man/systemd-nspawn.xml

   1 <?xml version='1.0'?> <!--*-nxml-*-->
   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 2010 Lennart Poettering
   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-nspawn"
  25           xmlns:xi="http://www.w3.org/2001/XInclude">
  26
  27         <refentryinfo>
  28                 <title>systemd-nspawn</title>
  29                 <productname>systemd</productname>
  30
  31                 <authorgroup>
  32                         <author>
  33                                 <contrib>Developer</contrib>
  34                                 <firstname>Lennart</firstname>
  35                                 <surname>Poettering</surname>
  36                                 <email>lennart@poettering.net</email>
  37                         </author>
  38                 </authorgroup>
  39         </refentryinfo>
  40
  41         <refmeta>
  42                 <refentrytitle>systemd-nspawn</refentrytitle>
  43                 <manvolnum>1</manvolnum>
  44         </refmeta>
  45
  46         <refnamediv>
  47                 <refname>systemd-nspawn</refname>
  48                 <refpurpose>Spawn a namespace container for debugging, testing and building</refpurpose>
  49         </refnamediv>
  50
  51         <refsynopsisdiv>
  52                 <cmdsynopsis>
  53                         <command>systemd-nspawn</command>
  54                         <arg choice="opt" rep="repeat">OPTIONS</arg>
  55                         <arg choice="opt"><replaceable>COMMAND</replaceable>
  56                         <arg choice="opt" rep="repeat">ARGS</arg>
  57                         </arg>
  58                 </cmdsynopsis>
  59                 <cmdsynopsis>
  60                         <command>systemd-nspawn</command>
  61                         <arg choice="plain">-b</arg>
  62                         <arg choice="opt" rep="repeat">OPTIONS</arg>
  63                         <arg choice="opt" rep="repeat">ARGS</arg>
  64                 </cmdsynopsis>
  65         </refsynopsisdiv>
  66
  67         <refsect1>
  68                 <title>Description</title>
  69
  70                 <para><command>systemd-nspawn</command> may be used to
  71                 run a command or OS in a light-weight namespace
  72                 container. In many ways it is similar to
  73                 <citerefentry project='man-pages'><refentrytitle>chroot</refentrytitle><manvolnum>1</manvolnum></citerefentry>,
  74                 but more powerful since it fully virtualizes the file
  75                 system hierarchy, as well as the process tree, the
  76                 various IPC subsystems and the host and domain
  77                 name.</para>
  78
  79                 <para><command>systemd-nspawn</command> limits access
  80                 to various kernel interfaces in the container to
  81                 read-only, such as <filename>/sys</filename>,
  82                 <filename>/proc/sys</filename> or
  83                 <filename>/sys/fs/selinux</filename>. Network
  84                 interfaces and the system clock may not be changed
  85                 from within the container. Device nodes may not be
  86                 created. The host system cannot be rebooted and kernel
  87                 modules may not be loaded from within the
  88                 container.</para>
  89
  90                 <para>Note that even though these security precautions
  91                 are taken <command>systemd-nspawn</command> is not
  92                 suitable for secure container setups. Many of the
  93                 security features may be circumvented and are hence
  94                 primarily useful to avoid accidental changes to the
  95                 host system from the container. The intended use of
  96                 this program is debugging and testing as well as
  97                 building of packages, distributions and software
  98                 involved with boot and systems management.</para>
  99
 100                 <para>In contrast to
 101                 <citerefentry project='man-pages'><refentrytitle>chroot</refentrytitle><manvolnum>1</manvolnum></citerefentry> <command>systemd-nspawn</command>
 102                 may be used to boot full Linux-based operating systems
 103                 in a container.</para>
 104
 105                 <para>Use a tool like
 106                 <citerefentry project='die-net'><refentrytitle>yum</refentrytitle><manvolnum>8</manvolnum></citerefentry>,
 107                 <citerefentry project='die-net'><refentrytitle>debootstrap</refentrytitle><manvolnum>8</manvolnum></citerefentry>,
 108                 or
 109                 <citerefentry project='archlinux'><refentrytitle>pacman</refentrytitle><manvolnum>8</manvolnum></citerefentry>
 110                 to set up an OS directory tree suitable as file system
 111                 hierarchy for <command>systemd-nspawn</command>
 112                 containers.</para>
 113
 114                 <para>Note that <command>systemd-nspawn</command> will
 115                 mount file systems private to the container to
 116                 <filename>/dev</filename>,
 117                 <filename>/run</filename> and similar. These will
 118                 not be visible outside of the container, and their
 119                 contents will be lost when the container exits.</para>
 120
 121                 <para>Note that running two
 122                 <command>systemd-nspawn</command> containers from the
 123                 same directory tree will not make processes in them
 124                 see each other. The PID namespace separation of the
 125                 two containers is complete and the containers will
 126                 share very few runtime objects except for the
 127                 underlying file system. Use
 128                 <citerefentry><refentrytitle>machinectl</refentrytitle><manvolnum>1</manvolnum></citerefentry>'s
 129                 <command>login</command> command to request an
 130                 additional login prompt in a running container.</para>
 131
 132                 <para><command>systemd-nspawn</command> implements the
 133                 <ulink
