. Note that the accuracy of timer units is
also affected by the configured timer slack for PID 1, see
Configures the default timeouts for starting, stopping and aborting of units, as well
as the default time to sleep between automatic restarts of units, as configured per-unit in
RestartSec= (for services, see
for details on the per-unit settings). For non-service units,
DefaultTimeoutStartSec= sets the default
default to 90 s in the system manager and 90 s in the user manager.
DefaultTimeoutAbortSec= is not set by default so that all units fall back to
DefaultRestartSec= defaults to 100 ms.
Configures the default timeout for waiting for devices. It can be changed per
device via the
x-systemd.device-timeout= option in
Defaults to 90 s in the system manager and 90 s in the user manager.
Configure the default unit start rate
limiting, as configured per-service by
for details on the per-service settings.
DefaultStartLimitIntervalSec= defaults to
DefaultStartLimitBurst= defaults to
Configures environment variables passed to all executed processes. Takes a space-separated list of variable assignments. See environ(7) for details about environment variables.
%"-specifier expansion is supported, see below for a list of supported
DefaultEnvironment="VAR1=word1 word2" VAR2=word3 "VAR3=word 5 6"
Sets three variables
Takes the same arguments as
DefaultEnvironment=, see above. Sets
environment variables just for the manager process itself. In contrast to user managers, these variables
are not inherited by processes spawned by the system manager, use
for that. Note that these variables are merged into the existing environment block. In particular, in
case of the system manager, this includes variables set by the kernel based on the kernel command line.
Setting environment variables for the manager process may be useful to modify its behaviour. See Known Environment Variables for a descriptions of some variables understood by systemd.
%"-specifier expansion is supported, see below for a list of supported
Configure the default resource accounting settings, as configured per-unit by
for details on the per-unit settings.
DefaultCPUAccounting= defaults to yes when running on kernel ≥4.15, and no on older versions.
DefaultMemoryAccounting= defaults to yes.
DefaultTasksAccounting= defaults to yes.
The other settings default to no.
Configure the default value for the per-unit
TasksMax= setting. See
for details. This setting applies to all unit types that support resource control settings, with the exception
of slice units. Defaults to 15% of the minimum of
and root cgroup
Kernel has a default value for
kernel.pid_max= and an algorithm of counting in case of more than 32 cores.
For example, with the default
DefaultTasksMax= defaults to 4915,
but might be greater in other systems or smaller in OS containers.
These settings control various default resource limits for processes executed by
details. These settings may be overridden in individual units using the corresponding
LimitXXX= directives and they accept the same parameter syntax,
for details. Note that these resource limits are only defaults
for units, they are not applied to the service manager process (i.e. PID 1) itself.
Most of these settings are unset, which means the resource limits are inherited from the kernel or, if invoked in a container, from the container manager. However, the following have defaults:
DefaultLimitNOFILE= defaults to 1024:524288.
DefaultLimitMEMLOCK= defaults to 8M.
DefaultLimitCORE= does not have a default but it is worth mentioning that
RLIMIT_CORE is set to "
infinity" by PID 1 which is inherited by its
Note that the service manager internally in PID 1 bumps
RLIMIT_MEMLOCK to higher values, however the limit is reverted to the mentioned
defaults for all child processes forked off.
Configure the default policy for reacting to processes being killed by the Linux
Out-Of-Memory (OOM) killer or systemd-oomd. This may be used to pick a global default for the per-unit
OOMPolicy= setting. See
for details. Note that this default is not used for services that have
Configures the default OOM score adjustments of processes run by the service
manager. This defaults to unset (meaning the forked off processes inherit the service manager's OOM
score adjustment value), except if the service manager is run for an unprivileged user, in which case
this defaults to the service manager's OOM adjustment value plus 100 (this makes service processes
slightly more likely to be killed under memory pressure than the manager itself). This may be used to
pick a global default for the per-unit
OOMScoreAdjust= setting. See
details. Note that this setting has no effect on the OOM score adjustment value of the service
manager process itself, it retains the original value set during its invocation.
SMACK64 security label as the argument. The process executed
by a unit will be started under this label if
SmackProcessLabel= is not set in the
unit. See systemd.exec(5)
for the details.
If the value is "
/", only labels specified with
are assigned and the compile-time default is ignored.
Rate limiting for daemon-reload requests. Default to unset, and any number of daemon-reload
operations can be requested at any time.
ReloadLimitIntervalSec= takes a value in seconds
to configure the rate limit window, and
ReloadLimitBurst= takes a positive integer to
configure the maximum allowed number of reloads within the configured time window.
Configures the default settings for the per-unit
for details. Defaults to "
auto" and "
200ms", respectively. This
also sets the memory pressure monitoring threshold for the service manager itself.
Specifiers may be used in the
ManagerEnvironment= settings. The following expansions are understood:
Table 1. Specifiers available
|"||Architecture||A short string identifying the architecture of the local system. A string such as |
|"||Operating system image version||The operating system image version identifier of the running system, as read from the |
|"||Boot ID||The boot ID of the running system, formatted as string. See random(4) for more information.|
|"||Operating system build ID||The operating system build identifier of the running system, as read from the |
|"||Host name||The hostname of the running system.|
|"||Short host name||The hostname of the running system, truncated at the first dot to remove any domain component.|
|"||Machine ID||The machine ID of the running system, formatted as string. See machine-id(5) for more information.|
|"||Operating system image identifier||The operating system image identifier of the running system, as read from the |
|"||Operating system ID||The operating system identifier of the running system, as read from the |
|"||Kernel release||Identical to uname -r output.|
|"||Operating system version ID||The operating system version identifier of the running system, as read from the |
|"||Operating system variant ID||The operating system variant identifier of the running system, as read from the |
|"||Directory for temporary files||This is either |
|"||Directory for larger and persistent temporary files||This is either |
|"||User home directory||This is the home directory of the user running the service manager instance.|
|"||Username||This is the username of the user running the service manager instance.|
|"||User id||This is the user id of the user running the service manager instance.|
|"||Primary group||This is the primary group of the user running the service manager instance.|
|"||Primary group id||This is the primary group id of the user running the service manager instance.|
|"||User shell||This is the shell of the user running the service manager instance.|
|"||Single percent sign||Use "|
DefaultBlockIOAccounting= was deprecated. Please switch
to the unified cgroup hierarchy.