1 systemd System and Service Manager
5 * modprobe.d drop-in is now shipped by default that sets bonding module
6 option max_bonds=0. This overrides the kernel default, to avoid
7 conflicts and ambiguity as to whether or not bond0 should be managed
8 by networkd or not. This resolves multiple bugs of bond0 properties
9 not being applied, when bond0 is configured with
10 networkd. Distributors may choose to not package this, however in
11 that case users will be prevented from correctly managing bond0
12 interface using networkd.
14 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
15 which print the logging level and target of the system manager,
16 respectively. They complement the existing "set-log-level" and
17 "set-log-target" verbs, which can be used to change those values.
21 * Meson is now supported as build system in addition to Automake. It is
22 our plan to remove Automake in one of our next releases, so that
23 Meson becomes our exclusive build system. Hence, please start using
24 the Meson build system in your downstream packaging. There's plenty
25 of documentation around how to use Meson, the extremely brief
28 ./autogen.sh && ./configure && make && sudo make install
32 meson build && ninja -C build && sudo ninja -C build install
34 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
35 which permits configuring a timeout on the time a job is
36 running. This is particularly useful for setting timeouts on jobs for
39 * Unit files gained two new options ConditionUser= and ConditionGroup=
40 for conditionalizing units based on the identity of the user/group
41 running a systemd user instance.
43 * systemd-networkd now understands a new FlowLabel= setting in the
44 [VXLAN] section of .network files, as well as a Priority= in
45 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
46 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
47 gained support for configuration of GENEVE links, and IPv6 address
48 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
50 * .link files now understand a new Port= setting.
52 * systemd-networkd's DHCP support gained support for DHCP option 119
55 * systemd-networkd gained support for serving IPv6 address ranges using
56 the Router Advertisment protocol. The new .network configuration
57 section [IPv6Prefix] may be used to configure the ranges to
58 serve. This is implemented based on a new, minimal, native server
61 * journalctl's --output= switch gained support for a new parameter
62 "short-iso-precise" for a mode where timestamps are shown as precise
65 * systemd-udevd's "net_id" builtin may now generate stable network
66 interface names from IBM PowerVM VIO devices as well as ACPI platform
69 * MulticastDNS support in systemd-resolved may now be explicitly
70 enabled/disabled using the new MulticastDNS= configuration file
73 * systemd-resolved may now optionally use libidn2 instead of the libidn
74 for processing internationalized domain names. Support for libidn2
75 should be considered experimental and should not be enabled by
78 * "machinectl pull-tar" and related call may now do verification of
79 downloaded images using SUSE-style .sha256 checksum files in addition
80 to the already existing support for validating using Ubuntu-style
83 * sd-bus gained support for a new sd_bus_message_appendv() call which
84 is va_list equivalent of sd_bus_message_append().
86 * sd-boot gained support for validating images using SHIM/MOK.
88 * The SMACK code learnt support for "onlycap".
90 * systemd-mount --umount is now much smarter in figuring out how to
91 properly unmount a device given its mount or device path.
93 * The code to call libnss_dns as a fallback from libnss_resolve when
94 the communication with systemd-resolved fails was removed. This
95 fallback was redundant and interfered with the [!UNAVAIL=return]
96 suffix. See nss-resolve(8) for the recommended configuration.
98 * systemd-logind may now be restarted without losing state. It stores
99 the file descriptors for devices it manages in the system manager
100 using the FDSTORE= mechanism. Please note that further changes in
101 other components may be required to make use of this (for example
102 Xorg has code to listen for stops of systemd-logind and terminate
103 itself when logind is stopped or restarted, in order to avoid using
104 stale file descriptors for graphical devices, which is now
105 counterproductive and must be reverted in order for restarts of
106 systemd-logind to be safe. See
107 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
109 * All kernel install plugins are called with the environment variable
110 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
111 /etc/machine-id. If the file is missing or empty, the variable is
112 empty and BOOT_DIR_ABS is the path of a temporary directory which is
113 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
114 is empty, all plugins should not put anything in BOOT_DIR_ABS.
116 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
117 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
118 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
119 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
120 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
121 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
122 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
123 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
124 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
125 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
126 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
127 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
128 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
129 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
130 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
131 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
132 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
133 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
134 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
135 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
136 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
137 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
138 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
139 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
140 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
141 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
142 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
143 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
144 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
151 * The "hybrid" control group mode has been modified to improve
152 compatibility with "legacy" cgroups-v1 setups. Specifically, the
153 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
154 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
155 cgroups-v1 hierarchy), the only externally visible change being that
156 the cgroups-v2 hierarchy is also mounted, to
157 /sys/fs/cgroup/unified. This should provide a large degree of
158 compatibility with "legacy" cgroups-v1, while taking benefit of the
159 better management capabilities of cgroups-v2.
161 * The default control group setup mode may be selected both a boot-time
162 via a set of kernel command line parameters (specifically:
163 systemd.unified_cgroup_hierarchy= and
164 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
165 default selected on the configure command line
166 (--with-default-hierarchy=). The upstream default is "hybrid"
167 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
168 this will change in a future systemd version to be "unified" (pure
169 cgroups-v2 mode). The third option for the compile time option is
170 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
171 distributions to default to "hybrid" mode for release distributions,
172 starting with v233. We recommend "unified" for development
173 distributions (specifically: distributions such as Fedora's rawhide)
174 as that's where things are headed in the long run. Use "legacy" for
175 greatest stability and compatibility only.
177 * Note one current limitation of "unified" and "hybrid" control group
178 setup modes: the kernel currently does not permit the systemd --user
179 instance (i.e. unprivileged code) to migrate processes between two
180 disconnected cgroup subtrees, even if both are managed and owned by
181 the user. This effectively means "systemd-run --user --scope" doesn't
182 work when invoked from outside of any "systemd --user" service or
183 scope. Specifically, it is not supported from session scopes. We are
184 working on fixing this in a future systemd version. (See #3388 for
185 further details about this.)
187 * DBus policy files are now installed into /usr rather than /etc. Make
188 sure your system has dbus >= 1.9.18 running before upgrading to this
189 version, or override the install path with --with-dbuspolicydir= .
191 * All python scripts shipped with systemd (specifically: the various
192 tests written in Python) now require Python 3.
194 * systemd unit tests can now run standalone (without the source or
195 build directories), and can be installed into /usr/lib/systemd/tests/
196 with 'make install-tests'.
198 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
199 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
202 * Support for the %c, %r, %R specifiers in unit files has been
203 removed. Specifiers are not supposed to be dependent on configuration
204 in the unit file itself (so that they resolve the same regardless
205 where used in the unit files), but these specifiers were influenced
206 by the Slice= option.
208 * The shell invoked by debug-shell.service now defaults to /bin/sh in
209 all cases. If distributions want to use a different shell for this
210 purpose (for example Fedora's /sbin/sushell) they need to specify
211 this explicitly at configure time using --with-debug-shell=.
213 * The confirmation spawn prompt has been reworked to offer the
216 (c)ontinue, proceed without asking anymore
217 (D)ump, show the state of the unit
218 (f)ail, don't execute the command and pretend it failed
220 (i)nfo, show a short summary of the unit
221 (j)obs, show jobs that are in progress
222 (s)kip, don't execute the command and pretend it succeeded
223 (y)es, execute the command
225 The 'n' choice for the confirmation spawn prompt has been removed,
226 because its meaning was confusing.
228 The prompt may now also be redirected to an alternative console by
229 specifying the console as parameter to systemd.confirm_spawn=.
231 * Services of Type=notify require a READY=1 notification to be sent
232 during startup. If no such message is sent, the service now fails,
233 even if the main process exited with a successful exit code.
235 * Services that fail to start up correctly now always have their
236 ExecStopPost= commands executed. Previously, they'd enter "failed"
237 state directly, without executing these commands.
239 * The option MulticastDNS= of network configuration files has acquired
240 an actual implementation. With MulticastDNS=yes a host can resolve
241 names of remote hosts and reply to mDNS A and AAAA requests.
243 * When units are about to be started an additional check is now done to
244 ensure that all dependencies of type BindsTo= (when used in
245 combination with After=) have been started.
247 * systemd-analyze gained a new verb "syscall-filter" which shows which
248 system call groups are defined for the SystemCallFilter= unit file
249 setting, and which system calls they contain.
251 * A new system call filter group "@filesystem" has been added,
252 consisting of various file system related system calls. Group
253 "@reboot" has been added, covering reboot, kexec and shutdown related
254 calls. Finally, group "@swap" has been added covering swap
255 configuration related calls.
257 * A new unit file option RestrictNamespaces= has been added that may be
258 used to restrict access to the various process namespace types the
259 Linux kernel provides. Specifically, it may be used to take away the
260 right for a service unit to create additional file system, network,
261 user, and other namespaces. This sandboxing option is particularly
262 relevant due to the high amount of recently discovered namespacing
263 related vulnerabilities in the kernel.
265 * systemd-udev's .link files gained support for a new AutoNegotiation=
266 setting for configuring Ethernet auto-negotiation.
268 * systemd-networkd's .network files gained support for a new
269 ListenPort= setting in the [DHCP] section to explicitly configure the
270 UDP client port the DHCP client shall listen on.
272 * .network files gained a new Unmanaged= boolean setting for explicitly
273 excluding one or more interfaces from management by systemd-networkd.
275 * The systemd-networkd ProxyARP= option has been renamed to
276 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
277 renamed to ReduceARPProxy=. The old names continue to be available
280 * systemd-networkd gained support for configuring IPv6 Proxy NDP
281 addresses via the new IPv6ProxyNDPAddress= .network file setting.
283 * systemd-networkd's bonding device support gained support for two new
284 configuration options ActiveSlave= and PrimarySlave=.
286 * The various options in the [Match] section of .network files gained
287 support for negative matching.
289 * New systemd-specific mount options are now understood in /etc/fstab:
291 x-systemd.mount-timeout= may be used to configure the maximum
292 permitted runtime of the mount command.
294 x-systemd.device-bound may be set to bind a mount point to its
295 backing device unit, in order to automatically remove a mount point
296 if its backing device is unplugged. This option may also be
297 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
298 on the block device, which is now automatically set for all CDROM
299 drives, so that mounted CDs are automatically unmounted when they are
300 removed from the drive.
302 x-systemd.after= and x-systemd.before= may be used to explicitly
303 order a mount after or before another unit or mount point.
305 * Enqueued start jobs for device units are now automatically garbage
306 collected if there are no jobs waiting for them anymore.
308 * systemctl list-jobs gained two new switches: with --after, for every
309 queued job the jobs it's waiting for are shown; with --before the
310 jobs which it's blocking are shown.
312 * systemd-nspawn gained support for ephemeral boots from disk images
313 (or in other words: --ephemeral and --image= may now be
314 combined). Moreover, ephemeral boots are now supported for normal
315 directories, even if the backing file system is not btrfs. Of course,
316 if the file system does not support file system snapshots or
317 reflinks, the initial copy operation will be relatively expensive, but
318 this should still be suitable for many use cases.
320 * Calendar time specifications in .timer units now support
321 specifications relative to the end of a month by using "~" instead of
322 "-" as separator between month and day. For example, "*-02~03" means
323 "the third last day in February". In addition a new syntax for
324 repeated events has been added using the "/" character. For example,
325 "9..17/2:00" means "every two hours from 9am to 5pm".
327 * systemd-socket-proxyd gained a new parameter --connections-max= for
328 configuring the maximum number of concurrent connections.
