man/systemd.resource-control.xml

   1 <?xml version='1.0'?> <!--*- Mode: nxml; nxml-child-indent: 2; indent-tabs-mode: nil -*-->
   2 <!DOCTYPE refentry PUBLIC "-//OASIS//DTD DocBook XML V4.2//EN"
   3 "http://www.oasis-open.org/docbook/xml/4.2/docbookx.dtd">
   4
   5 <!--
   6   This file is part of systemd.
   7
   8   Copyright 2013 Zbigniew Jędrzejewski-Szmek
   9
  10   systemd is free software; you can redistribute it and/or modify it
  11   under the terms of the GNU Lesser General Public License as published by
  12   the Free Software Foundation; either version 2.1 of the License, or
  13   (at your option) any later version.
  14
  15   systemd is distributed in the hope that it will be useful, but
  16   WITHOUT ANY WARRANTY; without even the implied warranty of
  17   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  18   Lesser General Public License for more details.
  19
  20   You should have received a copy of the GNU Lesser General Public License
  21   along with systemd; If not, see <http://www.gnu.org/licenses/>.
  22 -->
  23
  24 <refentry id="systemd.resource-control">
  25   <refentryinfo>
  26     <title>systemd.resource-control</title>
  27     <productname>systemd</productname>
  28
  29     <authorgroup>
  30       <author>
  31         <contrib>Developer</contrib>
  32         <firstname>Lennart</firstname>
  33         <surname>Poettering</surname>
  34         <email>lennart@poettering.net</email>
  35       </author>
  36     </authorgroup>
  37   </refentryinfo>
  38
  39   <refmeta>
  40     <refentrytitle>systemd.resource-control</refentrytitle>
  41     <manvolnum>5</manvolnum>
  42   </refmeta>
  43
  44   <refnamediv>
  45     <refname>systemd.resource-control</refname>
  46     <refpurpose>Resource control unit settings</refpurpose>
  47   </refnamediv>
  48
  49   <refsynopsisdiv>
  50     <para>
  51       <filename><replaceable>slice</replaceable>.slice</filename>,
  52       <filename><replaceable>scope</replaceable>.scope</filename>,
  53       <filename><replaceable>service</replaceable>.service</filename>,
  54       <filename><replaceable>socket</replaceable>.socket</filename>,
  55       <filename><replaceable>mount</replaceable>.mount</filename>,
  56       <filename><replaceable>swap</replaceable>.swap</filename>
  57     </para>
  58   </refsynopsisdiv>
  59
  60   <refsect1>
  61     <title>Description</title>
  62
  63     <para>Unit configuration files for services, slices, scopes,
  64     sockets, mount points, and swap devices share a subset of
  65     configuration options for resource control of spawned
  66     processes. Internally, this relies on the Control Groups
  67     kernel concept for organizing processes in a hierarchical tree of
  68     named groups for the purpose of resource management.</para>
  69
  70     <para>This man page lists the configuration options shared by
  71     those six unit types. See
  72     <citerefentry><refentrytitle>systemd.unit</refentrytitle><manvolnum>5</manvolnum></citerefentry>
  73     for the common options of all unit configuration files, and
  74     <citerefentry><refentrytitle>systemd.slice</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
  75     <citerefentry><refentrytitle>systemd.scope</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
  76     <citerefentry><refentrytitle>systemd.service</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
  77     <citerefentry><refentrytitle>systemd.socket</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
  78     <citerefentry><refentrytitle>systemd.mount</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
  79     and
  80     <citerefentry><refentrytitle>systemd.swap</refentrytitle><manvolnum>5</manvolnum></citerefentry>
  81     for more information on the specific unit configuration files. The
  82     resource control configuration options are configured in the
  83     [Slice], [Scope], [Service], [Socket], [Mount], or [Swap]
  84     sections, depending on the unit type.</para>
  85
  86     <para>See the <ulink
  87     url="http://www.freedesktop.org/wiki/Software/systemd/ControlGroupInterface/">New
  88     Control Group Interfaces</ulink> for an introduction on how to make
  89     use of resource control APIs from programs.</para>
  90   </refsect1>
  91
  92   <refsect1>
  93     <title>Automatic Dependencies</title>
  94
  95     <para>Units with the <varname>Slice=</varname> setting set automatically acquire <varname>Requires=</varname> and
  96     <varname>After=</varname> dependencies on the specified slice unit.</para>
  97   </refsect1>
  98
  99   <refsect1>
 100     <title>Unified and Legacy Control Group Hierarchies</title>
 101
 102     <para>The unified control group hierarchy is the new version of kernel control group interface, see <ulink