 134                 url="http://www.freedesktop.org/wiki/Software/systemd/ContainerInterface">Container
 135                 Interface</ulink> specification.</para>
 136
 137                 <para>As a safety check
 138                 <command>systemd-nspawn</command> will verify the
 139                 existence of <filename>/usr/lib/os-release</filename>
 140                 or <filename>/etc/os-release</filename> in the
 141                 container tree before starting the container (see
 142                 <citerefentry><refentrytitle>os-release</refentrytitle><manvolnum>5</manvolnum></citerefentry>). It
 143                 might be necessary to add this file to the container
 144                 tree manually if the OS of the container is too old to
 145                 contain this file out-of-the-box.</para>
 146         </refsect1>
 147
 148         <refsect1>
 149                 <title>Options</title>
 150
 151                 <para>If option <option>-b</option> is specified, the
 152                 arguments are used as arguments for the init
 153                 binary. Otherwise, <replaceable>COMMAND</replaceable>
 154                 specifies the program to launch in the container, and
 155                 the remaining arguments are used as arguments for this
 156                 program. If <option>-b</option> is not used and no
 157                 arguments are specifed, a shell is launched in the
 158                 container.</para>
 159
 160                 <para>The following options are understood:</para>
 161
 162                 <variablelist>
 163                         <varlistentry>
 164                                 <term><option>-D</option></term>
 165                                 <term><option>--directory=</option></term>
 166
 167                                 <listitem><para>Directory to use as
 168                                 file system root for the container.</para>
 169
 170                                 <para>If neither
 171                                 <option>--directory=</option>, nor
 172                                 <option>--image=</option> is specified
 173                                 the directory is determined as
 174                                 <filename>/var/lib/machines/</filename>
 175                                 suffixed by the machine name as
 176                                 specified with
 177                                 <option>--machine=</option>. If
 178                                 neither <option>--directory=</option>,
 179                                 <option>--image=</option>, nor
 180                                 <option>--machine=</option> are
 181                                 specified, the current directory will
 182                                 be used. May not be specified together
 183                                 with
 184                                 <option>--image=</option>.</para></listitem>
 185                         </varlistentry>
 186
 187                         <varlistentry>
 188                                 <term><option>--template=</option></term>
 189
 190                                 <listitem><para>Directory or
 191                                 <literal>btrfs</literal> subvolume to
 192                                 use as template for the container's
 193                                 root directory. If this is specified
 194                                 and the container's root directory (as
 195                                 configured by
 196                                 <option>--directory=</option>) does
 197                                 not yet exist it is created as
 198                                 <literal>btrfs</literal> subvolume and
 199                                 populated from this template
 200                                 tree. Ideally, the specified template
 201                                 path refers to the root of a
 202                                 <literal>btrfs</literal> subvolume, in
 203                                 which case a simple copy-on-write
 204                                 snapshot is taken, and populating the
 205                                 root directory is instant. If the
 206                                 specified template path does not refer
 207                                 to the root of a
 208                                 <literal>btrfs</literal> subvolume (or
 209                                 not even to a <literal>btrfs</literal>
 210                                 file system at all), the tree is
 211                                 copied, which can be substantially
 212                                 more time-consuming. Note that if this
 213                                 option is used the container's root
 214                                 directory (in contrast to the template
 215                                 directory!) must be located on a
 216                                 <literal>btrfs</literal> file system,
 217                                 so that the <literal>btrfs</literal>
 218                                 subvolume may be created. May not be
 219                                 specified together with
 220                                 <option>--image=</option> or
 221                                 <option>--ephemeral</option>.</para></listitem>
 222                         </varlistentry>
 223
 224                         <varlistentry>
 225                                 <term><option>-x</option></term>
 226                                 <term><option>--ephemeral</option></term>
 227
 228                                 <listitem><para>If specified, the
 229                                 container is run with a temporary
 230                                 <literal>btrfs</literal> snapshot of
 231                                 its root directory (as configured with
 232                                 <option>--directory=</option>), that
 233                                 is removed immediately when the
 234                                 container terminates. This option is
 235                                 only supported if the root file system
 236                                 is <literal>btrfs</literal>. May not
