1SYSTEMD-NSPAWN(1) systemd-nspawn SYSTEMD-NSPAWN(1)
2
6 systemd-nspawn - Spawn a command or OS in a light-weight container
7
9 systemd-nspawn [OPTIONS...] [COMMAND [ARGS...]]
10 systemd-nspawn --boot [OPTIONS...] [ARGS...]
12
14 systemd-nspawn may be used to run a command or OS in a light-weight
15 namespace container. In many ways it is similar to chroot(1), but more
16 powerful since it fully virtualizes the file system hierarchy, as well
17 as the process tree, the various IPC subsystems and the host and domain
18 name.
19 systemd-nspawn may be invoked on any directory tree containing an
21 operating system tree, using the --directory= command line option. By
22 using the --machine= option an OS tree is automatically searched for in
23 a couple of locations, most importantly in /var/lib/machines/, the
24 suggested directory to place OS container images installed on the
25 system.
26 In contrast to chroot(1) systemd-nspawn may be used to boot full
28 Linux-based operating systems in a container.
29 systemd-nspawn limits access to various kernel interfaces in the
31 container to read-only, such as /sys/, /proc/sys/ or /sys/fs/selinux/.
32 The host's network interfaces and the system clock may not be changed
33 from within the container. Device nodes may not be created. The host
34 system cannot be rebooted and kernel modules may not be loaded from
35 within the container.
36 Use a tool like dnf(8), debootstrap(8), or pacman(8) to set up an OS
38 directory tree suitable as file system hierarchy for systemd-nspawn
39 containers. See the Examples section below for details on suitable
40 invocation of these commands.
41 As a safety check systemd-nspawn will verify the existence of
43 /usr/lib/os-release or /etc/os-release in the container tree before
44 booting a container (see os-release(5)). It might be necessary to add
45 this file to the container tree manually if the OS of the container is
46 too old to contain this file out-of-the-box.
47 systemd-nspawn may be invoked directly from the interactive command
49 line or run as system service in the background. In this mode each
50 container instance runs as its own service instance; a default template
51 unit file systemd-nspawn@.service is provided to make this easy, taking
52 the container name as instance identifier. Note that different default
53 options apply when systemd-nspawn is invoked by the template unit file
54 than interactively on the command line. Most importantly the template
55 unit file makes use of the --boot option which is not the default in
56 case systemd-nspawn is invoked from the interactive command line.
57 Further differences with the defaults are documented along with the
58 various supported options below.
59 The machinectl(1) tool may be used to execute a number of operations on
61 containers. In particular it provides easy-to-use commands to run
62 containers as system services using the systemd-nspawn@.service
63 template unit file.
64 Along with each container a settings file with the .nspawn suffix may
66 exist, containing additional settings to apply when running the
67 container. See systemd.nspawn(5) for details. Settings files override
68 the default options used by the systemd-nspawn@.service template unit
69 file, making it usually unnecessary to alter this template file
70 directly.
71 Note that systemd-nspawn will mount file systems private to the
73 container to /dev/, /run/ and similar. These will not be visible
74 outside of the container, and their contents will be lost when the
75 container exits.
76 Note that running two systemd-nspawn containers from the same directory
78 tree will not make processes in them see each other. The PID namespace
79 separation of the two containers is complete and the containers will
80 share very few runtime objects except for the underlying file system.
81 Rather use machinectl(1)'s login or shell commands to request an
82 additional login session in a running container.
83 systemd-nspawn implements the Container Interface[1] specification.
85 While running, containers invoked with systemd-nspawn are registered
87 with the systemd-machined(8) service that keeps track of running
88 containers, and provides programming interfaces to interact with them.
89
91 If option --boot is specified, the arguments are used as arguments for
92 the init program. Otherwise, COMMAND specifies the program to launch in
93 the container, and the remaining arguments are used as arguments for
94 this program. If --boot is not used and no arguments are specified, a
95 shell is launched in the container.
96 The following options are understood:
98 -q, --quiet
100 Turns off any status output by the tool itself. When this switch is
101 used, the only output from nspawn will be the console output of the
102 container OS itself.
103 --settings=MODE
105 Controls whether systemd-nspawn shall search for and use additional
106 per-container settings from .nspawn files. Takes a boolean or the
107 special values override or trusted.
108 If enabled (the default), a settings file named after the machine
110 (as specified with the --machine= setting, or derived from the
111 directory or image file name) with the suffix .nspawn is searched
112 in /etc/systemd/nspawn/ and /run/systemd/nspawn/. If it is found
113 there, its settings are read and used. If it is not found there, it
114 is subsequently searched in the same directory as the image file or
115 in the immediate parent of the root directory of the container. In
116 this case, if the file is found, its settings will be also read and
117 used, but potentially unsafe settings are ignored. Note that in
118 both these cases, settings on the command line take precedence over
119 the corresponding settings from loaded .nspawn files, if both are
120 specified. Unsafe settings are considered all settings that elevate
121 the container's privileges or grant access to additional resources
122 such as files or directories of the host. For details about the
123 format and contents of .nspawn files, consult systemd.nspawn(5).
124 If this option is set to override, the file is searched, read and
126 used the same way, however, the order of precedence is reversed:
127 settings read from the .nspawn file will take precedence over the
128 corresponding command line options, if both are specified.
129 If this option is set to trusted, the file is searched, read and
131 used the same way, but regardless of being found in
132 /etc/systemd/nspawn/, /run/systemd/nspawn/ or next to the image
133 file or container root directory, all settings will take effect,
134 however, command line arguments still take precedence over
135 corresponding settings.
136 If disabled, no .nspawn file is read and no settings except the
138 ones on the command line are in effect.
139 Image Options
141 -D, --directory=
142 Directory to use as file system root for the container.
143 If neither --directory=, nor --image= is specified the directory is
145 determined by searching for a directory named the same as the
146 machine name specified with --machine=. See machinectl(1) section
147 "Files and Directories" for the precise search path.
148 If neither --directory=, --image=, nor --machine= are specified,
150 the current directory will be used. May not be specified together
151 with --image=.
152 --template=
154 Directory or "btrfs" subvolume to use as template for the
155 container's root directory. If this is specified and the
156 container's root directory (as configured by --directory=) does not
157 yet exist it is created as "btrfs" snapshot (if supported) or plain
158 directory (otherwise) and populated from this template tree.
159 Ideally, the specified template path refers to the root of a
160 "btrfs" subvolume, in which case a simple copy-on-write snapshot is
161 taken, and populating the root directory is instant. If the
162 specified template path does not refer to the root of a "btrfs"
163 subvolume (or not even to a "btrfs" file system at all), the tree
164 is copied (though possibly in a 'reflink' copy-on-write scheme — if
165 the file system supports that), which can be substantially more
166 time-consuming. Note that the snapshot taken is of the specified
167 directory or subvolume, including all subdirectories and subvolumes
168 below it, but excluding any sub-mounts. May not be specified
169 together with --image= or --ephemeral.
170 Note that this switch leaves hostname, machine ID and all other
172 settings that could identify the instance unmodified.
173 -x, --ephemeral
175 If specified, the container is run with a temporary snapshot of its
176 file system that is removed immediately when the container
177 terminates. May not be specified together with --template=.
178 Note that this switch leaves hostname, machine ID and all other
180 settings that could identify the instance unmodified. Please note
181 that — as with --template= — taking the temporary snapshot is more
182 efficient on file systems that support subvolume snapshots or
183 'reflinks' natively ("btrfs" or new "xfs") than on more traditional
184 file systems that do not ("ext4"). Note that the snapshot taken is
185 of the specified directory or subvolume, including all
186 subdirectories and subvolumes below it, but excluding any
187 sub-mounts.
188 With this option no modifications of the container image are
190 retained. Use --volatile= (described below) for other mechanisms to
191 restrict persistency of container images during runtime.
192 -i, --image=
194 Disk image to mount the root directory for the container from.
195 Takes a path to a regular file or to a block device node. The file
196 or block device must contain either:
197 • An MBR partition table with a single partition of type 0x83
199 that is marked bootable.
200 • A GUID partition table (GPT) with a single partition of type
202 0fc63daf-8483-4772-8e79-3d69d8477de4.
203 • A GUID partition table (GPT) with a marked root partition which
205 is mounted as the root directory of the container. Optionally,
206 GPT images may contain a home and/or a server data partition
207 which are mounted to the appropriate places in the container.
208 All these partitions must be identified by the partition types
209 defined by the Discoverable Partitions Specification[2].
210 • No partition table, and a single file system spanning the whole
212 image.
213 On GPT images, if an EFI System Partition (ESP) is discovered, it
215 is automatically mounted to /efi (or /boot as fallback) in case a
216 directory by this name exists and is empty.
217 Partitions encrypted with LUKS are automatically decrypted. Also,
219 on GPT images dm-verity data integrity hash partitions are set up
220 if the root hash for them is specified using the --root-hash=
221 option.
