2 <!DOCTYPE refentry PUBLIC
"-//OASIS//DTD DocBook XML V4.5//EN"
3 "http://www.oasis-open.org/docbook/xml/4.2/docbookx.dtd">
4 <!-- SPDX-License-Identifier: LGPL-2.1+ -->
6 <refentry id=
"systemd.resource-control">
8 <title>systemd.resource-control
</title>
9 <productname>systemd
</productname>
13 <refentrytitle>systemd.resource-control
</refentrytitle>
14 <manvolnum>5</manvolnum>
18 <refname>systemd.resource-control
</refname>
19 <refpurpose>Resource control unit settings
</refpurpose>
24 <filename><replaceable>slice
</replaceable>.slice
</filename>,
25 <filename><replaceable>scope
</replaceable>.scope
</filename>,
26 <filename><replaceable>service
</replaceable>.service
</filename>,
27 <filename><replaceable>socket
</replaceable>.socket
</filename>,
28 <filename><replaceable>mount
</replaceable>.mount
</filename>,
29 <filename><replaceable>swap
</replaceable>.swap
</filename>
34 <title>Description
</title>
36 <para>Unit configuration files for services, slices, scopes, sockets, mount points, and swap devices share a subset
37 of configuration options for resource control of spawned processes. Internally, this relies on the Linux Control
38 Groups (cgroups) kernel concept for organizing processes in a hierarchical tree of named groups for the purpose of
39 resource management.
</para>
41 <para>This man page lists the configuration options shared by
42 those six unit types. See
43 <citerefentry><refentrytitle>systemd.unit
</refentrytitle><manvolnum>5</manvolnum></citerefentry>
44 for the common options of all unit configuration files, and
45 <citerefentry><refentrytitle>systemd.slice
</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
46 <citerefentry><refentrytitle>systemd.scope
</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
47 <citerefentry><refentrytitle>systemd.service
</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
48 <citerefentry><refentrytitle>systemd.socket
</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
49 <citerefentry><refentrytitle>systemd.mount
</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
51 <citerefentry><refentrytitle>systemd.swap
</refentrytitle><manvolnum>5</manvolnum></citerefentry>
52 for more information on the specific unit configuration files. The
53 resource control configuration options are configured in the
54 [Slice], [Scope], [Service], [Socket], [Mount], or [Swap]
55 sections, depending on the unit type.
</para>
57 <para>In addition, options which control resources available to programs
58 <emphasis>executed
</emphasis> by systemd are listed in
59 <citerefentry><refentrytitle>systemd.exec
</refentrytitle><manvolnum>5</manvolnum></citerefentry>.
60 Those options complement options listed here.
</para>
63 url=
"https://www.freedesktop.org/wiki/Software/systemd/ControlGroupInterface/">New
64 Control Group Interfaces
</ulink> for an introduction on how to make
65 use of resource control APIs from programs.
</para>
69 <title>Implicit Dependencies
</title>
71 <para>The following dependencies are implicitly added:
</para>
74 <listitem><para>Units with the
<varname>Slice=
</varname> setting set automatically acquire
75 <varname>Requires=
</varname> and
<varname>After=
</varname> dependencies on the specified
76 slice unit.
</para></listitem>
80 <!-- We don't have any default dependency here. -->
83 <title>Unified and Legacy Control Group Hierarchies
</title>
85 <para>The unified control group hierarchy is the new version of kernel control group interface, see
86 <ulink url=
"https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v2.html">Control Groups v2
</ulink>.
87 Depending on the resource type, there are differences in resource control capabilities. Also, because of
88 interface changes, some resource types have separate set of options on the unified hierarchy.
</para>
96 <para><varname>CPUWeight=
</varname> and
<varname>StartupCPUWeight=
</varname> replace
97 <varname>CPUShares=
</varname> and
<varname>StartupCPUShares=
</varname>, respectively.
</para>
99 <para>The
<literal>cpuacct
</literal> controller does not exist separately on the unified hierarchy.
</para>
106 <para><varname>MemoryMax=
</varname> replaces
<varname>MemoryLimit=
</varname>.
<varname>MemoryLow=
</varname>
107 and
<varname>MemoryHigh=
</varname> are effective only on unified hierarchy.
</para>
114 <para><literal>IO
</literal>-prefixed settings are a superset of and replace
115 <literal>BlockIO
</literal>-prefixed ones. On unified hierarchy, IO resource control also applies
116 to buffered writes.
</para>
123 <para>To ease the transition, there is best-effort translation between the two versions of settings. For each
124 controller, if any of the settings for the unified hierarchy are present, all settings for the legacy hierarchy are
125 ignored. If the resulting settings are for the other type of hierarchy, the configurations are translated before
128 <para>Legacy control group hierarchy (see
<ulink
129 url=
"https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v1/">Control Groups version
1</ulink>),
130 also called cgroup-v1, doesn't allow safe delegation of controllers to unprivileged processes. If the
131 system uses the legacy control group hierarchy, resource control is disabled for the systemd user
133 <citerefentry><refentrytitle>systemd
</refentrytitle><manvolnum>1</manvolnum></citerefentry>.
</para>
137 <title>Options
</title>
139 <para>Units of the types listed above can have settings
140 for resource control configuration:
</para>
142 <variablelist class='unit-directives'
>
145 <term><varname>CPUAccounting=
</varname></term>
148 <para>Turn on CPU usage accounting for this unit. Takes a
149 boolean argument. Note that turning on CPU accounting for
150 one unit will also implicitly turn it on for all units
151 contained in the same slice and for all its parent slices
152 and the units contained therein. The system default for this
153 setting may be controlled with
154 <varname>DefaultCPUAccounting=
</varname> in
155 <citerefentry><refentrytitle>systemd-system.conf
</refentrytitle><manvolnum>5</manvolnum></citerefentry>.