 103     url="https://www.kernel.org/doc/Documentation/cgroup-v2.txt">cgroup-v2.txt</ulink>. Depending on the resource type,
 104     there are differences in resource control capabilities.  Also, because of interface changes, some resource types
 105     have separate set of options on the unified hierarchy.</para>
 106
 107     <para>
 108       <variablelist>
 109
 110         <varlistentry>
 111           <term><option>CPU</option></term>
 112           <listitem>
 113             <para>Due to the lack of consensus in the kernel community, the CPU controller support on the unified
 114             cgroup hierarchy requires out-of-tree kernel patches. See <ulink
 115             url="https://git.kernel.org/cgit/linux/kernel/git/tj/cgroup.git/tree/Documentation/cgroup-v2-cpu.txt?h=cgroup-v2-cpu">cgroup-v2-cpu.txt</ulink>.</para>
 116
 117             <para><varname>CPUWeight=</varname> and <varname>StartupCPUWeight=</varname> replace
 118             <varname>CPUShares=</varname> and <varname>StartupCPUShares=</varname>, respectively.</para>
 119
 120             <para>The <literal>cpuacct</literal> controller does not exist separately on the unified hierarchy.</para>
 121           </listitem>
 122         </varlistentry>
 123
 124         <varlistentry>
 125           <term><option>Memory</option></term>
 126           <listitem>
 127             <para><varname>MemoryMax=</varname> replaces <varname>MemoryLimit=</varname>. <varname>MemoryLow=</varname>
 128             and <varname>MemoryHigh=</varname> are effective only on unified hierarchy.</para>
 129           </listitem>
 130         </varlistentry>
 131
 132         <varlistentry>
 133           <term><option>IO</option></term>
 134           <listitem>
 135             <para><varname>IO</varname> prefixed settings are superset of and replace <varname>BlockIO</varname>
 136             prefixed ones. On unified hierarchy, IO resource control also applies to buffered writes.</para>
 137           </listitem>
 138         </varlistentry>
 139
 140       </variablelist>
 141     </para>
 142
 143     <para>To ease the transition, there is best-effort translation between the two versions of settings. If all
 144     settings of a unit for a given resource type are for the other hierarchy type, the settings are translated and
 145     applied. If there are any valid settings for the hierarchy in use, all translations are disabled for the resource
 146     type. Mixing the two types of settings on a unit can lead to confusing results.</para>
 147
 148     <para>Legacy control group hierarchy (see <ulink
 149     url="https://www.kernel.org/doc/Documentation/cgroup-v1/cgroups.txt">cgroups.txt</ulink>), also called cgroup-v1,
 150     doesn't allow safe delegation of controllers to unprivileged processes. If the system uses the legacy control group
 151     hierarchy, resource control is disabled for systemd user instance, see
 152     <citerefentry><refentrytitle>systemd</refentrytitle><manvolnum>1</manvolnum></citerefentry>.
 153     </para>
 154   </refsect1>
 155
 156   <refsect1>
 157     <title>Options</title>
 158
 159     <para>Units of the types listed above can have settings
 160     for resource control configuration:</para>
 161
 162     <variablelist class='unit-directives'>
 163
 164       <varlistentry>
 165         <term><varname>CPUAccounting=</varname></term>
 166
 167         <listitem>
 168           <para>Turn on CPU usage accounting for this unit. Takes a
 169           boolean argument. Note that turning on CPU accounting for
 170           one unit will also implicitly turn it on for all units
 171           contained in the same slice and for all its parent slices
 172           and the units contained therein. The system default for this
 173           setting may be controlled with
 174           <varname>DefaultCPUAccounting=</varname> in
 175           <citerefentry><refentrytitle>systemd-system.conf</refentrytitle><manvolnum>5</manvolnum></citerefentry>.</para>
 176         </listitem>
 177       </varlistentry>
 178
 179       <varlistentry>
 180         <term><varname>CPUWeight=<replaceable>weight</replaceable></varname></term>
 181         <term><varname>StartupCPUWeight=<replaceable>weight</replaceable></varname></term>
 182
 183         <listitem>
 184           <para>Assign the specified CPU time weight to the processes executed, if the unified control group hierarchy
 185           is used on the system. These options take an integer value and control the <literal>cpu.weight</literal>
 186           control group attribute. The allowed range is 1 to 10000. Defaults to 100. For details about this control
 187           group attribute, see <ulink
 188           url="https://www.kernel.org/doc/Documentation/cgroup-v2.txt">cgroup-v2.txt</ulink> and <ulink
 189           url="https://www.kernel.org/doc/Documentation/scheduler/sched-design-CFS.txt">sched-design-CFS.txt</ulink>.