222 Single file system images (i.e. file systems without a surrounding
224 partition table) can be opened using dm-verity if the integrity
225 data is passed using the --root-hash= and --verity-data= (and
226 optionally --root-hash-sig=) options.
227 Any other partitions, such as foreign partitions or swap partitions
229 are not mounted. May not be specified together with --directory=,
230 --template=.
231 --oci-bundle=
233 Takes the path to an OCI runtime bundle to invoke, as specified in
234 the OCI Runtime Specification[3]. In this case no .nspawn file is
235 loaded, and the root directory and various settings are read from
236 the OCI runtime JSON data (but data passed on the command line
237 takes precedence).
238 --read-only
240 Mount the container's root file system (and any other file systems
241 container in the container image) read-only. This has no effect on
242 additional mounts made with --bind=, --tmpfs= and similar options.
243 This mode is implied if the container image file or directory is
244 marked read-only itself. It is also implied if --volatile= is used.
245 In this case the container image on disk is strictly read-only,
246 while changes are permitted but kept non-persistently in memory
247 only. For further details, see below.
248 --volatile, --volatile=MODE
250 Boots the container in volatile mode. When no mode parameter is
251 passed or when mode is specified as yes, full volatile mode is
252 enabled. This means the root directory is mounted as a mostly
253 unpopulated "tmpfs" instance, and /usr/ from the OS tree is mounted
254 into it in read-only mode (the system thus starts up with read-only
255 OS image, but pristine state and configuration, any changes are
256 lost on shutdown). When the mode parameter is specified as state,
257 the OS tree is mounted read-only, but /var/ is mounted as a
258 writable "tmpfs" instance into it (the system thus starts up with
259 read-only OS resources and configuration, but pristine state, and
260 any changes to the latter are lost on shutdown). When the mode
261 parameter is specified as overlay the read-only root file system is
262 combined with a writable tmpfs instance through "overlayfs", so
263 that it appears at it normally would, but any changes are applied
264 to the temporary file system only and lost when the container is
265 terminated. When the mode parameter is specified as no (the
266 default), the whole OS tree is made available writable (unless
267 --read-only is specified, see above).
268 Note that if one of the volatile modes is chosen, its effect is
270 limited to the root file system (or /var/ in case of state), and
271 any other mounts placed in the hierarchy are unaffected —
272 regardless if they are established automatically (e.g. the EFI
273 system partition that might be mounted to /efi/ or /boot/) or
274 explicitly (e.g. through an additional command line option such as
275 --bind=, see below). This means, even if --volatile=overlay is used
276 changes to /efi/ or /boot/ are prohibited in case such a partition
277 exists in the container image operated on, and even if
278 --volatile=state is used the hypothetical file /etc/foobar is
279 potentially writable if --bind=/etc/foobar if used to mount it from
280 outside the read-only container /etc/ directory.
281 The --ephemeral option is closely related to this setting, and
283 provides similar behaviour by making a temporary, ephemeral copy of
284 the whole OS image and executing that. For further details, see
285 above.
286 The --tmpfs= and --overlay= options provide similar functionality,
288 but for specific sub-directories of the OS image only. For details,
289 see below.
290 This option provides similar functionality for containers as the
292 "systemd.volatile=" kernel command line switch provides for host
293 systems. See kernel-command-line(7) for details.
294 Note that setting this option to yes or state will only work
296 correctly with operating systems in the container that can boot up
297 with only /usr/ mounted, and are able to automatically populate
298 /var/ (and /etc/ in case of "--volatile=yes"). Specifically, this
299 means that operating systems that follow the historic split of
300 /bin/ and /lib/ (and related directories) from /usr/ (i.e. where
301 the former are not symlinks into the latter) are not supported by
302 "--volatile=yes" as container payload. The overlay option does not
303 require any particular preparations in the OS, but do note that
304 "overlayfs" behaviour differs from regular file systems in a number
305 of ways, and hence compatibility is limited.
306 --root-hash=
308 Takes a data integrity (dm-verity) root hash specified in
309 hexadecimal. This option enables data integrity checks using
310 dm-verity, if the used image contains the appropriate integrity
311 data (see above). The specified hash must match the root hash of
312 integrity data, and is usually at least 256 bits (and hence 64
313 formatted hexadecimal characters) long (in case of SHA256 for
314 example). If this option is not specified, but the image file
315 carries the "user.verity.roothash" extended file attribute (see
316 xattr(7)), then the root hash is read from it, also as formatted
317 hexadecimal characters. If the extended file attribute is not found
318 (or is not supported by the underlying file system), but a file
319 with the .roothash suffix is found next to the image file, bearing
320 otherwise the same name (except if the image has the .raw suffix,
321 in which case the root hash file must not have it in its name), the
322 root hash is read from it and automatically used, also as formatted
323 hexadecimal characters.
324 Note that this configures the root hash for the root file system.
326 Disk images may also contain separate file systems for the /usr/
327 hierarchy, which may be Verity protected as well. The root hash for
328 this protection may be configured via the "user.verity.usrhash"
329 extended file attribute or via a .usrhash file adjacent to the disk
330 image, following the same format and logic as for the root hash for
331 the root file system described here. Note that there's currently no
332 switch to configure the root hash for the /usr/ from the command
333 line.
334 Also see the RootHash= option in systemd.exec(5).
336 --root-hash-sig=
338 Takes a PKCS7 signature of the --root-hash= option. The semantics
339 are the same as for the RootHashSignature= option, see
340 systemd.exec(5).
341 --verity-data=
343 Takes the path to a data integrity (dm-verity) file. This option
344 enables data integrity checks using dm-verity, if a root-hash is
345 passed and if the used image itself does not contain the integrity
346 data. The integrity data must be matched by the root hash. If this
347 option is not specified, but a file with the .verity suffix is
348 found next to the image file, bearing otherwise the same name
349 (except if the image has the .raw suffix, in which case the verity
350 data file must not have it in its name), the verity data is read
351 from it and automatically used.
352 --pivot-root=
354 Pivot the specified directory to / inside the container, and either
355 unmount the container's old root, or pivot it to another specified
356 directory. Takes one of: a path argument — in which case the
357 specified path will be pivoted to / and the old root will be
358 unmounted; or a colon-separated pair of new root path and pivot
359 destination for the old root. The new root path will be pivoted to
360 /, and the old / will be pivoted to the other directory. Both paths
361 must be absolute, and are resolved in the container's file system
362 namespace.
363 This is for containers which have several bootable directories in
365 them; for example, several OSTree[4] deployments. It emulates the
366 behavior of the boot loader and the initrd which normally select
367 which directory to mount as the root and start the container's PID
368 1 in.
369 Execution Options
371 -a, --as-pid2
372 Invoke the shell or specified program as process ID (PID) 2 instead
373 of PID 1 (init). By default, if neither this option nor --boot is
374 used, the selected program is run as the process with PID 1, a mode
375 only suitable for programs that are aware of the special semantics
376 that the process with PID 1 has on UNIX. For example, it needs to
377 reap all processes reparented to it, and should implement sysvinit
378 compatible signal handling (specifically: it needs to reboot on
379 SIGINT, reexecute on SIGTERM, reload configuration on SIGHUP, and
380 so on). With --as-pid2 a minimal stub init process is run as PID 1
381 and the selected program is executed as PID 2 (and hence does not
382 need to implement any special semantics). The stub init process
383 will reap processes as necessary and react appropriately to
384 signals. It is recommended to use this mode to invoke arbitrary
385 commands in containers, unless they have been modified to run
386 correctly as PID 1. Or in other words: this switch should be used
387 for pretty much all commands, except when the command refers to an
388 init or shell implementation, as these are generally capable of
389 running correctly as PID 1. This option may not be combined with
390 --boot.
391 -b, --boot
393 Automatically search for an init program and invoke it as PID 1,
394 instead of a shell or a user supplied program. If this option is
395 used, arguments specified on the command line are used as arguments
396 for the init program. This option may not be combined with
397 --as-pid2.
398 The following table explains the different modes of invocation and
400 relationship to --as-pid2 (see above):
401 Table 1. Invocation Mode
403 ┌──────────────────────┬────────────────────────────┐
404 │Switch │ Explanation │
405 ├──────────────────────┼────────────────────────────┤
406 │Neither --as-pid2 nor │ The passed parameters are │
407 │--boot specified │ interpreted as the command │
408 │ │ line, which is executed as │
409 │ │ PID 1 in the container. │
410 ├──────────────────────┼────────────────────────────┤
411 │--as-pid2 specified │ The passed parameters are │
412 │ │ interpreted as the command │
413 │ │ line, which is executed as │
414 │ │ PID 2 in the container. A │
415 │ │ stub init process is run │
416 │ │ as PID 1. │
417 ├──────────────────────┼────────────────────────────┤
418 │--boot specified │ An init program is │
419 │ │ automatically searched for │
420 │ │ and run as PID 1 in the │
421 │ │ container. The passed │
422 │ │ parameters are used as │
423 │ │ invocation parameters for │
424 │ │ this process. │
425 └──────────────────────┴────────────────────────────┘
426 Note that --boot is the default mode of operation if the
427 systemd-nspawn@.service template unit file is used.