</para>
160 <term><varname>CPUWeight=
<replaceable>weight
</replaceable></varname></term>
161 <term><varname>StartupCPUWeight=
<replaceable>weight
</replaceable></varname></term>
164 <para>Assign the specified CPU time weight to the processes executed, if the unified control group hierarchy
165 is used on the system. These options take an integer value and control the
<literal>cpu.weight
</literal>
166 control group attribute. The allowed range is
1 to
10000. Defaults to
100. For details about this control
167 group attribute, see
<ulink
168 url=
"https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v2.html">Control Groups v2
</ulink> and
<ulink
169 url=
"https://www.kernel.org/doc/html/latest/scheduler/sched-design-CFS.html">CFS Scheduler
</ulink>.
170 The available CPU time is split up among all units within one slice relative to their CPU time weight.
</para>
172 <para>While
<varname>StartupCPUWeight=
</varname> only applies to the startup phase of the system,
173 <varname>CPUWeight=
</varname> applies to normal runtime of the system, and if the former is not set also to
174 the startup phase. Using
<varname>StartupCPUWeight=
</varname> allows prioritizing specific services at
175 boot-up differently than during normal runtime.
</para>
177 <para>These settings replace
<varname>CPUShares=
</varname> and
<varname>StartupCPUShares=
</varname>.
</para>
182 <term><varname>CPUQuota=
</varname></term>
185 <para>Assign the specified CPU time quota to the processes executed. Takes a percentage value, suffixed with
186 "%". The percentage specifies how much CPU time the unit shall get at maximum, relative to the total CPU time
187 available on one CPU. Use values
> 100% for allotting CPU time on more than one CPU. This controls the
188 <literal>cpu.max
</literal> attribute on the unified control group hierarchy and
189 <literal>cpu.cfs_quota_us
</literal> on legacy. For details about these control group attributes, see
<ulink
190 url=
"https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v2.html">Control Groups v2
</ulink> and
<ulink
191 url=
"https://www.kernel.org/doc/Documentation/scheduler/sched-bwc.txt">sched-bwc.txt
</ulink>.
</para>
193 <para>Example:
<varname>CPUQuota=
20%
</varname> ensures that the executed processes will never get more than
194 20% CPU time on one CPU.
</para>
200 <term><varname>CPUQuotaPeriodSec=
</varname></term>
203 <para>Assign the duration over which the CPU time quota specified by
<varname>CPUQuota=
</varname> is measured.
204 Takes a time duration value in seconds, with an optional suffix such as
"ms" for milliseconds (or
"s" for seconds.)
205 The default setting is
100ms. The period is clamped to the range supported by the kernel, which is [
1ms,
1000ms].
206 Additionally, the period is adjusted up so that the quota interval is also at least
1ms.
207 Setting
<varname>CPUQuotaPeriodSec=
</varname> to an empty value resets it to the default.
</para>
209 <para>This controls the second field of
<literal>cpu.max
</literal> attribute on the unified control group hierarchy
210 and
<literal>cpu.cfs_period_us
</literal> on legacy. For details about these control group attributes, see
211 <ulink url=
"https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v2.html">Control Groups v2
</ulink> and
212 <ulink url=
"https://www.kernel.org/doc/html/latest/scheduler/sched-design-CFS.html">CFS Scheduler
</ulink>.
</para>
214 <para>Example:
<varname>CPUQuotaPeriodSec=
10ms
</varname> to request that the CPU quota is measured in periods of
10ms.
</para>
219 <term><varname>AllowedCPUs=
</varname></term>
222 <para>Restrict processes to be executed on specific CPUs. Takes a list of CPU indices or ranges separated by either
223 whitespace or commas. CPU ranges are specified by the lower and upper CPU indices separated by a dash.
</para>
225 <para>Setting
<varname>AllowedCPUs=
</varname> doesn't guarantee that all of the CPUs will be used by the processes
226 as it may be limited by parent units. The effective configuration is reported as
<varname>EffectiveCPUs=
</varname>.
</para>
228 <para>This setting is supported only with the unified control group hierarchy.
</para>
233 <term><varname>AllowedMemoryNodes=
</varname></term>
236 <para>Restrict processes to be executed on specific memory NUMA nodes. Takes a list of memory NUMA nodes indices
237 or ranges separated by either whitespace or commas. Memory NUMA nodes ranges are specified by the lower and upper
238 CPU indices separated by a dash.
</para>
240 <para>Setting
<varname>AllowedMemoryNodes=
</varname> doesn't guarantee that all of the memory NUMA nodes will
241 be used by the processes as it may be limited by parent units. The effective configuration is reported as
242 <varname>EffectiveMemoryNodes=
</varname>.
</para>
244 <para>This setting is supported only with the unified control group hierarchy.
</para>
249 <term><varname>MemoryAccounting=
</varname></term>
252 <para>Turn on process and kernel memory accounting for this
253 unit. Takes a boolean argument. Note that turning on memory
254 accounting for one unit will also implicitly turn it on for
255 all units contained in the same slice and for all its parent
256 slices and the units contained therein. The system default
257 for this setting may be controlled with
258 <varname>DefaultMemoryAccounting=
</varname> in
259 <citerefentry><refentrytitle>systemd-system.conf
</refentrytitle><manvolnum>5</manvolnum></citerefentry>.
</para>
264 <term><varname>MemoryMin=
<replaceable>bytes
</replaceable></varname></term>
267 <para>Specify the memory usage protection of the executed processes in this unit. If the memory usages of
268 this unit and all its ancestors are below their minimum boundaries, this unit's memory won't be reclaimed.
</para>
270 <para>Takes a memory size in bytes. If the value is suffixed with K, M, G or T, the specified memory size is
271 parsed as Kilobytes, Megabytes, Gigabytes, or Terabytes (with the base
1024), respectively. Alternatively, a
272 percentage value may be specified, which is taken relative to the installed physical memory on the
273 system. If assigned the special value
<literal>infinity
</literal>, all available memory is protected, which may be
274 useful in order to always inherit all of the protection afforded by ancestors.
275 This controls the
<literal>memory.min
</literal> control group attribute. For details about this
276 control group attribute, see
<ulink
277 url=
"https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v2.html">Control Groups v2
</ulink>.
</para>
279 <para>This setting is supported only if the unified control group hierarchy is used and disables
280 <varname>MemoryLimit=
</varname>.