 237                                 be specified together with
 238                                 <option>--image=</option> or
 239                                 <option>--template=</option>.</para></listitem>
 240                         </varlistentry>
 241
 242                         <varlistentry>
 243                                 <term><option>-i</option></term>
 244                                 <term><option>--image=</option></term>
 245
 246                                 <listitem><para>Disk image to mount
 247                                 the root directory for the container
 248                                 from. Takes a path to a regular file
 249                                 or to a block device node. The file or
 250                                 block device must contain either an
 251                                 MBR partition table with a single
 252                                 partition of type 0x83 that is marked
 253                                 bootable, or a GUID partition table
 254                                 with a root partition which is mounted
 255                                 as the root directory of the
 256                                 container. Optionally, GPT images may
 257                                 contain a home and/or a server data
 258                                 partition which are mounted to the
 259                                 appropriate places in the
 260                                 container. All these partitions must
 261                                 be identified by the partition types
 262                                 defined by the <ulink
 263                                 url="http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec/">Discoverable
 264                                 Partitions Specification</ulink>. Any
 265                                 other partitions, such as foreign
 266                                 partitions, swap partitions or EFI
 267                                 system partitions are not mounted. May
 268                                 not be specified together with
 269                                 <option>--directory=</option>,
 270                                 <option>--template=</option> or
 271                                 <option>--ephemeral</option>.</para></listitem>
 272                         </varlistentry>
 273
 274                         <varlistentry>
 275                                 <term><option>-b</option></term>
 276                                 <term><option>--boot</option></term>
 277
 278                                 <listitem><para>Automatically search
 279                                 for an init binary and invoke it
 280                                 instead of a shell or a user supplied
 281                                 program. If this option is used,
 282                                 arguments specified on the command
 283                                 line are used as arguments for the
 284                                 init binary. This option may not be
 285                                 combined with
 286                                 <option>--share-system</option>.
 287                                 </para></listitem>
 288                         </varlistentry>
 289
 290                         <varlistentry>
 291                                 <term><option>-u</option></term>
 292                                 <term><option>--user=</option></term>
 293
 294                                 <listitem><para>After transitioning
 295                                 into the container, change to the
 296                                 specified user-defined in the
 297                                 container's user database. Like all
 298                                 other systemd-nspawn features, this is
 299                                 not a security feature and provides
 300                                 protection against accidental
 301                                 destructive operations
 302                                 only.</para></listitem>
 303                         </varlistentry>
 304
 305                         <varlistentry>
 306                                 <term><option>-M</option></term>
 307                                 <term><option>--machine=</option></term>
 308
 309                                 <listitem><para>Sets the machine name
 310                                 for this container. This name may be
 311                                 used to identify this container during
 312                                 its runtime (for example in tools like
 313                                 <citerefentry><refentrytitle>machinectl</refentrytitle><manvolnum>1</manvolnum></citerefentry>
 314                                 and similar), and is used to
 315                                 initialize the container's hostname
 316                                 (which the container can choose to
 317                                 override, however). If not specified,
 318                                 the last component of the root
 319                                 directory path of the container is
 320                                 used, possibly suffixed with a random
 321                                 identifier in case
 322                                 <option>--ephemeral</option> mode is
 323                                 selected. If the root directory
 324                                 selected is the host's root directory
 325                                 the host's hostname is used as default
 326                                 instead.</para></listitem>
 327                         </varlistentry>
 328
 329                         <varlistentry>
 330                                 <term><option>--uuid=</option></term>
 331
 332                                 <listitem><para>Set the specified UUID
 333                                 for the container. The init system
 334                                 will initialize
 335                                 <filename>/etc/machine-id</filename>
 336                                 from this if this file is not set yet.