428 --chdir=
430 Change to the specified working directory before invoking the
431 process in the container. Expects an absolute path in the
432 container's file system namespace.
433 -E NAME[=VALUE], --setenv=NAME[=VALUE]
435 Specifies an environment variable to pass to the init process in
436 the container. This may be used to override the default variables
437 or to set additional variables. It may be used more than once to
438 set multiple variables. When "=" and VALUE are omitted, the value
439 of the variable with the same name in the program environment will
440 be used.
441 -u, --user=
443 After transitioning into the container, change to the specified
444 user defined in the container's user database. Like all other
445 systemd-nspawn features, this is not a security feature and
446 provides protection against accidental destructive operations only.
447 --kill-signal=
449 Specify the process signal to send to the container's PID 1 when
450 nspawn itself receives SIGTERM, in order to trigger an orderly
451 shutdown of the container. Defaults to SIGRTMIN+3 if --boot is used
452 (on systemd-compatible init systems SIGRTMIN+3 triggers an orderly
453 shutdown). If --boot is not used and this option is not specified
454 the container's processes are terminated abruptly via SIGKILL. For
455 a list of valid signals, see signal(7).
456 --notify-ready=
458 Configures support for notifications from the container's init
459 process. --notify-ready= takes a boolean (no and yes). With option
460 no systemd-nspawn notifies systemd with a "READY=1" message when
461 the init process is created. With option yes systemd-nspawn waits
462 for the "READY=1" message from the init process in the container
463 before sending its own to systemd. For more details about
464 notifications see sd_notify(3).
465 --suppress-sync=
467 Expects a boolean argument. If true, turns off any form of on-disk
468 file system synchronization for the container payload. This means
469 all system calls such as sync(2), fsync(), syncfs(), ... will
470 execute no operation, and the O_SYNC/O_DSYNC flags to open(2) and
471 related calls will be made unavailable. This is potentially
472 dangerous, as assumed data integrity guarantees to the container
473 payload are not actually enforced (i.e. data assumed to have been
474 written to disk might be lost if the system is shut down
475 abnormally). However, this can dramatically improve container
476 runtime performance – as long as these guarantees are not required
477 or desirable, for example because any data written by the container
478 is of temporary, redundant nature, or just an intermediary artifact
479 that will be further processed and finalized by a later step in a
480 pipeline. Defaults to false.
481 System Identity Options
483 -M, --machine=
484 Sets the machine name for this container. This name may be used to
485 identify this container during its runtime (for example in tools
486 like machinectl(1) and similar), and is used to initialize the
487 container's hostname (which the container can choose to override,
488 however). If not specified, the last component of the root
489 directory path of the container is used, possibly suffixed with a
490 random identifier in case --ephemeral mode is selected. If the root
491 directory selected is the host's root directory the host's hostname
492 is used as default instead.
493 --hostname=
495 Controls the hostname to set within the container, if different
496 from the machine name. Expects a valid hostname as argument. If
497 this option is used, the kernel hostname of the container will be
498 set to this value, otherwise it will be initialized to the machine
499 name as controlled by the --machine= option described above. The
500 machine name is used for various aspect of identification of the
501 container from the outside, the kernel hostname configurable with
502 this option is useful for the container to identify itself from the
503 inside. It is usually a good idea to keep both forms of
504 identification synchronized, in order to avoid confusion. It is
505 hence recommended to avoid usage of this option, and use --machine=
506 exclusively. Note that regardless whether the container's hostname
507 is initialized from the name set with --hostname= or the one set
508 with --machine=, the container can later override its kernel
509 hostname freely on its own as well.
510 --uuid=
512 Set the specified UUID for the container. The init system will
513 initialize /etc/machine-id from this if this file is not set yet.
514 Note that this option takes effect only if /etc/machine-id in the
515 container is unpopulated.
516 Property Options
518 -S, --slice=
519 Make the container part of the specified slice, instead of the
520 default machine.slice. This applies only if the machine is run in
521 its own scope unit, i.e. if --keep-unit isn't used.
522 --property=
524 Set a unit property on the scope unit to register for the machine.
525 This applies only if the machine is run in its own scope unit, i.e.
526 if --keep-unit isn't used. Takes unit property assignments in the
527 same format as systemctl set-property. This is useful to set memory
528 limits and similar for the container.
529 --register=
531 Controls whether the container is registered with systemd-
532 machined(8). Takes a boolean argument, which defaults to "yes".
533 This option should be enabled when the container runs a full
534 Operating System (more specifically: a system and service manager
535 as PID 1), and is useful to ensure that the container is accessible
536 via machinectl(1) and shown by tools such as ps(1). If the
537 container does not run a service manager, it is recommended to set
538 this option to "no".
539 --keep-unit
541 Instead of creating a transient scope unit to run the container in,
542 simply use the service or scope unit systemd-nspawn has been
543 invoked in. If --register=yes is set this unit is registered with
544 systemd-machined(8). This switch should be used if systemd-nspawn
545 is invoked from within a service unit, and the service unit's sole
546 purpose is to run a single systemd-nspawn container. This option is
547 not available if run from a user session.
548 Note that passing --keep-unit disables the effect of --slice= and
550 --property=. Use --keep-unit and --register=no in combination to
551 disable any kind of unit allocation or registration with
552 systemd-machined.
553 User Namespacing Options
555 --private-users=
556 Controls user namespacing. If enabled, the container will run with
557 its own private set of UNIX user and group ids (UIDs and GIDs).
558 This involves mapping the private UIDs/GIDs used in the container
559 (starting with the container's root user 0 and up) to a range of
560 UIDs/GIDs on the host that are not used for other purposes (usually
561 in the range beyond the host's UID/GID 65536). The parameter may be
562 specified as follows:
563 1. If one or two colon-separated numbers are specified, user
565 namespacing is turned on. The first parameter specifies the
566 first host UID/GID to assign to the container, the second
567 parameter specifies the number of host UIDs/GIDs to assign to
568 the container. If the second parameter is omitted, 65536
569 UIDs/GIDs are assigned.
570 2. If the parameter is "yes", user namespacing is turned on. The
572 UID/GID range to use is determined automatically from the file
573 ownership of the root directory of the container's directory
574 tree. To use this option, make sure to prepare the directory
575 tree in advance, and ensure that all files and directories in
576 it are owned by UIDs/GIDs in the range you'd like to use. Also,
577 make sure that used file ACLs exclusively reference UIDs/GIDs
578 in the appropriate range. In this mode, the number of UIDs/GIDs
579 assigned to the container is 65536, and the owner UID/GID of
580 the root directory must be a multiple of 65536.
581 3. If the parameter is "no", user namespacing is turned off. This
583 is the default.
584 4. If the parameter is "identity", user namespacing is employed
586 with an identity mapping for the first 65536 UIDs/GIDs. This is
587 mostly equivalent to --private-users=0:65536. While it does not
588 provide UID/GID isolation, since all host and container
589 UIDs/GIDs are chosen identically it does provide process
590 capability isolation, and hence is often a good choice if
591 proper user namespacing with distinct UID maps is not
592 appropriate.
593 5. The special value "pick" turns on user namespacing. In this
595 case the UID/GID range is automatically chosen. As first step,
596 the file owner UID/GID of the root directory of the container's
597 directory tree is read, and it is checked that no other
598 container is currently using it. If this check is successful,
599 the UID/GID range determined this way is used, similarly to the
600 behavior if "yes" is specified. If the check is not successful
601 (and thus the UID/GID range indicated in the root directory's
602 file owner is already used elsewhere) a new – currently unused
603 – UID/GID range of 65536 UIDs/GIDs is randomly chosen between
604 the host UID/GIDs of 524288 and 1878982656, always starting at
605 a multiple of 65536, and, if possible, consistently hashed from
606 the machine name. This setting implies
607 --private-users-ownership=auto (see below), which possibly has
608 the effect that the files and directories in the container's
609 directory tree will be owned by the appropriate users of the
610 range picked. Using this option makes user namespace behavior
611 fully automatic. Note that the first invocation of a previously
612 unused container image might result in picking a new UID/GID
613 range for it, and thus in the (possibly expensive) file
614 ownership adjustment operation. However, subsequent invocations
615 of the container will be cheap (unless of course the picked
616 UID/GID range is assigned to a different use by then).
617 It is recommended to assign at least 65536 UIDs/GIDs to each
619 container, so that the usable UID/GID range in the container covers
620 16 bit. For best security, do not assign overlapping UID/GID ranges
621 to multiple containers. It is hence a good idea to use the upper 16
622 bit of the host 32-bit UIDs/GIDs as container identifier, while the
623 lower 16 bit encode the container UID/GID used. This is in fact the
624 behavior enforced by the --private-users=pick option.