</para>
282 <para>Units may have their children use a default
<literal>memory.min
</literal> value by specifying
283 <varname>DefaultMemoryMin=
</varname>, which has the same semantics as
<varname>MemoryMin=
</varname>. This setting
284 does not affect
<literal>memory.min
</literal> in the unit itself.
</para>
289 <term><varname>MemoryLow=
<replaceable>bytes
</replaceable></varname></term>
292 <para>Specify the best-effort memory usage protection of the executed processes in this unit. If the memory
293 usages of this unit and all its ancestors are below their low boundaries, this unit's memory won't be
294 reclaimed as long as memory can be reclaimed from unprotected units.
</para>
296 <para>Takes a memory size in bytes. If the value is suffixed with K, M, G or T, the specified memory size is
297 parsed as Kilobytes, Megabytes, Gigabytes, or Terabytes (with the base
1024), respectively. Alternatively, a
298 percentage value may be specified, which is taken relative to the installed physical memory on the
299 system. If assigned the special value
<literal>infinity
</literal>, all available memory is protected, which may be
300 useful in order to always inherit all of the protection afforded by ancestors.
301 This controls the
<literal>memory.low
</literal> control group attribute. For details about this
302 control group attribute, see
<ulink
303 url=
"https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v2.html">Control Groups v2
</ulink>.
</para>
305 <para>This setting is supported only if the unified control group hierarchy is used and disables
306 <varname>MemoryLimit=
</varname>.
</para>
308 <para>Units may have their children use a default
<literal>memory.low
</literal> value by specifying
309 <varname>DefaultMemoryLow=
</varname>, which has the same semantics as
<varname>MemoryLow=
</varname>. This setting
310 does not affect
<literal>memory.low
</literal> in the unit itself.
</para>
315 <term><varname>MemoryHigh=
<replaceable>bytes
</replaceable></varname></term>
318 <para>Specify the throttling limit on memory usage of the executed processes in this unit. Memory usage may go
319 above the limit if unavoidable, but the processes are heavily slowed down and memory is taken away
320 aggressively in such cases. This is the main mechanism to control memory usage of a unit.
</para>
322 <para>Takes a memory size in bytes. If the value is suffixed with K, M, G or T, the specified memory size is
323 parsed as Kilobytes, Megabytes, Gigabytes, or Terabytes (with the base
1024), respectively. Alternatively, a
324 percentage value may be specified, which is taken relative to the installed physical memory on the
325 system. If assigned the
326 special value
<literal>infinity
</literal>, no memory throttling is applied. This controls the
327 <literal>memory.high
</literal> control group attribute. For details about this control group attribute, see
328 <ulink url=
"https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v2.html">Control Groups v2
</ulink>.
</para>
330 <para>This setting is supported only if the unified control group hierarchy is used and disables
331 <varname>MemoryLimit=
</varname>.
</para>
336 <term><varname>MemoryMax=
<replaceable>bytes
</replaceable></varname></term>
339 <para>Specify the absolute limit on memory usage of the executed processes in this unit. If memory usage
340 cannot be contained under the limit, out-of-memory killer is invoked inside the unit. It is recommended to
341 use
<varname>MemoryHigh=
</varname> as the main control mechanism and use
<varname>MemoryMax=
</varname> as the
342 last line of defense.
</para>
344 <para>Takes a memory size in bytes. If the value is suffixed with K, M, G or T, the specified memory size is
345 parsed as Kilobytes, Megabytes, Gigabytes, or Terabytes (with the base
1024), respectively. Alternatively, a
346 percentage value may be specified, which is taken relative to the installed physical memory on the system. If
347 assigned the special value
<literal>infinity
</literal>, no memory limit is applied. This controls the
348 <literal>memory.max
</literal> control group attribute. For details about this control group attribute, see
349 <ulink url=
"https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v2.html">Control Groups v2
</ulink>.
</para>
351 <para>This setting replaces
<varname>MemoryLimit=
</varname>.
</para>
356 <term><varname>MemorySwapMax=
<replaceable>bytes
</replaceable></varname></term>
359 <para>Specify the absolute limit on swap usage of the executed processes in this unit.
</para>
361 <para>Takes a swap size in bytes. If the value is suffixed with K, M, G or T, the specified swap size is
362 parsed as Kilobytes, Megabytes, Gigabytes, or Terabytes (with the base
1024), respectively. If assigned the
363 special value
<literal>infinity
</literal>, no swap limit is applied. This controls the
364 <literal>memory.swap.max
</literal> control group attribute. For details about this control group attribute,
365 see
<ulink url=
"https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v2.html">Control Groups v2
</ulink>.
</para>
367 <para>This setting is supported only if the unified control group hierarchy is used and disables
368 <varname>MemoryLimit=
</varname>.
</para>
373 <term><varname>TasksAccounting=
</varname></term>
376 <para>Turn on task accounting for this unit. Takes a
377 boolean argument. If enabled, the system manager will keep
378 track of the number of tasks in the unit. The number of
379 tasks accounted this way includes both kernel threads and
380 userspace processes, with each thread counting
381 individually. Note that turning on tasks accounting for one
382 unit will also implicitly turn it on for all units contained
383 in the same slice and for all its parent slices and the
384 units contained therein. The system default for this setting
385 may be controlled with
386 <varname>DefaultTasksAccounting=
</varname> in
387 <citerefentry><refentrytitle>systemd-system.conf
</refentrytitle><manvolnum>5</manvolnum></citerefentry>.
</para>
392 <term><varname>TasksMax=
<replaceable>N
</replaceable></varname></term>
395 <para>Specify the maximum number of tasks that may be created in the unit. This ensures that the number of
396 tasks accounted for the unit (see above) stays below a specific limit. This either takes an absolute number
397 of tasks or a percentage value that is taken relative to the configured maximum number of tasks on the
398 system. If assigned the special value
<literal>infinity
</literal>, no tasks limit is applied. This controls
399 the
<literal>pids.max
</literal> control group attribute. For details about this control group attribute, see
400 <ulink url=
"https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v1/pids.html">Process Number Controller
</ulink>.