 190           The available CPU time is split up among all units within one slice relative to their CPU time weight.</para>
 191
 192           <para>While <varname>StartupCPUWeight=</varname> only applies to the startup phase of the system,
 193           <varname>CPUWeight=</varname> applies to normal runtime of the system, and if the former is not set also to
 194           the startup phase. Using <varname>StartupCPUWeight=</varname> allows prioritizing specific services at
 195           boot-up differently than during normal runtime.</para>
 196
 197           <para>Implies <literal>CPUAccounting=true</literal>.</para>
 198
 199           <para>These settings are supported only if the unified control group hierarchy is used.</para>
 200         </listitem>
 201       </varlistentry>
 202
 203       <varlistentry>
 204         <term><varname>CPUShares=<replaceable>weight</replaceable></varname></term>
 205         <term><varname>StartupCPUShares=<replaceable>weight</replaceable></varname></term>
 206
 207         <listitem>
 208           <para>Assign the specified CPU time share weight to the processes executed. These options take an integer
 209           value and control the <literal>cpu.shares</literal> control group attribute. The allowed range is 2 to
 210           262144. Defaults to 1024. For details about this control group attribute, see <ulink
 211           url="https://www.kernel.org/doc/Documentation/scheduler/sched-design-CFS.txt">sched-design-CFS.txt</ulink>.
 212           The available CPU time is split up among all units within one slice relative to their CPU time share
 213           weight.</para>
 214
 215           <para>While <varname>StartupCPUShares=</varname> only applies to the startup phase of the system,
 216           <varname>CPUShares=</varname> applies to normal runtime of the system, and if the former is not set also to
 217           the startup phase. Using <varname>StartupCPUShares=</varname> allows prioritizing specific services at
 218           boot-up differently than during normal runtime.</para>
 219
 220           <para>Implies <literal>CPUAccounting=true</literal>.</para>
 221
 222           <para>These settings are supported only if the legacy control group hierarchy is used.</para>
 223         </listitem>
 224       </varlistentry>
 225
 226       <varlistentry>
 227         <term><varname>CPUQuota=</varname></term>
 228
 229         <listitem>
 230           <para>Assign the specified CPU time quota to the processes executed. Takes a percentage value, suffixed with
 231           "%". The percentage specifies how much CPU time the unit shall get at maximum, relative to the total CPU time
 232           available on one CPU. Use values &gt; 100% for allotting CPU time on more than one CPU. This controls the
 233           <literal>cpu.max</literal> attribute on the unified control group hierarchy and
 234           <literal>cpu.cfs_quota_us</literal> on legacy. For details about these control group attributes, see <ulink
 235           url="https://www.kernel.org/doc/Documentation/cgroup-v2.txt">cgroup-v2.txt</ulink> and <ulink
 236           url="https://www.kernel.org/doc/Documentation/scheduler/sched-design-CFS.txt">sched-design-CFS.txt</ulink>.</para>
 237
 238           <para>Example: <varname>CPUQuota=20%</varname> ensures that the executed processes will never get more than
 239           20% CPU time on one CPU.</para>
 240
 241           <para>Implies <literal>CPUAccounting=true</literal>.</para>
 242
 243           <para>This setting is supported on both unified and legacy control group hierarchies.</para>
 244         </listitem>
 245       </varlistentry>
 246
 247       <varlistentry>
 248         <term><varname>MemoryAccounting=</varname></term>
 249
 250         <listitem>
 251           <para>Turn on process and kernel memory accounting for this
 252           unit. Takes a boolean argument. Note that turning on memory
 253           accounting for one unit will also implicitly turn it on for
 254           all units contained in the same slice and for all its parent
 255           slices and the units contained therein. The system default
 256           for this setting may be controlled with
 257           <varname>DefaultMemoryAccounting=</varname> in
 258           <citerefentry><refentrytitle>systemd-system.conf</refentrytitle><manvolnum>5</manvolnum></citerefentry>.</para>
 259         </listitem>
 260       </varlistentry>
 261
 262       <varlistentry>
 263         <term><varname>MemoryLow=<replaceable>bytes</replaceable></varname></term>
 264
 265         <listitem>
 266           <para>Specify the best-effort memory usage protection of the executed processes in this unit. If the memory
 267           usages of this unit and all its ancestors are below their low boundaries, this unit's memory won't be
 268           reclaimed as long as memory can be reclaimed from unprotected units.</para>
 269
 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. This controls the <literal>memory.low</literal> control group attribute. For details about this
 274           control group attribute, see <ulink
 275           url="https://www.kernel.org/doc/Documentation/cgroup-v2.txt">cgroup-v2.txt</ulink>.</para>
 276
 277           <para>Implies <literal>MemoryAccounting=true</literal>.</para>
 278
 279           <para>This setting is supported only if the unified control group hierarchy is used.</para>
 280         </listitem>
 281       </varlistentry>
 282
 283       <varlistentry>
 284         <term><varname>MemoryHigh=<replaceable>bytes</replaceable></varname></term>
 285
 286         <listitem>
 287           <para>Specify the high limit on memory usage of the executed processes in this unit. Memory usage may go
 288           above the limit if unavoidable, but the processes are heavily slowed down and memory is taken away
 289           aggressively in such cases. This is the main mechanism to control memory usage of a unit.</para>
 290
