2 <!DOCTYPE refentry PUBLIC
"-//OASIS//DTD DocBook XML V4.5//EN"
3 "http://www.oasis-open.org/docbook/xml/4.2/docbookx.dtd" [
4 <!ENTITY fedora_latest_version
"30">
5 <!ENTITY fedora_cloud_release
"1.2">
7 <!-- SPDX-License-Identifier: LGPL-2.1+ -->
9 <refentry id=
"systemd-nspawn"
10 xmlns:
xi=
"http://www.w3.org/2001/XInclude">
13 <title>systemd-nspawn
</title>
14 <productname>systemd
</productname>
18 <refentrytitle>systemd-nspawn
</refentrytitle>
19 <manvolnum>1</manvolnum>
23 <refname>systemd-nspawn
</refname>
24 <refpurpose>Spawn a command or OS in a light-weight container
</refpurpose>
29 <command>systemd-nspawn
</command>
30 <arg choice=
"opt" rep=
"repeat">OPTIONS
</arg>
31 <arg choice=
"opt"><replaceable>COMMAND
</replaceable>
32 <arg choice=
"opt" rep=
"repeat">ARGS
</arg>
36 <command>systemd-nspawn
</command>
37 <arg choice=
"plain">--boot
</arg>
38 <arg choice=
"opt" rep=
"repeat">OPTIONS
</arg>
39 <arg choice=
"opt" rep=
"repeat">ARGS
</arg>
44 <title>Description
</title>
46 <para><command>systemd-nspawn
</command> may be used to run a command or OS in a light-weight namespace
47 container. In many ways it is similar to
<citerefentry
48 project='man-pages'
><refentrytitle>chroot
</refentrytitle><manvolnum>1</manvolnum></citerefentry>, but more powerful
49 since it fully virtualizes the file system hierarchy, as well as the process tree, the various IPC subsystems and
50 the host and domain name.
</para>
52 <para><command>systemd-nspawn
</command> may be invoked on any directory tree containing an operating system tree,
53 using the
<option>--directory=
</option> command line option. By using the
<option>--machine=
</option> option an OS
54 tree is automatically searched for in a couple of locations, most importantly in
55 <filename>/var/lib/machines
</filename>, the suggested directory to place OS container images installed on the
58 <para>In contrast to
<citerefentry
59 project='man-pages'
><refentrytitle>chroot
</refentrytitle><manvolnum>1</manvolnum></citerefentry> <command>systemd-nspawn
</command>
60 may be used to boot full Linux-based operating systems in a container.
</para>
62 <para><command>systemd-nspawn
</command> limits access to various kernel interfaces in the container to read-only,
63 such as
<filename>/sys
</filename>,
<filename>/proc/sys
</filename> or
<filename>/sys/fs/selinux
</filename>. The
64 host's network interfaces and the system clock may not be changed from within the container. Device nodes may not
65 be created. The host system cannot be rebooted and kernel modules may not be loaded from within the
68 <para>Use a tool like
<citerefentry
69 project='mankier'
><refentrytitle>dnf
</refentrytitle><manvolnum>8</manvolnum></citerefentry>,
<citerefentry
70 project='die-net'
><refentrytitle>debootstrap
</refentrytitle><manvolnum>8</manvolnum></citerefentry>, or
71 <citerefentry project='archlinux'
><refentrytitle>pacman
</refentrytitle><manvolnum>8</manvolnum></citerefentry> to
72 set up an OS directory tree suitable as file system hierarchy for
<command>systemd-nspawn
</command> containers. See
73 the Examples section below for details on suitable invocation of these commands.
</para>
75 <para>As a safety check
<command>systemd-nspawn
</command> will verify the existence of
76 <filename>/usr/lib/os-release
</filename> or
<filename>/etc/os-release
</filename> in the container tree before
77 starting the container (see
78 <citerefentry><refentrytitle>os-release
</refentrytitle><manvolnum>5</manvolnum></citerefentry>). It might be
79 necessary to add this file to the container tree manually if the OS of the container is too old to contain this
80 file out-of-the-box.
</para>
82 <para><command>systemd-nspawn
</command> may be invoked directly from the interactive command line or run as system
83 service in the background. In this mode each container instance runs as its own service instance; a default
84 template unit file
<filename>systemd-nspawn@.service
</filename> is provided to make this easy, taking the container
85 name as instance identifier. Note that different default options apply when
<command>systemd-nspawn
</command> is
86 invoked by the template unit file than interactively on the command line. Most importantly the template unit file
87 makes use of the
<option>--boot
</option> which is not the default in case
<command>systemd-nspawn
</command> is
88 invoked from the interactive command line. Further differences with the defaults are documented along with the
89 various supported options below.
</para>
91 <para>The
<citerefentry><refentrytitle>machinectl
</refentrytitle><manvolnum>1</manvolnum></citerefentry> tool may
92 be used to execute a number of operations on containers. In particular it provides easy-to-use commands to run
93 containers as system services using the
<filename>systemd-nspawn@.service
</filename> template unit
96 <para>Along with each container a settings file with the
<filename>.nspawn
</filename> suffix may exist, containing
97 additional settings to apply when running the container. See
98 <citerefentry><refentrytitle>systemd.nspawn
</refentrytitle><manvolnum>5</manvolnum></citerefentry> for
99 details. Settings files override the default options used by the
<filename>systemd-nspawn@.service
</filename>
100 template unit file, making it usually unnecessary to alter this template file directly.
</para>
102 <para>Note that
<command>systemd-nspawn
</command> will mount file systems private to the container to
103 <filename>/dev
</filename>,
<filename>/run
</filename> and similar. These will not be visible outside of the
104 container, and their contents will be lost when the container exits.
</para>
106 <para>Note that running two
<command>systemd-nspawn
</command> containers from the same directory tree will not make
107 processes in them see each other. The PID namespace separation of the two containers is complete and the containers
108 will share very few runtime objects except for the underlying file system. Use
109 <citerefentry><refentrytitle>machinectl
</refentrytitle><manvolnum>1</manvolnum></citerefentry>'s
110 <command>login
</command> or
<command>shell
</command> commands to request an additional login session in a running
113 <para><command>systemd-nspawn
</command> implements the
<ulink
114 url=
"https://www.freedesktop.org/wiki/Software/systemd/ContainerInterface">Container Interface
</ulink>
115 specification.
</para>
117 <para>While running, containers invoked with
<command>systemd-nspawn
</command> are registered with the
118 <citerefentry><refentrytitle>systemd-machined
</refentrytitle><manvolnum>8</manvolnum></citerefentry> service that
119 keeps track of running containers, and provides programming interfaces to interact with them.
</para>
123 <title>Options
</title>
125 <para>If option
<option>-b
</option> is specified, the arguments
126 are used as arguments for the init program. Otherwise,
127 <replaceable>COMMAND
</replaceable> specifies the program to launch
128 in the container, and the remaining arguments are used as
129 arguments for this program. If
<option>--boot
</option> is not used and
130 no arguments are specified, a shell is launched in the
133 <para>The following options are understood:
</para>
138 <term><option>-q
</option></term>
139 <term><option>--quiet
</option></term>
141 <listitem><para>Turns off any status output by the tool
142 itself. When this switch is used, the only output from nspawn
143 will be the console output of the container OS
144 itself.
</para></listitem>
148 <term><option>--settings=
</option><replaceable>MODE
</replaceable></term>
150 <listitem><para>Controls whether
151 <command>systemd-nspawn
</command> shall search for and use
152 additional per-container settings from
153 <filename>.nspawn
</filename> files. Takes a boolean or the
154 special values
<option>override
</option> or
155 <option>trusted
</option>.
</para>
157 <para>If enabled (the default), a settings file named after the
158 machine (as specified with the
<option>--machine=
</option>
159 setting, or derived from the directory or image file name)
160 with the suffix
<filename>.nspawn
</filename> is searched in
161 <filename>/etc/systemd/nspawn/
</filename> and
162 <filename>/run/systemd/nspawn/
</filename>. If it is found
163 there, its settings are read and used. If it is not found
164 there, it is subsequently searched in the same directory as the
165 image file or in the immediate parent of the root directory of
166 the container. In this case, if the file is found, its settings
167 will be also read and used, but potentially unsafe settings
168 are ignored. Note that in both these cases, settings on the
169 command line take precedence over the corresponding settings
170 from loaded
<filename>.nspawn
</filename> files, if both are
171 specified. Unsafe settings are considered all settings that
172 elevate the container's privileges or grant access to
173 additional resources such as files or directories of the
174 host. For details about the format and contents of
175 <filename>.nspawn
</filename> files, consult
176 <citerefentry><refentrytitle>systemd.nspawn
</refentrytitle><manvolnum>5</manvolnum></citerefentry>.
</para>
178 <para>If this option is set to
<option>override
</option>, the
179 file is searched, read and used the same way, however, the order of
180 precedence is reversed: settings read from the
181 <filename>.nspawn
</filename> file will take precedence over
182 the corresponding command line options, if both are
185 <para>If this option is set to
<option>trusted
</option>, the
186 file is searched, read and used the same way, but regardless
187 of being found in
<filename>/etc/systemd/nspawn/
</filename>,
188 <filename>/run/systemd/nspawn/
</filename> or next to the image
189 file or container root directory, all settings will take
190 effect, however, command line arguments still take precedence
191 over corresponding settings.