330 * sd-id128 gained a new API for generating unique IDs for the host in a
331 way that does not leak the machine ID. Specifically,
332 sd_id128_get_machine_app_specific() derives an ID based on the
333 machine ID a in well-defined, non-reversible, stable way. This is
334 useful whenever an identifier for the host is needed but where the
335 identifier shall not be useful to identify the system beyond the
336 scope of the application itself. (Internally this uses HMAC-SHA256 as
337 keyed hash function using the machine ID as input.)
339 * NotifyAccess= gained a new supported value "exec". When set
340 notifications are accepted from all processes systemd itself invoked,
341 including all control processes.
343 * .nspawn files gained support for defining overlay mounts using the
344 Overlay= and OverlayReadOnly= options. Previously this functionality
345 was only available on the systemd-nspawn command line.
347 * systemd-nspawn's --bind= and --overlay= options gained support for
348 bind/overlay mounts whose source lies within the container tree by
349 prefixing the source path with "+".
351 * systemd-nspawn's --bind= and --overlay= options gained support for
352 automatically allocating a temporary source directory in /var/tmp
353 that is removed when the container dies. Specifically, if the source
354 directory is specified as empty string this mechanism is selected. An
355 example usage is --overlay=+/var::/var, which creates an overlay
356 mount based on the original /var contained in the image, overlayed
357 with a temporary directory in the host's /var/tmp. This way changes
358 to /var are automatically flushed when the container shuts down.
360 * systemd-nspawn --image= option does now permit raw file system block
361 devices (in addition to images containing partition tables, as
364 * The disk image dissection logic in systemd-nspawn gained support for
365 automatically setting up LUKS encrypted as well as Verity protected
366 partitions. When a container is booted from an encrypted image the
367 passphrase is queried at start-up time. When a container with Verity
368 data is started, the root hash is search in a ".roothash" file
369 accompanying the disk image (alternatively, pass the root hash via
370 the new --root-hash= command line option).
372 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
373 be used to dissect disk images the same way as systemd-nspawn does
374 it, following the Bootable Partition Specification. It may even be
375 used to mount disk images with complex partition setups (including
376 LUKS and Verity partitions) to a local host directory, in order to
377 inspect them. This tool is not considered public API (yet), and is
378 thus not installed into /usr/bin. Please do not rely on its
379 existence, since it might go away or be changed in later systemd
382 * A new generator "systemd-verity-generator" has been added, similar in
383 style to "systemd-cryptsetup-generator", permitting automatic setup of
384 Verity root partitions when systemd boots up. In order to make use of
385 this your partition setup should follow the Discoverable Partitions
386 Specification, and the GPT partition ID of the root file system
387 partition should be identical to the upper 128bit of the Verity root
388 hash. The GPT partition ID of the Verity partition protecting it
389 should be the lower 128bit of the Verity root hash. If the partition
390 image follows this model it is sufficient to specify a single
391 "roothash=" kernel command line argument to both configure which root
392 image and verity partition to use as well as the root hash for
393 it. Note that systemd-nspawn's Verity support follows the same
394 semantics, meaning that disk images with proper Verity data in place
395 may be booted in containers with systemd-nspawn as well as on
396 physical systems via the verity generator. Also note that the "mkosi"
397 tool available at https://github.com/systemd/mkosi has been updated
398 to generate Verity protected disk images following this scheme. In
399 fact, it has been updated to generate disk images that optionally
400 implement a complete UEFI SecureBoot trust chain, involving a signed
401 kernel and initrd image that incorporates such a root hash as well as
402 a Verity-enabled root partition.
404 * The hardware database (hwdb) udev supports has been updated to carry
405 accelerometer quirks.
407 * All system services are now run with a fresh kernel keyring set up
408 for them. The invocation ID is stored by default in it, thus
409 providing a safe, non-overridable way to determine the invocation
412 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
413 options for bind mounting arbitrary paths in a service-specific
414 way. When these options are used, arbitrary host or service files and
415 directories may be mounted to arbitrary locations in the service's
418 * Documentation has been added that lists all of systemd's low-level
419 environment variables:
421 https://github.com/systemd/systemd/blob/master/ENVIRONMENT.md
423 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
424 whether a specific socket file descriptor matches a specified socket
427 * systemd-firstboot has been updated to check for the
428 systemd.firstboot= kernel command line option. It accepts a boolean
429 and when set to false the first boot questions are skipped.
431 * systemd-fstab-generator has been updated to check for the
432 systemd.volatile= kernel command line option, which either takes an
433 optional boolean parameter or the special value "state". If used the
434 system may be booted in a "volatile" boot mode. Specifically,
435 "systemd.volatile" is used, the root directory will be mounted as
436 tmpfs, and only /usr is mounted from the actual root file system. If
437 "systemd.volatile=state" is used, the root directory will be mounted
438 as usual, but /var is mounted as tmpfs. This concept provides similar
439 functionality as systemd-nspawn's --volatile= option, but provides it
440 on physical boots. Use this option for implementing stateless
441 systems, or testing systems with all state and/or configuration reset
442 to the defaults. (Note though that many distributions are not
443 prepared to boot up without a populated /etc or /var, though.)
445 * systemd-gpt-auto-generator gained support for LUKS encrypted root
446 partitions. Previously it only supported LUKS encrypted partitions
447 for all other uses, except for the root partition itself.
449 * Socket units gained support for listening on AF_VSOCK sockets for
450 communication in virtualized QEMU environments.
452 * The "configure" script gained a new option --with-fallback-hostname=
453 for specifying the fallback hostname to use if none is configured in
454 /etc/hostname. For example, by specifying
455 --with-fallback-hostname=fedora it is possible to default to a
456 hostname of "fedora" on pristine installations.
458 * systemd-cgls gained support for a new --unit= switch for listing only
459 the control groups of a specific unit. Similar --user-unit= has been
460 added for listing only the control groups of a specific user unit.
462 * systemd-mount gained a new --umount switch for unmounting a mount or
463 automount point (and all mount/automount points below it).
465 * systemd will now refuse full configuration reloads (via systemctl
466 daemon-reload and related calls) unless at least 16MiB of free space
467 are available in /run. This is a safety precaution in order to ensure
468 that generators can safely operate after the reload completed.
470 * A new unit file option RootImage= has been added, which has a similar
471 effect as RootDirectory= but mounts the service's root directory from
472 a disk image instead of plain directory. This logic reuses the same
473 image dissection and mount logic that systemd-nspawn already uses,
474 and hence supports any disk images systemd-nspawn supports, including
475 those following the Discoverable Partition Specification, as well as
476 Verity enabled images. This option enables systemd to run system
477 services directly off disk images acting as resource bundles,
478 possibly even including full integrity data.
480 * A new MountAPIVFS= unit file option has been added, taking a boolean
481 argument. If enabled /proc, /sys and /dev (collectively called the
482 "API VFS") will be mounted for the service. This is only relevant if
483 RootDirectory= or RootImage= is used for the service, as these mounts
484 are of course in place in the host mount namespace anyway.
486 * systemd-nspawn gained support for a new --pivot-root= switch. If
487 specified the root directory within the container image is pivoted to
488 the specified mount point, while the original root disk is moved to a
489 different place. This option enables booting of ostree images
490 directly with systemd-nspawn.
492 * The systemd build scripts will no longer complain if the NTP server
493 addresses are not changed from the defaults. Google now supports
494 these NTP servers officially. We still recommend downstreams to
495 properly register an NTP pool with the NTP pool project though.
497 * coredumpctl gained a new "--reverse" option for printing the list
498 of coredumps in reverse order.
500 * coredumpctl will now show additional information about truncated and
501 inaccessible coredumps, as well as coredumps that are still being
502 processed. It also gained a new --quiet switch for suppressing
503 additional informational message in its output.
505 * coredumpctl gained support for only showing coredumps newer and/or
506 older than specific timestamps, using the new --since= and --until=
507 options, reminiscent of journalctl's options by the same name.
509 * The systemd-coredump logic has been improved so that it may be reused
510 to collect backtraces in non-compiled languages, for example in
511 scripting languages such as Python.
513 * machinectl will now show the UID shift of local containers, if user
514 namespacing is enabled for them.
516 * systemd will now optionally run "environment generator" binaries at
517 configuration load time. They may be used to add environment
518 variables to the environment block passed to services invoked. One
519 user environment generator is shipped by default that sets up
520 environment variables based on files dropped into /etc/environment.d
521 and ~/.config/environment.d/.
523 * systemd-resolved now includes the new, recently published 2017 DNSSEC
526 * hostnamed has been updated to report a new chassis type of
527 "convertible" to cover "foldable" laptops that can both act as a
528 tablet and as a laptop, such as various Lenovo Yoga devices.
530 Contributions from: Adrián López, Alexander Galanin, Alexander
531 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
532 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
533 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
534 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
535 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
536 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
537 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
538 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
539 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
540 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
541 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
542 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
543 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
544 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
545 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
546 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
547 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
548 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
549 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
550 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
551 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
552 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
553 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
554 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
555 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
556 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
557 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
564 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
565 RestrictAddressFamilies= enabled. These sandboxing options should
566 generally be compatible with the various external udev call-out
567 binaries we are aware of, however there may be exceptions, in
568 particular when exotic languages for these call-outs are used. In
569 this case, consider turning off these settings locally.
571 * The new RemoveIPC= option can be used to remove IPC objects owned by
572 the user or group of a service when that service exits.
574 * The new ProtectKernelModules= option can be used to disable explicit
575 load and unload operations of kernel modules by a service. In
576 addition access to /usr/lib/modules is removed if this option is set.
578 * ProtectSystem= option gained a new value "strict", which causes the
579 whole file system tree with the exception of /dev, /proc, and /sys,
580 to be remounted read-only for a service.
582 * The new ProtectKernelTunables= option can be used to disable
583 modification of configuration files in /sys and /proc by a service.
584 Various directories and files are remounted read-only, so access is
585 restricted even if the file permissions would allow it.
587 * The new ProtectControlGroups= option can be used to disable write
588 access by a service to /sys/fs/cgroup.
590 * Various systemd services have been hardened with
591 ProtectKernelTunables=yes, ProtectControlGroups=yes,
592 RestrictAddressFamilies=.
594 * Support for dynamically creating users for the lifetime of a service
595 has been added. If DynamicUser=yes is specified, user and group IDs
596 will be allocated from the range 61184..65519 for the lifetime of the
597 service. They can be resolved using the new nss-systemd.so NSS
598 module. The module must be enabled in /etc/nsswitch.conf. Services
599 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
600 any resources allocated by the service will be cleaned up when the
601 service exits. They also have ProtectHome=read-only and
602 ProtectSystem=strict enabled, so they are not able to make any
603 permanent modifications to the system.
605 * The nss-systemd module also always resolves root and nobody, making
606 it possible to have no /etc/passwd or /etc/group files in minimal
607 container or chroot environments.
609 * Services may be started with their own user namespace using the new
610 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
611 under which the service is running are mapped. All other users are
614 * Support for the cgroup namespace has been added to systemd-nspawn. If
615 supported by kernel, the container system started by systemd-nspawn
616 will have its own view of the cgroup hierarchy. This new behaviour
617 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
619 * The new MemorySwapMax= option can be used to limit the maximum swap
620 usage under the unified cgroup hierarchy.
622 * Support for the CPU controller in the unified cgroup hierarchy has
623 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
624 options. This controller requires out-of-tree patches for the kernel
625 and the support is provisional.
627 * Mount and automount units may now be created transiently
628 (i.e. dynamically at runtime via the bus API, instead of requiring
629 unit files in the file system).