 291           <para>Takes a memory size in bytes. If the value is suffixed with K, M, G or T, the specified memory size is
 292           parsed as Kilobytes, Megabytes, Gigabytes, or Terabytes (with the base 1024), respectively. Alternatively, a
 293           percentage value may be specified, which is taken relative to the installed physical memory on the
 294           system. If assigned the
 295           special value <literal>infinity</literal>, no memory limit is applied. This controls the
 296           <literal>memory.high</literal> control group attribute. For details about this control group attribute, see
 297           <ulink url="https://www.kernel.org/doc/Documentation/cgroup-v2.txt">cgroup-v2.txt</ulink>.</para>
 298
 299           <para>Implies <literal>MemoryAccounting=true</literal>.</para>
 300
 301           <para>This setting is supported only if the unified control group hierarchy is used.</para>
 302         </listitem>
 303       </varlistentry>
 304
 305       <varlistentry>
 306         <term><varname>MemoryMax=<replaceable>bytes</replaceable></varname></term>
 307
 308         <listitem>
 309           <para>Specify the absolute limit on memory usage of the executed processes in this unit. If memory usage
 310           cannot be contained under the limit, out-of-memory killer is invoked inside the unit. It is recommended to
 311           use <varname>MemoryHigh=</varname> as the main control mechanism and use <varname>MemoryMax=</varname> as the
 312           last line of defense.</para>
 313
 314           <para>Takes a memory size in bytes. If the value is suffixed with K, M, G or T, the specified memory size is
 315           parsed as Kilobytes, Megabytes, Gigabytes, or Terabytes (with the base 1024), respectively. Alternatively, a
 316           percentage value may be specified, which is taken relative to the installed physical memory on the system. If
 317           assigned the special value <literal>infinity</literal>, no memory limit is applied. This controls the
 318           <literal>memory.max</literal> control group attribute. For details about this control group attribute, see
 319           <ulink url="https://www.kernel.org/doc/Documentation/cgroup-v2.txt">cgroup-v2.txt</ulink>.</para>
 320
 321           <para>Implies <literal>MemoryAccounting=true</literal>.</para>
 322
 323           <para>This setting is supported only if the unified control group hierarchy is used. Use
 324           <varname>MemoryLimit=</varname> on systems using the legacy control group hierarchy.</para>
 325         </listitem>
 326       </varlistentry>
 327
 328       <varlistentry>
 329         <term><varname>MemoryLimit=<replaceable>bytes</replaceable></varname></term>
 330
 331         <listitem>
 332           <para>Specify the limit on maximum memory usage of the executed processes. The limit specifies how much
 333           process and kernel memory can be used by tasks in this unit. Takes a memory size in bytes. If the value is
 334           suffixed with K, M, G or T, the specified memory size is parsed as Kilobytes, Megabytes, Gigabytes, or
 335           Terabytes (with the base 1024), respectively. Alternatively, a percentage value may be specified, which is
 336           taken relative to the installed physical memory on the system. If assigned the special value
 337           <literal>infinity</literal>, no memory limit is applied. This controls the
 338           <literal>memory.limit_in_bytes</literal> control group attribute. For details about this control group
 339           attribute, see <ulink
 340           url="https://www.kernel.org/doc/Documentation/cgroup-v1/memory.txt">memory.txt</ulink>.</para>
 341
 342           <para>Implies <literal>MemoryAccounting=true</literal>.</para>
 343
 344           <para>This setting is supported only if the legacy control group hierarchy is used. Use
 345           <varname>MemoryMax=</varname> on systems using the unified control group hierarchy.</para>
 346         </listitem>
 347       </varlistentry>
 348
 349       <varlistentry>
 350         <term><varname>TasksAccounting=</varname></term>
 351
 352         <listitem>
 353           <para>Turn on task accounting for this unit. Takes a
 354           boolean argument. If enabled, the system manager will keep
 355           track of the number of tasks in the unit. The number of
 356           tasks accounted this way includes both kernel threads and
 357           userspace processes, with each thread counting
 358           individually. Note that turning on tasks accounting for one
 359           unit will also implicitly turn it on for all units contained
 360           in the same slice and for all its parent slices and the
 361           units contained therein. The system default for this setting
 362           may be controlled with
 363           <varname>DefaultTasksAccounting=</varname> in
 364           <citerefentry><refentrytitle>systemd-system.conf</refentrytitle><manvolnum>5</manvolnum></citerefentry>.</para>
 365         </listitem>
 366       </varlistentry>
 367
 368       <varlistentry>
 369         <term><varname>TasksMax=<replaceable>N</replaceable></varname></term>
 370
 371         <listitem>
 372           <para>Specify the maximum number of tasks that may be created in the unit. This ensures that the number of
 373           tasks accounted for the unit (see above) stays below a specific limit. This either takes an absolute number
 374           of tasks or a percentage value that is taken relative to the configured maximum number of tasks on the
 375           system.  If assigned the special value <literal>infinity</literal>, no tasks limit is applied. This controls
 376           the <literal>pids.max</literal> control group attribute. For details about this control group attribute, see
 377           <ulink url="https://www.kernel.org/doc/Documentation/cgroup-v1/pids.txt">pids.txt</ulink>.</para>
 378
 379           <para>Implies <literal>TasksAccounting=true</literal>. The