 337                                 </para></listitem>
 338                         </varlistentry>
 339
 340                         <varlistentry>
 341                                 <term><option>--slice=</option></term>
 342
 343                                 <listitem><para>Make the container
 344                                 part of the specified slice, instead
 345                                 of the default
 346                                 <filename>machine.slice</filename>.</para>
 347                                 </listitem>
 348                         </varlistentry>
 349
 350                         <varlistentry>
 351                                 <term><option>--private-network</option></term>
 352
 353                                 <listitem><para>Disconnect networking
 354                                 of the container from the host. This
 355                                 makes all network interfaces
 356                                 unavailable in the container, with the
 357                                 exception of the loopback device and
 358                                 those specified with
 359                                 <option>--network-interface=</option>
 360                                 and configured with
 361                                 <option>--network-veth</option>. If
 362                                 this option is specified, the
 363                                 CAP_NET_ADMIN capability will be added
 364                                 to the set of capabilities the
 365                                 container retains. The latter may be
 366                                 disabled by using
 367                                 <option>--drop-capability=</option>.</para></listitem>
 368                         </varlistentry>
 369
 370                         <varlistentry>
 371                                 <term><option>--network-interface=</option></term>
 372
 373                                 <listitem><para>Assign the specified
 374                                 network interface to the
 375                                 container. This will remove the
 376                                 specified interface from the calling
 377                                 namespace and place it in the
 378                                 container. When the container
 379                                 terminates, it is moved back to the
 380                                 host namespace. Note that
 381                                 <option>--network-interface=</option>
 382                                 implies
 383                                 <option>--private-network</option>. This
 384                                 option may be used more than once to
 385                                 add multiple network interfaces to the
 386                                 container.</para></listitem>
 387                         </varlistentry>
 388
 389                         <varlistentry>
 390                                 <term><option>--network-macvlan=</option></term>
 391
 392                                 <listitem><para>Create a
 393                                 <literal>macvlan</literal> interface
 394                                 of the specified Ethernet network
 395                                 interface and add it to the
 396                                 container. A
 397                                 <literal>macvlan</literal> interface
 398                                 is a virtual interface that adds a
 399                                 second MAC address to an existing
 400                                 physical Ethernet link. The interface
 401                                 in the container will be named after
 402                                 the interface on the host, prefixed
 403                                 with <literal>mv-</literal>. Note that
 404                                 <option>--network-macvlan=</option>
 405                                 implies
 406                                 <option>--private-network</option>. This
 407                                 option may be used more than once to
 408                                 add multiple network interfaces to the
 409                                 container.</para></listitem>
 410                         </varlistentry>
 411
 412                         <varlistentry>
 413                                 <term><option>--network-ipvlan=</option></term>
 414
 415                                 <listitem><para>Create a
 416                                 <literal>ipvlan</literal> interface
 417                                 of the specified Ethernet network
 418                                 interface and add it to the
 419                                 container. An
 420                                 <literal>ipvlan</literal> interface
 421                                 is a virtual interface, similar to a
 422                                 <literal>macvlan</literal> interface, which
 423                                 uses the same MAC address as the underlying
 424                                 interface. The interface
 425                                 in the container will be named after
 426                                 the interface on the host, prefixed
 427                                 with <literal>iv-</literal>. Note that
 428                                 <option>--network-ipvlan=</option>
 429                                 implies
 430                                 <option>--private-network</option>. This
 431                                 option may be used more than once to
 432                                 add multiple network interfaces to the
 433                                 container.</para></listitem>
 434                         </varlistentry>
 435
 436                         <varlistentry>
 437                                 <term><option>-n</option></term>
 438                                 <term><option>--network-veth</option></term>
 439
 440                                 <listitem><para>Create a virtual
 441                                 Ethernet link
 442                                 (<literal>veth</literal>) between host
 443                                 and container. The host side of the
 444                                 Ethernet link will be available as a
 445                                 network interface named after the
 446                                 container's name (as specified with
 447                                 <option>--machine=</option>), prefixed
 448                                 with <literal>ve-</literal>. The
 449                                 container side of the Ethernet
 450                                 link will be named
 451                                 <literal>host0</literal>. Note that
 452                                 <option>--network-veth</option>
 453                                 implies
 454                                 <option>--private-network</option>.</para></listitem>
 455                         </varlistentry>
 456
 457                         <varlistentry>
 458                                 <term><option>--network-bridge=</option></term>
 459
 460                                 <listitem><para>Adds the host side of
 461                                 the Ethernet link created with
 462                                 <option>--network-veth</option> to the
 463                                 specified bridge. Note that
 464                                 <option>--network-bridge=</option>
 465                                 implies
 466                                 <option>--network-veth</option>. If
 467                                 this option is used, the host side of
 468                                 the Ethernet link will use the
 469                                 <literal>vb-</literal> prefix instead
 470                                 of <literal>ve-</literal>.</para></listitem>
 471                         </varlistentry>
 472
 473                         <varlistentry>
 474                                 <term><option>-p</option></term>
 475                                 <term><option>--port=</option></term>
 476
 477                                 <listitem><para>If private networking
 478                                 is enabled, maps an IP port on the
 479                                 host onto an IP port on the
 480                                 container. Takes a protocol specifier
 481                                 (either <literal>tcp</literal> or
 482                                 <literal>udp</literal>), separated by
 483                                 a colon from a host port number in the
 484                                 range 1 to 65535, separated by a colon
 485                                 from a container port number in the
 486                                 range from 1 to 65535. The protocol
 487                                 specifier and its separating colon may
 488                                 be omitted, in which case
 489                                 <literal>tcp</literal> is assumed.