625 When user namespaces are used, the GID range assigned to each
627 container is always chosen identical to the UID range.
628 In most cases, using --private-users=pick is the recommended option
630 as it enhances container security massively and operates fully
631 automatically in most cases.
632 Note that the picked UID/GID range is not written to /etc/passwd or
634 /etc/group. In fact, the allocation of the range is not stored
635 persistently anywhere, except in the file ownership of the files
636 and directories of the container.
637 Note that when user namespacing is used file ownership on disk
639 reflects this, and all of the container's files and directories are
640 owned by the container's effective user and group IDs. This means
641 that copying files from and to the container image requires
642 correction of the numeric UID/GID values, according to the UID/GID
643 shift applied.
644 --private-users-ownership=
646 Controls how to adjust the container image's UIDs and GIDs to match
647 the UID/GID range chosen with --private-users=, see above. Takes
648 one of "off" (to leave the image as is), "chown" (to recursively
649 chown() the container's directory tree as needed), "map" (in order
650 to use transparent ID mapping mounts) or "auto" for automatically
651 using "map" where available and "chown" where not.
652 If "chown" is selected, all files and directories in the
654 container's directory tree will be adjusted so that they are owned
655 by the appropriate UIDs/GIDs selected for the container (see
656 above). This operation is potentially expensive, as it involves
657 iterating through the full directory tree of the container. Besides
658 actual file ownership, file ACLs are adjusted as well.
659 Typically "map" is the best choice, since it transparently maps
661 UIDs/GIDs in memory as needed without modifying the image, and
662 without requiring an expensive recursive adjustment operation.
663 However, it is not available for all file systems, currently.
664 The --private-users-ownership=auto option is implied if
666 --private-users=pick is used. This option has no effect if user
667 namespacing is not used.
668 -U
670 If the kernel supports the user namespaces feature, equivalent to
671 --private-users=pick --private-users-ownership=auto, otherwise
672 equivalent to --private-users=no.
673 Note that -U is the default if the systemd-nspawn@.service template
675 unit file is used.
676 Note: it is possible to undo the effect of
678 --private-users-ownership=chown (or -U) on the file system by
679 redoing the operation with the first UID of 0:
680 systemd-nspawn ... --private-users=0 --private-users-ownership=chown
682 Networking Options
684 --private-network
685 Disconnect networking of the container from the host. This makes
686 all network interfaces unavailable in the container, with the
687 exception of the loopback device and those specified with
688 --network-interface= and configured with --network-veth. If this
689 option is specified, the CAP_NET_ADMIN capability will be added to
690 the set of capabilities the container retains. The latter may be
691 disabled by using --drop-capability=. If this option is not
692 specified (or implied by one of the options listed below), the
693 container will have full access to the host network.
694 --network-interface=
696 Assign the specified network interface to the container. This will
697 remove the specified interface from the calling namespace and place
698 it in the container. When the container terminates, it is moved
699 back to the calling namespace. Note that --network-interface=
700 implies --private-network. This option may be used more than once
701 to add multiple network interfaces to the container.
702 Note that any network interface specified this way must already
704 exist at the time the container is started. If the container shall
705 be started automatically at boot via a systemd-nspawn@.service unit
706 file instance, it might hence make sense to add a unit file drop-in
707 to the service instance (e.g.
708 /etc/systemd/system/systemd-nspawn@foobar.service.d/50-network.conf)
709 with contents like the following:
710 [Unit]
712 Wants=sys-subsystem-net-devices-ens1.device
713 After=sys-subsystem-net-devices-ens1.device
714 This will make sure that activation of the container service will
716 be delayed until the "ens1" network interface has shown up. This is
717 required since hardware probing is fully asynchronous, and network
718 interfaces might be discovered only later during the boot process,
719 after the container would normally be started without these
720 explicit dependencies.
721 --network-macvlan=
723 Create a "macvlan" interface of the specified Ethernet network
724 interface and add it to the container. A "macvlan" interface is a
725 virtual interface that adds a second MAC address to an existing
726 physical Ethernet link. The interface in the container will be
727 named after the interface on the host, prefixed with "mv-". Note
728 that --network-macvlan= implies --private-network. This option may
729 be used more than once to add multiple network interfaces to the
730 container.
731 As with --network-interface=, the underlying Ethernet network
733 interface must already exist at the time the container is started,
734 and thus similar unit file drop-ins as described above might be
735 useful.
736 --network-ipvlan=
738 Create an "ipvlan" interface of the specified Ethernet network
739 interface and add it to the container. An "ipvlan" interface is a
740 virtual interface, similar to a "macvlan" interface, which uses the
741 same MAC address as the underlying interface. The interface in the
742 container will be named after the interface on the host, prefixed
743 with "iv-". Note that --network-ipvlan= implies --private-network.
744 This option may be used more than once to add multiple network
745 interfaces to the container.
746 As with --network-interface=, the underlying Ethernet network
748 interface must already exist at the time the container is started,
749 and thus similar unit file drop-ins as described above might be
750 useful.
751 -n, --network-veth
753 Create a virtual Ethernet link ("veth") between host and container.
754 The host side of the Ethernet link will be available as a network
755 interface named after the container's name (as specified with
756 --machine=), prefixed with "ve-". The container side of the
757 Ethernet link will be named "host0". The --network-veth option
758 implies --private-network.
759 Note that systemd-networkd.service(8) includes by default a network
761 file /usr/lib/systemd/network/80-container-ve.network matching the
762 host-side interfaces created this way, which contains settings to
763 enable automatic address provisioning on the created virtual link
764 via DHCP, as well as automatic IP routing onto the host's external
765 network interfaces. It also contains
766 /usr/lib/systemd/network/80-container-host0.network matching the
767 container-side interface created this way, containing settings to
768 enable client side address assignment via DHCP. In case
769 systemd-networkd is running on both the host and inside the
770 container, automatic IP communication from the container to the
771 host is thus available, with further connectivity to the external
772 network.
773 Note that --network-veth is the default if the
775 systemd-nspawn@.service template unit file is used.
776 Note that on Linux network interface names may have a length of 15
778 characters at maximum, while container names may have a length up
779 to 64 characters. As this option derives the host-side interface
780 name from the container name the name is possibly truncated. Thus,
781 care needs to be taken to ensure that interface names remain unique
782 in this case, or even better container names are generally not
783 chosen longer than 12 characters, to avoid the truncation. If the
784 name is truncated, systemd-nspawn will automatically append a
785 4-digit hash value to the name to reduce the chance of collisions.
786 However, the hash algorithm is not collision-free. (See
787 systemd.net-naming-scheme(7) for details on older naming algorithms
788 for this interface). Alternatively, the --network-veth-extra=
789 option may be used, which allows free configuration of the
790 host-side interface name independently of the container name — but
791 might require a bit more additional configuration in case bridging
792 in a fashion similar to --network-bridge= is desired.
793 --network-veth-extra=
795 Adds an additional virtual Ethernet link between host and
796 container. Takes a colon-separated pair of host interface name and
797 container interface name. The latter may be omitted in which case
798 the container and host sides will be assigned the same name. This
799 switch is independent of --network-veth, and — in contrast — may be
800 used multiple times, and allows configuration of the network
801 interface names. Note that --network-bridge= has no effect on
802 interfaces created with --network-veth-extra=.
803 --network-bridge=
805 Adds the host side of the Ethernet link created with --network-veth
806 to the specified Ethernet bridge interface. Expects a valid network
807 interface name of a bridge device as argument. Note that
808 --network-bridge= implies --network-veth. If this option is used,
809 the host side of the Ethernet link will use the "vb-" prefix
810 instead of "ve-". Regardless of the used naming prefix the same
811 network interface name length limits imposed by Linux apply, along
812 with the complications this creates (for details see above).
813 As with --network-interface=, the underlying bridge network
815 interface must already exist at the time the container is started,
816 and thus similar unit file drop-ins as described above might be
817 useful.
818 --network-zone=
820 Creates a virtual Ethernet link ("veth") to the container and adds
821 it to an automatically managed Ethernet bridge interface. The
822 bridge interface is named after the passed argument, prefixed with
823 "vz-". The bridge interface is automatically created when the first
824 container configured for its name is started, and is automatically
825 removed when the last container configured for its name exits.
826 Hence, each bridge interface configured this way exists only as
827 long as there's at least one container referencing it running. This
828 option is very similar to --network-bridge=, besides this automatic
829 creation/removal of the bridge device.
830 This setting makes it easy to place multiple related containers on
832 a common, virtual Ethernet-based broadcast domain, here called a
833 "zone". Each container may only be part of one zone, but each zone
834 may contain any number of containers. Each zone is referenced by
835 its name. Names may be chosen freely (as long as they form valid
836 network interface names when prefixed with "vz-"), and it is
837 sufficient to pass the same name to the --network-zone= switch of
838 the various concurrently running containers to join them in one
839 zone.