 380           system default for this setting may be controlled with
 381           <varname>DefaultTasksMax=</varname> in
 382           <citerefentry><refentrytitle>systemd-system.conf</refentrytitle><manvolnum>5</manvolnum></citerefentry>.</para>
 383         </listitem>
 384       </varlistentry>
 385
 386       <varlistentry>
 387         <term><varname>IOAccounting=</varname></term>
 388
 389         <listitem>
 390           <para>Turn on Block I/O accounting for this unit, if the unified control group hierarchy is used on the
 391           system. Takes a boolean argument. Note that turning on block I/O accounting for one unit will also implicitly
 392           turn it on for all units contained in the same slice and all for its parent slices and the units contained
 393           therein. The system default for this setting may be controlled with <varname>DefaultIOAccounting=</varname>
 394           in
 395           <citerefentry><refentrytitle>systemd-system.conf</refentrytitle><manvolnum>5</manvolnum></citerefentry>.</para>
 396
 397           <para>This setting is supported only if the unified control group hierarchy is used. Use
 398           <varname>BlockIOAccounting=</varname> on systems using the legacy control group hierarchy.</para>
 399         </listitem>
 400       </varlistentry>
 401
 402       <varlistentry>
 403         <term><varname>IOWeight=<replaceable>weight</replaceable></varname></term>
 404         <term><varname>StartupIOWeight=<replaceable>weight</replaceable></varname></term>
 405
 406         <listitem>
 407           <para>Set the default overall block I/O weight for the executed processes, if the unified control group
 408           hierarchy is used on the system. Takes a single weight value (between 1 and 10000) to set the default block
 409           I/O weight. This controls the <literal>io.weight</literal> control group attribute, which defaults to
 410           100. For details about this control group attribute, see <ulink
 411           url="https://www.kernel.org/doc/Documentation/cgroup-v2.txt">cgroup-v2.txt</ulink>.  The available I/O
 412           bandwidth is split up among all units within one slice relative to their block I/O weight.</para>
 413
 414           <para>While <varname>StartupIOWeight=</varname> only applies
 415           to the startup phase of the system,
 416           <varname>IOWeight=</varname> applies to the later runtime of
 417           the system, and if the former is not set also to the startup
 418           phase. This allows prioritizing specific services at boot-up
 419           differently than during runtime.</para>
 420
 421           <para>Implies <literal>IOAccounting=true</literal>.</para>
 422
 423           <para>This setting is supported only if the unified control group hierarchy is used. Use
 424           <varname>BlockIOWeight=</varname> and <varname>StartupBlockIOWeight=</varname> on systems using the legacy
 425           control group hierarchy.</para>
 426         </listitem>
 427       </varlistentry>
 428
 429       <varlistentry>
 430         <term><varname>IODeviceWeight=<replaceable>device</replaceable> <replaceable>weight</replaceable></varname></term>
 431
 432         <listitem>
 433           <para>Set the per-device overall block I/O weight for the executed processes, if the unified control group
 434           hierarchy is used on the system. Takes a space-separated pair of a file path and a weight value to specify
 435           the device specific weight value, between 1 and 10000. (Example: "/dev/sda 1000"). The file path may be
 436           specified as path to a block device node or as any other file, in which case the backing block device of the
 437           file system of the file is determined. This controls the <literal>io.weight</literal> control group
 438           attribute, which defaults to 100. Use this option multiple times to set weights for multiple devices. For
 439           details about this control group attribute, see <ulink
 440           url="https://www.kernel.org/doc/Documentation/cgroup-v2.txt">cgroup-v2.txt</ulink>.</para>
 441
 442           <para>Implies <literal>IOAccounting=true</literal>.</para>
 443
 444           <para>This setting is supported only if the unified control group hierarchy is used. Use
 445           <varname>BlockIODeviceWeight=</varname> on systems using the legacy control group hierarchy.</para>
 446         </listitem>
 447       </varlistentry>
 448
 449       <varlistentry>
 450         <term><varname>IOReadBandwidthMax=<replaceable>device</replaceable> <replaceable>bytes</replaceable></varname></term>
 451         <term><varname>IOWriteBandwidthMax=<replaceable>device</replaceable> <replaceable>bytes</replaceable></varname></term>
 452
 453         <listitem>
 454           <para>Set the per-device overall block I/O bandwidth maximum limit for the executed processes, if the unified
 455           control group hierarchy is used on the system. This limit is not work-conserving and the executed processes
 456           are not allowed to use more even if the device has idle capacity.  Takes a space-separated pair of a file
 457           path and a bandwidth value (in bytes per second) to specify the device specific bandwidth. The file path may
 458           be a path to a block device node, or as any other file in which case the backing block device of the file
 459           system of the file is used. If the bandwidth is suffixed with K, M, G, or T, the specified bandwidth is
 460           parsed as Kilobytes, Megabytes, Gigabytes, or Terabytes, respectively, to the base of 1000. (Example:
 461           "/dev/disk/by-path/pci-0000:00:1f.2-scsi-0:0:0:0 5M"). This controls the <literal>io.max</literal> control
 462           group attributes. Use this option multiple times to set bandwidth limits for multiple devices. For details
 463           about this control group attribute, see <ulink
 464           url="https://www.kernel.org/doc/Documentation/cgroup-v2.txt">cgroup-v2.txt</ulink>.