631 * systemd-mount is a new tool which may mount file systems – much like
632 mount(8), optionally pulling in additional dependencies through
633 transient .mount and .automount units. For example, this tool
634 automatically runs fsck on a backing block device before mounting,
635 and allows the automount logic to be used dynamically from the
636 command line for establishing mount points. This tool is particularly
637 useful when dealing with removable media, as it will ensure fsck is
638 run – if necessary – before the first access and that the file system
639 is quickly unmounted after each access by utilizing the automount
640 logic. This maximizes the chance that the file system on the
641 removable media stays in a clean state, and if it isn't in a clean
642 state is fixed automatically.
644 * LazyUnmount=yes option for mount units has been added to expose the
645 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
648 * /efi will be used as the mount point of the EFI boot partition, if
649 the directory is present, and the mount point was not configured
650 through other means (e.g. fstab). If /efi directory does not exist,
651 /boot will be used as before. This makes it easier to automatically
652 mount the EFI partition on systems where /boot is used for something
655 * When operating on GPT disk images for containers, systemd-nspawn will
656 now mount the ESP to /boot or /efi according to the same rules as PID
657 1 running on a host. This allows tools like "bootctl" to operate
658 correctly within such containers, in order to make container images
659 bootable on physical systems.
661 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
663 * Two new user session targets have been added to support running
664 graphical sessions under the systemd --user instance:
665 graphical-session.target and graphical-session-pre.target. See
666 systemd.special(7) for a description of how those targets should be
669 * The vconsole initialization code has been significantly reworked to
670 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
671 support unicode keymaps. Font and keymap configuration will now be
672 copied to all allocated virtual consoles.
674 * FreeBSD's bhyve virtualization is now detected.
676 * Information recorded in the journal for core dumps now includes the
677 contents of /proc/mountinfo and the command line of the process at
678 the top of the process hierarchy (which is usually the init process
681 * systemd-journal-gatewayd learned the --directory= option to serve
682 files from the specified location.
684 * journalctl --root=… can be used to peruse the journal in the
685 /var/log/ directories inside of a container tree. This is similar to
686 the existing --machine= option, but does not require the container to
689 * The hardware database has been extended to support
690 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
693 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
694 specify the click rate for mice which include a horizontal wheel with
695 a click rate that is different than the one for the vertical wheel.
697 * systemd-run gained a new --wait option that makes service execution
698 synchronous. (Specifically, the command will not return until the
699 specified service binary exited.)
701 * systemctl gained a new --wait option that causes the start command to
702 wait until the units being started have terminated again.
704 * A new journal output mode "short-full" has been added which displays
705 timestamps with abbreviated English day names and adds a timezone
706 suffix. Those timestamps include more information than the default
707 "short" output mode, and can be passed directly to journalctl's
708 --since= and --until= options.
710 * /etc/resolv.conf will be bind-mounted into containers started by
711 systemd-nspawn, if possible, so any changes to resolv.conf contents
712 are automatically propagated to the container.
714 * The number of instances for socket-activated services originating
715 from a single IP address can be limited with
716 MaxConnectionsPerSource=, extending the existing setting of
719 * systemd-networkd gained support for vcan ("Virtual CAN") interface
722 * .netdev and .network configuration can now be extended through
725 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
726 Segmentation Offload, Generic Receive Offload, Large Receive Offload
727 can be enabled and disabled using the new UDPSegmentationOffload=,
728 TCPSegmentationOffload=, GenericSegmentationOffload=,
729 GenericReceiveOffload=, LargeReceiveOffload= options in the
730 [Link] section of .link files.
732 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
733 Port VLAN ID can be configured for bridge devices using the new STP=,
734 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
735 section of .netdev files.
737 * The route table to which routes received over DHCP or RA should be
738 added can be configured with the new RouteTable= option in the [DHCP]
739 and [IPv6AcceptRA] sections of .network files.
741 * The Address Resolution Protocol can be disabled on links managed by
742 systemd-networkd using the ARP=no setting in the [Link] section of
745 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
746 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
747 encode information about the result and exit codes of the current
748 service runtime cycle.
750 * systemd-sysctl will now configure kernel parameters in the order
751 they occur in the configuration files. This matches what sysctl
752 has been traditionally doing.
754 * kernel-install "plugins" that are executed to perform various
755 tasks after a new kernel is added and before an old one is removed
756 can now return a special value to terminate the procedure and
757 prevent any later plugins from running.
759 * Journald's SplitMode=login setting has been deprecated. It has been
760 removed from documentation, and its use is discouraged. In a future
761 release it will be completely removed, and made equivalent to current
762 default of SplitMode=uid.
764 * Storage=both option setting in /etc/systemd/coredump.conf has been
765 removed. With fast LZ4 compression storing the core dump twice is not
768 * The --share-system systemd-nspawn option has been replaced with an
769 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
770 this functionality is discouraged. In addition the variables
771 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
772 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
773 individual namespaces.
775 * "machinectl list" now shows the IP address of running containers in
776 the output, as well as OS release information.
778 * "loginctl list" now shows the TTY of each session in the output.
780 * sd-bus gained new API calls sd_bus_track_set_recursive(),
781 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
782 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
783 tracking objects in a "recursive" mode, where a single client can be
784 counted multiple times, if it takes multiple references.
786 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
787 sd_bus_get_exit_on_disconnect(). They may be used to to make a
788 process using sd-bus automatically exit if the bus connection is
791 * Bus clients of the service manager may now "pin" loaded units into
792 memory, by taking an explicit reference on them. This is useful to
793 ensure the client can retrieve runtime data about the service even
794 after the service completed execution. Taking such a reference is
795 available only for privileged clients and should be helpful to watch
796 running services in a race-free manner, and in particular collect
797 information about exit statuses and results.
799 * The nss-resolve module has been changed to strictly return UNAVAIL
800 when communication via D-Bus with resolved failed, and NOTFOUND when
801 a lookup completed but was negative. This means it is now possible to
802 neatly configure fallbacks using nsswitch.conf result checking
803 expressions. Taking benefit of this, the new recommended
804 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
806 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
808 * A new setting CtrlAltDelBurstAction= has been added to
809 /etc/systemd/system.conf which may be used to configure the precise
810 behaviour if the user on the console presses Ctrl-Alt-Del more often
811 than 7 times in 2s. Previously this would unconditionally result in
812 an expedited, immediate reboot. With this new setting the precise
813 operation may be configured in more detail, and also turned off
816 * In .netdev files two new settings RemoteChecksumTx= and
817 RemoteChecksumRx= are now understood that permit configuring the
818 remote checksumming logic for VXLAN networks.
820 * The service manager learnt a new "invocation ID" concept for invoked
821 services. Each runtime cycle of a service will get a new invocation
822 ID (a 128bit random UUID) assigned that identifies the current
823 run of the service uniquely and globally. A new invocation ID
824 is generated each time a service starts up. The journal will store
825 the invocation ID of a service along with any logged messages, thus
826 making the invocation ID useful for matching the online runtime of a
827 service with the offline log data it generated in a safe way without
828 relying on synchronized timestamps. In many ways this new service
829 invocation ID concept is similar to the kernel's boot ID concept that
830 uniquely and globally identifies the runtime of each boot. The
831 invocation ID of a service is passed to the service itself via an
832 environment variable ($INVOCATION_ID). A new bus call
833 GetUnitByInvocationID() has been added that is similar to GetUnit()
834 but instead of retrieving the bus path for a unit by its name
835 retrieves it by its invocation ID. The returned path is valid only as
836 long as the passed invocation ID is current.
838 * systemd-resolved gained a new "DNSStubListener" setting in
839 resolved.conf. It either takes a boolean value or the special values
840 "udp" and "tcp", and configures whether to enable the stub DNS
841 listener on 127.0.0.53:53.
843 * IP addresses configured via networkd may now carry additional
844 configuration settings supported by the kernel. New options include:
845 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
846 PrefixRoute=, AutoJoin=.
848 * The PAM configuration fragment file for "user@.service" shipped with
849 systemd (i.e. the --user instance of systemd) has been stripped to
850 the minimum necessary to make the system boot. Previously, it
851 contained Fedora-specific stanzas that did not apply to other
852 distributions. It is expected that downstream distributions add
853 additional configuration lines, matching their needs to this file,
854 using it only as rough template of what systemd itself needs. Note
855 that this reduced fragment does not even include an invocation of
856 pam_limits which most distributions probably want to add, even though
857 systemd itself does not need it. (There's also the new build time
858 option --with-pamconfdir=no to disable installation of the PAM
861 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
862 capability is now also dropped from its set (in addition to
863 CAP_SYS_MKNOD as before).
865 * In service unit files it is now possible to connect a specific named
866 file descriptor with stdin/stdout/stdout of an executed service. The
867 name may be specified in matching .socket units using the
868 FileDescriptorName= setting.
870 * A number of journal settings may now be configured on the kernel
871 command line. Specifically, the following options are now understood:
872 systemd.journald.max_level_console=,
873 systemd.journald.max_level_store=,
874 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
875 systemd.journald.max_level_wall=.
877 * "systemctl is-enabled --full" will now show by which symlinks a unit
878 file is enabled in the unit dependency tree.
880 * Support for VeraCrypt encrypted partitions has been added to the
881 "cryptsetup" logic and /etc/crypttab.
883 * systemd-detect-virt gained support for a new --private-users switch
884 that checks whether the invoking processes are running inside a user
885 namespace. Similar, a new special value "private-users" for the
886 existing ConditionVirtualization= setting has been added, permitting
887 skipping of specific units in user namespace environments.
889 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
890 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
891 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
892 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
893 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
894 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
895 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
896 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
897 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
898 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
899 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
900 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
901 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
902 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
903 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
904 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
905 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
906 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
907 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
908 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
909 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
910 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
911 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
912 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
913 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
914 Jędrzejewski-Szmek, Zeal Jagannatha
916 — Santa Fe, 2016-11-03
920 * In service units the various ExecXYZ= settings have been extended
921 with an additional special character as first argument of the
922 assigned value: if the character '+' is used the specified command
923 line it will be run with full privileges, regardless of User=,
924 Group=, CapabilityBoundingSet= and similar options. The effect is
925 similar to the existing PermissionsStartOnly= option, but allows
926 configuration of this concept for each executed command line
929 * Services may now alter the service watchdog timeout at runtime by
930 sending a WATCHDOG_USEC= message via sd_notify().
932 * MemoryLimit= and related unit settings now optionally take percentage
933 specifications. The percentage is taken relative to the amount of
934 physical memory in the system (or in case of containers, the assigned
935 amount of memory). This allows scaling service resources neatly with
936 the amount of RAM available on the system. Similarly, systemd-logind's
937 RuntimeDirectorySize= option now also optionally takes percentage
940 * In similar fashion TasksMax= takes percentage values now, too. The
941 value is taken relative to the configured maximum number of processes
942 on the system. The per-service task maximum has been changed to 15%
943 using this functionality. (Effectively this is an increase of 512 →
944 4915 for service units, given the kernel's default pid_max setting.)
946 * Calendar time specifications in .timer units now understand a ".."
947 syntax for time ranges. Example: "4..7:10" may now be used for
948 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
951 * The InaccessableDirectories=, ReadOnlyDirectories= and
952 ReadWriteDirectories= unit file settings have been renamed to
953 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
954 applied to all kinds of file nodes, and not just directories, with
955 the exception of symlinks. Specifically these settings may now be
956 used on block and character device nodes, UNIX sockets and FIFOS as
957 well as regular files. The old names of these settings remain
958 available for compatibility.
960 * systemd will now log about all service processes it kills forcibly
961 (using SIGKILL) because they remained after the clean shutdown phase
962 of the service completed. This should help identifying services that
963 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
964 systemd-logind's configuration a similar log message is generated for
965 processes killed at the end of each session due to this setting.