840 Note that systemd-networkd.service(8) includes by default a network
842 file /usr/lib/systemd/network/80-container-vz.network matching the
843 bridge interfaces created this way, which contains settings to
844 enable automatic address provisioning on the created virtual
845 network via DHCP, as well as automatic IP routing onto the host's
846 external network interfaces. Using --network-zone= is hence in most
847 cases fully automatic and sufficient to connect multiple local
848 containers in a joined broadcast domain to the host, with further
849 connectivity to the external network.
850 --network-namespace-path=
852 Takes the path to a file representing a kernel network namespace
853 that the container shall run in. The specified path should refer to
854 a (possibly bind-mounted) network namespace file, as exposed by the
855 kernel below /proc/$PID/ns/net. This makes the container enter the
856 given network namespace. One of the typical use cases is to give a
857 network namespace under /run/netns created by ip-netns(8), for
858 example, --network-namespace-path=/run/netns/foo. Note that this
859 option cannot be used together with other network-related options,
860 such as --private-network or --network-interface=.
861 -p, --port=
863 If private networking is enabled, maps an IP port on the host onto
864 an IP port on the container. Takes a protocol specifier (either
865 "tcp" or "udp"), separated by a colon from a host port number in
866 the range 1 to 65535, separated by a colon from a container port
867 number in the range from 1 to 65535. The protocol specifier and its
868 separating colon may be omitted, in which case "tcp" is assumed.
869 The container port number and its colon may be omitted, in which
870 case the same port as the host port is implied. This option is only
871 supported if private networking is used, such as with
872 --network-veth, --network-zone= --network-bridge=.
873 Security Options
875 --capability=
876 List one or more additional capabilities to grant the container.
877 Takes a comma-separated list of capability names, see
878 capabilities(7) for more information. Note that the following
879 capabilities will be granted in any way: CAP_AUDIT_CONTROL,
880 CAP_AUDIT_WRITE, CAP_CHOWN, CAP_DAC_OVERRIDE, CAP_DAC_READ_SEARCH,
881 CAP_FOWNER, CAP_FSETID, CAP_IPC_OWNER, CAP_KILL, CAP_LEASE,
882 CAP_LINUX_IMMUTABLE, CAP_MKNOD, CAP_NET_BIND_SERVICE,
883 CAP_NET_BROADCAST, CAP_NET_RAW, CAP_SETFCAP, CAP_SETGID,
884 CAP_SETPCAP, CAP_SETUID, CAP_SYS_ADMIN, CAP_SYS_BOOT,
885 CAP_SYS_CHROOT, CAP_SYS_NICE, CAP_SYS_PTRACE, CAP_SYS_RESOURCE,
886 CAP_SYS_TTY_CONFIG. Also CAP_NET_ADMIN is retained if
887 --private-network is specified. If the special value "all" is
888 passed, all capabilities are retained.
889 If the special value of "help" is passed, the program will print
891 known capability names and exit.
892 This option sets the bounding set of capabilities which also limits
894 the ambient capabilities as given with the --ambient-capability=.
895 --drop-capability=
897 Specify one or more additional capabilities to drop for the
898 container. This allows running the container with fewer
899 capabilities than the default (see above).
900 If the special value of "help" is passed, the program will print
902 known capability names and exit.
903 This option sets the bounding set of capabilities which also limits
905 the ambient capabilities as given with the --ambient-capability=.
906 --ambient-capability=
908 Specify one or more additional capabilities to pass in the
909 inheritable and ambient set to the program started within the
910 container. The value "all" is not supported for this setting.
911 All capabilities specified here must be in the set allowed with the
913 --capability= and --drop-capability= options. Otherwise, an error
914 message will be shown.
915 This option cannot be combined with the boot mode of the container
917 (as requested via --boot).
918 If the special value of "help" is passed, the program will print
920 known capability names and exit.
921 --no-new-privileges=
923 Takes a boolean argument. Specifies the value of the
924 PR_SET_NO_NEW_PRIVS flag for the container payload. Defaults to
925 off. When turned on the payload code of the container cannot
926 acquire new privileges, i.e. the "setuid" file bit as well as file
927 system capabilities will not have an effect anymore. See prctl(2)
928 for details about this flag.
929 --system-call-filter=
931 Alter the system call filter applied to containers. Takes a
932 space-separated list of system call names or group names (the
933 latter prefixed with "@", as listed by the syscall-filter command
934 of systemd-analyze(1)). Passed system calls will be permitted. The
935 list may optionally be prefixed by "~", in which case all listed
936 system calls are prohibited. If this command line option is used
937 multiple times the configured lists are combined. If both a
938 positive and a negative list (that is one system call list without
939 and one with the "~" prefix) are configured, the negative list
940 takes precedence over the positive list. Note that systemd-nspawn
941 always implements a system call allow list (as opposed to a deny
942 list!), and this command line option hence adds or removes entries
943 from the default allow list, depending on the "~" prefix. Note that
944 the applied system call filter is also altered implicitly if
945 additional capabilities are passed using the --capabilities=.
946 -Z, --selinux-context=
948 Sets the SELinux security context to be used to label processes in
949 the container.
950 -L, --selinux-apifs-context=
952 Sets the SELinux security context to be used to label files in the
953 virtual API file systems in the container.
954 Resource Options
956 --rlimit=
957 Sets the specified POSIX resource limit for the container payload.
958 Expects an assignment of the form "LIMIT=SOFT:HARD" or
959 "LIMIT=VALUE", where LIMIT should refer to a resource limit type,
960 such as RLIMIT_NOFILE or RLIMIT_NICE. The SOFT and HARD fields
961 should refer to the numeric soft and hard resource limit values. If
962 the second form is used, VALUE may specify a value that is used
963 both as soft and hard limit. In place of a numeric value the
964 special string "infinity" may be used to turn off resource limiting
965 for the specific type of resource. This command line option may be
966 used multiple times to control limits on multiple limit types. If
967 used multiple times for the same limit type, the last use wins. For
968 details about resource limits see setrlimit(2). By default resource
969 limits for the container's init process (PID 1) are set to the same
970 values the Linux kernel originally passed to the host init system.
971 Note that some resource limits are enforced on resources counted
972 per user, in particular RLIMIT_NPROC. This means that unless user
973 namespacing is deployed (i.e. --private-users= is used, see
974 above), any limits set will be applied to the resource usage of the
975 same user on all local containers as well as the host. This means
976 particular care needs to be taken with these limits as they might
977 be triggered by possibly less trusted code. Example:
978 "--rlimit=RLIMIT_NOFILE=8192:16384".
979 --oom-score-adjust=
981 Changes the OOM ("Out Of Memory") score adjustment value for the
982 container payload. This controls /proc/self/oom_score_adj which
983 influences the preference with which this container is terminated
984 when memory becomes scarce. For details see proc(5). Takes an
985 integer in the range -1000...1000.
986 --cpu-affinity=
988 Controls the CPU affinity of the container payload. Takes a comma
989 separated list of CPU numbers or number ranges (the latter's start
990 and end value separated by dashes). See sched_setaffinity(2) for
991 details.
992 --personality=
994 Control the architecture ("personality") reported by uname(2) in
995 the container. Currently, only "x86" and "x86-64" are supported.
996 This is useful when running a 32-bit container on a 64-bit host. If
997 this setting is not used, the personality reported in the container
998 is the same as the one reported on the host.
999 Integration Options
1001 --resolv-conf=
1002 Configures how /etc/resolv.conf inside of the container shall be
1003 handled (i.e. DNS configuration synchronization from host to
1004 container). Takes one of "off", "copy-host", "copy-static",
1005 "copy-uplink", "copy-stub", "replace-host", "replace-static",
1006 "replace-uplink", "replace-stub", "bind-host", "bind-static",
1007 "bind-uplink", "bind-stub", "delete" or "auto".
1008 If set to "off" the /etc/resolv.conf file in the container is left
1010 as it is included in the image, and neither modified nor bind
1011 mounted over.
1012 If set to "copy-host", the /etc/resolv.conf file from the host is
1014 copied into the container, unless the file exists already and is
1015 not a regular file (e.g. a symlink). Similarly, if "replace-host"
1016 is used the file is copied, replacing any existing inode, including
1017 symlinks. Similarly, if "bind-host" is used, the file is bind
1018 mounted from the host into the container.
1019 If set to "copy-static", "replace-static" or "bind-static" the
1021 static resolv.conf file supplied with systemd-resolved.service(8)
1022 (specifically: /usr/lib/systemd/resolv.conf) is copied or bind
1023 mounted into the container.
1024 If set to "copy-uplink", "replace-uplink" or "bind-uplink" the
1026 uplink resolv.conf file managed by systemd-resolved.service
1027 (specifically: /run/systemd/resolve/resolv.conf) is copied or bind
1028 mounted into the container.
1029 If set to "copy-stub", "replace-stub" or "bind-stub" the stub
1031 resolv.conf file managed by systemd-resolved.service (specifically:
1032 /run/systemd/resolve/stub-resolv.conf) is copied or bind mounted
1033 into the container.