 465           </para>
 466
 467           <para>Implies <literal>IOAccounting=true</literal>.</para>
 468
 469           <para>This setting is supported only if the unified control group hierarchy is used. Use
 470           <varname>BlockIOAccounting=</varname> on systems using the legacy control group hierarchy.</para>
 471         </listitem>
 472       </varlistentry>
 473
 474       <varlistentry>
 475         <term><varname>IOReadIOPSMax=<replaceable>device</replaceable> <replaceable>IOPS</replaceable></varname></term>
 476         <term><varname>IOWriteIOPSMax=<replaceable>device</replaceable> <replaceable>IOPS</replaceable></varname></term>
 477
 478         <listitem>
 479           <para>Set the per-device overall block I/O IOs-Per-Second maximum limit for the executed processes, if the
 480           unified control group hierarchy is used on the system. This limit is not work-conserving and the executed
 481           processes are not allowed to use more even if the device has idle capacity.  Takes a space-separated pair of
 482           a file path and an IOPS value to specify the device specific IOPS. The file path may be a path to a block
 483           device node, or as any other file in which case the backing block device of the file system of the file is
 484           used. If the IOPS is suffixed with K, M, G, or T, the specified IOPS is parsed as KiloIOPS, MegaIOPS,
 485           GigaIOPS, or TeraIOPS, respectively, to the base of 1000. (Example:
 486           "/dev/disk/by-path/pci-0000:00:1f.2-scsi-0:0:0:0 1K"). This controls the <literal>io.max</literal> control
 487           group attributes. Use this option multiple times to set IOPS limits for multiple devices. For details about
 488           this control group attribute, see <ulink
 489           url="https://www.kernel.org/doc/Documentation/cgroup-v2.txt">cgroup-v2.txt</ulink>.
 490           </para>
 491
 492           <para>Implies <literal>IOAccounting=true</literal>.</para>
 493
 494           <para>This setting is supported only if the unified control group hierarchy is used.</para>
 495         </listitem>
 496       </varlistentry>
 497
 498       <varlistentry>
 499         <term><varname>BlockIOAccounting=</varname></term>
 500
 501         <listitem>
 502           <para>Turn on Block I/O accounting for this unit, if the legacy control group hierarchy is used on the
 503           system. Takes a boolean argument. Note that turning on block I/O accounting for one unit will also implicitly
 504           turn it on for all units contained in the same slice and all for its parent slices and the units contained
 505           therein. The system default for this setting may be controlled with
 506           <varname>DefaultBlockIOAccounting=</varname> in
 507           <citerefentry><refentrytitle>systemd-system.conf</refentrytitle><manvolnum>5</manvolnum></citerefentry>.</para>
 508
 509           <para>This setting is supported only if the legacy control group hierarchy is used. Use
 510           <varname>IOAccounting=</varname> on systems using the unified control group hierarchy.</para>
 511         </listitem>
 512       </varlistentry>
 513
 514       <varlistentry>
 515         <term><varname>BlockIOWeight=<replaceable>weight</replaceable></varname></term>
 516         <term><varname>StartupBlockIOWeight=<replaceable>weight</replaceable></varname></term>
 517
 518         <listitem><para>Set the default overall block I/O weight for the executed processes, if the legacy control
 519         group hierarchy is used on the system. Takes a single weight value (between 10 and 1000) to set the default
 520         block I/O weight. This controls the <literal>blkio.weight</literal> control group attribute, which defaults to
 521         500. For details about this control group attribute, see <ulink
 522         url="https://www.kernel.org/doc/Documentation/cgroup-v1/blkio-controller.txt">blkio-controller.txt</ulink>.