 490                                 The container port number and its
 491                                 colon may be ommitted, in which case
 492                                 the same port as the host port is
 493                                 implied. This option is only supported
 494                                 if private networking is used, such as
 495                                 <option>--network-veth</option> or
 496                                 <option>--network-bridge=</option>.</para></listitem>
 497                         </varlistentry>
 498
 499                         <varlistentry>
 500                                 <term><option>-Z</option></term>
 501                                 <term><option>--selinux-context=</option></term>
 502
 503                                 <listitem><para>Sets the SELinux
 504                                 security context to be used to label
 505                                 processes in the container.</para>
 506                                 </listitem>
 507                         </varlistentry>
 508
 509                         <varlistentry>
 510                                 <term><option>-L</option></term>
 511                                 <term><option>--selinux-apifs-context=</option></term>
 512
 513                                 <listitem><para>Sets the SELinux security
 514                                 context to be used to label files in
 515                                 the virtual API file systems in the
 516                                 container.</para>
 517                                 </listitem>
 518                         </varlistentry>
 519
 520                         <varlistentry>
 521                                 <term><option>--capability=</option></term>
 522
 523                                 <listitem><para>List one or more
 524                                 additional capabilities to grant the
 525                                 container. Takes a comma-separated
 526                                 list of capability names, see
 527                                 <citerefentry project='man-pages'><refentrytitle>capabilities</refentrytitle><manvolnum>7</manvolnum></citerefentry>
 528                                 for more information. Note that the
 529                                 following capabilities will be granted
 530                                 in any way: CAP_CHOWN,
 531                                 CAP_DAC_OVERRIDE, CAP_DAC_READ_SEARCH,
 532                                 CAP_FOWNER, CAP_FSETID, CAP_IPC_OWNER,
 533                                 CAP_KILL, CAP_LEASE,
 534                                 CAP_LINUX_IMMUTABLE,
 535                                 CAP_NET_BIND_SERVICE,
 536                                 CAP_NET_BROADCAST, CAP_NET_RAW,
 537                                 CAP_SETGID, CAP_SETFCAP, CAP_SETPCAP,
 538                                 CAP_SETUID, CAP_SYS_ADMIN,
 539                                 CAP_SYS_CHROOT, CAP_SYS_NICE,
 540                                 CAP_SYS_PTRACE, CAP_SYS_TTY_CONFIG,
 541                                 CAP_SYS_RESOURCE, CAP_SYS_BOOT,
 542                                 CAP_AUDIT_WRITE,
 543                                 CAP_AUDIT_CONTROL. Also CAP_NET_ADMIN
 544                                 is retained if
 545                                 <option>--private-network</option> is
 546                                 specified. If the special value
 547                                 <literal>all</literal> is passed, all
 548                                 capabilities are
 549                                 retained.</para></listitem>
 550                         </varlistentry>
 551
 552                         <varlistentry>
 553                                 <term><option>--drop-capability=</option></term>
 554
 555                                 <listitem><para>Specify one or more
 556                                 additional capabilities to drop for
 557                                 the container. This allows running the
 558                                 container with fewer capabilities than
 559                                 the default (see above).</para></listitem>
 560                         </varlistentry>
 561
 562                         <varlistentry>
 563                                 <term><option>--link-journal=</option></term>
 564
 565                                 <listitem><para>Control whether the
 566                                 container's journal shall be made
 567                                 visible to the host system. If enabled,
 568                                 allows viewing the container's journal
 569                                 files from the host (but not vice
 570                                 versa). Takes one of
 571                                 <literal>no</literal>,
 572                                 <literal>host</literal>,
 573                                 <literal>try-host</literal>,
 574                                 <literal>guest</literal>,
 575                                 <literal>try-guest</literal>,
 576                                 <literal>auto</literal>. If
 577                                 <literal>no</literal>, the journal is
 578                                 not linked. If <literal>host</literal>,
 579                                 the journal files are stored on the
 580                                 host file system (beneath
 581                                 <filename>/var/log/journal/<replaceable>machine-id</replaceable></filename>)
 582                                 and the subdirectory is bind-mounted
 583                                 into the container at the same
 584                                 location. If <literal>guest</literal>,
 585                                 the journal files are stored on the
 586                                 guest file system (beneath
 587                                 <filename>/var/log/journal/<replaceable>machine-id</replaceable></filename>)
 588                                 and the subdirectory is symlinked into the host
 589                                 at the same location. <literal>try-host</literal>
 590                                 and <literal>try-guest</literal> do the same
 591                                 but do not fail if the host does not have
 592                                 persistent journalling enabled.