967 * systemd will now set the $JOURNAL_STREAM environment variable for all
968 services whose stdout/stderr are connected to the Journal (which
969 effectively means by default: all services). The variable contains
970 the device and inode number of the file descriptor used for
971 stdout/stderr. This may be used by invoked programs to detect whether
972 their stdout/stderr is connected to the Journal, in which case they
973 can switch over to direct Journal communication, thus being able to
974 pass extended, structured metadata along with their log messages. As
975 one example, this is now used by glib's logging primitives.
977 * When using systemd's default tmp.mount unit for /tmp, the mount point
978 will now be established with the "nosuid" and "nodev" options. This
979 avoids privilege escalation attacks that put traps and exploits into
980 /tmp. However, this might cause problems if you e. g. put container
981 images or overlays into /tmp; if you need this, override tmp.mount's
982 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
985 * systemd now supports the "memory" cgroup controller also on
988 * The systemd-cgtop tool now optionally takes a control group path as
989 command line argument. If specified, the control group list shown is
990 limited to subgroups of that group.
992 * The SystemCallFilter= unit file setting gained support for
993 pre-defined, named system call filter sets. For example
994 SystemCallFilter=@clock is now an effective way to make all clock
995 changing-related system calls unavailable to a service. A number of
996 similar pre-defined groups are defined. Writing system call filters
997 for system services is simplified substantially with this new
998 concept. Accordingly, all of systemd's own, long-running services now
999 enable system call filtering based on this, by default.
1001 * A new service setting MemoryDenyWriteExecute= has been added, taking
1002 a boolean value. If turned on, a service may no longer create memory
1003 mappings that are writable and executable at the same time. This
1004 enhances security for services where this is enabled as it becomes
1005 harder to dynamically write and then execute memory in exploited
1006 service processes. This option has been enabled for all of systemd's
1007 own long-running services.
1009 * A new RestrictRealtime= service setting has been added, taking a
1010 boolean argument. If set the service's processes may no longer
1011 acquire realtime scheduling. This improves security as realtime
1012 scheduling may otherwise be used to easily freeze the system.
1014 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
1015 value. This may be used for requesting that the system manager inside
1016 of the container reports start-up completion to nspawn which then
1017 propagates this notification further to the service manager
1018 supervising nspawn itself. A related option NotifyReady= in .nspawn
1019 files has been added too. This functionality allows ordering of the
1020 start-up of multiple containers using the usual systemd ordering
1023 * machinectl gained a new command "stop" that is an alias for
1026 * systemd-resolved gained support for contacting DNS servers on
1027 link-local IPv6 addresses.
1029 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
1030 its caches. A method call for requesting the same operation has been
1031 added to the bus API too, and is made available via "systemd-resolve
1034 * systemd-resolve gained a new --status switch. If passed a brief
1035 summary of the used DNS configuration with per-interface information
1038 * resolved.conf gained a new Cache= boolean option, defaulting to
1039 on. If turned off local DNS caching is disabled. This comes with a
1040 performance penalty in particular when DNSSEC is enabled. Note that
1041 resolved disables its internal caching implicitly anyway, when the
1042 configured DNS server is on a host-local IP address such as ::1 or
1043 127.0.0.1, thus automatically avoiding double local caching.
1045 * systemd-resolved now listens on the local IP address 127.0.0.53:53
1046 for DNS requests. This improves compatibility with local programs
1047 that do not use the libc NSS or systemd-resolved's bus APIs for name
1048 resolution. This minimal DNS service is only available to local
1049 programs and does not implement the full DNS protocol, but enough to
1050 cover local DNS clients. A new, static resolv.conf file, listing just
1051 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
1052 now recommended to make /etc/resolv.conf a symlink to this file in
1053 order to route all DNS lookups to systemd-resolved, regardless if
1054 done via NSS, the bus API or raw DNS packets. Note that this local
1055 DNS service is not as fully featured as the libc NSS or
1056 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
1057 used to deliver link-local address information (as this implies
1058 sending a local interface index along), LLMNR/mDNS support via this
1059 interface is severely restricted. It is thus strongly recommended for
1060 all applications to use the libc NSS API or native systemd-resolved
1063 * systemd-networkd's bridge support learned a new setting
1064 VLANFiltering= for controlling VLAN filtering. Moreover a new section
1065 in .network files has been added for configuring VLAN bridging in
1066 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
1068 * systemd-networkd's IPv6 Router Advertisement code now makes use of
1069 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
1070 now be acquired without relying on DHCPv6. Two new options
1071 UseDomains= and UseDNS= have been added to configure this behaviour.
1073 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
1074 renamed IPv6AcceptRA=, without altering its behaviour. The old
1075 setting name remains available for compatibility reasons.
1077 * The systemd-networkd VTI/VTI6 tunneling support gained new options
1078 Key=, InputKey= and OutputKey=.
1080 * systemd-networkd gained support for VRF ("Virtual Routing Function")
1081 interface configuration.
1083 * "systemctl edit" may now be used to create new unit files by
1084 specifying the --force switch.
1086 * sd-event gained a new function sd_event_get_iteration() for
1087 requesting the current iteration counter of the event loop. It starts
1088 at zero and is increased by one with each event loop iteration.
1090 * A new rpm macro %systemd_ordering is provided by the macros.systemd
1091 file. It can be used in lieu of %systemd_requires in packages which
1092 don't use any systemd functionality and are intended to be installed
1093 in minimal containers without systemd present. This macro provides
1094 ordering dependencies to ensure that if the package is installed in
1095 the same rpm transaction as systemd, systemd will be installed before
1096 the scriptlets for the package are executed, allowing unit presets
1099 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
1100 been added to simplify packaging of generators.
1102 * The os-release file gained VERSION_CODENAME field for the
1103 distribution nickname (e.g. VERSION_CODENAME=woody).
1105 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
1106 can be set to disable parsing of metadata and the creation
1107 of persistent symlinks for that device.
1109 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
1110 to make them available to logged-in users has been reverted.
1112 * Much of the common code of the various systemd components is now
1113 built into an internal shared library libsystemd-shared-231.so
1114 (incorporating the systemd version number in the name, to be updated
1115 with future releases) that the components link to. This should
1116 decrease systemd footprint both in memory during runtime and on
1117 disk. Note that the shared library is not for public use, and is
1118 neither API not ABI stable, but is likely to change with every new
1119 released update. Packagers need to make sure that binaries
1120 linking to libsystemd-shared.so are updated in step with the
1123 * Configuration for "mkosi" is now part of the systemd
1124 repository. mkosi is a tool to easily build legacy-free OS images,
1125 and is available on github: https://github.com/systemd/mkosi. If
1126 "mkosi" is invoked in the build tree a new raw OS image is generated
1127 incorporating the systemd sources currently being worked on and a
1128 clean, fresh distribution installation. The generated OS image may be
1129 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
1130 UEFI PC. This functionality is particularly useful to easily test
1131 local changes made to systemd in a pristine, defined environment. See
1132 HACKING for details.
1134 * configure learned the --with-support-url= option to specify the
1135 distribution's bugtracker.
1137 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
1138 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
1139 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
1140 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
1141 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
1142 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
1143 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
1144 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
1145 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
1146 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
1147 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
1148 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
1149 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
1150 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
1151 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
1152 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
1153 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
1154 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
1155 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
1157 — Berlin, 2016-07-25
1161 * DNSSEC is now turned on by default in systemd-resolved (in
1162 "allow-downgrade" mode), but may be turned off during compile time by
1163 passing "--with-default-dnssec=no" to "configure" (and of course,
1164 during runtime with DNSSEC= in resolved.conf). We recommend
1165 downstreams to leave this on at least during development cycles and
1166 report any issues with the DNSSEC logic upstream. We are very
1167 interested in collecting feedback about the DNSSEC validator and its
1168 limitations in the wild. Note however, that DNSSEC support is
1169 probably nothing downstreams should turn on in stable distros just
1170 yet, as it might create incompatibilities with a few DNS servers and
1171 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
1172 automatically whenever we detect such incompatible setups, but there
1173 might be systems we do not cover yet. Hence: please help us testing
1174 the DNSSEC code, leave this on where you can, report back, but then
1175 again don't consider turning this on in your stable, LTS or
1176 production release just yet. (Note that you have to enable
1177 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
1178 and its DNSSEC mode for host name resolution from local
1181 * systemd-resolve conveniently resolves DANE records with the --tlsa
1182 option and OPENPGPKEY records with the --openpgp option. It also
1183 supports dumping raw DNS record data via the new --raw= switch.
1185 * systemd-logind will now by default terminate user processes that are
1186 part of the user session scope unit (session-XX.scope) when the user
1187 logs out. This behavior is controlled by the KillUserProcesses=
1188 setting in logind.conf, and the previous default of "no" is now
1189 changed to "yes". This means that user sessions will be properly
1190 cleaned up after, but additional steps are necessary to allow
1191 intentionally long-running processes to survive logout.
1193 While the user is logged in at least once, user@.service is running,
1194 and any service that should survive the end of any individual login
1195 session can be started at a user service or scope using systemd-run.
1196 systemd-run(1) man page has been extended with an example which shows
1197 how to run screen in a scope unit underneath user@.service. The same
1198 command works for tmux.
1200 After the user logs out of all sessions, user@.service will be
1201 terminated too, by default, unless the user has "lingering" enabled.
1202 To effectively allow users to run long-term tasks even if they are
1203 logged out, lingering must be enabled for them. See loginctl(1) for
1204 details. The default polkit policy was modified to allow users to
1205 set lingering for themselves without authentication.
1207 Previous defaults can be restored at compile time by the
1208 --without-kill-user-processes option to "configure".
1210 * systemd-logind gained new configuration settings SessionsMax= and
1211 InhibitorsMax=, both with a default of 8192. It will not register new
1212 user sessions or inhibitors above this limit.
1214 * systemd-logind will now reload configuration on SIGHUP.
1216 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
1217 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
1218 enable. Also, support for the "io" cgroup controller in the unified
1219 hierarchy has been added, so that the "memory", "pids" and "io" are
1220 now the controllers that are supported on the unified hierarchy.
1222 WARNING: it is not possible to use previous systemd versions with
1223 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
1224 is necessary to also update systemd in the initramfs if using the
1225 unified hierarchy. An updated SELinux policy is also required.
1227 * LLDP support has been extended, and both passive (receive-only) and
1228 active (sender) modes are supported. Passive mode ("routers-only") is
1229 enabled by default in systemd-networkd. Active LLDP mode is enabled
1230 by default for containers on the internal network. The "networkctl
1231 lldp" command may be used to list information gathered. "networkctl
1232 status" will also show basic LLDP information on connected peers now.
1234 * The IAID and DUID unique identifier sent in DHCP requests may now be
1235 configured for the system and each .network file managed by
1236 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
1238 * systemd-networkd gained support for configuring proxy ARP support for
1239 each interface, via the ProxyArp= setting in .network files. It also
1240 gained support for configuring the multicast querier feature of
1241 bridge devices, via the new MulticastQuerier= setting in .netdev
1242 files. Similarly, snooping on the IGMP traffic can be controlled
1243 via the new setting MulticastSnooping=.
1245 A new setting PreferredLifetime= has been added for addresses
1246 configured in .network file to configure the lifetime intended for an
1249 The systemd-networkd DHCP server gained the option EmitRouter=, which
1250 defaults to yes, to configure whether the DHCP Option 3 (Router)
1253 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
1254 systemd-socket-activate and installed into /usr/bin. It is now fully
1257 * systemd-journald now uses separate threads to flush changes to disk
1258 when closing journal files, thus reducing impact of slow disk I/O on
1259 logging performance.