</para>
193 <para>If disabled, no
<filename>.nspawn
</filename> file is read
194 and no settings except the ones on the command line are in
195 effect.
</para></listitem>
201 <title>Image Options
</title>
206 <term><option>-D
</option></term>
207 <term><option>--directory=
</option></term>
209 <listitem><para>Directory to use as file system root for the
212 <para>If neither
<option>--directory=
</option>, nor
213 <option>--image=
</option> is specified the directory is
214 determined by searching for a directory named the same as the
215 machine name specified with
<option>--machine=
</option>. See
216 <citerefentry><refentrytitle>machinectl
</refentrytitle><manvolnum>1</manvolnum></citerefentry>
217 section
"Files and Directories" for the precise search path.
</para>
219 <para>If neither
<option>--directory=
</option>,
220 <option>--image=
</option>, nor
<option>--machine=
</option>
221 are specified, the current directory will
222 be used. May not be specified together with
223 <option>--image=
</option>.
</para></listitem>
227 <term><option>--template=
</option></term>
229 <listitem><para>Directory or
<literal>btrfs
</literal> subvolume to use as template for the
230 container's root directory. If this is specified and the container's root directory (as configured by
231 <option>--directory=
</option>) does not yet exist it is created as
<literal>btrfs
</literal> snapshot
232 (if supported) or plain directory (otherwise) and populated from this template tree. Ideally, the
233 specified template path refers to the root of a
<literal>btrfs
</literal> subvolume, in which case a
234 simple copy-on-write snapshot is taken, and populating the root directory is instant. If the
235 specified template path does not refer to the root of a
<literal>btrfs
</literal> subvolume (or not
236 even to a
<literal>btrfs
</literal> file system at all), the tree is copied (though possibly in a
237 'reflink' copy-on-write scheme — if the file system supports that), which can be substantially more
238 time-consuming. Note that the snapshot taken is of the specified directory or subvolume, including
239 all subdirectories and subvolumes below it, but excluding any sub-mounts. May not be specified
240 together with
<option>--image=
</option> or
<option>--ephemeral
</option>.
</para>
242 <para>Note that this switch leaves host name, machine ID and
243 all other settings that could identify the instance
244 unmodified.
</para></listitem>
248 <term><option>-x
</option></term>
249 <term><option>--ephemeral
</option></term>
251 <listitem><para>If specified, the container is run with a temporary snapshot of its file system that is removed
252 immediately when the container terminates. May not be specified together with
253 <option>--template=
</option>.
</para>
254 <para>Note that this switch leaves host name, machine ID and all other settings that could identify
255 the instance unmodified. Please note that — as with
<option>--template=
</option> — taking the
256 temporary snapshot is more efficient on file systems that support subvolume snapshots or 'reflinks'
257 natively (
<literal>btrfs
</literal> or new
<literal>xfs
</literal>) than on more traditional file
258 systems that do not (
<literal>ext4
</literal>). Note that the snapshot taken is of the specified
259 directory or subvolume, including all subdirectories and subvolumes below it, but excluding any
262 <para>With this option no modifications of the container image are retained. Use
263 <option>--volatile=
</option> (described below) for other mechanisms to restrict persistency of
264 container images during runtime.
</para>
269 <term><option>-i
</option></term>
270 <term><option>--image=
</option></term>
272 <listitem><para>Disk image to mount the root directory for the
273 container from. Takes a path to a regular file or to a block
274 device node. The file or block device must contain
278 <listitem><para>An MBR partition table with a single
279 partition of type
0x83 that is marked
280 bootable.
</para></listitem>
282 <listitem><para>A GUID partition table (GPT) with a single
284 0fc63daf-
8483-
4772-
8e79-
3d69d8477de4.
</para></listitem>
286 <listitem><para>A GUID partition table (GPT) with a marked
287 root partition which is mounted as the root directory of the
288 container. Optionally, GPT images may contain a home and/or
289 a server data partition which are mounted to the appropriate
290 places in the container. All these partitions must be
291 identified by the partition types defined by the
<ulink
292 url=
"https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec/">Discoverable
293 Partitions Specification
</ulink>.
</para></listitem>
295 <listitem><para>No partition table, and a single file system spanning the whole image.
</para></listitem>
298 <para>On GPT images, if an EFI System Partition (ESP) is discovered, it is automatically mounted to
299 <filename>/efi
</filename> (or
<filename>/boot
</filename> as fallback) in case a directory by this name exists
302 <para>Partitions encrypted with LUKS are automatically decrypted. Also, on GPT images dm-verity data integrity
303 hash partitions are set up if the root hash for them is specified using the
<option>--root-hash=
</option>
306 <para>Any other partitions, such as foreign partitions or swap partitions are not mounted. May not be specified
307 together with
<option>--directory=
</option>,
<option>--template=
</option>.
</para></listitem>
311 <term><option>--oci-bundle=
</option></term>
313 <listitem><para>Takes the path to an OCI runtime bundle to invoke, as specified in the
<ulink
314 url=
"https://github.com/opencontainers/runtime-spec/blob/master/spec.md">OCI Runtime Specification
</ulink>. In
315 this case no
<filename>.nspawn
</filename> file is loaded, and the root directory and various settings are read
316 from the OCI runtime JSON data (but data passed on the command line takes precedence).
</para></listitem>
320 <term><option>--read-only
</option></term>
322 <listitem><para>Mount the container's root file system (and any other file systems container in the container
323 image) read-only. This has no effect on additional mounts made with
<option>--bind=
</option>,
324 <option>--tmpfs=
</option> and similar options. This mode is implied if the container image file or directory is
325 marked read-only itself. It is also implied if
<option>--volatile=
</option> is used. In this case the container
326 image on disk is strictly read-only, while changes are permitted but kept non-persistently in memory only. For
327 further details, see below.
</para></listitem>
331 <term><option>--volatile
</option></term>
332 <term><option>--volatile=
</option><replaceable>MODE
</replaceable></term>
334 <listitem><para>Boots the container in volatile mode. When no mode parameter is passed or when mode is
335 specified as
<option>yes
</option>, full volatile mode is enabled. This means the root directory is mounted as a
336 mostly unpopulated
<literal>tmpfs
</literal> instance, and
<filename>/usr/
</filename> from the OS tree is
337 mounted into it in read-only mode (the system thus starts up with read-only OS image, but pristine state and
338 configuration, any changes are lost on shutdown). When the mode parameter is specified as
339 <option>state
</option>, the OS tree is mounted read-only, but
<filename>/var/
</filename> is mounted as a
340 writable
<literal>tmpfs
</literal> instance into it (the system thus starts up with read-only OS resources and
341 configuration, but pristine state, and any changes to the latter are lost on shutdown). When the mode parameter
342 is specified as
<option>overlay
</option> the read-only root file system is combined with a writable
343 <filename>tmpfs
</filename> instance through
<literal>overlayfs
</literal>, so that it appears at it normally
344 would, but any changes are applied to the temporary file system only and lost when the container is
345 terminated. When the mode parameter is specified as
<option>no
</option> (the default), the whole OS tree is
346 made available writable (unless
<option>--read-only
</option> is specified, see above).
</para>
348 <para>Note that if one of the volatile modes is chosen, its effect is limited to the root file system (or
349 <filename>/var/
</filename> in case of
<option>state
</option>), and any other mounts placed in the hierarchy are
350 unaffected — regardless if they are established automatically (e.g. the EFI system partition that might be
351 mounted to
<filename>/efi/
</filename> or
<filename>/boot/
</filename>) or explicitly (e.g. through an additional
352 command line option such as
<option>--bind=
</option>, see below). This means, even if
353 <option>--volatile=overlay
</option> is used changes to
<filename>/efi/
</filename> or
354 <filename>/boot/
</filename> are prohibited in case such a partition exists in the container image operated on,
355 and even if
<option>--volatile=state
</option> is used the hypothetical file
<filename>/etc/foobar
</filename> is
356 potentially writable if
<option>--bind=/etc/foobar
</option> if used to mount it from outside the read-only
357 container
<filename>/etc
</filename> directory.
</para>
359 <para>The
<option>--ephemeral
</option> option is closely related to this setting, and provides similar
360 behaviour by making a temporary, ephemeral copy of the whole OS image and executing that. For further details,
363 <para>The
<option>--tmpfs=
</option> and
<option>--overlay=
</option> options provide similar functionality, but
364 for specific sub-directories of the OS image only. For details, see below.
</para>
366 <para>This option provides similar functionality for containers as the
<literal>systemd.volatile=
</literal>
367 kernel command line switch provides for host systems. See
368 <citerefentry><refentrytitle>kernel-command-line
</refentrytitle><manvolnum>7</manvolnum></citerefentry> for
371 <para>Note that setting this option to
<option>yes
</option> or
<option>state
</option> will only work
372 correctly with operating systems in the container that can boot up with only
373 <filename>/usr/
</filename> mounted, and are able to automatically populate
<filename>/var/
</filename>
374 (and
<filename>/etc/
</filename> in case of
<literal>--volatile=yes
</literal>). Specifically, this
375 means that operating systems that follow the historic split of
<filename>/bin/
</filename> and
376 <filename>/lib/
</filename> (and related directories) from
<filename>/usr/
</filename> (i.e. where the
377 former are not symlinks into the latter) are not supported by
<literal>--volatile=yes
</literal> as
378 container payload. The
<option>overlay
</option> option does not require any particular preparations
379 in the OS, but do note that
<literal>overlayfs
</literal> behaviour differs from regular file systems
380 in a number of ways, and hence compatibility is limited.