 593                                 If <literal>auto</literal> (the default),
 594                                 and the right subdirectory of
 595                                 <filename>/var/log/journal</filename>
 596                                 exists, it will be bind mounted
 597                                 into the container. If the
 598                                 subdirectory does not exist, no
 599                                 linking is performed. Effectively,
 600                                 booting a container once with
 601                                 <literal>guest</literal> or
 602                                 <literal>host</literal> will link the
 603                                 journal persistently if further on
 604                                 the default of <literal>auto</literal>
 605                                 is used.</para></listitem>
 606                         </varlistentry>
 607
 608                         <varlistentry>
 609                                 <term><option>-j</option></term>
 610
 611                                 <listitem><para>Equivalent to
 612                                 <option>--link-journal=try-guest</option>.</para></listitem>
 613                         </varlistentry>
 614
 615                         <varlistentry>
 616                                 <term><option>--read-only</option></term>
 617
 618                                 <listitem><para>Mount the root file
 619                                 system read-only for the
 620                                 container.</para></listitem>
 621                         </varlistentry>
 622
 623                         <varlistentry>
 624                                 <term><option>--bind=</option></term>
 625                                 <term><option>--bind-ro=</option></term>
 626
 627                                 <listitem><para>Bind mount a file or
 628                                 directory from the host into the
 629                                 container. Either takes a path
 630                                 argument -- in which case the
 631                                 specified path will be mounted from
 632                                 the host to the same path in the
 633                                 container --, or a colon-separated
 634                                 pair of paths -- in which case the
 635                                 first specified path is the source in
 636                                 the host, and the second path is the
 637                                 destination in the container. The
 638                                 <option>--bind-ro=</option> option
 639                                 creates read-only bind
 640                                 mounts.</para></listitem>
 641                         </varlistentry>
 642
 643                         <varlistentry>
 644                                 <term><option>--tmpfs=</option></term>
 645
 646                                 <listitem><para>Mount a tmpfs file
 647                                 system into the container. Takes a
 648                                 single absolute path argument that
 649                                 specifies where to mount the tmpfs
 650                                 instance to (in which case the
 651                                 directory access mode will be chosen
 652                                 as 0755, owned by root/root), or
 653                                 optionally a colon-separated pair of
 654                                 path and mount option string, that is
 655                                 used for mounting (in which case the
 656                                 kernel default for access mode and
 657                                 owner will be chosen, unless otherwise
 658                                 specified). This option is
 659                                 particularly useful for mounting
 660                                 directories such as
 661                                 <filename>/var</filename> as tmpfs, to
 662                                 allow state-less systems, in
 663                                 particular when combined with
 664                                 <option>--read-only</option>.</para></listitem>
 665                         </varlistentry>
 666
 667                         <varlistentry>
 668                                 <term><option>--setenv=</option></term>
 669
 670                                 <listitem><para>Specifies an
 671                                 environment variable assignment to
 672                                 pass to the init process in the
 673                                 container, in the format
 674                                 <literal>NAME=VALUE</literal>. This
 675                                 may be used to override the default
 676                                 variables or to set additional
 677                                 variables. This parameter may be used
 678                                 more than once.</para></listitem>
 679                         </varlistentry>
 680
 681                         <varlistentry>
 682                                 <term><option>--share-system</option></term>
 683
 684                                 <listitem><para>Allows the container
 685                                 to share certain system facilities
 686                                 with the host. More specifically, this
 687                                 turns off PID namespacing, UTS
 688                                 namespacing and IPC namespacing, and
 689                                 thus allows the guest to see and
 690                                 interact more easily with processes
 691                                 outside of the container. Note that
 692                                 using this option makes it impossible
 693                                 to start up a full Operating System in
 694                                 the container, as an init system
 695                                 cannot operate in this mode. It is
 696                                 only useful to run specific programs
 697                                 or applications this way, without
 698                                 involving an init system in the
 699                                 container. This option implies
 700                                 <option>--register=no</option>. This
 701                                 option may not be combined with
 702                                 <option>--boot</option>.</para></listitem>
 703                         </varlistentry>
 704
 705                         <varlistentry>
 706                                 <term><option>--register=</option></term>
 707
 708                                 <listitem><para>Controls whether the
 709                                 container is registered with
 710                                 <citerefentry><refentrytitle>systemd-machined</refentrytitle><manvolnum>8</manvolnum></citerefentry>. Takes
 711                                 a boolean argument, defaults to
 712                                 <literal>yes</literal>. This option
 713                                 should be enabled when the container
 714                                 runs a full Operating System (more
 715                                 specifically: an init system), and is
 716                                 useful to ensure that the container is
 717                                 accessible via
 718                                 <citerefentry><refentrytitle>machinectl</refentrytitle><manvolnum>1</manvolnum></citerefentry>
 719                                 and shown by tools such as
 720                                 <citerefentry project='man-pages'><refentrytitle>ps</refentrytitle><manvolnum>1</manvolnum></citerefentry>. If
 721                                 the container does not run an init
 722                                 system, it is recommended to set this
 723                                 option to <literal>no</literal>. Note
 724                                 that <option>--share-system</option>
 725                                 implies
 726                                 <option>--register=no</option>.