1261 * The sd-journal API gained two new calls
1262 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
1263 can be used to open journal files using file descriptors instead of
1264 file or directory paths. sd_journal_open_container() has been
1265 deprecated, sd_journal_open_directory_fd() should be used instead
1266 with the flag SD_JOURNAL_OS_ROOT.
1268 * journalctl learned a new output mode "-o short-unix" that outputs log
1269 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
1270 UTC). It also gained support for a new --no-hostname setting to
1271 suppress the hostname column in the family of "short" output modes.
1273 * systemd-ask-password now optionally skips printing of the password to
1274 stdout with --no-output which can be useful in scripts.
1276 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
1277 (devices tagged with ID_MAKER_TOOL) are now tagged with
1278 "uaccess" and are available to logged in users.
1280 * The DeviceAllow= unit setting now supports specifiers (with "%").
1282 * "systemctl show" gained a new --value switch, which allows print a
1283 only the contents of a specific unit property, without also printing
1284 the property's name. Similar support was added to "show*" verbs
1285 of loginctl and machinectl that output "key=value" lists.
1287 * A new unit type "generated" was added for files dynamically generated
1288 by generator tools. Similarly, a new unit type "transient" is used
1289 for unit files created using the runtime API. "systemctl enable" will
1290 refuse to operate on such files.
1292 * A new command "systemctl revert" has been added that may be used to
1293 revert to the vendor version of a unit file, in case local changes
1294 have been made by adding drop-ins or overriding the unit file.
1296 * "machinectl clean" gained a new verb to automatically remove all or
1297 just hidden container images.
1299 * systemd-tmpfiles gained support for a new line type "e" for emptying
1300 directories, if they exist, without creating them if they don't.
1302 * systemd-nspawn gained support for automatically patching the UID/GIDs
1303 of the owners and the ACLs of all files and directories in a
1304 container tree to match the UID/GID user namespacing range selected
1305 for the container invocation. This mode is enabled via the new
1306 --private-users-chown switch. It also gained support for
1307 automatically choosing a free, previously unused UID/GID range when
1308 starting a container, via the new --private-users=pick setting (which
1309 implies --private-users-chown). Together, these options for the first
1310 time make user namespacing for nspawn containers fully automatic and
1311 thus deployable. The systemd-nspawn@.service template unit file has
1312 been changed to use this functionality by default.
1314 * systemd-nspawn gained a new --network-zone= switch, that allows
1315 creating ad-hoc virtual Ethernet links between multiple containers,
1316 that only exist as long as at least one container referencing them is
1317 running. This allows easy connecting of multiple containers with a
1318 common link that implements an Ethernet broadcast domain. Each of
1319 these network "zones" may be named relatively freely by the user, and
1320 may be referenced by any number of containers, but each container may
1321 only reference one of these "zones". On the lower level, this is
1322 implemented by an automatically managed bridge network interface for
1323 each zone, that is created when the first container referencing its
1324 zone is created and removed when the last one referencing its zone
1327 * The default start timeout may now be configured on the kernel command
1328 line via systemd.default_timeout_start_sec=. It was already
1329 configurable via the DefaultTimeoutStartSec= option in
1330 /etc/systemd/system.conf.
1332 * Socket units gained a new TriggerLimitIntervalSec= and
1333 TriggerLimitBurst= setting to configure a limit on the activation
1334 rate of the socket unit.
1336 * The LimitNICE= setting now optionally takes normal UNIX nice values
1337 in addition to the raw integer limit value. If the specified
1338 parameter is prefixed with "+" or "-" and is in the range -20..19 the
1339 value is understood as UNIX nice value. If not prefixed like this it
1340 is understood as raw RLIMIT_NICE limit.
1342 * Note that the effect of the PrivateDevices= unit file setting changed
1343 slightly with this release: the per-device /dev file system will be
1344 mounted read-only from this version on, and will have "noexec"
1345 set. This (minor) change of behavior might cause some (exceptional)
1346 legacy software to break, when PrivateDevices=yes is set for its
1347 service. Please leave PrivateDevices= off if you run into problems
1350 * systemd-bootchart has been split out to a separate repository:
1351 https://github.com/systemd/systemd-bootchart
1353 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
1354 merged into the kernel in its current form.
1356 * The compatibility libraries libsystemd-daemon.so,
1357 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
1358 which have been deprecated since systemd-209 have been removed along
1359 with the corresponding pkg-config files. All symbols provided by
1360 those libraries are provided by libsystemd.so.
1362 * The Capabilities= unit file setting has been removed (it is ignored
1363 for backwards compatibility). AmbientCapabilities= and
1364 CapabilityBoundingSet= should be used instead.
1366 * A new special target has been added, initrd-root-device.target,
1367 which creates a synchronization point for dependencies of the root
1368 device in early userspace. Initramfs builders must ensure that this
1369 target is now included in early userspace.
1371 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
1372 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
1373 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
1374 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
1375 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
1376 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
1377 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
1378 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
1379 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
1380 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
1381 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
1382 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
1383 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
1384 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
1385 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
1386 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
1387 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
1388 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
1389 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
1390 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
1391 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
1392 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
1393 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
1394 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
1395 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
1398 — Fairfax, 2016-05-21
1402 * The systemd-resolved DNS resolver service has gained a substantial
1403 set of new features, most prominently it may now act as a DNSSEC
1404 validating stub resolver. DNSSEC mode is currently turned off by
1405 default, but is expected to be turned on by default in one of the
1406 next releases. For now, we invite everybody to test the DNSSEC logic
1407 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
1408 service also gained a full set of D-Bus interfaces, including calls
1409 to configure DNS and DNSSEC settings per link (for use by external
1410 network management software). systemd-resolved and systemd-networkd
1411 now distinguish between "search" and "routing" domains. The former
1412 are used to qualify single-label names, the latter are used purely
1413 for routing lookups within certain domains to specific links.
1414 resolved now also synthesizes RRs for all entries from /etc/hosts.
1416 * The systemd-resolve tool (which is a client utility for
1417 systemd-resolved) has been improved considerably and is now fully
1418 supported and documented. Hence it has moved from /usr/lib/systemd to
1421 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
1424 * The coredump collection logic has been reworked: when a coredump is
1425 collected it is now written to disk, compressed and processed
1426 (including stacktrace extraction) from a new instantiated service
1427 systemd-coredump@.service, instead of directly from the
1428 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
1429 processing large coredumps can take up a substantial amount of
1430 resources and time, and this previously happened entirely outside of
1431 systemd's service supervision. With the new logic the core_pattern
1432 hook only does minimal metadata collection before passing off control
1433 to the new instantiated service, which is configured with a time
1434 limit, a nice level and other settings to minimize negative impact on
1435 the rest of the system. Also note that the new logic will honour the
1436 RLIMIT_CORE setting of the crashed process, which now allows users
1437 and processes to turn off coredumping for their processes by setting
1440 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
1441 and all forked processes by default. Previously, PID 1 would leave
1442 the setting at "0" for all processes, as set by the kernel. Note that
1443 the resource limit traditionally has no effect on the generated
1444 coredumps on the system if the /proc/sys/kernel/core_pattern hook
1445 logic is used. Since the limit is now honoured (see above) its
1446 default has been changed so that the coredumping logic is enabled by
1447 default for all processes, while allowing specific opt-out.
1449 * When the stacktrace is extracted from processes of system users, this
1450 is now done as "systemd-coredump" user, in order to sandbox this
1451 potentially security sensitive parsing operation. (Note that when
1452 processing coredumps of normal users this is done under the user ID
1453 of process that crashed, as before.) Packagers should take notice
1454 that it is now necessary to create the "systemd-coredump" system user
1455 and group at package installation time.
1457 * The systemd-activate socket activation testing tool gained support
1458 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
1459 and --seqpacket switches. It also has been extended to support both
1460 new-style and inetd-style file descriptor passing. Use the new
1461 --inetd switch to request inetd-style file descriptor passing.
1463 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
1464 variable, which takes a boolean value. If set to false, ANSI color
1465 output is disabled in the tools even when run on a terminal that
1468 * The VXLAN support in networkd now supports two new settings
1469 DestinationPort= and PortRange=.
1471 * A new systemd.machine_id= kernel command line switch has been added,
1472 that may be used to set the machine ID in /etc/machine-id if it is
1473 not initialized yet. This command line option has no effect if the
1474 file is already initialized.
1476 * systemd-nspawn gained a new --as-pid2 switch that invokes any
1477 specified command line as PID 2 rather than PID 1 in the
1478 container. In this mode PID 1 is a minimal stub init process that
1479 implements the special POSIX and Linux semantics of PID 1 regarding
1480 signal and child process management. Note that this stub init process
1481 is implemented in nspawn itself and requires no support from the
1482 container image. This new logic is useful to support running
1483 arbitrary commands in the container, as normal processes are
1484 generally not prepared to run as PID 1.
1486 * systemd-nspawn gained a new --chdir= switch for setting the current
1487 working directory for the process started in the container.
1489 * "journalctl /dev/sda" will now output all kernel log messages for
1490 specified device from the current boot, in addition to all devices
1491 that are parents of it. This should make log output about devices
1492 pretty useful, as long as kernel drivers attach enough metadata to
1493 the log messages. (The usual SATA drivers do.)
1495 * The sd-journal API gained two new calls
1496 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
1497 that report whether log data from /run or /var has been found.
1499 * journalctl gained a new switch "--fields" that prints all journal
1500 record field names currently in use in the journal. This is backed
1501 by two new sd-journal API calls sd_journal_enumerate_fields() and
1502 sd_journal_restart_fields().
1504 * Most configurable timeouts in systemd now expect an argument of
1505 "infinity" to turn them off, instead of "0" as before. The semantics
1506 from now on is that a timeout of "0" means "now", and "infinity"
1507 means "never". To maintain backwards compatibility, "0" continues to
1508 turn off previously existing timeout settings.
1510 * "systemctl reload-or-try-restart" has been renamed to "systemctl
1511 try-reload-or-restart" to clarify what it actually does: the "try"
1512 logic applies to both reloading and restarting, not just restarting.
1513 The old name continues to be accepted for compatibility.
1515 * On boot-up, when PID 1 detects that the system clock is behind the
1516 release date of the systemd version in use, the clock is now set
1517 to the latter. Previously, this was already done in timesyncd, in order
1518 to avoid running with clocks set to the various clock epochs such as
1519 1902, 1938 or 1970. With this change the logic is now done in PID 1
1520 in addition to timesyncd during early boot-up, so that it is enforced
1521 before the first process is spawned by systemd. Note that the logic
1522 in timesyncd remains, as it is more comprehensive and ensures
1523 clock monotonicity by maintaining a persistent timestamp file in
1524 /var. Since /var is generally not available in earliest boot or the
1525 initrd, this part of the logic remains in timesyncd, and is not done
1528 * Support for tweaking details in net_cls.class_id through the
1529 NetClass= configuration directive has been removed, as the kernel
1530 people have decided to deprecate that controller in cgroup v2.
1531 Userspace tools such as nftables are moving over to setting rules
1532 that are specific to the full cgroup path of a task, which obsoletes
1533 these controllers anyway. The NetClass= directive is kept around for
1534 legacy compatibility reasons. For a more in-depth description of the
1535 kernel change, please refer to the respective upstream commit:
1537 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
1539 * A new service setting RuntimeMaxSec= has been added that may be used
1540 to specify a maximum runtime for a service. If the timeout is hit, the
1541 service is terminated and put into a failure state.