</para></listitem>
384 <term><option>--root-hash=
</option></term>
386 <listitem><para>Takes a data integrity (dm-verity) root hash specified in hexadecimal. This option enables data
387 integrity checks using dm-verity, if the used image contains the appropriate integrity data (see above). The
388 specified hash must match the root hash of integrity data, and is usually at least
256 bits (and hence
64
389 formatted hexadecimal characters) long (in case of SHA256 for example). If this option is not specified, but
390 the image file carries the
<literal>user.verity.roothash
</literal> extended file attribute (see
<citerefentry
391 project='man-pages'
><refentrytitle>xattr
</refentrytitle><manvolnum>7</manvolnum></citerefentry>), then the root
392 hash is read from it, also as formatted hexadecimal characters. If the extended file attribute is not found (or
393 is not supported by the underlying file system), but a file with the
<filename>.roothash
</filename> suffix is
394 found next to the image file, bearing otherwise the same name, the root hash is read from it and automatically
395 used, also as formatted hexadecimal characters.
</para></listitem>
399 <term><option>--pivot-root=
</option></term>
401 <listitem><para>Pivot the specified directory to
<filename>/
</filename> inside the container, and either unmount the
402 container's old root, or pivot it to another specified directory. Takes one of: a path argument — in which case the
403 specified path will be pivoted to
<filename>/
</filename> and the old root will be unmounted; or a colon-separated pair
404 of new root path and pivot destination for the old root. The new root path will be pivoted to
<filename>/
</filename>,
405 and the old
<filename>/
</filename> will be pivoted to the other directory. Both paths must be absolute, and are resolved
406 in the container's file system namespace.
</para>
408 <para>This is for containers which have several bootable directories in them; for example, several
409 <ulink url=
"https://ostree.readthedocs.io/en/latest/">OSTree
</ulink> deployments. It emulates the behavior of
410 the boot loader and initial RAM disk which normally select which directory to mount as the root and start the
411 container's PID
1 in.
</para></listitem>
415 </refsect2><refsect2>
416 <title>Execution Options
</title>
420 <term><option>-a
</option></term>
421 <term><option>--as-pid2
</option></term>
423 <listitem><para>Invoke the shell or specified program as process ID (PID)
2 instead of PID
1 (init). By
424 default, if neither this option nor
<option>--boot
</option> is used, the selected program is run as the process
425 with PID
1, a mode only suitable for programs that are aware of the special semantics that the process with
426 PID
1 has on UNIX. For example, it needs to reap all processes reparented to it, and should implement
427 <command>sysvinit
</command> compatible signal handling (specifically: it needs to reboot on SIGINT, reexecute
428 on SIGTERM, reload configuration on SIGHUP, and so on). With
<option>--as-pid2
</option> a minimal stub init
429 process is run as PID
1 and the selected program is executed as PID
2 (and hence does not need to implement any
430 special semantics). The stub init process will reap processes as necessary and react appropriately to
431 signals. It is recommended to use this mode to invoke arbitrary commands in containers, unless they have been
432 modified to run correctly as PID
1. Or in other words: this switch should be used for pretty much all commands,
433 except when the command refers to an init or shell implementation, as these are generally capable of running
434 correctly as PID
1. This option may not be combined with
<option>--boot
</option>.
</para>
439 <term><option>-b
</option></term>
440 <term><option>--boot
</option></term>
442 <listitem><para>Automatically search for an init program and invoke it as PID
1, instead of a shell or a user
443 supplied program. If this option is used, arguments specified on the command line are used as arguments for the
444 init program. This option may not be combined with
<option>--as-pid2
</option>.
</para>
446 <para>The following table explains the different modes of invocation and relationship to
447 <option>--as-pid2
</option> (see above):
</para>
450 <title>Invocation Mode
</title>
451 <tgroup cols='
2' align='left' colsep='
1' rowsep='
1'
>
452 <colspec colname=
"switch" />
453 <colspec colname=
"explanation" />
456 <entry>Switch
</entry>
457 <entry>Explanation
</entry>
462 <entry>Neither
<option>--as-pid2
</option> nor
<option>--boot
</option> specified
</entry>
463 <entry>The passed parameters are interpreted as the command line, which is executed as PID
1 in the container.
</entry>
467 <entry><option>--as-pid2
</option> specified
</entry>
468 <entry>The passed parameters are interpreted as the command line, which is executed as PID
2 in the container. A stub init process is run as PID
1.
</entry>
472 <entry><option>--boot
</option> specified
</entry>
473 <entry>An init program is automatically searched for and run as PID
1 in the container. The passed parameters are used as invocation parameters for this process.
</entry>
480 <para>Note that
<option>--boot
</option> is the default mode of operation if the
481 <filename>systemd-nspawn@.service
</filename> template unit file is used.
</para>
486 <term><option>--chdir=
</option></term>
488 <listitem><para>Change to the specified working directory before invoking the process in the container. Expects
489 an absolute path in the container's file system namespace.
</para></listitem>
493 <term><option>-E
<replaceable>NAME
</replaceable>=
<replaceable>VALUE
</replaceable></option></term>
494 <term><option>--setenv=
<replaceable>NAME
</replaceable>=
<replaceable>VALUE
</replaceable></option></term>
496 <listitem><para>Specifies an environment variable assignment
497 to pass to the init process in the container, in the format
498 <literal>NAME=VALUE
</literal>. This may be used to override
499 the default variables or to set additional variables. This
500 parameter may be used more than once.
</para></listitem>
504 <term><option>-u
</option></term>
505 <term><option>--user=
</option></term>
507 <listitem><para>After transitioning into the container, change
508 to the specified user-defined in the container's user
509 database. Like all other systemd-nspawn features, this is not
510 a security feature and provides protection against accidental
511 destructive operations only.
</para></listitem>
515 <term><option>--kill-signal=
</option></term>
517 <listitem><para>Specify the process signal to send to the container's PID
1 when nspawn itself receives
518 <constant>SIGTERM
</constant>, in order to trigger an orderly shutdown of the container. Defaults to
519 <constant>SIGRTMIN+
3</constant> if
<option>--boot
</option> is used (on systemd-compatible init systems
520 <constant>SIGRTMIN+
3</constant> triggers an orderly shutdown). If
<option>--boot
</option> is not used and this
521 option is not specified the container's processes are terminated abruptly via
<constant>SIGKILL
</constant>. For
522 a list of valid signals, see
<citerefentry
523 project='man-pages'
><refentrytitle>signal
</refentrytitle><manvolnum>7</manvolnum></citerefentry>.
</para></listitem>
527 <term><option>--notify-ready=
</option></term>
529 <listitem><para>Configures support for notifications from the container's init process.
530 <option>--notify-ready=
</option> takes a boolean (
<option>no
</option> and
<option>yes
</option>).
531 With option
<option>no
</option> systemd-nspawn notifies systemd
532 with a
<literal>READY=
1</literal> message when the init process is created.
533 With option
<option>yes
</option> systemd-nspawn waits for the
534 <literal>READY=
1</literal> message from the init process in the container
535 before sending its own to systemd. For more details about notifications
536 see
<citerefentry><refentrytitle>sd_notify
</refentrytitle><manvolnum>3</manvolnum></citerefentry>).
</para></listitem>
540 </refsect2><refsect2>
541 <title>System Identity Options
</title>
545 <term><option>-M
</option></term>
546 <term><option>--machine=
</option></term>
548 <listitem><para>Sets the machine name for this container. This
549 name may be used to identify this container during its runtime
550 (for example in tools like
551 <citerefentry><refentrytitle>machinectl
</refentrytitle><manvolnum>1</manvolnum></citerefentry>
552 and similar), and is used to initialize the container's
553 hostname (which the container can choose to override,
554 however). If not specified, the last component of the root
555 directory path of the container is used, possibly suffixed
556 with a random identifier in case
<option>--ephemeral
</option>
557 mode is selected. If the root directory selected is the host's
558 root directory the host's hostname is used as default
559 instead.
</para></listitem>
563 <term><option>--hostname=
</option></term>
565 <listitem><para>Controls the hostname to set within the container, if different from the machine name. Expects
566 a valid hostname as argument. If this option is used, the kernel hostname of the container will be set to this
567 value, otherwise it will be initialized to the machine name as controlled by the
<option>--machine=
</option>
568 option described above. The machine name is used for various aspect of identification of the container from the
569 outside, the kernel hostname configurable with this option is useful for the container to identify itself from
570 the inside. It is usually a good idea to keep both forms of identification synchronized, in order to avoid
571 confusion. It is hence recommended to avoid usage of this option, and use
<option>--machine=
</option>
572 exclusively. Note that regardless whether the container's hostname is initialized from the name set with
573 <option>--hostname=
</option> or the one set with
<option>--machine=
</option>, the container can later override
574 its kernel hostname freely on its own as well.
</para>
579 <term><option>--uuid=
</option></term>
581 <listitem><para>Set the specified UUID for the container. The
582 init system will initialize
583 <filename>/etc/machine-id
</filename> from this if this file is
584 not set yet. Note that this option takes effect only if
585 <filename>/etc/machine-id
</filename> in the container is
586 unpopulated.