403 <para>The system default for this setting may be controlled with
404 <varname>DefaultTasksMax=
</varname> in
405 <citerefentry><refentrytitle>systemd-system.conf
</refentrytitle><manvolnum>5</manvolnum></citerefentry>.
</para>
410 <term><varname>IOAccounting=
</varname></term>
413 <para>Turn on Block I/O accounting for this unit, if the unified control group hierarchy is used on the
414 system. Takes a boolean argument. Note that turning on block I/O accounting for one unit will also implicitly
415 turn it on for all units contained in the same slice and all for its parent slices and the units contained
416 therein. The system default for this setting may be controlled with
<varname>DefaultIOAccounting=
</varname>
418 <citerefentry><refentrytitle>systemd-system.conf
</refentrytitle><manvolnum>5</manvolnum></citerefentry>.
</para>
420 <para>This setting replaces
<varname>BlockIOAccounting=
</varname> and disables settings prefixed with
421 <varname>BlockIO
</varname> or
<varname>StartupBlockIO
</varname>.
</para>
426 <term><varname>IOWeight=
<replaceable>weight
</replaceable></varname></term>
427 <term><varname>StartupIOWeight=
<replaceable>weight
</replaceable></varname></term>
430 <para>Set the default overall block I/O weight for the executed processes, if the unified control group
431 hierarchy is used on the system. Takes a single weight value (between
1 and
10000) to set the default block
432 I/O weight. This controls the
<literal>io.weight
</literal> control group attribute, which defaults to
433 100. For details about this control group attribute, see
<ulink
434 url=
"https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v2.html">Control Groups v2
</ulink>. The available I/O
435 bandwidth is split up among all units within one slice relative to their block I/O weight.
</para>
437 <para>While
<varname>StartupIOWeight=
</varname> only applies
438 to the startup phase of the system,
439 <varname>IOWeight=
</varname> applies to the later runtime of
440 the system, and if the former is not set also to the startup
441 phase. This allows prioritizing specific services at boot-up
442 differently than during runtime.
</para>
444 <para>These settings replace
<varname>BlockIOWeight=
</varname> and
<varname>StartupBlockIOWeight=
</varname>
445 and disable settings prefixed with
<varname>BlockIO
</varname> or
<varname>StartupBlockIO
</varname>.
</para>
450 <term><varname>IODeviceWeight=
<replaceable>device
</replaceable> <replaceable>weight
</replaceable></varname></term>
453 <para>Set the per-device overall block I/O weight for the executed processes, if the unified control group
454 hierarchy is used on the system. Takes a space-separated pair of a file path and a weight value to specify
455 the device specific weight value, between
1 and
10000. (Example:
<literal>/dev/sda
1000</literal>). The file
456 path may be specified as path to a block device node or as any other file, in which case the backing block
457 device of the file system of the file is determined. This controls the
<literal>io.weight
</literal> control
458 group attribute, which defaults to
100. Use this option multiple times to set weights for multiple devices.
459 For details about this control group attribute, see
<ulink
460 url=
"https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v2.html">Control Groups v2
</ulink>.
</para>
462 <para>This setting replaces
<varname>BlockIODeviceWeight=
</varname> and disables settings prefixed with
463 <varname>BlockIO
</varname> or
<varname>StartupBlockIO
</varname>.
</para>
468 <term><varname>IOReadBandwidthMax=
<replaceable>device
</replaceable> <replaceable>bytes
</replaceable></varname></term>
469 <term><varname>IOWriteBandwidthMax=
<replaceable>device
</replaceable> <replaceable>bytes
</replaceable></varname></term>
472 <para>Set the per-device overall block I/O bandwidth maximum limit for the executed processes, if the unified
473 control group hierarchy is used on the system. This limit is not work-conserving and the executed processes
474 are not allowed to use more even if the device has idle capacity. Takes a space-separated pair of a file
475 path and a bandwidth value (in bytes per second) to specify the device specific bandwidth. The file path may
476 be a path to a block device node, or as any other file in which case the backing block device of the file
477 system of the file is used. If the bandwidth is suffixed with K, M, G, or T, the specified bandwidth is
478 parsed as Kilobytes, Megabytes, Gigabytes, or Terabytes, respectively, to the base of
1000. (Example:
479 "/dev/disk/by-path/pci-0000:00:1f.2-scsi-0:0:0:0 5M"). This controls the
<literal>io.max
</literal> control
480 group attributes. Use this option multiple times to set bandwidth limits for multiple devices. For details
481 about this control group attribute, see
<ulink
482 url=
"https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v2.html">Control Groups v2
</ulink>.
485 <para>These settings replace
<varname>BlockIOReadBandwidth=
</varname> and
486 <varname>BlockIOWriteBandwidth=
</varname> and disable settings prefixed with
<varname>BlockIO
</varname> or
487 <varname>StartupBlockIO
</varname>.
</para>
492 <term><varname>IOReadIOPSMax=
<replaceable>device
</replaceable> <replaceable>IOPS
</replaceable></varname></term>
493 <term><varname>IOWriteIOPSMax=
<replaceable>device
</replaceable> <replaceable>IOPS
</replaceable></varname></term>
496 <para>Set the per-device overall block I/O IOs-Per-Second maximum limit for the executed processes, if the
497 unified control group hierarchy is used on the system. This limit is not work-conserving and the executed
498 processes are not allowed to use more even if the device has idle capacity. Takes a space-separated pair of
499 a file path and an IOPS value to specify the device specific IOPS. The file path may be a path to a block
500 device node, or as any other file in which case the backing block device of the file system of the file is
501 used. If the IOPS is suffixed with K, M, G, or T, the specified IOPS is parsed as KiloIOPS, MegaIOPS,
502 GigaIOPS, or TeraIOPS, respectively, to the base of
1000. (Example:
503 "/dev/disk/by-path/pci-0000:00:1f.2-scsi-0:0:0:0 1K"). This controls the
<literal>io.max
</literal> control
504 group attributes. Use this option multiple times to set IOPS limits for multiple devices. For details about
505 this control group attribute, see
<ulink
506 url=
"https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v2.html">Control Groups v2
</ulink>.