 523         The available I/O bandwidth is split up among all units within one slice relative to their block I/O
 524         weight.</para>
 525
 526         <para>While <varname>StartupBlockIOWeight=</varname> only
 527         applies to the startup phase of the system,
 528         <varname>BlockIOWeight=</varname> applies to the later runtime
 529         of the system, and if the former is not set also to the
 530         startup phase. This allows prioritizing specific services at
 531         boot-up differently than during runtime.</para>
 532
 533         <para>Implies
 534         <literal>BlockIOAccounting=true</literal>.</para>
 535
 536         <para>This setting is supported only if the legacy control group hierarchy is used. Use
 537         <varname>IOWeight=</varname> and <varname>StartupIOWeight=</varname> on systems using the unified control group
 538         hierarchy.</para>
 539
 540       </listitem>
 541       </varlistentry>
 542
 543       <varlistentry>
 544         <term><varname>BlockIODeviceWeight=<replaceable>device</replaceable> <replaceable>weight</replaceable></varname></term>
 545
 546         <listitem>
 547           <para>Set the per-device overall block I/O weight for the executed processes, if the legacy control group
 548           hierarchy is used on the system. Takes a space-separated pair of a file path and a weight value to specify
 549           the device specific weight value, between 10 and 1000. (Example: "/dev/sda 500"). The file path may be
 550           specified as path to a block device node or as any other file, in which case the backing block device of the
 551           file system of the file is determined. This controls the <literal>blkio.weight_device</literal> control group
 552           attribute, which defaults to 1000. Use this option multiple times to set weights for multiple devices. For
 553           details about this control group attribute, see <ulink
 554           url="https://www.kernel.org/doc/Documentation/cgroup-v1/blkio-controller.txt">blkio-controller.txt</ulink>.</para>
 555
 556           <para>Implies
 557           <literal>BlockIOAccounting=true</literal>.</para>
 558
 559           <para>This setting is supported only if the legacy control group hierarchy is used. Use
 560           <varname>IODeviceWeight=</varname> on systems using the unified control group hierarchy.</para>
 561         </listitem>
 562       </varlistentry>
 563
 564       <varlistentry>
 565         <term><varname>BlockIOReadBandwidth=<replaceable>device</replaceable> <replaceable>bytes</replaceable></varname></term>
 566         <term><varname>BlockIOWriteBandwidth=<replaceable>device</replaceable> <replaceable>bytes</replaceable></varname></term>
 567
 568         <listitem>
 569           <para>Set the per-device overall block I/O bandwidth limit for the executed processes, if the legacy control
 570           group hierarchy is used on the system. Takes a space-separated pair of a file path and a bandwidth value (in
 571           bytes per second) to specify the device specific bandwidth. The file path may be a path to a block device
 572           node, or as any other file in which case the backing block device of the file system of the file is used. If
 573           the bandwidth is suffixed with K, M, G, or T, the specified bandwidth is parsed as Kilobytes, Megabytes,
 574           Gigabytes, or Terabytes, respectively, to the base of 1000. (Example:
 575           "/dev/disk/by-path/pci-0000:00:1f.2-scsi-0:0:0:0 5M"). This controls the
 576           <literal>blkio.throttle.read_bps_device</literal> and <literal>blkio.throttle.write_bps_device</literal>
 577           control group attributes. Use this option multiple times to set bandwidth limits for multiple devices. For
 578           details about these control group attributes, see <ulink
 579           url="https://www.kernel.org/doc/Documentation/cgroup-v1/blkio-controller.txt">blkio-controller.txt</ulink>.
 580           </para>
 581
 582           <para>Implies
 583           <literal>BlockIOAccounting=true</literal>.</para>
 584
 585           <para>This setting is supported only if the legacy control group hierarchy is used. Use
 586           <varname>IOReadBandwidthMax=</varname> and <varname>IOWriteBandwidthMax=</varname> on systems using the
 587           unified control group hierarchy.</para>
 588         </listitem>
 589       </varlistentry>
 590
 591       <varlistentry>
 592         <term><varname>DeviceAllow=</varname></term>
 593
 594         <listitem>
 595           <para>Control access to specific device nodes by the
 596           executed processes. Takes two space-separated strings: a
 597           device node specifier followed by a combination of
 598           <constant>r</constant>, <constant>w</constant>,
 599           <constant>m</constant> to control
 600           <emphasis>r</emphasis>eading, <emphasis>w</emphasis>riting,
 601           or creation of the specific device node(s) by the unit
 602           (<emphasis>m</emphasis>knod), respectively. This controls
 603           the <literal>devices.allow</literal> and
 604           <literal>devices.deny</literal> control group
 605           attributes. For details about these control group
 606           attributes, see <ulink
 607           url="https://www.kernel.org/doc/Documentation/cgroup-v1/devices.txt">devices.txt</ulink>.</para>
 608
 609           <para>The device node specifier is either a path to a device
 610           node in the file system, starting with
 611           <filename>/dev/</filename>, or a string starting with either
 612           <literal>char-</literal> or <literal>block-</literal>
 613           followed by a device group name, as listed in
 614           <filename>/proc/devices</filename>. The latter is useful to
 615           whitelist all current and future devices belonging to a
 616           specific device group at once. The device group is matched
 617           according to file name globbing rules, you may hence use the
 618           <literal>*</literal> and <literal>?</literal>
 619           wildcards. Examples: <filename>/dev/sda5</filename> is a
 620           path to a device node, referring to an ATA or SCSI block
 621           device. <literal>char-pts</literal> and
 622           <literal>char-alsa</literal> are specifiers for all pseudo
 623           TTYs and all ALSA sound devices,
 624           respectively. <literal>char-cpu/*</literal> is a specifier
 625           matching all CPU related device groups.</para>
 626         </listitem>
 627       </varlistentry>
 628
 629       <varlistentry>
 630         <term><varname>DevicePolicy=auto|closed|strict</varname></term>
 631
 632         <listitem>
 633           <para>
 634             Control the policy for allowing device access:
 635           </para>
 636           <variablelist>
 637             <varlistentry>
 638               <term><option>strict</option></term>
 639               <listitem>
 640                 <para>means to only allow types of access that are
 641                 explicitly specified.</para>
 642               </listitem>
 643             </varlistentry>
 644
 645             <varlistentry>
 646               <term><option>closed</option></term>
 647               <listitem>
 648                 <para>in addition, allows access to standard pseudo
 649                 devices including
 650                 <filename>/dev/null</filename>,
 651                 <filename>/dev/zero</filename>,
 652                 <filename>/dev/full</filename>,
 653                 <filename>/dev/random</filename>, and
 654                 <filename>/dev/urandom</filename>.