</para></listitem>
590 </refsect2><refsect2>
591 <title>Property Options
</title>
595 <term><option>-S
</option></term>
596 <term><option>--slice=
</option></term>
598 <listitem><para>Make the container part of the specified slice, instead of the default
599 <filename>machine.slice
</filename>. This applies only if the machine is run in its own scope unit, i.e. if
600 <option>--keep-unit
</option> isn't used.
</para>
605 <term><option>--property=
</option></term>
607 <listitem><para>Set a unit property on the scope unit to register for the machine. This applies only if the
608 machine is run in its own scope unit, i.e. if
<option>--keep-unit
</option> isn't used. Takes unit property
609 assignments in the same format as
<command>systemctl set-property
</command>. This is useful to set memory
610 limits and similar for container.
</para>
615 <term><option>--register=
</option></term>
617 <listitem><para>Controls whether the container is registered with
618 <citerefentry><refentrytitle>systemd-machined
</refentrytitle><manvolnum>8</manvolnum></citerefentry>. Takes a
619 boolean argument, which defaults to
<literal>yes
</literal>. This option should be enabled when the container
620 runs a full Operating System (more specifically: a system and service manager as PID
1), and is useful to
621 ensure that the container is accessible via
622 <citerefentry><refentrytitle>machinectl
</refentrytitle><manvolnum>1</manvolnum></citerefentry> and shown by
623 tools such as
<citerefentry
624 project='man-pages'
><refentrytitle>ps
</refentrytitle><manvolnum>1</manvolnum></citerefentry>. If the container
625 does not run a service manager, it is recommended to set this option to
626 <literal>no
</literal>.
</para></listitem>
630 <term><option>--keep-unit
</option></term>
632 <listitem><para>Instead of creating a transient scope unit to run the container in, simply use the service or
633 scope unit
<command>systemd-nspawn
</command> has been invoked in. If
<option>--register=yes
</option> is set
634 this unit is registered with
635 <citerefentry><refentrytitle>systemd-machined
</refentrytitle><manvolnum>8</manvolnum></citerefentry>. This
636 switch should be used if
<command>systemd-nspawn
</command> is invoked from within a service unit, and the
637 service unit's sole purpose is to run a single
<command>systemd-nspawn
</command> container. This option is not
638 available if run from a user session.
</para>
639 <para>Note that passing
<option>--keep-unit
</option> disables the effect of
<option>--slice=
</option> and
640 <option>--property=
</option>. Use
<option>--keep-unit
</option> and
<option>--register=no
</option> in
641 combination to disable any kind of unit allocation or registration with
642 <command>systemd-machined
</command>.
</para></listitem>
646 </refsect2><refsect2>
647 <title>User Namespacing Options
</title>
651 <term><option>--private-users=
</option></term>
653 <listitem><para>Controls user namespacing. If enabled, the container will run with its own private set of UNIX
654 user and group ids (UIDs and GIDs). This involves mapping the private UIDs/GIDs used in the container (starting
655 with the container's root user
0 and up) to a range of UIDs/GIDs on the host that are not used for other
656 purposes (usually in the range beyond the host's UID/GID
65536). The parameter may be specified as follows:
</para>
659 <listitem><para>If one or two colon-separated numbers are specified, user namespacing is turned on. The first
660 parameter specifies the first host UID/GID to assign to the container, the second parameter specifies the
661 number of host UIDs/GIDs to assign to the container. If the second parameter is omitted,
65536 UIDs/GIDs are
662 assigned.
</para></listitem>
664 <listitem><para>If the parameter is omitted, or true, user namespacing is turned on. The UID/GID range to
665 use is determined automatically from the file ownership of the root directory of the container's directory
666 tree. To use this option, make sure to prepare the directory tree in advance, and ensure that all files and
667 directories in it are owned by UIDs/GIDs in the range you'd like to use. Also, make sure that used file ACLs
668 exclusively reference UIDs/GIDs in the appropriate range. If this mode is used the number of UIDs/GIDs
669 assigned to the container for use is
65536, and the UID/GID of the root directory must be a multiple of
670 65536.
</para></listitem>
672 <listitem><para>If the parameter is false, user namespacing is turned off. This is the default.
</para>
675 <listitem><para>The special value
<literal>pick
</literal> turns on user namespacing. In this case the UID/GID
676 range is automatically chosen. As first step, the file owner of the root directory of the container's
677 directory tree is read, and it is checked that it is currently not used by the system otherwise (in
678 particular, that no other container is using it). If this check is successful, the UID/GID range determined
679 this way is used, similar to the behavior if
"yes" is specified. If the check is not successful (and thus
680 the UID/GID range indicated in the root directory's file owner is already used elsewhere) a new – currently
681 unused – UID/GID range of
65536 UIDs/GIDs is randomly chosen between the host UID/GIDs of
524288 and
682 1878982656, always starting at a multiple of
65536. This setting implies
683 <option>--private-users-chown
</option> (see below), which has the effect that the files and directories in
684 the container's directory tree will be owned by the appropriate users of the range picked. Using this option
685 makes user namespace behavior fully automatic. Note that the first invocation of a previously unused
686 container image might result in picking a new UID/GID range for it, and thus in the (possibly expensive) file
687 ownership adjustment operation. However, subsequent invocations of the container will be cheap (unless of
688 course the picked UID/GID range is assigned to a different use by then).
</para></listitem>
691 <para>It is recommended to assign at least
65536 UIDs/GIDs to each container, so that the usable UID/GID range in the
692 container covers
16 bit. For best security, do not assign overlapping UID/GID ranges to multiple containers. It is
693 hence a good idea to use the upper
16 bit of the host
32-bit UIDs/GIDs as container identifier, while the lower
16
694 bit encode the container UID/GID used. This is in fact the behavior enforced by the
695 <option>--private-users=pick
</option> option.
</para>
697 <para>When user namespaces are used, the GID range assigned to each container is always chosen identical to the
700 <para>In most cases, using
<option>--private-users=pick
</option> is the recommended option as it enhances
701 container security massively and operates fully automatically in most cases.
</para>
703 <para>Note that the picked UID/GID range is not written to
<filename>/etc/passwd
</filename> or
704 <filename>/etc/group
</filename>. In fact, the allocation of the range is not stored persistently anywhere,
705 except in the file ownership of the files and directories of the container.
</para>
707 <para>Note that when user namespacing is used file ownership on disk reflects this, and all of the container's
708 files and directories are owned by the container's effective user and group IDs. This means that copying files
709 from and to the container image requires correction of the numeric UID/GID values, according to the UID/GID
710 shift applied.
</para></listitem>
714 <term><option>--private-users-chown
</option></term>
716 <listitem><para>If specified, all files and directories in the container's directory tree will be
717 adjusted so that they are owned by the appropriate UIDs/GIDs selected for the container (see above).
718 This operation is potentially expensive, as it involves iterating through the full directory tree of
719 the container. Besides actual file ownership, file ACLs are adjusted as well.
</para>
721 <para>This option is implied if
<option>--private-users=pick
</option> is used. This option has no effect if
722 user namespacing is not used.
</para></listitem>
726 <term><option>-U
</option></term>
728 <listitem><para>If the kernel supports the user namespaces feature, equivalent to
729 <option>--private-users=pick --private-users-chown
</option>, otherwise equivalent to
730 <option>--private-users=no
</option>.
</para>
732 <para>Note that
<option>-U
</option> is the default if the
733 <filename>systemd-nspawn@.service
</filename> template unit file is used.
</para>
735 <para>Note: it is possible to undo the effect of
<option>--private-users-chown
</option> (or
736 <option>-U
</option>) on the file system by redoing the operation with the first UID of
0:
</para>
738 <programlisting>systemd-nspawn … --private-users=
0 --private-users-chown
</programlisting>
744 </refsect2><refsect2>
745 <title>Networking Options
</title>
750 <term><option>--private-network
</option></term>
752 <listitem><para>Disconnect networking of the container from
753 the host. This makes all network interfaces unavailable in the
754 container, with the exception of the loopback device and those
755 specified with
<option>--network-interface=
</option> and
756 configured with
<option>--network-veth
</option>. If this
757 option is specified, the
<constant>CAP_NET_ADMIN
</constant> capability will be
758 added to the set of capabilities the container retains. The
759 latter may be disabled by using
<option>--drop-capability=
</option>.
760 If this option is not specified (or implied by one of the options
761 listed below), the container will have full access to the host network.
766 <term><option>--network-interface=
</option></term>
768 <listitem><para>Assign the specified network interface to the
769 container. This will remove the specified interface from the
770 calling namespace and place it in the container. When the
771 container terminates, it is moved back to the host namespace.
772 Note that
<option>--network-interface=
</option> implies
773 <option>--private-network
</option>. This option may be used
774 more than once to add multiple network interfaces to the
775 container.
</para></listitem>
779 <term><option>--network-macvlan=
</option></term>
781 <listitem><para>Create a
<literal>macvlan
</literal> interface
782 of the specified Ethernet network interface and add it to the
783 container. A
<literal>macvlan
</literal> interface is a virtual
784 interface that adds a second MAC address to an existing
785 physical Ethernet link. The interface in the container will be
786 named after the interface on the host, prefixed with
787 <literal>mv-
</literal>. Note that
788 <option>--network-macvlan=
</option> implies
789 <option>--private-network
</option>. This option may be used
790 more than once to add multiple network interfaces to the
791 container.