509 <para>These settings are supported only if the unified control group hierarchy is used and disable settings
510 prefixed with
<varname>BlockIO
</varname> or
<varname>StartupBlockIO
</varname>.
</para>
515 <term><varname>IODeviceLatencyTargetSec=
<replaceable>device
</replaceable> <replaceable>target
</replaceable></varname></term>
518 <para>Set the per-device average target I/O latency for the executed processes, if the unified control group
519 hierarchy is used on the system. Takes a file path and a timespan separated by a space to specify
520 the device specific latency target. (Example:
"/dev/sda 25ms"). The file path may be specified
521 as path to a block device node or as any other file, in which case the backing block device of the file
522 system of the file is determined. This controls the
<literal>io.latency
</literal> control group
523 attribute. Use this option multiple times to set latency target for multiple devices. For details about this
524 control group attribute, see
<ulink
525 url=
"https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v2.html">Control Groups v2
</ulink>.
</para>
527 <para>Implies
<literal>IOAccounting=yes
</literal>.
</para>
529 <para>These settings are supported only if the unified control group hierarchy is used.
</para>
534 <term><varname>IPAccounting=
</varname></term>
537 <para>Takes a boolean argument. If true, turns on IPv4 and IPv6 network traffic accounting for packets sent
538 or received by the unit. When this option is turned on, all IPv4 and IPv6 sockets created by any process of
539 the unit are accounted for.
</para>
541 <para>When this option is used in socket units, it applies to all IPv4 and IPv6 sockets
542 associated with it (including both listening and connection sockets where this applies). Note that for
543 socket-activated services, this configuration setting and the accounting data of the service unit and the
544 socket unit are kept separate, and displayed separately. No propagation of the setting and the collected
545 statistics is done, in either direction. Moreover, any traffic sent or received on any of the socket unit's
546 sockets is accounted to the socket unit — and never to the service unit it might have activated, even if the
547 socket is used by it.
</para>
549 <para>The system default for this setting may be controlled with
<varname>DefaultIPAccounting=
</varname> in
550 <citerefentry><refentrytitle>systemd-system.conf
</refentrytitle><manvolnum>5</manvolnum></citerefentry>.
</para>
555 <term><varname>IPAddressAllow=
<replaceable>ADDRESS[/PREFIXLENGTH]…
</replaceable></varname></term>
556 <term><varname>IPAddressDeny=
<replaceable>ADDRESS[/PREFIXLENGTH]…
</replaceable></varname></term>
559 <para>Turn on address range network traffic filtering for IP packets sent and received over
560 <constant>AF_INET
</constant> and
<constant>AF_INET6
</constant> sockets. Both directives take a
561 space separated list of IPv4 or IPv6 addresses, each optionally suffixed with an address prefix
562 length in bits (separated by a
<literal>/
</literal> character). If the latter is omitted, the
563 address is considered a host address, i.e. the prefix covers the whole address (
32 for IPv4,
128
566 <para>The access lists configured with this option are applied to all sockets created by processes
567 of this unit (or in the case of socket units, associated with it). The lists are implicitly
568 combined with any lists configured for any of the parent slice units this unit might be a member
569 of. By default all access lists are empty. Both ingress and egress traffic is filtered by these
570 settings. In case of ingress traffic the source IP address is checked against these access lists,
571 in case of egress traffic the destination IP address is checked. When configured the lists are
572 enforced as follows:
</para>
575 <listitem><para>Access will be granted in case an IP packet's destination/source address matches
576 any entry in the
<varname>IPAddressAllow=
</varname> setting.
</para></listitem>
578 <listitem><para>Otherwise, access will be denied in case its destination/source address matches
579 any entry in the
<varname>IPAddressDeny=
</varname> setting.
</para></listitem>
581 <listitem><para>Otherwise, access will be granted.
</para></listitem>
584 <para>In order to implement a whitelisting IP firewall, it is recommended to use a
585 <varname>IPAddressDeny=
</varname><constant>any
</constant> setting on an upper-level slice unit (such as the
586 root slice
<filename>-.slice
</filename> or the slice containing all system services
587 <filename>system.slice
</filename> – see
588 <citerefentry><refentrytitle>systemd.special
</refentrytitle><manvolnum>7</manvolnum></citerefentry> for
589 details on these slice units), plus individual per-service
<varname>IPAddressAllow=
</varname> lines
590 permitting network access to relevant services, and only them.
</para>
592 <para>Note that for socket-activated services, the IP access list configured on the socket unit applies to
593 all sockets associated with it directly, but not to any sockets created by the ultimately activated services
594 for it. Conversely, the IP access list configured for the service is not applied to any sockets passed into
595 the service via socket activation. Thus, it is usually a good idea, to replicate the IP access lists on both
596 the socket and the service unit, however it often makes sense to maintain one list more open and the other
597 one more restricted, depending on the usecase.
</para>
599 <para>If these settings are used multiple times in the same unit the specified lists are combined. If an
600 empty string is assigned to these settings the specific access list is reset and all previous settings undone.
</para>
602 <para>In place of explicit IPv4 or IPv6 address and prefix length specifications a small set of symbolic
603 names may be used. The following names are defined:
</para>
606 <title>Special address/network names
</title>
609 <colspec colname='name'
/>
610 <colspec colname='definition'
/>
611 <colspec colname='meaning'
/>
615 <entry>Symbolic Name
</entry>
616 <entry>Definition
</entry>
617 <entry>Meaning
</entry>
623 <entry><constant>any
</constant></entry>
624 <entry>0.0.0.0/
0 ::/
0</entry>
625 <entry>Any host
</entry>
629 <entry><constant>localhost
</constant></entry>
630 <entry>127.0.0.0/
8 ::
1/
128</entry>
631 <entry>All addresses on the local loopback
</entry>
635 <entry><constant>link-local
</constant></entry>
636 <entry>169.254.0.0/
16 fe80::/
64</entry>
637 <entry>All link-local IP addresses
</entry>
641 <entry><constant>multicast
</constant></entry>
642 <entry>224.0.0.0/
4 ff00::/
8</entry>
643 <entry>All IP multicasting addresses
</entry>
649 <para>Note that these settings might not be supported on some systems (for example if eBPF control group
650 support is not enabled in the underlying kernel or container manager). These settings will have no effect in
651 that case. If compatibility with such systems is desired it is hence recommended to not exclusively rely on
652 them for IP security.