1543 * A new service setting AmbientCapabilities= has been added. It allows
1544 configuration of additional Linux process capabilities that are
1545 passed to the activated processes. This is only available on very
1548 * The process resource limit settings in service units may now be used
1549 to configure hard and soft limits individually.
1551 * The various libsystemd APIs such as sd-bus or sd-event now publicly
1552 expose support for gcc's __attribute__((cleanup())) C extension.
1553 Specifically, for many object destructor functions alternative
1554 versions have been added that have names suffixed with "p" and take a
1555 pointer to a pointer to the object to destroy, instead of just a
1556 pointer to the object itself. This is useful because these destructor
1557 functions may be used directly as parameters to the cleanup
1558 construct. Internally, systemd has been a heavy user of this GCC
1559 extension for a long time, and with this change similar support is
1560 now available to consumers of the library outside of systemd. Note
1561 that by using this extension in your sources compatibility with old
1562 and strictly ANSI compatible C compilers is lost. However, all gcc or
1563 LLVM versions of recent years support this extension.
1565 * Timer units gained support for a new setting RandomizedDelaySec= that
1566 allows configuring some additional randomized delay to the configured
1567 time. This is useful to spread out timer events to avoid load peaks in
1568 clusters or larger setups.
1570 * Calendar time specifications now support sub-second accuracy.
1572 * Socket units now support listening on SCTP and UDP-lite protocol
1575 * The sd-event API now comes with a full set of man pages.
1577 * Older versions of systemd contained experimental support for
1578 compressing journal files and coredumps with the LZ4 compressor that
1579 was not compatible with the lz4 binary (due to API limitations of the
1580 lz4 library). This support has been removed; only support for files
1581 compatible with the lz4 binary remains. This LZ4 logic is now
1582 officially supported and no longer considered experimental.
1584 * The dkr image import logic has been removed again from importd. dkr's
1585 micro-services focus doesn't fit into the machine image focus of
1586 importd, and quickly got out of date with the upstream dkr API.
1588 * Creation of the /run/lock/lockdev/ directory was dropped from
1589 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
1590 been available for many years. If you still need this, you need to
1591 create your own tmpfiles.d config file with:
1593 d /run/lock/lockdev 0775 root lock -
1595 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
1596 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
1597 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
1598 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
1599 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
1600 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
1601 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
1602 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
1603 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
1604 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
1605 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
1606 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
1607 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
1608 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
1609 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
1610 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
1611 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
1612 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
1613 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1615 — Berlin, 2016-02-11
1619 * A number of properties previously only settable in unit
1620 files are now also available as properties to set when
1621 creating transient units programmatically via the bus, as it
1622 is exposed with systemd-run's --property=
1623 setting. Specifically, these are: SyslogIdentifier=,
1624 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
1625 EnvironmentFile=, ReadWriteDirectories=,
1626 ReadOnlyDirectories=, InaccessibleDirectories=,
1627 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
1629 * When creating transient services via the bus API it is now
1630 possible to pass in a set of file descriptors to use as
1631 STDIN/STDOUT/STDERR for the invoked process.
1633 * Slice units may now be created transiently via the bus APIs,
1634 similar to the way service and scope units may already be
1635 created transiently.
1637 * Wherever systemd expects a calendar timestamp specification
1638 (like in journalctl's --since= and --until= switches) UTC
1639 timestamps are now supported. Timestamps suffixed with "UTC"
1640 are now considered to be in Universal Time Coordinated
1641 instead of the local timezone. Also, timestamps may now
1642 optionally be specified with sub-second accuracy. Both of
1643 these additions also apply to recurring calendar event
1644 specification, such as OnCalendar= in timer units.
1646 * journalctl gained a new "--sync" switch that asks the
1647 journal daemon to write all so far unwritten log messages to
1648 disk and sync the files, before returning.
1650 * systemd-tmpfiles learned two new line types "q" and "Q" that
1651 operate like "v", but also set up a basic btrfs quota
1652 hierarchy when used on a btrfs file system with quota
1655 * tmpfiles' "v", "q" and "Q" will now create a plain directory
1656 instead of a subvolume (even on a btrfs file system) if the
1657 root directory is a plain directory, and not a
1658 subvolume. This should simplify things with certain chroot()
1659 environments which are not aware of the concept of btrfs
1662 * systemd-detect-virt gained a new --chroot switch to detect
1663 whether execution takes place in a chroot() environment.
1665 * CPUAffinity= now takes CPU index ranges in addition to
1668 * The various memory-related resource limit settings (such as
1669 LimitAS=) now understand the usual K, M, G, ... suffixes to
1670 the base of 1024 (IEC). Similar, the time-related resource
1671 limit settings understand the usual min, h, day, ...
1674 * There's a new system.conf setting DefaultTasksMax= to
1675 control the default TasksMax= setting for services and
1676 scopes running on the system. (TasksMax= is the primary
1677 setting that exposes the "pids" cgroup controller on systemd
1678 and was introduced in the previous systemd release.) The
1679 setting now defaults to 512, which means services that are
1680 not explicitly configured otherwise will only be able to
1681 create 512 processes or threads at maximum, from this
1682 version on. Note that this means that thread- or
1683 process-heavy services might need to be reconfigured to set
1684 TasksMax= to a higher value. It is sufficient to set
1685 TasksMax= in these specific unit files to a higher value, or
1686 even "infinity". Similar, there's now a logind.conf setting
1687 UserTasksMax= that defaults to 4096 and limits the total
1688 number of processes or tasks each user may own
1689 concurrently. nspawn containers also have the TasksMax=
1690 value set by default now, to 8192. Note that all of this
1691 only has an effect if the "pids" cgroup controller is
1692 enabled in the kernel. The general benefit of these changes
1693 should be a more robust and safer system, that provides a
1694 certain amount of per-service fork() bomb protection.
1696 * systemd-nspawn gained the new --network-veth-extra= switch
1697 to define additional and arbitrarily-named virtual Ethernet
1698 links between the host and the container.
1700 * A new service execution setting PassEnvironment= has been
1701 added that allows importing select environment variables
1702 from PID1's environment block into the environment block of
1705 * Timer units gained support for a new RemainAfterElapse=
1706 setting which takes a boolean argument. It defaults to on,
1707 exposing behaviour unchanged to previous releases. If set to
1708 off, timer units are unloaded after they elapsed if they
1709 cannot elapse again. This is particularly useful for
1710 transient timer units, which shall not stay around longer
1711 than until they first elapse.
1713 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
1714 default now (the kernel default is 16). This is beneficial
1715 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
1716 allows substantially larger numbers of queued
1717 datagrams. This should increase the capability of systemd to
1718 parallelize boot-up, as logging and sd_notify() are unlikely
1719 to stall execution anymore. If you need to change the value
1720 from the new defaults, use the usual sysctl.d/ snippets.
1722 * The compression framing format used by the journal or
1723 coredump processing has changed to be in line with what the
1724 official LZ4 tools generate. LZ4 compression support in
1725 systemd was considered unsupported previously, as the format
1726 was not compatible with the normal tools. With this release
1727 this has changed now, and it is hence safe for downstream
1728 distributions to turn it on. While not compressing as well
1729 as the XZ, LZ4 is substantially faster, which makes
1730 it a good default choice for the compression logic in the
1731 journal and in coredump handling.
1733 * Any reference to /etc/mtab has been dropped from
1734 systemd. The file has been obsolete since a while, but
1735 systemd refused to work on systems where it was incorrectly
1736 set up (it should be a symlink or non-existent). Please make
1737 sure to update to util-linux 2.27.1 or newer in conjunction
1738 with this systemd release, which also drops any reference to
1739 /etc/mtab. If you maintain a distribution make sure that no
1740 software you package still references it, as this is a
1741 likely source of bugs. There's also a glibc bug pending,
1742 asking for removal of any reference to this obsolete file:
1744 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
1746 Note that only util-linux versions built with
1747 --enable-libmount-force-mountinfo are supported.
1749 * Support for the ".snapshot" unit type has been removed. This
1750 feature turned out to be little useful and little used, and
1751 has now been removed from the core and from systemctl.
1753 * The dependency types RequiresOverridable= and
1754 RequisiteOverridable= have been removed from systemd. They
1755 have been used only very sparingly to our knowledge and
1756 other options that provide a similar effect (such as
1757 systemctl --mode=ignore-dependencies) are much more useful
1758 and commonly used. Moreover, they were only half-way
1759 implemented as the option to control behaviour regarding
1760 these dependencies was never added to systemctl. By removing
1761 these dependency types the execution engine becomes a bit
1762 simpler. Unit files that use these dependencies should be
1763 changed to use the non-Overridable dependency types
1764 instead. In fact, when parsing unit files with these
1765 options, that's what systemd will automatically convert them
1766 too, but it will also warn, asking users to fix the unit
1767 files accordingly. Removal of these dependency types should
1768 only affect a negligible number of unit files in the wild.
1770 * Behaviour of networkd's IPForward= option changed
1771 (again). It will no longer maintain a per-interface setting,
1772 but propagate one way from interfaces where this is enabled
1773 to the global kernel setting. The global setting will be
1774 enabled when requested by a network that is set up, but
1775 never be disabled again. This change was made to make sure
1776 IPv4 and IPv6 behaviour regarding packet forwarding is
1777 similar (as the Linux IPv6 stack does not support
1778 per-interface control of this setting) and to minimize
1781 * In unit files the behaviour of %u, %U, %h, %s has
1782 changed. These specifiers will now unconditionally resolve
1783 to the various user database fields of the user that the
1784 systemd instance is running as, instead of the user
1785 configured in the specific unit via User=. Note that this
1786 effectively doesn't change much, as resolving of these
1787 specifiers was already turned off in the --system instance
1788 of systemd, as we cannot do NSS lookups from PID 1. In the
1789 --user instance of systemd these specifiers where correctly
1790 resolved, but hardly made any sense, since the user instance
1791 lacks privileges to do user switches anyway, and User= is
1792 hence useless. Moreover, even in the --user instance of
1793 systemd behaviour was awkward as it would only take settings
1794 from User= assignment placed before the specifier into
1795 account. In order to unify and simplify the logic around
1796 this the specifiers will now always resolve to the
1797 credentials of the user invoking the manager (which in case
1798 of PID 1 is the root user).
1800 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
1801 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
1802 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
1803 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
1804 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
1805 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
1806 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
1807 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
1808 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
1809 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
1810 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
1811 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
1812 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
1813 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
1816 — Berlin, 2015-11-18
1820 * systemd now depends on util-linux v2.27. More specifically,
1821 the newly added mount monitor feature in libmount now
1822 replaces systemd's former own implementation.
1824 * libmount mandates /etc/mtab not to be regular file, and
1825 systemd now enforces this condition at early boot.
1826 /etc/mtab has been deprecated and warned about for a very
1827 long time, so systems running systemd should already have
1828 stopped having this file around as anything else than a
1829 symlink to /proc/self/mounts.
1831 * Support for the "pids" cgroup controller has been added. It
1832 allows accounting the number of tasks in a cgroup and
1833 enforcing limits on it. This adds two new setting
1834 TasksAccounting= and TasksMax= to each unit, as well as a
1835 global option DefaultTasksAccounting=.
1837 * Support for the "net_cls" cgroup controller has been added.
1838 It allows assigning a net class ID to each task in the
1839 cgroup, which can then be used in firewall rules and traffic
1840 shaping configurations. Note that the kernel netfilter net
1841 class code does not currently work reliably for ingress
1842 packets on unestablished sockets.