</para></listitem>
795 <term><option>--network-ipvlan=
</option></term>
797 <listitem><para>Create an
<literal>ipvlan
</literal> interface
798 of the specified Ethernet network interface and add it to the
799 container. An
<literal>ipvlan
</literal> interface is a virtual
800 interface, similar to a
<literal>macvlan
</literal> interface,
801 which uses the same MAC address as the underlying interface.
802 The interface in the container will be named after the
803 interface on the host, prefixed with
<literal>iv-
</literal>.
804 Note that
<option>--network-ipvlan=
</option> implies
805 <option>--private-network
</option>. This option may be used
806 more than once to add multiple network interfaces to the
807 container.
</para></listitem>
811 <term><option>-n
</option></term>
812 <term><option>--network-veth
</option></term>
814 <listitem><para>Create a virtual Ethernet link (
<literal>veth
</literal>) between host and container. The host
815 side of the Ethernet link will be available as a network interface named after the container's name (as
816 specified with
<option>--machine=
</option>), prefixed with
<literal>ve-
</literal>. The container side of the
817 Ethernet link will be named
<literal>host0
</literal>. The
<option>--network-veth
</option> option implies
818 <option>--private-network
</option>.
</para>
821 <citerefentry><refentrytitle>systemd-networkd.service
</refentrytitle><manvolnum>8</manvolnum></citerefentry>
822 includes by default a network file
<filename>/usr/lib/systemd/network/
80-container-ve.network
</filename>
823 matching the host-side interfaces created this way, which contains settings to enable automatic address
824 provisioning on the created virtual link via DHCP, as well as automatic IP routing onto the host's external
825 network interfaces. It also contains
<filename>/usr/lib/systemd/network/
80-container-host0.network
</filename>
826 matching the container-side interface created this way, containing settings to enable client side address
827 assignment via DHCP. In case
<filename>systemd-networkd
</filename> is running on both the host and inside the
828 container, automatic IP communication from the container to the host is thus available, with further
829 connectivity to the external network.
</para>
831 <para>Note that
<option>--network-veth
</option> is the default if the
832 <filename>systemd-nspawn@.service
</filename> template unit file is used.
</para>
834 <para>Note that on Linux network interface names may have a length of
15 characters at maximum, while
835 container names may have a length up to
64 characters. As this option derives the host-side interface
836 name from the container name the name is possibly truncated. Thus, care needs to be taken to ensure
837 that interface names remain unique in this case, or even better container names are generally not
838 chosen longer than
12 characters, to avoid the truncation. If the name is truncated,
839 <command>systemd-nspawn
</command> will automatically append a
4-digit hash value to the name to
840 reduce the chance of collisions. However, the hash algorithm is not collision-free. (See
841 <citerefentry><refentrytitle>systemd.net-naming-scheme
</refentrytitle><manvolnum>7</manvolnum></citerefentry>
842 for details on older naming algorithms for this interface). Alternatively, the
843 <option>--network-veth-extra=
</option> option may be used, which allows free configuration of the
844 host-side interface name independently of the container name — but might require a bit more
845 additional configuration in case bridging in a fashion similar to
<option>--network-bridge=
</option>
851 <term><option>--network-veth-extra=
</option></term>
853 <listitem><para>Adds an additional virtual Ethernet link
854 between host and container. Takes a colon-separated pair of
855 host interface name and container interface name. The latter
856 may be omitted in which case the container and host sides will
857 be assigned the same name. This switch is independent of
858 <option>--network-veth
</option>, and — in contrast — may be
859 used multiple times, and allows configuration of the network
860 interface names. Note that
<option>--network-bridge=
</option>
861 has no effect on interfaces created with
862 <option>--network-veth-extra=
</option>.
</para></listitem>
866 <term><option>--network-bridge=
</option></term>
868 <listitem><para>Adds the host side of the Ethernet link created with
<option>--network-veth
</option>
869 to the specified Ethernet bridge interface. Expects a valid network interface name of a bridge device
870 as argument. Note that
<option>--network-bridge=
</option> implies
<option>--network-veth
</option>. If
871 this option is used, the host side of the Ethernet link will use the
<literal>vb-
</literal> prefix
872 instead of
<literal>ve-
</literal>. Regardless of the used naming prefix the same network interface
873 name length limits imposed by Linux apply, along with the complications this creates (for details see
874 above).
</para></listitem>
878 <term><option>--network-zone=
</option></term>
880 <listitem><para>Creates a virtual Ethernet link (
<literal>veth
</literal>) to the container and adds it to an
881 automatically managed Ethernet bridge interface. The bridge interface is named after the passed argument,
882 prefixed with
<literal>vz-
</literal>. The bridge interface is automatically created when the first container
883 configured for its name is started, and is automatically removed when the last container configured for its
884 name exits. Hence, each bridge interface configured this way exists only as long as there's at least one
885 container referencing it running. This option is very similar to
<option>--network-bridge=
</option>, besides
886 this automatic creation/removal of the bridge device.
</para>
888 <para>This setting makes it easy to place multiple related containers on a common, virtual Ethernet-based
889 broadcast domain, here called a
"zone". Each container may only be part of one zone, but each zone may contain
890 any number of containers. Each zone is referenced by its name. Names may be chosen freely (as long as they form
891 valid network interface names when prefixed with
<literal>vz-
</literal>), and it is sufficient to pass the same
892 name to the
<option>--network-zone=
</option> switch of the various concurrently running containers to join
893 them in one zone.
</para>
896 <citerefentry><refentrytitle>systemd-networkd.service
</refentrytitle><manvolnum>8</manvolnum></citerefentry>
897 includes by default a network file
<filename>/usr/lib/systemd/network/
80-container-vz.network
</filename>
898 matching the bridge interfaces created this way, which contains settings to enable automatic address
899 provisioning on the created virtual network via DHCP, as well as automatic IP routing onto the host's external
900 network interfaces. Using
<option>--network-zone=
</option> is hence in most cases fully automatic and
901 sufficient to connect multiple local containers in a joined broadcast domain to the host, with further
902 connectivity to the external network.
</para>
907 <term><option>--network-namespace-path=
</option></term>
909 <listitem><para>Takes the path to a file representing a kernel
910 network namespace that the container shall run in. The specified path
911 should refer to a (possibly bind-mounted) network namespace file, as
912 exposed by the kernel below
<filename>/proc/$PID/ns/net
</filename>.
913 This makes the container enter the given network namespace. One of the
914 typical use cases is to give a network namespace under
915 <filename>/run/netns
</filename> created by
<citerefentry
916 project='man-pages'
><refentrytitle>ip-netns
</refentrytitle><manvolnum>8</manvolnum></citerefentry>,
917 for example,
<option>--network-namespace-path=/run/netns/foo
</option>.
918 Note that this option cannot be used together with other
919 network-related options, such as
<option>--private-network
</option>
920 or
<option>--network-interface=
</option>.
</para></listitem>
924 <term><option>-p
</option></term>
925 <term><option>--port=
</option></term>
927 <listitem><para>If private networking is enabled, maps an IP
928 port on the host onto an IP port on the container. Takes a
929 protocol specifier (either
<literal>tcp
</literal> or
930 <literal>udp
</literal>), separated by a colon from a host port
931 number in the range
1 to
65535, separated by a colon from a
932 container port number in the range from
1 to
65535. The
933 protocol specifier and its separating colon may be omitted, in
934 which case
<literal>tcp
</literal> is assumed. The container
935 port number and its colon may be omitted, in which case the
936 same port as the host port is implied. This option is only
937 supported if private networking is used, such as with
938 <option>--network-veth
</option>,
<option>--network-zone=
</option>
939 <option>--network-bridge=
</option>.
</para></listitem>
943 </refsect2><refsect2>
944 <title>Security Options
</title>
948 <term><option>--capability=
</option></term>
950 <listitem><para>List one or more additional capabilities to grant the container. Takes a
951 comma-separated list of capability names, see
<citerefentry
952 project='man-pages'
><refentrytitle>capabilities
</refentrytitle><manvolnum>7</manvolnum></citerefentry>
953 for more information. Note that the following capabilities will be granted in any way:
954 <constant>CAP_AUDIT_CONTROL
</constant>,
<constant>CAP_AUDIT_WRITE
</constant>,
955 <constant>CAP_CHOWN
</constant>,
<constant>CAP_DAC_OVERRIDE
</constant>,
956 <constant>CAP_DAC_READ_SEARCH
</constant>,
<constant>CAP_FOWNER
</constant>,
957 <constant>CAP_FSETID
</constant>,
<constant>CAP_IPC_OWNER
</constant>,
<constant>CAP_KILL
</constant>,
958 <constant>CAP_LEASE
</constant>,
<constant>CAP_LINUX_IMMUTABLE
</constant>,
959 <constant>CAP_MKNOD
</constant>,
<constant>CAP_NET_BIND_SERVICE
</constant>,
960 <constant>CAP_NET_BROADCAST
</constant>,
<constant>CAP_NET_RAW
</constant>,
961 <constant>CAP_SETFCAP
</constant>,
<constant>CAP_SETGID
</constant>,
<constant>CAP_SETPCAP
</constant>,
962 <constant>CAP_SETUID
</constant>,
<constant>CAP_SYS_ADMIN
</constant>,
963 <constant>CAP_SYS_BOOT
</constant>,
<constant>CAP_SYS_CHROOT
</constant>,
964 <constant>CAP_SYS_NICE
</constant>,
<constant>CAP_SYS_PTRACE
</constant>,
965 <constant>CAP_SYS_RESOURCE
</constant>,
<constant>CAP_SYS_TTY_CONFIG
</constant>. Also
966 <constant>CAP_NET_ADMIN
</constant> is retained if
<option>--private-network
</option> is specified.