</para>
657 <term><varname>IPIngressFilterPath=
<replaceable>BPF_FS_PROGRAMM_PATH
</replaceable></varname></term>
658 <term><varname>IPEgressFilterPath=
<replaceable>BPF_FS_PROGRAMM_PATH
</replaceable></varname></term>
661 <para>Add custom network traffic filters implemented as BPF programs, applying to all IP packets
662 sent and received over
<constant>AF_INET
</constant> and
<constant>AF_INET6
</constant> sockets.
663 Takes an absolute path to a pinned BPF program in the BPF virtual filesystem (
<filename>/sys/fs/bpf/
</filename>).
666 <para>The filters configured with this option are applied to all sockets created by processes
667 of this unit (or in the case of socket units, associated with it). The filters are loaded in addition
668 to filters any of the parent slice units this unit might be a member of as well as any
669 <varname>IPAddressAllow=
</varname> and
<varname>IPAddressDeny=
</varname> filters in any of these units.
670 By default there are no filters specified.
</para>
672 <para>If these settings are used multiple times in the same unit all the specified programs are attached. If an
673 empty string is assigned to these settings the program list is reset and all previous specified programs ignored.
</para>
675 <para>Note that for socket-activated services, the IP filter programs configured on the socket unit apply to
676 all sockets associated with it directly, but not to any sockets created by the ultimately activated services
677 for it. Conversely, the IP filter programs configured for the service are not applied to any sockets passed into
678 the service via socket activation. Thus, it is usually a good idea, to replicate the IP filter programs on both
679 the socket and the service unit, however it often makes sense to maintain one configuration more open and the other
680 one more restricted, depending on the usecase.
</para>
682 <para>Note that these settings might not be supported on some systems (for example if eBPF control group
683 support is not enabled in the underlying kernel or container manager). These settings will fail the service in
684 that case. If compatibility with such systems is desired it is hence recommended to attach your filter manually
685 (requires
<varname>Delegate=
</varname><constant>yes
</constant>) instead of using this setting.
</para>
690 <term><varname>DeviceAllow=
</varname></term>
693 <para>Control access to specific device nodes by the executed processes. Takes two space-separated
694 strings: a device node specifier followed by a combination of
<constant>r
</constant>,
695 <constant>w
</constant>,
<constant>m
</constant> to control
<emphasis>r
</emphasis>eading,
696 <emphasis>w
</emphasis>riting, or creation of the specific device node(s) by the unit
697 (
<emphasis>m
</emphasis>knod), respectively. On cgroup-v1 this controls the
698 <literal>devices.allow
</literal> control group attribute. For details about this control group
699 attribute, see
<ulink
700 url=
"https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v1/devices.html">Device Whitelist Controller
</ulink>.
701 In the unified cgroup hierarchy this functionality is implemented using eBPF filtering.
</para>
703 <para>The device node specifier is either a path to a device node in the file system, starting with
704 <filename>/dev/
</filename>, or a string starting with either
<literal>char-
</literal> or
705 <literal>block-
</literal> followed by a device group name, as listed in
706 <filename>/proc/devices
</filename>. The latter is useful to whitelist all current and future
707 devices belonging to a specific device group at once. The device group is matched according to
708 filename globbing rules, you may hence use the
<literal>*
</literal> and
<literal>?
</literal>
709 wildcards. (Note that such globbing wildcards are not available for device node path
710 specifications!) In order to match device nodes by numeric major/minor, use device node paths in
711 the
<filename>/dev/char/
</filename> and
<filename>/dev/block/
</filename> directories. However,
712 matching devices by major/minor is generally not recommended as assignments are neither stable nor
713 portable between systems or different kernel versions.
</para>
715 <para>Examples:
<filename>/dev/sda5
</filename> is a path to a device node, referring to an ATA or
716 SCSI block device.
<literal>char-pts
</literal> and
<literal>char-alsa
</literal> are specifiers for
717 all pseudo TTYs and all ALSA sound devices, respectively.
<literal>char-cpu/*
</literal> is a
718 specifier matching all CPU related device groups.
</para>
720 <para>Note that whitelists defined this way should only reference device groups which are
721 resolvable at the time the unit is started. Any device groups not resolvable then are not added to
722 the device whitelist. In order to work around this limitation, consider extending service units
723 with a pair of
<command>After=modprobe@xyz.service
</command> and
724 <command>Wants=modprobe@xyz.service
</command> lines that load the necessary kernel module
725 implementing the device group if missing.
726 Example:
<programlisting>…
728 Wants=modprobe@loop.service
729 After=modprobe@loop.service
732 DeviceAllow=block-loop
733 DeviceAllow=/dev/loop-control
734 …
</programlisting></para>
740 <term><varname>DevicePolicy=auto|closed|strict
</varname></term>
744 Control the policy for allowing device access:
748 <term><option>strict
</option></term>
750 <para>means to only allow types of access that are
751 explicitly specified.
</para>
756 <term><option>closed
</option></term>
758 <para>in addition, allows access to standard pseudo
760 <filename>/dev/null
</filename>,
761 <filename>/dev/zero
</filename>,
762 <filename>/dev/full
</filename>,
763 <filename>/dev/random
</filename>, and
764 <filename>/dev/urandom
</filename>.
770 <term><option>auto
</option></term>
773 in addition, allows access to all devices if no
774 explicit
<varname>DeviceAllow=
</varname> is present.
784 <term><varname>Slice=
</varname></term>
787 <para>The name of the slice unit to place the unit
788 in. Defaults to
<filename>system.slice
</filename> for all
789 non-instantiated units of all unit types (except for slice
790 units themselves see below). Instance units are by default
791 placed in a subslice of
<filename>system.slice
</filename>
792 that is named after the template name.