 727                                 </para></listitem>
 728                         </varlistentry>
 729
 730                         <varlistentry>
 731                                 <term><option>--keep-unit</option></term>
 732
 733                                 <listitem><para>Instead of creating a
 734                                 transient scope unit to run the
 735                                 container in, simply register the
 736                                 service or scope unit
 737                                 <command>systemd-nspawn</command> has
 738                                 been invoked in with
 739                                 <citerefentry><refentrytitle>systemd-machined</refentrytitle><manvolnum>8</manvolnum></citerefentry>. This
 740                                 has no effect if
 741                                 <option>--register=no</option> is
 742                                 used. This switch should be used if
 743                                 <command>systemd-nspawn</command> is
 744                                 invoked from within a service unit,
 745                                 and the service unit's sole purpose
 746                                 is to run a single
 747                                 <command>systemd-nspawn</command>
 748                                 container. This option is not
 749                                 available if run from a user
 750                                 session.</para></listitem>
 751                         </varlistentry>
 752
 753                         <varlistentry>
 754                                 <term><option>--personality=</option></term>
 755
 756                                 <listitem><para>Control the
 757                                 architecture ("personality") reported
 758                                 by
 759                                 <citerefentry><refentrytitle>uname</refentrytitle><manvolnum>2</manvolnum></citerefentry>
 760                                 in the container. Currently, only
 761                                 <literal>x86</literal> and
 762                                 <literal>x86-64</literal> are
 763                                 supported. This is useful when running
 764                                 a 32-bit container on a 64-bit
 765                                 host. If this setting is not used,
 766                                 the personality reported in the
 767                                 container is the same as the one
 768                                 reported on the
 769                                 host.</para></listitem>
 770                         </varlistentry>
 771
 772                         <varlistentry>
 773                                 <term><option>-q</option></term>
 774                                 <term><option>--quiet</option></term>
 775
 776                                 <listitem><para>Turns off any status
 777                                 output by the tool itself. When this
 778                                 switch is used, the only output
 779                                 from nspawn will be the console output
 780                                 of the container OS itself.</para></listitem>
 781                         </varlistentry>
 782
 783                         <varlistentry>
 784                                 <term><option>--volatile</option><replaceable>=MODE</replaceable></term>
 785
 786                                 <listitem><para>Boots the container in
 787                                 volatile mode. When no mode parameter
 788                                 is passed or when mode is specified as
 789                                 <literal>yes</literal> full volatile
 790                                 mode is enabled. This means the root
 791                                 directory is mounted as mostly
 792                                 unpopulated <literal>tmpfs</literal>
 793                                 instance, and
 794                                 <filename>/usr</filename> from the OS
 795                                 tree is mounted into it, read-only
 796                                 (the system thus starts up with
 797                                 read-only OS resources, but pristine
 798                                 state and configuration, any changes
 799                                 to the either are lost on
 800                                 shutdown). When the mode parameter is
 801                                 specified as <literal>state</literal>
 802                                 the OS tree is mounted read-only, but
 803                                 <filename>/var</filename> is mounted
 804                                 as <literal>tmpfs</literal> instance
 805                                 into it (the system thus starts up
 806                                 with read-only OS resources and
 807                                 configuration, but pristine state, any
 808                                 changes to the latter are lost on
 809                                 shutdown). When the mode parameter is
 810                                 specified as <literal>no</literal>
 811                                 (the default) the whole OS tree is
 812                                 made available writable.</para>
 813
 814                                 <para>Note that setting this to
 815                                 <literal>yes</literal> or
 816                                 <literal>state</literal> will only
 817                                 work correctly with operating systems
 818                                 in the container that can boot up with