 655                 </para>
 656               </listitem>
 657             </varlistentry>
 658
 659             <varlistentry>
 660               <term><option>auto</option></term>
 661               <listitem>
 662                 <para>
 663                   in addition, allows access to all devices if no
 664                   explicit <varname>DeviceAllow=</varname> is present.
 665                   This is the default.
 666                 </para>
 667               </listitem>
 668             </varlistentry>
 669           </variablelist>
 670         </listitem>
 671       </varlistentry>
 672
 673       <varlistentry>
 674         <term><varname>Slice=</varname></term>
 675
 676         <listitem>
 677           <para>The name of the slice unit to place the unit
 678           in. Defaults to <filename>system.slice</filename> for all
 679           non-instantiated units of all unit types (except for slice
 680           units themselves see below). Instance units are by default
 681           placed in a subslice of <filename>system.slice</filename>
 682           that is named after the template name.</para>
 683
 684           <para>This option may be used to arrange systemd units in a
 685           hierarchy of slices each of which might have resource
 686           settings applied.</para>
 687
 688           <para>For units of type slice, the only accepted value for
 689           this setting is the parent slice. Since the name of a slice
 690           unit implies the parent slice, it is hence redundant to ever
 691           set this parameter directly for slice units.</para>
 692
 693           <para>Special care should be taken when relying on the default slice assignment in templated service units
 694           that have <varname>DefaultDependencies=no</varname> set, see
 695           <citerefentry><refentrytitle>systemd.service</refentrytitle><manvolnum>5</manvolnum></citerefentry>, section
 696           "Automatic Dependencies" for details.</para>
 697
 698         </listitem>
 699       </varlistentry>
 700
 701       <varlistentry>
 702         <term><varname>Delegate=</varname></term>
 703
 704         <listitem>
 705           <para>Turns on delegation of further resource control
 706           partitioning to processes of the unit. For unprivileged
 707           services (i.e. those using the <varname>User=</varname>
 708           setting), this allows processes to create a subhierarchy
 709           beneath its control group path. For privileged services and
 710           scopes, this ensures the processes will have all control
 711           group controllers enabled.</para>
 712         </listitem>
 713       </varlistentry>
 714
 715     </variablelist>
 716   </refsect1>
 717
 718   <refsect1>
 719     <title>See Also</title>
 720     <para>
 721       <citerefentry><refentrytitle>systemd</refentrytitle><manvolnum>1</manvolnum></citerefentry>,
 722       <citerefentry><refentrytitle>systemd.unit</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
 723       <citerefentry><refentrytitle>systemd.service</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
 724       <citerefentry><refentrytitle>systemd.slice</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
 725       <citerefentry><refentrytitle>systemd.scope</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
 726       <citerefentry><refentrytitle>systemd.socket</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
 727       <citerefentry><refentrytitle>systemd.mount</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
 728       <citerefentry><refentrytitle>systemd.swap</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
 729       <citerefentry><refentrytitle>systemd.directives</refentrytitle><manvolnum>7</manvolnum></citerefentry>,
 730       <citerefentry><refentrytitle>systemd.special</refentrytitle><manvolnum>7</manvolnum></citerefentry>,
 731       The documentation for control groups and specific controllers in the Linux kernel:
 732       <ulink url="https://www.kernel.org/doc/Documentation/cgroup-v1/cgroups.txt">cgroups.txt</ulink>,
 733       <ulink url="https://www.kernel.org/doc/Documentation/cgroup-v1/cpuacct.txt">cpuacct.txt</ulink>,
 734       <ulink url="https://www.kernel.org/doc/Documentation/cgroup-v1/memory.txt">memory.txt</ulink>,
 735       <ulink url="https://www.kernel.org/doc/Documentation/cgroup-v1/blkio-controller.txt">blkio-controller.txt</ulink>.
 736     </para>
 737   </refsect1>
 738 </refentry>