</para>
794 <para>This option may be used to arrange systemd units in a
795 hierarchy of slices each of which might have resource
796 settings applied.
</para>
798 <para>For units of type slice, the only accepted value for
799 this setting is the parent slice. Since the name of a slice
800 unit implies the parent slice, it is hence redundant to ever
801 set this parameter directly for slice units.
</para>
803 <para>Special care should be taken when relying on the default slice assignment in templated service units
804 that have
<varname>DefaultDependencies=no
</varname> set, see
805 <citerefentry><refentrytitle>systemd.service
</refentrytitle><manvolnum>5</manvolnum></citerefentry>, section
806 "Default Dependencies" for details.
</para>
812 <term><varname>Delegate=
</varname></term>
815 <para>Turns on delegation of further resource control partitioning to processes of the unit. Units where this
816 is enabled may create and manage their own private subhierarchy of control groups below the control group of
817 the unit itself. For unprivileged services (i.e. those using the
<varname>User=
</varname> setting) the unit's
818 control group will be made accessible to the relevant user. When enabled the service manager will refrain
819 from manipulating control groups or moving processes below the unit's control group, so that a clear concept
820 of ownership is established: the control group tree above the unit's control group (i.e. towards the root
821 control group) is owned and managed by the service manager of the host, while the control group tree below
822 the unit's control group is owned and managed by the unit itself. Takes either a boolean argument or a list
823 of control group controller names. If true, delegation is turned on, and all supported controllers are
824 enabled for the unit, making them available to the unit's processes for management. If false, delegation is
825 turned off entirely (and no additional controllers are enabled). If set to a list of controllers, delegation
826 is turned on, and the specified controllers are enabled for the unit. Note that additional controllers than
827 the ones specified might be made available as well, depending on configuration of the containing slice unit
828 or other units contained in it. Note that assigning the empty string will enable delegation, but reset the
829 list of controllers, all assignments prior to this will have no effect. Defaults to false.
</para>
831 <para>Note that controller delegation to less privileged code is only safe on the unified control group
832 hierarchy. Accordingly, access to the specified controllers will not be granted to unprivileged services on
833 the legacy hierarchy, even when requested.
</para>
835 <para>The following controller names may be specified:
<option>cpu
</option>,
<option>cpuacct
</option>,
836 <option>io
</option>,
<option>blkio
</option>,
<option>memory
</option>,
<option>devices
</option>,
837 <option>pids
</option>. Not all of these controllers are available on all kernels however, and some are
838 specific to the unified hierarchy while others are specific to the legacy hierarchy. Also note that the
839 kernel might support further controllers, which aren't covered here yet as delegation is either not supported
840 at all for them or not defined cleanly.
</para>
842 <para>For further details on the delegation model consult
<ulink
843 url=
"https://systemd.io/CGROUP_DELEGATION">Control Group APIs and Delegation
</ulink>.
</para>
848 <term><varname>DisableControllers=
</varname></term>
851 <para>Disables controllers from being enabled for a unit's children. If a controller listed is already in use
852 in its subtree, the controller will be removed from the subtree. This can be used to avoid child units being
853 able to implicitly or explicitly enable a controller. Defaults to not disabling any controllers.
</para>
855 <para>It may not be possible to successfully disable a controller if the unit or any child of the unit in
856 question delegates controllers to its children, as any delegated subtree of the cgroup hierarchy is unmanaged
859 <para>Multiple controllers may be specified, separated by spaces. You may also pass
860 <varname>DisableControllers=
</varname> multiple times, in which case each new instance adds another controller
861 to disable. Passing
<varname>DisableControllers=
</varname> by itself with no controller name present resets
862 the disabled controller list.
</para>
864 <para>Valid controllers are
<option>cpu
</option>,
<option>cpuacct
</option>,
<option>io
</option>,
865 <option>blkio
</option>,
<option>memory
</option>,
<option>devices
</option>, and
<option>pids
</option>.
</para>
872 <title>Deprecated Options
</title>
874 <para>The following options are deprecated. Use the indicated superseding options instead:
</para>
876 <variablelist class='unit-directives'
>
879 <term><varname>CPUShares=
<replaceable>weight
</replaceable></varname></term>
880 <term><varname>StartupCPUShares=
<replaceable>weight
</replaceable></varname></term>
883 <para>Assign the specified CPU time share weight to the processes executed. These options take an integer
884 value and control the
<literal>cpu.shares
</literal> control group attribute. The allowed range is
2 to
885 262144. Defaults to
1024. For details about this control group attribute, see
<ulink
886 url=
"https://www.kernel.org/doc/html/latest/scheduler/sched-design-CFS.html">CFS Scheduler
</ulink>.
887 The available CPU time is split up among all units within one slice relative to their CPU time share
890 <para>While
<varname>StartupCPUShares=
</varname> only applies to the startup phase of the system,
891 <varname>CPUShares=
</varname> applies to normal runtime of the system, and if the former is not set also to
892 the startup phase. Using
<varname>StartupCPUShares=
</varname> allows prioritizing specific services at
893 boot-up differently than during normal runtime.
</para>
895 <para>Implies
<literal>CPUAccounting=yes
</literal>.
</para>
897 <para>These settings are deprecated. Use
<varname>CPUWeight=
</varname> and
898 <varname>StartupCPUWeight=
</varname> instead.
</para>
903 <term><varname>MemoryLimit=
<replaceable>bytes
</replaceable></varname></term>
906 <para>Specify the limit on maximum memory usage of the executed processes. The limit specifies how much
907 process and kernel memory can be used by tasks in this unit. Takes a memory size in bytes. If the value is
908 suffixed with K, M, G or T, the specified memory size is parsed as Kilobytes, Megabytes, Gigabytes, or
909 Terabytes (with the base
1024), respectively. Alternatively, a percentage value may be specified, which is
910 taken relative to the installed physical memory on the system. If assigned the special value
911 <literal>infinity
</literal>, no memory limit is applied. This controls the
912 <literal>memory.limit_in_bytes
</literal> control group attribute. For details about this control group
913 attribute, see
<ulink
914 url=
"https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v1/memory.html">Memory Resource Controller
</ulink>.