1034 If set to "delete" the /etc/resolv.conf file in the container is
1036 deleted if it exists.
1037 Finally, if set to "auto" the file is left as it is if private
1039 networking is turned on (see --private-network). Otherwise, if
1040 systemd-resolved.service is running its stub resolv.conf file is
1041 used, and if not the host's /etc/resolv.conf file. In the latter
1042 cases the file is copied if the image is writable, and bind mounted
1043 otherwise.
1044 It's recommended to use "copy-..." or "replace-..." if the
1046 container shall be able to make changes to the DNS configuration on
1047 its own, deviating from the host's settings. Otherwise "bind" is
1048 preferable, as it means direct changes to /etc/resolv.conf in the
1049 container are not allowed, as it is a read-only bind mount (but
1050 note that if the container has enough privileges, it might simply
1051 go ahead and unmount the bind mount anyway). Note that both if the
1052 file is bind mounted and if it is copied no further propagation of
1053 configuration is generally done after the one-time early
1054 initialization (this is because the file is usually updated through
1055 copying and renaming). Defaults to "auto".
1056 --timezone=
1058 Configures how /etc/localtime inside of the container (i.e. local
1059 timezone synchronization from host to container) shall be handled.
1060 Takes one of "off", "copy", "bind", "symlink", "delete" or "auto".
1061 If set to "off" the /etc/localtime file in the container is left as
1062 it is included in the image, and neither modified nor bind mounted
1063 over. If set to "copy" the /etc/localtime file of the host is
1064 copied into the container. Similarly, if "bind" is used, the file
1065 is bind mounted from the host into the container. If set to
1066 "symlink", a symlink is created pointing from /etc/localtime in the
1067 container to the timezone file in the container that matches the
1068 timezone setting on the host. If set to "delete", the file in the
1069 container is deleted, should it exist. If set to "auto" and the
1070 /etc/localtime file of the host is a symlink, then "symlink" mode
1071 is used, and "copy" otherwise, except if the image is read-only in
1072 which case "bind" is used instead. Defaults to "auto".
1073 --link-journal=
1075 Control whether the container's journal shall be made visible to
1076 the host system. If enabled, allows viewing the container's journal
1077 files from the host (but not vice versa). Takes one of "no",
1078 "host", "try-host", "guest", "try-guest", "auto". If "no", the
1079 journal is not linked. If "host", the journal files are stored on
1080 the host file system (beneath /var/log/journal/machine-id) and the
1081 subdirectory is bind-mounted into the container at the same
1082 location. If "guest", the journal files are stored on the guest
1083 file system (beneath /var/log/journal/machine-id) and the
1084 subdirectory is symlinked into the host at the same location.
1085 "try-host" and "try-guest" do the same but do not fail if the host
1086 does not have persistent journaling enabled. If "auto" (the
1087 default), and the right subdirectory of /var/log/journal exists, it
1088 will be bind mounted into the container. If the subdirectory does
1089 not exist, no linking is performed. Effectively, booting a
1090 container once with "guest" or "host" will link the journal
1091 persistently if further on the default of "auto" is used.
1092 Note that --link-journal=try-guest is the default if the
1094 systemd-nspawn@.service template unit file is used.
1095 -j
1097 Equivalent to --link-journal=try-guest.
1098 Mount Options
1100 --bind=, --bind-ro=
1101 Bind mount a file or directory from the host into the container.
1102 Takes one of: a path argument — in which case the specified path
1103 will be mounted from the host to the same path in the container, or
1104 a colon-separated pair of paths — in which case the first specified
1105 path is the source in the host, and the second path is the
1106 destination in the container, or a colon-separated triple of source
1107 path, destination path and mount options. The source path may
1108 optionally be prefixed with a "+" character. If so, the source path
1109 is taken relative to the image's root directory. This permits
1110 setting up bind mounts within the container image. The source path
1111 may be specified as empty string, in which case a temporary
1112 directory below the host's /var/tmp/ directory is used. It is
1113 automatically removed when the container is shut down. If the
1114 source path is not absolute, it is resolved relative to the current
1115 working directory. The --bind-ro= option creates read-only bind
1116 mounts. Backslash escapes are interpreted, so "\:" may be used to
1117 embed colons in either path. This option may be specified multiple
1118 times for creating multiple independent bind mount points.
1119 Mount options are comma-separated. rbind and norbind control
1121 whether to create a recursive or a regular bind mount. Defaults to
1122 "rbind". noidmap, idmap, and rootidmap control ID mapping.
1123 Using idmap or rootidmap requires support by the source filesystem
1125 for user/group ID mapped mounts. Defaults to "noidmap". With x
1126 being the container's UID range offset, y being the length of the
1127 container's UID range, and p being the owner UID of the bind mount
1128 source inode on the host:
1129 • If noidmap is used, any user z in the range 0 ... y seen from
1131 inside of the container is mapped to x + z in the x ... x + y
1132 range on the host. Other host users are mapped to nobody inside
1133 the container.
1134 • If idmap is used, any user z in the UID range 0 ... y as seen
1136 from inside the container is mapped to the same z in the same 0
1137 ... y range on the host. Other host users are mapped to nobody
1138 inside the container.
1139 • If rootidmap is used, the user 0 seen from inside of the
1141 container is mapped to p on the host. Other host users are
1142 mapped to nobody inside the container.
1143 Whichever ID mapping option is used, the same mapping will be used
1145 for users and groups IDs. If rootidmap is used, the group owning
1146 the bind mounted directory will have no effect.
1147 Note that when this option is used in combination with
1149 --private-users, the resulting mount points will be owned by the
1150 nobody user. That's because the mount and its files and directories
1151 continue to be owned by the relevant host users and groups, which
1152 do not exist in the container, and thus show up under the wildcard
1153 UID 65534 (nobody). If such bind mounts are created, it is
1154 recommended to make them read-only, using --bind-ro=. Alternatively
1155 you can use the "idmap" mount option to map the filesystem IDs.
1156 --bind-user=
1158 Binds the home directory of the specified user on the host into the
1159 container. Takes the name of an existing user on the host as
1160 argument. May be used multiple times to bind multiple users into
1161 the container. This does three things:
1162 1. The user's home directory is bind mounted from the host into
1164 /run/host/home/.
1165 2. An additional UID/GID mapping is added that maps the host
1167 user's UID/GID to a container UID/GID, allocated from the
1168 60514...60577 range.
1169 3. A JSON user and group record is generated in /run/userdb/ that
1171 describes the mapped user. It contains a minimized
1172 representation of the host's user record, adjusted to the
1173 UID/GID and home directory path assigned to the user in the
1174 container. The nss-systemd(8) glibc NSS module will pick up
1175 these records from there and make them available in the
1176 container's user/group databases.
1177 The combination of the three operations above ensures that it is
1179 possible to log into the container using the same account
1180 information as on the host. The user is only mapped transiently,
1181 while the container is running, and the mapping itself does not
1182 result in persistent changes to the container (except maybe for log
1183 messages generated at login time, and similar). Note that in
1184 particular the UID/GID assignment in the container is not made
1185 persistently. If the user is mapped transiently, it is best to not
1186 allow the user to make persistent changes to the container. If the
1187 user leaves files or directories owned by the user, and those
1188 UIDs/GIDs are reused during later container invocations (possibly
1189 with a different --bind-user= mapping), those files and directories
1190 will be accessible to the "new" user.
1191 The user/group record mapping only works if the container contains
1193 systemd 249 or newer, with nss-systemd properly configured in
1194 nsswitch.conf. See nss-systemd(8) for details.
1195 Note that the user record propagated from the host into the
1197 container will contain the UNIX password hash of the user, so that
1198 seamless logins in the container are possible. If the container is
1199 less trusted than the host it's hence important to use a strong
1200 UNIX password hash function (e.g. yescrypt or similar, with the
1201 "$y$" hash prefix).
1202 When binding a user from the host into the container checks are
1204 executed to ensure that the username is not yet known in the
1205 container. Moreover, it is checked that the UID/GID allocated for
1206 it is not currently defined in the user/group databases of the
1207 container. Both checks directly access the container's /etc/passwd
1208 and /etc/group, and thus might not detect existing accounts in
1209 other databases.
1210 This operation is only supported in combination with
1212 --private-users=/-U.
1213 --inaccessible=
1215 Make the specified path inaccessible in the container. This
1216 over-mounts the specified path (which must exist in the container)
1217 with a file node of the same type that is empty and has the most
1218 restrictive access mode supported. This is an effective way to mask
1219 files, directories and other file system objects from the container
1220 payload. This option may be used more than once in case all
1221 specified paths are masked.
1222 --tmpfs=
1224 Mount a tmpfs file system into the container. Takes a single
1225 absolute path argument that specifies where to mount the tmpfs
1226 instance to (in which case the directory access mode will be chosen
1227 as 0755, owned by root/root), or optionally a colon-separated pair
1228 of path and mount option string that is used for mounting (in which
1229 case the kernel default for access mode and owner will be chosen,
1230 unless otherwise specified). Backslash escapes are interpreted in
1231 the path, so "\:" may be used to embed colons in the path.