967 If the special value
<literal>all
</literal> is passed, all capabilities are retained.
</para>
969 <para>If the special value of
<literal>help
</literal> is passed, the program will print known
970 capability names and exit.
</para></listitem>
974 <term><option>--drop-capability=
</option></term>
976 <listitem><para>Specify one or more additional capabilities to
977 drop for the container. This allows running the container with
978 fewer capabilities than the default (see
981 <para>If the special value of
<literal>help
</literal> is passed, the program will print known
982 capability names and exit.
</para></listitem>
986 <term><option>--no-new-privileges=
</option></term>
988 <listitem><para>Takes a boolean argument. Specifies the value of the
<constant>PR_SET_NO_NEW_PRIVS
</constant>
989 flag for the container payload. Defaults to off. When turned on the payload code of the container cannot
990 acquire new privileges, i.e. the
"setuid" file bit as well as file system capabilities will not have an effect
991 anymore. See
<citerefentry
992 project='man-pages'
><refentrytitle>prctl
</refentrytitle><manvolnum>2</manvolnum></citerefentry> for details
993 about this flag.
</para></listitem>
997 <term><option>--system-call-filter=
</option></term>
999 <listitem><para>Alter the system call filter applied to containers. Takes a space-separated list of system call
1000 names or group names (the latter prefixed with
<literal>@
</literal>, as listed by the
1001 <command>syscall-filter
</command> command of
1002 <citerefentry><refentrytitle>systemd-analyze
</refentrytitle><manvolnum>1</manvolnum></citerefentry>). Passed
1003 system calls will be permitted. The list may optionally be prefixed by
<literal>~
</literal>, in which case all
1004 listed system calls are prohibited. If this command line option is used multiple times the configured lists are
1005 combined. If both a positive and a negative list (that is one system call list without and one with the
1006 <literal>~
</literal> prefix) are configured, the negative list takes precedence over the positive list. Note
1007 that
<command>systemd-nspawn
</command> always implements a system call whitelist (as opposed to a blacklist),
1008 and this command line option hence adds or removes entries from the default whitelist, depending on the
1009 <literal>~
</literal> prefix. Note that the applied system call filter is also altered implicitly if additional
1010 capabilities are passed using the
<command>--capabilities=
</command>.
</para></listitem>
1014 <term><option>-Z
</option></term>
1015 <term><option>--selinux-context=
</option></term>
1017 <listitem><para>Sets the SELinux security context to be used
1018 to label processes in the container.
</para>
1023 <term><option>-L
</option></term>
1024 <term><option>--selinux-apifs-context=
</option></term>
1026 <listitem><para>Sets the SELinux security context to be used
1027 to label files in the virtual API file systems in the
1033 </refsect2><refsect2>
1034 <title>Resource Options
</title>
1039 <term><option>--rlimit=
</option></term>
1041 <listitem><para>Sets the specified POSIX resource limit for the container payload. Expects an assignment of the
1043 <literal><replaceable>LIMIT
</replaceable>=
<replaceable>SOFT
</replaceable>:
<replaceable>HARD
</replaceable></literal>
1044 or
<literal><replaceable>LIMIT
</replaceable>=
<replaceable>VALUE
</replaceable></literal>, where
1045 <replaceable>LIMIT
</replaceable> should refer to a resource limit type, such as
1046 <constant>RLIMIT_NOFILE
</constant> or
<constant>RLIMIT_NICE
</constant>. The
<replaceable>SOFT
</replaceable> and
1047 <replaceable>HARD
</replaceable> fields should refer to the numeric soft and hard resource limit values. If the
1048 second form is used,
<replaceable>VALUE
</replaceable> may specify a value that is used both as soft and hard
1049 limit. In place of a numeric value the special string
<literal>infinity
</literal> may be used to turn off
1050 resource limiting for the specific type of resource. This command line option may be used multiple times to
1051 control limits on multiple limit types. If used multiple times for the same limit type, the last use
1052 wins. For details about resource limits see
<citerefentry
1053 project='man-pages'
><refentrytitle>setrlimit
</refentrytitle><manvolnum>2</manvolnum></citerefentry>. By default
1054 resource limits for the container's init process (PID
1) are set to the same values the Linux kernel originally
1055 passed to the host init system. Note that some resource limits are enforced on resources counted per user, in
1056 particular
<constant>RLIMIT_NPROC
</constant>. This means that unless user namespacing is deployed
1057 (i.e.
<option>--private-users=
</option> is used, see above), any limits set will be applied to the resource
1058 usage of the same user on all local containers as well as the host. This means particular care needs to be
1059 taken with these limits as they might be triggered by possibly less trusted code. Example:
1060 <literal>--rlimit=RLIMIT_NOFILE=
8192:
16384</literal>.
</para></listitem>
1064 <term><option>--oom-score-adjust=
</option></term>
1066 <listitem><para>Changes the OOM (
"Out Of Memory") score adjustment value for the container payload. This controls
1067 <filename>/proc/self/oom_score_adj
</filename> which influences the preference with which this container is
1068 terminated when memory becomes scarce. For details see
<citerefentry
1069 project='man-pages'
><refentrytitle>proc
</refentrytitle><manvolnum>5</manvolnum></citerefentry>. Takes an
1070 integer in the range -
1000…
1000.
</para></listitem>
1074 <term><option>--cpu-affinity=
</option></term>
1076 <listitem><para>Controls the CPU affinity of the container payload. Takes a comma separated list of CPU numbers
1077 or number ranges (the latter's start and end value separated by dashes). See
<citerefentry
1078 project='man-pages'
><refentrytitle>sched_setaffinity
</refentrytitle><manvolnum>2</manvolnum></citerefentry> for
1079 details.
</para></listitem>
1083 <term><option>--personality=
</option></term>
1085 <listitem><para>Control the architecture (
"personality")
1087 <citerefentry project='man-pages'
><refentrytitle>uname
</refentrytitle><manvolnum>2</manvolnum></citerefentry>
1088 in the container. Currently, only
<literal>x86
</literal> and
1089 <literal>x86-
64</literal> are supported. This is useful when
1090 running a
32-bit container on a
64-bit host. If this setting
1091 is not used, the personality reported in the container is the
1092 same as the one reported on the host.
</para></listitem>
1096 </refsect2><refsect2>
1097 <title>Integration Options
</title>
1101 <term><option>--resolv-conf=
</option></term>
1103 <listitem><para>Configures how
<filename>/etc/resolv.conf
</filename> inside of the container (i.e. DNS
1104 configuration synchronization from host to container) shall be handled. Takes one of
<literal>off
</literal>,
1105 <literal>copy-host
</literal>,
<literal>copy-static
</literal>,
<literal>bind-host
</literal>,
1106 <literal>bind-static
</literal>,
<literal>delete
</literal> or
<literal>auto
</literal>. If set to
1107 <literal>off
</literal> the
<filename>/etc/resolv.conf
</filename> file in the container is left as it is
1108 included in the image, and neither modified nor bind mounted over. If set to
<literal>copy-host
</literal>, the
1109 <filename>/etc/resolv.conf
</filename> file from the host is copied into the container. Similar, if
1110 <literal>bind-host
</literal> is used, the file is bind mounted from the host into the container. If set to
1111 <literal>copy-static
</literal> the static
<filename>resolv.conf
</filename> file supplied with
1112 <citerefentry><refentrytitle>systemd-resolved.service
</refentrytitle><manvolnum>8</manvolnum></citerefentry> is
1113 copied into the container, and correspondingly
<literal>bind-static
</literal> bind mounts it there. If set to
1114 <literal>delete
</literal> the
<filename>/etc/resolv.conf
</filename> file in the container is deleted if it
1115 exists. Finally, if set to
<literal>auto
</literal> the file is left as it is if private networking is turned on
1116 (see
<option>--private-network
</option>). Otherwise, if
<filename>systemd-resolved.service
</filename> is
1117 connectible its static
<filename>resolv.conf
</filename> file is used, and if not the host's
1118 <filename>/etc/resolv.conf
</filename> file is used. In the latter cases the file is copied if the image is
1119 writable, and bind mounted otherwise. It's recommended to use
<literal>copy
</literal> if the container shall be
1120 able to make changes to the DNS configuration on its own, deviating from the host's settings. Otherwise
1121 <literal>bind
</literal> is preferable, as it means direct changes to
<filename>/etc/resolv.conf
</filename> in
1122 the container are not allowed, as it is a read-only bind mount (but note that if the container has enough
1123 privileges, it might simply go ahead and unmount the bind mount anyway). Note that both if the file is bind
1124 mounted and if it is copied no further propagation of configuration is generally done after the one-time early
1125 initialization (this is because the file is usually updated through copying and renaming). Defaults to
1126 <literal>auto
</literal>.