</para>
916 <para>Implies
<literal>MemoryAccounting=yes
</literal>.
</para>
918 <para>This setting is deprecated. Use
<varname>MemoryMax=
</varname> instead.
</para>
923 <term><varname>BlockIOAccounting=
</varname></term>
926 <para>Turn on Block I/O accounting for this unit, if the legacy control group hierarchy is used on the
927 system. Takes a boolean argument. Note that turning on block I/O accounting for one unit will also implicitly
928 turn it on for all units contained in the same slice and all for its parent slices and the units contained
929 therein. The system default for this setting may be controlled with
930 <varname>DefaultBlockIOAccounting=
</varname> in
931 <citerefentry><refentrytitle>systemd-system.conf
</refentrytitle><manvolnum>5</manvolnum></citerefentry>.
</para>
933 <para>This setting is deprecated. Use
<varname>IOAccounting=
</varname> instead.
</para>
938 <term><varname>BlockIOWeight=
<replaceable>weight
</replaceable></varname></term>
939 <term><varname>StartupBlockIOWeight=
<replaceable>weight
</replaceable></varname></term>
941 <listitem><para>Set the default overall block I/O weight for the executed processes, if the legacy control
942 group hierarchy is used on the system. Takes a single weight value (between
10 and
1000) to set the default
943 block I/O weight. This controls the
<literal>blkio.weight
</literal> control group attribute, which defaults to
944 500. For details about this control group attribute, see
<ulink
945 url=
"https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v1/blkio-controller.html">Block IO Controller
</ulink>.
946 The available I/O bandwidth is split up among all units within one slice relative to their block I/O
949 <para>While
<varname>StartupBlockIOWeight=
</varname> only
950 applies to the startup phase of the system,
951 <varname>BlockIOWeight=
</varname> applies to the later runtime
952 of the system, and if the former is not set also to the
953 startup phase. This allows prioritizing specific services at
954 boot-up differently than during runtime.
</para>
957 <literal>BlockIOAccounting=yes
</literal>.
</para>
959 <para>These settings are deprecated. Use
<varname>IOWeight=
</varname> and
<varname>StartupIOWeight=
</varname>
966 <term><varname>BlockIODeviceWeight=
<replaceable>device
</replaceable> <replaceable>weight
</replaceable></varname></term>
969 <para>Set the per-device overall block I/O weight for the executed processes, if the legacy control group
970 hierarchy is used on the system. Takes a space-separated pair of a file path and a weight value to specify
971 the device specific weight value, between
10 and
1000. (Example:
"/dev/sda 500"). The file path may be
972 specified as path to a block device node or as any other file, in which case the backing block device of the
973 file system of the file is determined. This controls the
<literal>blkio.weight_device
</literal> control group
974 attribute, which defaults to
1000. Use this option multiple times to set weights for multiple devices. For
975 details about this control group attribute, see
<ulink
976 url=
"https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v1/blkio-controller.html">Block IO Controller
</ulink>.
</para>
979 <literal>BlockIOAccounting=yes
</literal>.
</para>
981 <para>This setting is deprecated. Use
<varname>IODeviceWeight=
</varname> instead.
</para>
986 <term><varname>BlockIOReadBandwidth=
<replaceable>device
</replaceable> <replaceable>bytes
</replaceable></varname></term>
987 <term><varname>BlockIOWriteBandwidth=
<replaceable>device
</replaceable> <replaceable>bytes
</replaceable></varname></term>
990 <para>Set the per-device overall block I/O bandwidth limit for the executed processes, if the legacy control
991 group hierarchy is used on the system. Takes a space-separated pair of a file path and a bandwidth value (in
992 bytes per second) to specify the device specific bandwidth. The file path may be a path to a block device
993 node, or as any other file in which case the backing block device of the file system of the file is used. If
994 the bandwidth is suffixed with K, M, G, or T, the specified bandwidth is parsed as Kilobytes, Megabytes,
995 Gigabytes, or Terabytes, respectively, to the base of
1000. (Example:
996 "/dev/disk/by-path/pci-0000:00:1f.2-scsi-0:0:0:0 5M"). This controls the
997 <literal>blkio.throttle.read_bps_device
</literal> and
<literal>blkio.throttle.write_bps_device
</literal>
998 control group attributes. Use this option multiple times to set bandwidth limits for multiple devices. For
999 details about these control group attributes, see
<ulink
1000 url=
"https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v1/blkio-controller.html">Block IO Controller
</ulink>.
1004 <literal>BlockIOAccounting=yes
</literal>.
</para>
1006 <para>These settings are deprecated. Use
<varname>IOReadBandwidthMax=
</varname> and
1007 <varname>IOWriteBandwidthMax=
</varname> instead.
</para>
1015 <title>See Also
</title>
1017 <citerefentry><refentrytitle>systemd
</refentrytitle><manvolnum>1</manvolnum></citerefentry>,
1018 <citerefentry><refentrytitle>systemd-system.conf
</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
1019 <citerefentry><refentrytitle>systemd.unit
</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
1020 <citerefentry><refentrytitle>systemd.service
</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
1021 <citerefentry><refentrytitle>systemd.slice
</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
1022 <citerefentry><refentrytitle>systemd.scope
</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
1023 <citerefentry><refentrytitle>systemd.socket
</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
1024 <citerefentry><refentrytitle>systemd.mount
</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
1025 <citerefentry><refentrytitle>systemd.swap
</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
1026 <citerefentry><refentrytitle>systemd.exec
</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
1027 <citerefentry><refentrytitle>systemd.directives
</refentrytitle><manvolnum>7</manvolnum></citerefentry>,
1028 <citerefentry><refentrytitle>systemd.special
</refentrytitle><manvolnum>7</manvolnum></citerefentry>,
1029 The documentation for control groups and specific controllers in the Linux kernel:
1030 <ulink url=
"https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v2.html">Control Groups v2
</ulink>.