1232 Note that this option cannot be used to replace the root file
1234 system of the container with a temporary file system. However, the
1235 --volatile= option described below provides similar functionality,
1236 with a focus on implementing stateless operating system images.
1237 --overlay=, --overlay-ro=
1239 Combine multiple directory trees into one overlay file system and
1240 mount it into the container. Takes a list of colon-separated paths
1241 to the directory trees to combine and the destination mount point.
1242 Backslash escapes are interpreted in the paths, so "\:" may be used
1244 to embed colons in the paths.
1245 If three or more paths are specified, then the last specified path
1247 is the destination mount point in the container, all paths
1248 specified before refer to directory trees on the host and are
1249 combined in the specified order into one overlay file system. The
1250 left-most path is hence the lowest directory tree, the
1251 second-to-last path the highest directory tree in the stacking
1252 order. If --overlay-ro= is used instead of --overlay=, a read-only
1253 overlay file system is created. If a writable overlay file system
1254 is created, all changes made to it are written to the highest
1255 directory tree in the stacking order, i.e. the second-to-last
1256 specified.
1257 If only two paths are specified, then the second specified path is
1259 used both as the top-level directory tree in the stacking order as
1260 seen from the host, as well as the mount point for the overlay file
1261 system in the container. At least two paths have to be specified.
1262 The source paths may optionally be prefixed with "+" character. If
1264 so they are taken relative to the image's root directory. The
1265 uppermost source path may also be specified as an empty string, in
1266 which case a temporary directory below the host's /var/tmp/ is
1267 used. The directory is removed automatically when the container is
1268 shut down. This behaviour is useful in order to make read-only
1269 container directories writable while the container is running. For
1270 example, use "--overlay=+/var::/var" in order to automatically
1271 overlay a writable temporary directory on a read-only /var/
1272 directory. If a source path is not absolute, it is resolved
1273 relative to the current working directory.
1274 For details about overlay file systems, see Overlay Filesystem[5].
1276 Note that the semantics of overlay file systems are substantially
1277 different from normal file systems, in particular regarding
1278 reported device and inode information. Device and inode information
1279 may change for a file while it is being written to, and processes
1280 might see out-of-date versions of files at times. Note that this
1281 switch automatically derives the "workdir=" mount option for the
1282 overlay file system from the top-level directory tree, making it a
1283 sibling of it. It is hence essential that the top-level directory
1284 tree is not a mount point itself (since the working directory must
1285 be on the same file system as the top-most directory tree). Also
1286 note that the "lowerdir=" mount option receives the paths to stack
1287 in the opposite order of this switch.
1288 Note that this option cannot be used to replace the root file
1290 system of the container with an overlay file system. However, the
1291 --volatile= option described above provides similar functionality,
1292 with a focus on implementing stateless operating system images.
1293 Input/Output Options
1295 --console=MODE
1296 Configures how to set up standard input, output and error output
1297 for the container payload, as well as the /dev/console device for
1298 the container. Takes one of interactive, read-only, passive, pipe
1299 or autopipe. If interactive, a pseudo-TTY is allocated and made
1300 available as /dev/console in the container. It is then
1301 bi-directionally connected to the standard input and output passed
1302 to systemd-nspawn. read-only is similar but only the output of the
1303 container is propagated and no input from the caller is read. If
1304 passive, a pseudo TTY is allocated, but it is not connected
1305 anywhere. In pipe mode no pseudo TTY is allocated, but the standard
1306 input, output and error output file descriptors passed to
1307 systemd-nspawn are passed on — as they are — to the container
1308 payload, see the following paragraph. Finally, autopipe mode
1309 operates like interactive when systemd-nspawn is invoked on a
1310 terminal, and like pipe otherwise. Defaults to interactive if
1311 systemd-nspawn is invoked from a terminal, and read-only otherwise.
1312 In pipe mode, /dev/console will not exist in the container. This
1314 means that the container payload generally cannot be a full init
1315 system as init systems tend to require /dev/console to be
1316 available. On the other hand, in this mode container invocations
1317 can be used within shell pipelines. This is because intermediary
1318 pseudo TTYs do not permit independent bidirectional propagation of
1319 the end-of-file (EOF) condition, which is necessary for shell
1320 pipelines to work correctly. Note that the pipe mode should be
1321 used carefully, as passing arbitrary file descriptors to less
1322 trusted container payloads might open up unwanted interfaces for
1323 access by the container payload. For example, if a passed file
1324 descriptor refers to a TTY of some form, APIs such as TIOCSTI may
1325 be used to synthesize input that might be used for escaping the
1326 container. Hence pipe mode should only be used if the payload is
1327 sufficiently trusted or when the standard input/output/error output
1328 file descriptors are known safe, for example pipes.
1329 --pipe, -P
1331 Equivalent to --console=pipe.
1332 Credentials
1334 --load-credential=ID:PATH, --set-credential=ID:VALUE
1335 Pass a credential to the container. These two options correspond to
1336 the LoadCredential= and SetCredential= settings in unit files. See
1337 systemd.exec(5) for details about these concepts, as well as the
1338 syntax of the option's arguments.
1339 Note: when systemd-nspawn runs as systemd system service it can
1341 propagate the credentials it received via
1342 LoadCredential=/SetCredential= to the container payload. A systemd
1343 service manager running as PID 1 in the container can further
1344 propagate them to the services it itself starts. It is thus
1345 possible to easily propagate credentials from a parent service
1346 manager to a container manager service and from there into its
1347 payload. This can even be done recursively.
1348 In order to embed binary data into the credential data for
1350 --set-credential=, use C-style escaping (i.e. "\n" to embed a
1351 newline, or "\x00" to embed a NUL byte). Note that the invoking
1352 shell might already apply unescaping once, hence this might require
1353 double escaping!.
1354 The systemd-sysusers.service(8) and systemd-firstboot(1) services
1356 read credentials configured this way for the purpose of configuring
1357 the container's root user's password and shell, as well as system
1358 locale, keymap and timezone during the first boot process of the
1359 container. This is particularly useful in combination with
1360 --volatile=yes where every single boot appears as first boot, since
1361 configuration applied to /etc/ is lost on container reboot cycles.
1362 See the respective man pages for details. Example:
1363 # systemd-nspawn -i image.raw \
1365 --volatile=yes \
1366 --set-credential=firstboot.locale:de_DE.UTF-8 \
1367 --set-credential=passwd.hashed-password.root:'$y$j9T$yAuRJu1o5HioZAGDYPU5d.$F64ni6J2y2nNQve90M/p0ZP0ECP/qqzipNyaY9fjGpC' \
1368 -b
1369 The above command line will invoke the specified image file
1371 image.raw in volatile mode, i.e. with empty /etc/ and /var/. The
1372 container payload will recognize this as a first boot, and will
1373 invoke systemd-firstboot.service, which then reads the two passed
1374 credentials to configure the system's initial locale and root
1375 password.
1376 Other
1378 --no-pager
1379 Do not pipe output into a pager.
1380 -h, --help
1382 Print a short help text and exit.
1383 --version
1385 Print a short version string and exit.
1386
1388 $SYSTEMD_LOG_LEVEL
1389 The maximum log level of emitted messages (messages with a higher
1390 log level, i.e. less important ones, will be suppressed). Either
1391 one of (in order of decreasing importance) emerg, alert, crit, err,
1392 warning, notice, info, debug, or an integer in the range 0...7. See
1393 syslog(3) for more information.
1394 $SYSTEMD_LOG_COLOR
1396 A boolean. If true, messages written to the tty will be colored
1397 according to priority.
1398 This setting is only useful when messages are written directly to
1400 the terminal, because journalctl(1) and other tools that display
1401 logs will color messages based on the log level on their own.
1402 $SYSTEMD_LOG_TIME
1404 A boolean. If true, console log messages will be prefixed with a
1405 timestamp.
1406 This setting is only useful when messages are written directly to
1408 the terminal or a file, because journalctl(1) and other tools that
1409 display logs will attach timestamps based on the entry metadata on
1410 their own.
1411 $SYSTEMD_LOG_LOCATION
1413 A boolean. If true, messages will be prefixed with a filename and
1414 line number in the source code where the message originates.
1415 Note that the log location is often attached as metadata to journal
1417 entries anyway. Including it directly in the message text can
1418 nevertheless be convenient when debugging programs.
1419 $SYSTEMD_LOG_TID
1421 A boolean. If true, messages will be prefixed with the current
1422 numerical thread ID (TID).
1423 Note that the this information is attached as metadata to journal
1425 entries anyway. Including it directly in the message text can
1426 nevertheless be convenient when debugging programs.