</para></listitem>
1130 <term><option>--timezone=
</option></term>
1132 <listitem><para>Configures how
<filename>/etc/localtime
</filename> inside of the container (i.e. local timezone
1133 synchronization from host to container) shall be handled. Takes one of
<literal>off
</literal>,
1134 <literal>copy
</literal>,
<literal>bind
</literal>,
<literal>symlink
</literal>,
<literal>delete
</literal> or
1135 <literal>auto
</literal>. If set to
<literal>off
</literal> the
<filename>/etc/localtime
</filename> file in the
1136 container is left as it is included in the image, and neither modified nor bind mounted over. If set to
1137 <literal>copy
</literal> the
<filename>/etc/localtime
</filename> file of the host is copied into the
1138 container. Similar, if
<literal>bind
</literal> is used, it is bind mounted from the host into the container. If
1139 set to
<literal>symlink
</literal> a symlink from
<filename>/etc/localtime
</filename> in the container is
1140 created pointing to the matching the timezone file of the container that matches the timezone setting on the
1141 host. If set to
<literal>delete
</literal> the file in the container is deleted, should it exist. If set to
1142 <literal>auto
</literal> and the
<filename>/etc/localtime
</filename> file of the host is a symlink, then
1143 <literal>symlink
</literal> mode is used, and
<literal>copy
</literal> otherwise, except if the image is
1144 read-only in which case
<literal>bind
</literal> is used instead. Defaults to
1145 <literal>auto
</literal>.
</para></listitem>
1149 <term><option>--link-journal=
</option></term>
1151 <listitem><para>Control whether the container's journal shall
1152 be made visible to the host system. If enabled, allows viewing
1153 the container's journal files from the host (but not vice
1154 versa). Takes one of
<literal>no
</literal>,
1155 <literal>host
</literal>,
<literal>try-host
</literal>,
1156 <literal>guest
</literal>,
<literal>try-guest
</literal>,
1157 <literal>auto
</literal>. If
<literal>no
</literal>, the journal
1158 is not linked. If
<literal>host
</literal>, the journal files
1159 are stored on the host file system (beneath
1160 <filename>/var/log/journal/
<replaceable>machine-id
</replaceable></filename>)
1161 and the subdirectory is bind-mounted into the container at the
1162 same location. If
<literal>guest
</literal>, the journal files
1163 are stored on the guest file system (beneath
1164 <filename>/var/log/journal/
<replaceable>machine-id
</replaceable></filename>)
1165 and the subdirectory is symlinked into the host at the same
1166 location.
<literal>try-host
</literal> and
1167 <literal>try-guest
</literal> do the same but do not fail if
1168 the host does not have persistent journaling enabled. If
1169 <literal>auto
</literal> (the default), and the right
1170 subdirectory of
<filename>/var/log/journal
</filename> exists,
1171 it will be bind mounted into the container. If the
1172 subdirectory does not exist, no linking is performed.
1173 Effectively, booting a container once with
1174 <literal>guest
</literal> or
<literal>host
</literal> will link
1175 the journal persistently if further on the default of
1176 <literal>auto
</literal> is used.
</para>
1178 <para>Note that
<option>--link-journal=try-guest
</option> is the default if the
1179 <filename>systemd-nspawn@.service
</filename> template unit file is used.
</para></listitem>
1183 <term><option>-j
</option></term>
1185 <listitem><para>Equivalent to
1186 <option>--link-journal=try-guest
</option>.
</para></listitem>
1191 </refsect2><refsect2>
1192 <title>Mount Options
</title>
1197 <term><option>--bind=
</option></term>
1198 <term><option>--bind-ro=
</option></term>
1200 <listitem><para>Bind mount a file or directory from the host into the container. Takes one of: a path
1201 argument — in which case the specified path will be mounted from the host to the same path in the container, or
1202 a colon-separated pair of paths — in which case the first specified path is the source in the host, and the
1203 second path is the destination in the container, or a colon-separated triple of source path, destination path
1204 and mount options. The source path may optionally be prefixed with a
<literal>+
</literal> character. If so, the
1205 source path is taken relative to the image's root directory. This permits setting up bind mounts within the
1206 container image. The source path may be specified as empty string, in which case a temporary directory below
1207 the host's
<filename>/var/tmp
</filename> directory is used. It is automatically removed when the container is
1208 shut down. Mount options are comma-separated and currently, only
<option>rbind
</option> and
1209 <option>norbind
</option> are allowed, controlling whether to create a recursive or a regular bind
1210 mount. Defaults to
"rbind". Backslash escapes are interpreted, so
<literal>\:
</literal> may be used to embed
1211 colons in either path. This option may be specified multiple times for creating multiple independent bind
1212 mount points. The
<option>--bind-ro=
</option> option creates read-only bind mounts.
</para>
1214 <para>Note that when this option is used in combination with
<option>--private-users
</option>, the resulting
1215 mount points will be owned by the
<constant>nobody
</constant> user. That's because the mount and its files and
1216 directories continue to be owned by the relevant host users and groups, which do not exist in the container,
1217 and thus show up under the wildcard UID
65534 (nobody). If such bind mounts are created, it is recommended to
1218 make them read-only, using
<option>--bind-ro=
</option>.
</para></listitem>
1222 <term><option>--inaccessible=
</option></term>
1224 <listitem><para>Make the specified path inaccessible in the container. This over-mounts the specified path
1225 (which must exist in the container) with a file node of the same type that is empty and has the most
1226 restrictive access mode supported. This is an effective way to mask files, directories and other file system
1227 objects from the container payload. This option may be used more than once in case all specified paths are
1228 masked.
</para></listitem>
1232 <term><option>--tmpfs=
</option></term>
1234 <listitem><para>Mount a tmpfs file system into the container. Takes a single absolute path argument that
1235 specifies where to mount the tmpfs instance to (in which case the directory access mode will be chosen as
0755,
1236 owned by root/root), or optionally a colon-separated pair of path and mount option string that is used for
1237 mounting (in which case the kernel default for access mode and owner will be chosen, unless otherwise
1238 specified). Backslash escapes are interpreted in the path, so
<literal>\:
</literal> may be used to embed colons
1241 <para>Note that this option cannot be used to replace the root file system of the container with a temporary
1242 file system. However, the
<option>--volatile=
</option> option described below provides similar
1243 functionality, with a focus on implementing stateless operating system images.
</para></listitem>
1247 <term><option>--overlay=
</option></term>
1248 <term><option>--overlay-ro=
</option></term>
1250 <listitem><para>Combine multiple directory trees into one
1251 overlay file system and mount it into the container. Takes a
1252 list of colon-separated paths to the directory trees to
1253 combine and the destination mount point.
</para>
1255 <para>Backslash escapes are interpreted in the paths, so
1256 <literal>\:
</literal> may be used to embed colons in the paths.
1259 <para>If three or more paths are specified, then the last
1260 specified path is the destination mount point in the
1261 container, all paths specified before refer to directory trees
1262 on the host and are combined in the specified order into one
1263 overlay file system. The left-most path is hence the lowest
1264 directory tree, the second-to-last path the highest directory
1265 tree in the stacking order. If
<option>--overlay-ro=
</option>
1266 is used instead of
<option>--overlay=
</option>, a read-only
1267 overlay file system is created. If a writable overlay file
1268 system is created, all changes made to it are written to the
1269 highest directory tree in the stacking order, i.e. the
1270 second-to-last specified.
</para>
1272 <para>If only two paths are specified, then the second
1273 specified path is used both as the top-level directory tree in
1274 the stacking order as seen from the host, as well as the mount
1275 point for the overlay file system in the container. At least
1276 two paths have to be specified.
</para>
1278 <para>The source paths may optionally be prefixed with
<literal>+
</literal> character. If so they are taken
1279 relative to the image's root directory. The uppermost source path may also be specified as empty string, in
1280 which case a temporary directory below the host's
<filename>/var/tmp
</filename> is used. The directory is
1281 removed automatically when the container is shut down. This behaviour is useful in order to make read-only
1282 container directories writable while the container is running. For example, use the
1283 <literal>--overlay=+/var::/var
</literal> option in order to automatically overlay a writable temporary
1284 directory on a read-only
<filename>/var
</filename> directory.
</para>
1286 <para>For details about overlay file systems, see
<ulink
1287 url=
"https://www.kernel.org/doc/Documentation/filesystems/overlayfs.txt">overlayfs.txt
</ulink>. Note
1288 that the semantics of overlay file systems are substantially
1289 different from normal file systems, in particular regarding
1290 reported device and inode information. Device and inode
1291 information may change for a file while it is being written
1292 to, and processes might see out-of-date versions of files at
1293 times. Note that this switch automatically derives the
1294 <literal>workdir=
</literal> mount option for the overlay file
1295 system from the top-level directory tree, making it a sibling
1296 of it. It is hence essential that the top-level directory tree
1297 is not a mount point itself (since the working directory must
1298 be on the same file system as the top-most directory
1299 tree). Also note that the
<literal>lowerdir=
</literal> mount
1300 option receives the paths to stack in the opposite order of
1303 <para>Note that this option cannot be used to replace the root file system of the container with an overlay
1304 file system. However, the
<option>--volatile=
</option> option described above provides similar functionality,
1305 with a focus on implementing stateless operating system images.