1844 This adds a new config directive called NetClass= to CGroup
1845 enabled units. Allowed values are positive numbers for fixed
1846 assignments and "auto" for picking a free value
1849 * 'systemctl is-system-running' now returns 'offline' if the
1850 system is not booted with systemd. This command can now be
1851 used as a substitute for 'systemd-notify --booted'.
1853 * Watchdog timeouts have been increased to 3 minutes for all
1854 in-tree service files. Apparently, disk IO issues are more
1855 frequent than we hoped, and user reported >1 minute waiting
1858 * 'machine-id-commit' functionality has been merged into
1859 'machine-id-setup --commit'. The separate binary has been
1862 * The WorkingDirectory= directive in unit files may now be set
1863 to the special value '~'. In this case, the working
1864 directory is set to the home directory of the user
1865 configured in User=.
1867 * "machinectl shell" will now open the shell in the home
1868 directory of the selected user by default.
1870 * The CrashChVT= configuration file setting is renamed to
1871 CrashChangeVT=, following our usual logic of not
1872 abbreviating unnecessarily. The old directive is still
1873 supported for compat reasons. Also, this directive now takes
1874 an integer value between 1 and 63, or a boolean value. The
1875 formerly supported '-1' value for disabling stays around for
1878 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
1879 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
1880 RootDirectory= properties can now be set for transient
1883 * The systemd-analyze tool gained a new "set-log-target" verb
1884 to change the logging target the system manager logs to
1885 dynamically during runtime. This is similar to how
1886 "systemd-analyze set-log-level" already changes the log
1889 * In nspawn /sys is now mounted as tmpfs, with only a selected
1890 set of subdirectories mounted in from the real sysfs. This
1891 enhances security slightly, and is useful for ensuring user
1892 namespaces work correctly.
1894 * Support for USB FunctionFS activation has been added. This
1895 allows implementation of USB gadget services that are
1896 activated as soon as they are requested, so that they don't
1897 have to run continuously, similar to classic socket
1900 * The "systemctl exit" command now optionally takes an
1901 additional parameter that sets the exit code to return from
1902 the systemd manager when exiting. This is only relevant when
1903 running the systemd user instance, or when running the
1904 system instance in a container.
1906 * sd-bus gained the new API calls sd_bus_path_encode_many()
1907 and sd_bus_path_decode_many() that allow easy encoding and
1908 decoding of multiple identifier strings inside a D-Bus
1909 object path. Another new call sd_bus_default_flush_close()
1910 has been added to flush and close per-thread default
1913 * systemd-cgtop gained support for a -M/--machine= switch to
1914 show the control groups within a certain container only.
1916 * "systemctl kill" gained support for an optional --fail
1917 switch. If specified the requested operation will fail of no
1918 processes have been killed, because the unit had no
1919 processes attached, or similar.
1921 * A new systemd.crash_reboot=1 kernel command line option has
1922 been added that triggers a reboot after crashing. This can
1923 also be set through CrashReboot= in systemd.conf.
1925 * The RuntimeDirectory= setting now understands unit
1926 specifiers like %i or %f.
1928 * A new (still internal) library API sd-ipv4acd has been added,
1929 that implements address conflict detection for IPv4. It's
1930 based on code from sd-ipv4ll, and will be useful for
1931 detecting DHCP address conflicts.
1933 * File descriptors passed during socket activation may now be
1934 named. A new API sd_listen_fds_with_names() is added to
1935 access the names. The default names may be overridden,
1936 either in the .socket file using the FileDescriptorName=
1937 parameter, or by passing FDNAME= when storing the file
1938 descriptors using sd_notify().
1940 * systemd-networkd gained support for:
1942 - Setting the IPv6 Router Advertisement settings via
1943 IPv6AcceptRouterAdvertisements= in .network files.
1945 - Configuring the HelloTimeSec=, MaxAgeSec= and
1946 ForwardDelaySec= bridge parameters in .netdev files.
1948 - Configuring PreferredSource= for static routes in
1951 * The "ask-password" framework used to query for LUKS harddisk
1952 passwords or SSL passwords during boot gained support for
1953 caching passwords in the kernel keyring, if it is
1954 available. This makes sure that the user only has to type in
1955 a passphrase once if there are multiple objects to unlock
1956 with the same one. Previously, such password caching was
1957 available only when Plymouth was used; this moves the
1958 caching logic into the systemd codebase itself. The
1959 "systemd-ask-password" utility gained a new --keyname=
1960 switch to control which kernel keyring key to use for
1961 caching a password in. This functionality is also useful for
1962 enabling display managers such as gdm to automatically
1963 unlock the user's GNOME keyring if its passphrase, the
1964 user's password and the harddisk password are the same, if
1965 gdm-autologin is used.
1967 * When downloading tar or raw images using "machinectl
1968 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
1969 file is now also downloaded, if it is available and stored
1970 next to the image file.
1972 * Units of type ".socket" gained a new boolean setting
1973 Writable= which is only useful in conjunction with
1974 ListenSpecial=. If true, enables opening the specified
1975 special file in O_RDWR mode rather than O_RDONLY mode.
1977 * systemd-rfkill has been reworked to become a singleton
1978 service that is activated through /dev/rfkill on each rfkill
1979 state change and saves the settings to disk. This way,
1980 systemd-rfkill is now compatible with devices that exist
1981 only intermittendly, and even restores state if the previous
1982 system shutdown was abrupt rather than clean.
1984 * The journal daemon gained support for vacuuming old journal
1985 files controlled by the number of files that shall remain,
1986 in addition to the already existing control by size and by
1987 date. This is useful as journal interleaving performance
1988 degrades with too many separate journal files, and allows
1989 putting an effective limit on them. The new setting defaults
1990 to 100, but this may be changed by setting SystemMaxFiles=
1991 and RuntimeMaxFiles= in journald.conf. Also, the
1992 "journalctl" tool gained the new --vacuum-files= switch to
1993 manually vacuum journal files to leave only the specified
1994 number of files in place.
1996 * udev will now create /dev/disk/by-path links for ATA devices
1997 on kernels where that is supported.
1999 * Galician, Serbian, Turkish and Korean translations were added.
2001 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
2002 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
2003 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
2004 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
2005 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
2006 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
2007 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
2008 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
2009 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
2010 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
2011 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
2012 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
2013 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
2014 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
2015 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
2016 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
2017 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
2018 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
2020 — Berlin, 2015-10-07
2024 * The DHCP implementation of systemd-networkd gained a set of
2027 - The DHCP server now supports emitting DNS and NTP
2028 information. It may be enabled and configured via
2029 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
2030 and NTP information is enabled, but no servers are
2031 configured, the corresponding uplink information (if there
2032 is any) is propagated.
2034 - Server and client now support transmission and reception
2035 of timezone information. It can be configured via the
2036 newly introduced network options UseTimezone=,
2037 EmitTimezone=, and Timezone=. Transmission of timezone
2038 information is enabled between host and containers by
2039 default now: the container will change its local timezone
2040 to what the host has set.
2042 - Lease timeouts can now be configured via
2043 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
2045 - The DHCP server improved on the stability of
2046 leases. Clients are more likely to get the same lease
2047 information back, even if the server loses state.
2049 - The DHCP server supports two new configuration options to
2050 control the lease address pool metrics, PoolOffset= and
2053 * The encapsulation limit of tunnels in systemd-networkd may
2054 now be configured via 'EncapsulationLimit='. It allows
2055 modifying the maximum additional levels of encapsulation
2056 that are permitted to be prepended to a packet.
2058 * systemd now supports the concept of user buses replacing
2059 session buses, if used with dbus-1.10 (and enabled via dbus
2060 --enable-user-session). It previously only supported this on
2061 kdbus-enabled systems, and this release expands this to
2062 'dbus-daemon' systems.
2064 * systemd-networkd now supports predictable interface names
2067 * systemd now optionally supports the new Linux kernel
2068 "unified" control group hierarchy. If enabled via the kernel
2069 command-line option 'systemd.unified_cgroup_hierarchy=1',
2070 systemd will try to mount the unified cgroup hierarchy
2071 directly on /sys/fs/cgroup. If not enabled, or not
2072 available, systemd will fall back to the legacy cgroup
2073 hierarchy setup, as before. Host system and containers can
2074 mix and match legacy and unified hierarchies as they
2075 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
2076 environment variable to individually select the hierarchy to
2077 use for executed containers. By default, nspawn will use the
2078 unified hierarchy for the containers if the host uses the
2079 unified hierarchy, and the legacy hierarchy otherwise.
2080 Please note that at this point the unified hierarchy is an
2081 experimental kernel feature and is likely to change in one
2082 of the next kernel releases. Therefore, it should not be
2083 enabled by default in downstream distributions yet. The
2084 minimum required kernel version for the unified hierarchy to
2085 work is 4.2. Note that when the unified hierarchy is used
2086 for the first time delegated access to controllers is
2087 safe. Because of this systemd-nspawn containers will get
2088 access to controllers now, as will systemd user
2089 sessions. This means containers and user sessions may now
2090 manage their own resources, partitioning up what the system
2093 * A new special scope unit "init.scope" has been introduced
2094 that encapsulates PID 1 of the system. It may be used to
2095 determine resource usage and enforce resource limits on PID
2096 1 itself. PID 1 hence moved out of the root of the control
2099 * The cgtop tool gained support for filtering out kernel
2100 threads when counting tasks in a control group. Also, the
2101 count of processes is now recursively summed up by
2102 default. Two options -k and --recursive= have been added to
2103 revert to old behaviour. The tool has also been updated to
2104 work correctly in containers now.
2106 * systemd-nspawn's --bind= and --bind-ro= options have been
2107 extended to allow creation of non-recursive bind mounts.
2109 * libsystemd gained two new calls sd_pid_get_cgroup() and
2110 sd_peer_get_cgroup() which return the control group path of
2111 a process or peer of a connected AF_UNIX socket. This
2112 function call is particularly useful when implementing
2113 delegated subtrees support in the control group hierarchy.
2115 * The "sd-event" event loop API of libsystemd now supports
2116 correct dequeuing of real-time signals, without losing
2119 * When systemd requests a PolicyKit decision when managing
2120 units it will now add additional fields to the request,
2121 including unit name and desired operation. This enables more
2122 powerful PolicyKit policies, that make decisions depending
2123 on these parameters.
2125 * nspawn learnt support for .nspawn settings files, that may
2126 accompany the image files or directories of containers, and
2127 may contain additional settings for the container. This is
2128 an alternative to configuring container parameters via the
2129 nspawn command line.
2131 Contributions from: Cristian Rodríguez, Daniel Mack, David
2132 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
2133 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
2134 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
2135 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
2136 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
2137 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
2138 Andersen, Tom Gundersen, Torstein Husebø
2140 — Berlin, 2015-09-08
2144 * machinectl gained a new verb 'shell' which opens a fresh
2145 shell on the target container or the host. It is similar to
2146 the existing 'login' command of machinectl, but spawns the
2147 shell directly without prompting for username or
2148 password. The pseudo machine '.host' now refers to the local
2149 host and is used by default. Hence, 'machinectl shell' can
2150 be used as replacement for 'su -' which spawns a session as
2151 a fresh systemd unit in a way that is fully isolated from
2152 the originating session.
2154 * systemd-networkd learned to cope with private-zone DHCP
2155 options and allows other programs to query the values.
2157 * SELinux access control when enabling/disabling units is no
2158 longer enforced with this release. The previous
2159 implementation was incorrect, and a new corrected
2160 implementation is not yet available. As unit file operations
2161 are still protected via PolicyKit and D-Bus policy this is
2162 not a security problem. Yet, distributions which care about
2163 optimal SELinux support should probably not stabilize on
2166 * sd-bus gained support for matches of type "arg0has=", that
2167 test for membership of strings in string arrays sent in bus
2170 * systemd-resolved now dumps the contents of its DNS and LLMNR
2171 caches to the logs on reception of the SIGUSR1 signal. This
2172 is useful to debug DNS behaviour.