1427 $SYSTEMD_LOG_TARGET
1429 The destination for log messages. One of console (log to the
1430 attached tty), console-prefixed (log to the attached tty but with
1431 prefixes encoding the log level and "facility", see syslog(3), kmsg
1432 (log to the kernel circular log buffer), journal (log to the
1433 journal), journal-or-kmsg (log to the journal if available, and to
1434 kmsg otherwise), auto (determine the appropriate log target
1435 automatically, the default), null (disable log output).
1436 $SYSTEMD_PAGER
1438 Pager to use when --no-pager is not given; overrides $PAGER. If
1439 neither $SYSTEMD_PAGER nor $PAGER are set, a set of well-known
1440 pager implementations are tried in turn, including less(1) and
1441 more(1), until one is found. If no pager implementation is
1442 discovered no pager is invoked. Setting this environment variable
1443 to an empty string or the value "cat" is equivalent to passing
1444 --no-pager.
1445 Note: if $SYSTEMD_PAGERSECURE is not set, $SYSTEMD_PAGER (as well
1447 as $PAGER) will be silently ignored.
1448 $SYSTEMD_LESS
1450 Override the options passed to less (by default "FRSXMK").
1451 Users might want to change two options in particular:
1453 K
1455 This option instructs the pager to exit immediately when Ctrl+C
1456 is pressed. To allow less to handle Ctrl+C itself to switch
1457 back to the pager command prompt, unset this option.
1458 If the value of $SYSTEMD_LESS does not include "K", and the
1460 pager that is invoked is less, Ctrl+C will be ignored by the
1461 executable, and needs to be handled by the pager.
1462 X
1464 This option instructs the pager to not send termcap
1465 initialization and deinitialization strings to the terminal. It
1466 is set by default to allow command output to remain visible in
1467 the terminal even after the pager exits. Nevertheless, this
1468 prevents some pager functionality from working, in particular
1469 paged output cannot be scrolled with the mouse.
1470 See less(1) for more discussion.
1472 $SYSTEMD_LESSCHARSET
1474 Override the charset passed to less (by default "utf-8", if the
1475 invoking terminal is determined to be UTF-8 compatible).
1476 $SYSTEMD_PAGERSECURE
1478 Takes a boolean argument. When true, the "secure" mode of the pager
1479 is enabled; if false, disabled. If $SYSTEMD_PAGERSECURE is not set
1480 at all, secure mode is enabled if the effective UID is not the same
1481 as the owner of the login session, see geteuid(2) and
1482 sd_pid_get_owner_uid(3). In secure mode, LESSSECURE=1 will be set
1483 when invoking the pager, and the pager shall disable commands that
1484 open or create new files or start new subprocesses. When
1485 $SYSTEMD_PAGERSECURE is not set at all, pagers which are not known
1486 to implement secure mode will not be used. (Currently only less(1)
1487 implements secure mode.)
1488 Note: when commands are invoked with elevated privileges, for
1490 example under sudo(8) or pkexec(1), care must be taken to ensure
1491 that unintended interactive features are not enabled. "Secure" mode
1492 for the pager may be enabled automatically as describe above.
1493 Setting SYSTEMD_PAGERSECURE=0 or not removing it from the inherited
1494 environment allows the user to invoke arbitrary commands. Note that
1495 if the $SYSTEMD_PAGER or $PAGER variables are to be honoured,
1496 $SYSTEMD_PAGERSECURE must be set too. It might be reasonable to
1497 completely disable the pager using --no-pager instead.
1498 $SYSTEMD_COLORS
1500 Takes a boolean argument. When true, systemd and related utilities
1501 will use colors in their output, otherwise the output will be
1502 monochrome. Additionally, the variable can take one of the
1503 following special values: "16", "256" to restrict the use of colors
1504 to the base 16 or 256 ANSI colors, respectively. This can be
1505 specified to override the automatic decision based on $TERM and
1506 what the console is connected to.
1507 $SYSTEMD_URLIFY
1509 The value must be a boolean. Controls whether clickable links
1510 should be generated in the output for terminal emulators supporting
1511 this. This can be specified to override the decision that systemd
1512 makes based on $TERM and other conditions.
1513
1515 Example 1. Download a Fedora image and start a shell in it
1516 # machinectl pull-raw --verify=no \
1518 https://download.fedoraproject.org/pub/fedora/linux/releases/37/Cloud/x86_64/images/Fedora-Cloud-Base-37-1.7.x86_64.raw.xz \
1519 Fedora-Cloud-Base-37-1.7.x86-64
1520 # systemd-nspawn -M Fedora-Cloud-Base-37-1.7.x86-64
1521 This downloads an image using machinectl(1) and opens a shell in it.
1523 Example 2. Build and boot a minimal Fedora distribution in a container
1525 # dnf -y --releasever=37 --installroot=/var/lib/machines/f37 \
1527 --repo=fedora --repo=updates --setopt=install_weak_deps=False install \
1528 passwd dnf fedora-release vim-minimal util-linux systemd systemd-networkd
1529 # systemd-nspawn -bD /var/lib/machines/f37
1530 This installs a minimal Fedora distribution into the directory
1532 /var/lib/machines/f37 and then boots that OS in a namespace container.
1533 Because the installation is located underneath the standard
1534 /var/lib/machines/ directory, it is also possible to start the machine
1535 using systemd-nspawn -M f37.
1536 Example 3. Spawn a shell in a container of a minimal Debian unstable
1538 distribution
1539 # debootstrap unstable ~/debian-tree/
1541 # systemd-nspawn -D ~/debian-tree/
1542 This installs a minimal Debian unstable distribution into the directory
1544 ~/debian-tree/ and then spawns a shell from this image in a namespace
1545 container.
1546 debootstrap supports Debian[7], Ubuntu[8], and Tanglu[9] out of the
1548 box, so the same command can be used to install any of those. For other
1549 distributions from the Debian family, a mirror has to be specified, see
1550 debootstrap(8).
1551 Example 4. Boot a minimal Arch Linux distribution in a container
1553 # pacstrap -c ~/arch-tree/ base
1555 # systemd-nspawn -bD ~/arch-tree/
1556 This installs a minimal Arch Linux distribution into the directory
1558 ~/arch-tree/ and then boots an OS in a namespace container in it.
1559 Example 5. Install the OpenSUSE Tumbleweed rolling distribution
1561 # zypper --root=/var/lib/machines/tumbleweed ar -c \
1563 https://download.opensuse.org/tumbleweed/repo/oss tumbleweed
1564 # zypper --root=/var/lib/machines/tumbleweed refresh
1565 # zypper --root=/var/lib/machines/tumbleweed install --no-recommends \
1566 systemd shadow zypper openSUSE-release vim
1567 # systemd-nspawn -M tumbleweed passwd root
1568 # systemd-nspawn -M tumbleweed -b
1569 Example 6. Boot into an ephemeral snapshot of the host system
1571 # systemd-nspawn -D / -xb
1573 This runs a copy of the host system in a snapshot which is removed
1575 immediately when the container exits. All file system changes made
1576 during runtime will be lost on shutdown, hence.
1577 Example 7. Run a container with SELinux sandbox security contexts
1579 # chcon system_u:object_r:svirt_sandbox_file_t:s0:c0,c1 -R /srv/container
1581 # systemd-nspawn -L system_u:object_r:svirt_sandbox_file_t:s0:c0,c1 \
1582 -Z system_u:system_r:svirt_lxc_net_t:s0:c0,c1 -D /srv/container /bin/sh
1583 Example 8. Run a container with an OSTree deployment
1585 # systemd-nspawn -b -i ~/image.raw \
1587 --pivot-root=/ostree/deploy/$OS/deploy/$CHECKSUM:/sysroot \
1588 --bind=+/sysroot/ostree/deploy/$OS/var:/var
1589
1591 The exit code of the program executed in the container is returned.
1592
1594 systemd(1), systemd.nspawn(5), chroot(1), dnf(8), debootstrap(8),
1595 pacman(8), zypper(8), systemd.slice(5), machinectl(1), btrfs(8)
1596
1598 1. Container Interface
1599 https://systemd.io/CONTAINER_INTERFACE
1600 2. Discoverable Partitions Specification
1602 https://uapi-group.org/specifications/specs/discoverable_partitions_specification
1603 3. OCI Runtime Specification
1605 https://github.com/opencontainers/runtime-spec/blob/master/spec.md
1606 4. OSTree
1608 https://ostree.readthedocs.io/en/latest/
1609 5. Overlay Filesystem
1611 https://docs.kernel.org/filesystems/overlayfs.html
1612 6. Fedora
1614 https://getfedora.org
1615 7. Debian
1617 https://www.debian.org
1618 8. Ubuntu
1620 https://www.ubuntu.com
1621 9. Tanglu
1623 https://www.tanglu.org
1624 10. Arch Linux
1626 https://www.archlinux.org
1627 11. OpenSUSE Tumbleweed
1629 https://software.opensuse.org/distributions/tumbleweed
1630systemd 253 SYSTEMD-NSPAWN(1)