</para></listitem>
1309 </refsect2><refsect2>
1310 <title>Input/Output Options
</title>
1314 <term><option>--console=
</option><replaceable>MODE
</replaceable></term>
1316 <listitem><para>Configures how to set up standard input, output and error output for the container
1317 payload, as well as the
<filename>/dev/console
</filename> device for the container. Takes one of
1318 <option>interactive
</option>,
<option>read-only
</option>,
<option>passive
</option>, or
1319 <option>pipe
</option>. If
<option>interactive
</option>, a pseudo-TTY is allocated and made available
1320 as
<filename>/dev/console
</filename> in the container. It is then bi-directionally connected to the
1321 standard input and output passed to
<command>systemd-nspawn
</command>.
<option>read-only
</option> is
1322 similar but only the output of the container is propagated and no input from the caller is read. If
1323 <option>passive
</option>, a pseudo TTY is allocated, but it is not connected anywhere. Finally, in
1324 <option>pipe
</option> mode no pseudo TTY is allocated, but the standard input, output and error
1325 output file descriptors passed to
<command>systemd-nspawn
</command> are passed on — as they are — to
1326 the container payload, see the following paragraph. Defaults to
<option>interactive
</option> if
1327 <command>systemd-nspawn
</command> is invoked from a terminal, and
<option>read-only
</option>
1330 <para>In
<option>pipe
</option> mode,
<filename>/dev/console
</filename> will not exist in the
1331 container. This means that the container payload generally cannot be a full init system as init
1332 systems tend to require
<filename>/dev/console
</filename> to be available. On the other hand, in this
1333 mode container invocations can be used within shell pipelines. This is because intermediary pseudo
1334 TTYs do not permit independent bidirectional propagation of the end-of-file (EOF) condition, which is
1335 necessary for shell pipelines to work correctly.
<emphasis>Note that the
<option>pipe
</option> mode
1336 should be used carefully
</emphasis>, as passing arbitrary file descriptors to less trusted container
1337 payloads might open up unwanted interfaces for access by the container payload. For example, if a
1338 passed file descriptor refers to a TTY of some form, APIs such as
<constant>TIOCSTI
</constant> may be
1339 used to synthesize input that might be used for escaping the container. Hence
<option>pipe
</option>
1340 mode should only be used if the payload is sufficiently trusted or when the standard
1341 input/output/error output file descriptors are known safe, for example pipes.
</para></listitem>
1345 <term><option>--pipe
</option></term>
1346 <term><option>-P
</option></term>
1348 <listitem><para>Equivalent to
<option>--console=pipe
</option>.
</para></listitem>
1351 <xi:include href=
"standard-options.xml" xpointer=
"no-pager" />
1352 <xi:include href=
"standard-options.xml" xpointer=
"help" />
1353 <xi:include href=
"standard-options.xml" xpointer=
"version" />
1358 <xi:include href=
"less-variables.xml" />
1361 <title>Examples
</title>
1365 <ulink url=
"https://getfedora.org">Fedora
</ulink> image and start a shell in it
</title>
1367 <programlisting># machinectl pull-raw --verify=no \
1368 https://download.fedoraproject.org/pub/fedora/linux/releases/
&fedora_latest_version;/Cloud/x86_64/images/Fedora-Cloud-Base-
&fedora_latest_version;-
&fedora_cloud_release;.x86_64.raw.xz \
1369 Fedora-Cloud-Base-
&fedora_latest_version;-
&fedora_cloud_release;.x86-
64
1370 # systemd-nspawn -M Fedora-Cloud-Base-
&fedora_latest_version;-
&fedora_cloud_release;.x86-
64</programlisting>
1372 <para>This downloads an image using
1373 <citerefentry><refentrytitle>machinectl
</refentrytitle><manvolnum>1</manvolnum></citerefentry>
1374 and opens a shell in it.
</para>
1378 <title>Build and boot a minimal Fedora distribution in a container
</title>
1380 <programlisting># dnf -y --releasever=
&fedora_latest_version; --installroot=/var/lib/machines/f
&fedora_latest_version; \
1381 --disablerepo='*' --enablerepo=fedora --enablerepo=updates install \
1382 systemd passwd dnf fedora-release vim-minimal
1383 # systemd-nspawn -bD /var/lib/machines/f
&fedora_latest_version;</programlisting>
1385 <para>This installs a minimal Fedora distribution into the
1386 directory
<filename index=
"false">/var/lib/machines/f
&fedora_latest_version;</filename>
1387 and then boots an OS in a namespace container in it. Because the installation
1388 is located underneath the standard
<filename>/var/lib/machines/
</filename>
1389 directory, it is also possible to start the machine using
1390 <command>systemd-nspawn -M f
&fedora_latest_version;</command>.
</para>
1394 <title>Spawn a shell in a container of a minimal Debian unstable distribution
</title>
1396 <programlisting># debootstrap unstable ~/debian-tree/
1397 # systemd-nspawn -D ~/debian-tree/
</programlisting>
1399 <para>This installs a minimal Debian unstable distribution into
1400 the directory
<filename>~/debian-tree/
</filename> and then
1401 spawns a shell in a namespace container in it.
</para>
1403 <para><command>debootstrap
</command> supports
1404 <ulink url=
"https://www.debian.org">Debian
</ulink>,
1405 <ulink url=
"https://www.ubuntu.com">Ubuntu
</ulink>,
1406 and
<ulink url=
"https://www.tanglu.org">Tanglu
</ulink>
1407 out of the box, so the same command can be used to install any of those. For other
1408 distributions from the Debian family, a mirror has to be specified, see
1409 <citerefentry project='die-net'
><refentrytitle>debootstrap
</refentrytitle><manvolnum>8</manvolnum></citerefentry>.
1414 <title>Boot a minimal
1415 <ulink url=
"https://www.archlinux.org">Arch Linux
</ulink> distribution in a container
</title>
1417 <programlisting># pacstrap -c ~/arch-tree/ base
1418 # systemd-nspawn -bD ~/arch-tree/
</programlisting>
1420 <para>This installs a minimal Arch Linux distribution into the
1421 directory
<filename>~/arch-tree/
</filename> and then boots an OS
1422 in a namespace container in it.
</para>
1427 <ulink url=
"https://software.opensuse.org/distributions/tumbleweed">OpenSUSE Tumbleweed
</ulink>
1428 rolling distribution
</title>
1430 <programlisting># zypper --root=/var/lib/machines/tumbleweed ar -c \
1431 https://download.opensuse.org/tumbleweed/repo/oss tumbleweed
1432 # zypper --root=/var/lib/machines/tumbleweed refresh
1433 # zypper --root=/var/lib/machines/tumbleweed install --no-recommends \
1434 systemd shadow zypper openSUSE-release vim
1435 # systemd-nspawn -M tumbleweed passwd root
1436 # systemd-nspawn -M tumbleweed -b
</programlisting>
1440 <title>Boot into an ephemeral snapshot of the host system
</title>
1442 <programlisting># systemd-nspawn -D / -xb
</programlisting>
1444 <para>This runs a copy of the host system in a snapshot which is removed immediately when the container
1445 exits. All file system changes made during runtime will be lost on shutdown, hence.
</para>
1449 <title>Run a container with SELinux sandbox security contexts
</title>
1451 <programlisting># chcon system_u:object_r:svirt_sandbox_file_t:s0:c0,c1 -R /srv/container
1452 # systemd-nspawn -L system_u:object_r:svirt_sandbox_file_t:s0:c0,c1 \
1453 -Z system_u:system_r:svirt_lxc_net_t:s0:c0,c1 -D /srv/container /bin/sh
</programlisting>
1457 <title>Run a container with an OSTree deployment
</title>
1459 <programlisting># systemd-nspawn -b -i ~/image.raw \
1460 --pivot-root=/ostree/deploy/$OS/deploy/$CHECKSUM:/sysroot \
1461 --bind=+/sysroot/ostree/deploy/$OS/var:/var
</programlisting>
1466 <title>Exit status
</title>
1468 <para>The exit code of the program executed in the container is
1473 <title>See Also
</title>
1475 <citerefentry><refentrytitle>systemd
</refentrytitle><manvolnum>1</manvolnum></citerefentry>,
1476 <citerefentry><refentrytitle>systemd.nspawn
</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
1477 <citerefentry project='man-pages'
><refentrytitle>chroot
</refentrytitle><manvolnum>1</manvolnum></citerefentry>,
1478 <citerefentry project='mankier'
><refentrytitle>dnf
</refentrytitle><manvolnum>8</manvolnum></citerefentry>,
1479 <citerefentry project='die-net'
><refentrytitle>debootstrap
</refentrytitle><manvolnum>8</manvolnum></citerefentry>,
1480 <citerefentry project='archlinux'
><refentrytitle>pacman
</refentrytitle><manvolnum>8</manvolnum></citerefentry>,
1481 <citerefentry project='mankier'
><refentrytitle>zypper
</refentrytitle><manvolnum>8</manvolnum></citerefentry>,
1482 <citerefentry><refentrytitle>systemd.slice
</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
1483 <citerefentry><refentrytitle>machinectl
</refentrytitle><manvolnum>1</manvolnum></citerefentry>,
1484 <citerefentry project='man-pages'
><refentrytitle>btrfs
</refentrytitle><manvolnum>8</manvolnum></citerefentry>
projects / thirdparty / systemd.git / blob