 819                                 only <filename>/usr</filename>
 820                                 mounted, and are able to populate
 821                                 <filename>/var</filename>
 822                                 automatically, as
 823                                 needed.</para></listitem>
 824                         </varlistentry>
 825
 826                         <xi:include href="standard-options.xml" xpointer="help" />
 827                         <xi:include href="standard-options.xml" xpointer="version" />
 828                 </variablelist>
 829
 830         </refsect1>
 831
 832         <refsect1>
 833                 <title>Examples</title>
 834                 <example>
 835                         <title>Boot a minimal Fedora distribution in a container</title>
 836
 837                         <programlisting># yum -y --releasever=21 --nogpg --installroot=/srv/mycontainer --disablerepo='*' --enablerepo=fedora install systemd passwd yum fedora-release vim-minimal
 838 # systemd-nspawn -bD /srv/mycontainer</programlisting>
 839
 840                         <para>This installs a minimal Fedora distribution into
 841                         the directory <filename noindex='true'>/srv/mycontainer/</filename> and
 842                         then boots an OS in a namespace container in
 843                         it.</para>
 844                 </example>
 845
 846                 <example>
 847                         <title>Spawn a shell in a container of a minimal Debian unstable distribution</title>
 848
 849                         <programlisting># debootstrap --arch=amd64 unstable ~/debian-tree/
 850 # systemd-nspawn -D ~/debian-tree/</programlisting>
 851
 852                         <para>This installs a minimal Debian unstable
 853                         distribution into the directory
 854                         <filename>~/debian-tree/</filename> and then spawns a
 855                         shell in a namespace container in it.</para>
 856                 </example>
 857
 858                 <example>
 859                         <title>Boot a minimal Arch Linux distribution in a container</title>
 860
 861                         <programlisting># pacstrap -c -d ~/arch-tree/ base
 862 # systemd-nspawn -bD ~/arch-tree/</programlisting>
 863
 864                         <para>This installs a mimimal Arch Linux distribution into
 865                         the directory <filename>~/arch-tree/</filename> and then
 866                         boots an OS in a namespace container in it.</para>
 867                 </example>
 868
 869                 <example>
 870                         <title>Enable Arch Linux container on boot</title>
 871
 872                         <programlisting># mv ~/arch-tree /var/lib/machines/arch
 873 # systemctl enable systemd-nspawn@arch.service
 874 # systemctl start systemd-nspawn@arch.service</programlisting>
 875
 876                         <para>This makes the Arch Linux container part of the
 877                         <filename>multi-user.target</filename> on the host.
 878                         </para>
 879                 </example>
 880
 881                 <example>
 882                         <title>Boot into an ephemeral <literal>btrfs</literal> snapshot of the host system</title>
 883
 884                         <programlisting># systemd-nspawn -D / -xb</programlisting>
 885
 886                         <para>This runs a copy of the host system in a
 887                         <literal>btrfs</literal> snapshot which is
 888                         removed immediately when the container
 889                         exits. All file system changes made during
 890                         runtime will be lost on shutdown,
 891                         hence.</para>
 892                 </example>
 893
 894                 <example>
 895                         <title>Run a container with SELinux sandbox security contexts</title>
 896
 897                         <programlisting># chcon system_u:object_r:svirt_sandbox_file_t:s0:c0,c1 -R /srv/container
 898 # systemd-nspawn -L system_u:object_r:svirt_sandbox_file_t:s0:c0,c1 -Z system_u:system_r:svirt_lxc_net_t:s0:c0,c1 -D /srv/container /bin/sh</programlisting>
 899                 </example>
 900         </refsect1>
 901
 902         <refsect1>
 903                 <title>Exit status</title>
 904
 905                 <para>The exit code of the program executed in the
 906                 container is returned.</para>
 907         </refsect1>
 908
 909         <refsect1>
 910                 <title>See Also</title>
 911                 <para>
 912                         <citerefentry><refentrytitle>systemd</refentrytitle><manvolnum>1</manvolnum></citerefentry>,
 913                         <citerefentry project='man-pages'><refentrytitle>chroot</refentrytitle><manvolnum>1</manvolnum></citerefentry>,
 914                         <citerefentry project='die-net'><refentrytitle>yum</refentrytitle><manvolnum>8</manvolnum></citerefentry>,
 915                         <citerefentry project='die-net'><refentrytitle>debootstrap</refentrytitle><manvolnum>8</manvolnum></citerefentry>,
 916                         <citerefentry project='archlinux'><refentrytitle>pacman</refentrytitle><manvolnum>8</manvolnum></citerefentry>,
 917                         <citerefentry><refentrytitle>systemd.slice</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
 918                         <citerefentry><refentrytitle>machinectl</refentrytitle><manvolnum>1</manvolnum></citerefentry>,
 919                         <citerefentry><refentrytitle>btrfs</refentrytitle><manvolnum>8</manvolnum></citerefentry>
 920                 </para>
 921         </refsect1>
 922
 923 </refentry>