2174 * The coredumpctl tool gained a new --directory= option to
2175 operate on journal files in a specific directory.
2177 * "systemctl reboot" and related commands gained a new
2178 "--message=" option which may be used to set a free-text
2179 wall message when shutting down or rebooting the
2180 system. This message is also logged, which is useful for
2181 figuring out the reason for a reboot or shutdown a
2184 * The "systemd-resolve-host" tool's -i switch now takes
2185 network interface numbers as alternative to interface names.
2187 * A new unit file setting for services has been introduced:
2188 UtmpMode= allows configuration of how precisely systemd
2189 handles utmp and wtmp entries for the service if this is
2190 enabled. This allows writing services that appear similar to
2191 user sessions in the output of the "w", "who", "last" and
2194 * systemd-resolved will now locally synthesize DNS resource
2195 records for the "localhost" and "gateway" domains as well as
2196 the local hostname. This should ensure that clients querying
2197 RRs via resolved will get similar results as those going via
2198 NSS, if nss-myhostname is enabled.
2200 Contributions from: Alastair Hughes, Alex Crawford, Daniel
2201 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
2202 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
2203 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
2204 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
2205 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
2206 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
2207 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
2208 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
2209 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
2210 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
2211 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2213 — Berlin, 2015-08-27
2217 * The systemd-efi-boot-generator functionality was merged into
2218 systemd-gpt-auto-generator.
2220 * systemd-networkd now supports Group Policy for vxlan
2221 devices. It can be enabled via the new boolean configuration
2222 option called 'GroupPolicyExtension='.
2224 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
2225 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
2226 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
2228 — Berlin, 2015-07-31
2232 * The python-systemd code has been removed from the systemd repository.
2233 A new repository has been created which accommodates the code from
2234 now on, and we kindly ask distributions to create a separate package
2235 for this: https://github.com/systemd/python-systemd
2237 * The systemd daemon will now reload its main configuration
2238 (/etc/systemd/system.conf) on daemon-reload.
2240 * sd-dhcp now exposes vendor specific extensions via
2241 sd_dhcp_lease_get_vendor_specific().
2243 * systemd-networkd gained a number of new configuration options.
2245 - A new boolean configuration option for TAP devices called
2246 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
2247 device, thus allowing to send and receive GSO packets.
2249 - A new tunnel configuration option called 'CopyDSCP='.
2250 If enabled, the DSCP field of ip6 tunnels is copied into the
2251 decapsulated packet.
2253 - A set of boolean bridge configuration options were added.
2254 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
2255 and 'UnicastFlood=' are now parsed by networkd and applied to the
2256 respective bridge link device via the respective IFLA_BRPORT_*
2259 - A new string configuration option to override the hostname sent
2260 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
2261 is true, networkd will use the configured hostname instead of the
2262 system hostname when sending DHCP requests.
2264 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
2265 networkd will configure the IPv6 flow-label of the tunnel device
2266 according to RFC2460.
2268 - The 'macvtap' virtual network devices are now supported, similar to
2269 the already supported 'macvlan' devices.
2271 * systemd-resolved now implements RFC5452 to improve resilience against
2272 cache poisoning. Additionally, source port randomization is enabled
2273 by default to further protect against DNS spoofing attacks.
2275 * nss-mymachines now supports translating UIDs and GIDs of running
2276 containers with user-namespaces enabled. If a container 'foo'
2277 translates a host uid 'UID' to the container uid 'TUID', then
2278 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
2279 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
2280 mapped as 'vg-foo-TGID'.
2282 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
2283 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
2284 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
2285 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
2286 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
2287 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
2288 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
2289 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
2290 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
2291 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
2293 — Berlin, 2015-07-29
2297 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
2298 There are no known issues with current sysfs, and udev does not need
2299 or should be used to work around such bugs.
2301 * udev does no longer enable USB HID power management. Several reports
2302 indicate, that some devices cannot handle that setting.
2304 * The udev accelerometer helper was removed. The functionality
2305 is now fully included in iio-sensor-proxy. But this means,
2306 older iio-sensor-proxy versions will no longer provide
2307 accelerometer/orientation data with this systemd version.
2308 Please upgrade iio-sensor-proxy to version 1.0.
2310 * networkd gained a new configuration option IPv6PrivacyExtensions=
2311 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
2312 for Stateless Address") on selected networks.
2314 * For the sake of fewer build-time dependencies and less code in the
2315 main repository, the python bindings are about to be removed in the
2316 next release. A new repository has been created which accommodates
2317 the code from now on, and we kindly ask distributions to create a
2318 separate package for this. The removal will take place in v223.
2320 https://github.com/systemd/python-systemd
2322 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
2323 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
2324 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
2325 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
2326 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
2327 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
2328 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
2329 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
2330 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2331 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2333 — Berlin, 2015-07-07
2337 * The sd-bus.h and sd-event.h APIs have now been declared
2338 stable and have been added to the official interface of
2339 libsystemd.so. sd-bus implements an alternative D-Bus client
2340 library, that is relatively easy to use, very efficient and
2341 supports both classic D-Bus as well as kdbus as transport
2342 backend. sd-event is a generic event loop abstraction that
2343 is built around Linux epoll, but adds features such as event
2344 prioritization or efficient timer handling. Both APIs are good
2345 choices for C programs looking for a bus and/or event loop
2346 implementation that is minimal and does not have to be
2347 portable to other kernels.
2349 * kdbus support is no longer compile-time optional. It is now
2350 always built-in. However, it can still be disabled at
2351 runtime using the kdbus=0 kernel command line setting, and
2352 that setting may be changed to default to off, by specifying
2353 --disable-kdbus at build-time. Note though that the kernel
2354 command line setting has no effect if the kdbus.ko kernel
2355 module is not installed, in which case kdbus is (obviously)
2356 also disabled. We encourage all downstream distributions to
2357 begin testing kdbus by adding it to the kernel images in the
2358 development distributions, and leaving kdbus support in
2361 * The minimal required util-linux version has been bumped to
2364 * Support for chkconfig (--enable-chkconfig) was removed in
2365 favor of calling an abstraction tool
2366 /lib/systemd/systemd-sysv-install. This needs to be
2367 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
2368 in README for details.
2370 * If there's a systemd unit and a SysV init script for the
2371 same service name, and the user executes "systemctl enable"
2372 for it (or a related call), then this will now enable both
2373 (or execute the related operation on both), not just the
2376 * The libudev API documentation has been converted from gtkdoc
2379 * gudev has been removed from the systemd tree, it is now an
2382 * The systemd-cgtop tool learnt a new --raw switch to generate
2383 "raw" (machine parsable) output.
2385 * networkd's IPForwarding= .network file setting learnt the
2386 new setting "kernel", which ensures that networkd does not
2387 change the IP forwarding sysctl from the default kernel
2390 * The systemd-logind bus API now exposes a new boolean
2391 property "Docked" that reports whether logind considers the
2392 system "docked", i.e. connected to a docking station or not.
2394 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
2395 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
2396 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
2397 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
2398 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
2399 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
2400 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
2401 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
2402 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
2403 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
2404 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
2405 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
2406 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
2407 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2408 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
2409 Fink, Zbigniew Jędrzejewski-Szmek
2411 — Berlin, 2015-06-19
2415 * The gudev library has been extracted into a separate repository
2416 available at: https://git.gnome.org/browse/libgudev/
2417 It is now managed as part of the Gnome project. Distributions
2418 are recommended to pass --disable-gudev to systemd and use
2419 gudev from the Gnome project instead. gudev is still included
2420 in systemd, for now. It will be removed soon, though. Please
2421 also see the announcement-thread on systemd-devel:
2422 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
2424 * systemd now exposes a CPUUsageNSec= property for each
2425 service unit on the bus, that contains the overall consumed
2426 CPU time of a service (the sum of what each process of the
2427 service consumed). This value is only available if
2428 CPUAccounting= is turned on for a service, and is then shown
2429 in the "systemctl status" output.
2431 * Support for configuring alternative mappings of the old SysV
2432 runlevels to systemd targets has been removed. They are now
2433 hardcoded in a way that runlevels 2, 3, 4 all map to
2434 multi-user.target and 5 to graphical.target (which
2435 previously was already the default behaviour).
2437 * The auto-mounter logic gained support for mount point
2438 expiry, using a new TimeoutIdleSec= setting in .automount
2439 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
2441 * The EFI System Partition (ESP) as mounted to /boot by
2442 systemd-efi-boot-generator will now be unmounted
2443 automatically after 2 minutes of not being used. This should
2444 minimize the risk of ESP corruptions.
2446 * New /etc/fstab options x-systemd.requires= and
2447 x-systemd.requires-mounts-for= are now supported to express
2448 additional dependencies for mounts. This is useful for
2449 journalling file systems that support external journal
2450 devices or overlay file systems that require underlying file
2451 systems to be mounted.
2453 * systemd does not support direct live-upgrades (via systemctl
2454 daemon-reexec) from versions older than v44 anymore. As no
2455 distribution we are aware of shipped such old versions in a
2456 stable release this should not be problematic.
2458 * When systemd forks off a new per-connection service instance
2459 it will now set the $REMOTE_ADDR environment variable to the
2460 remote IP address, and $REMOTE_PORT environment variable to
2461 the remote IP port. This behaviour is similar to the
2462 corresponding environment variables defined by CGI.
2464 * systemd-networkd gained support for uplink failure
2465 detection. The BindCarrier= option allows binding interface
2466 configuration dynamically to the link sense of other
2467 interfaces. This is useful to achieve behaviour like in
2470 * systemd-networkd gained support for configuring the DHCP
2471 client identifier to use when requesting leases.
2473 * systemd-networkd now has a per-network UseNTP= option to
2474 configure whether NTP server information acquired via DHCP
2475 is passed on to services like systemd-timesyncd.
2477 * systemd-networkd gained support for vti6 tunnels.
2479 * Note that systemd-networkd manages the sysctl variable
2480 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
2481 it is configured for since v219. The variable controls IP
2482 forwarding, and is a per-interface alternative to the global
2483 /proc/sys/net/ipv[46]/ip_forward. This setting is
2484 configurable in the IPForward= option, which defaults to
2485 "no". This means if networkd is used for an interface it is
2486 no longer sufficient to set the global sysctl option to turn
2487 on IP forwarding! Instead, the .network file option
2488 IPForward= needs to be turned on! Note that the
2489 implementation of this behaviour was broken in v219 and has
2492 * Many bonding and vxlan options are now configurable in
2495 * systemd-nspawn gained a new --property= setting to set unit
2496 properties for the container scope. This is useful for
2497 setting resource parameters (e.g. "CPUShares=500") on
2498 containers started from the command line.
2500 * systemd-nspawn gained a new --private-users= switch to make
2501 use of user namespacing available on recent Linux kernels.
2503 * systemd-nspawn may now be called as part of a shell pipeline
2504 in which case the pipes used for stdin and stdout are passed
2505 directly to the process invoked in the container, without
2506 indirection via a pseudo tty.
2508 * systemd-nspawn gained a new switch to control the UNIX
2509 signal to use when killing the init process of the container
2512 * systemd-nspawn gained a new --overlay= switch for mounting
2513 overlay file systems into the container using the new kernel
2516 * When a container image is imported via systemd-importd and
2517 the host file system is not btrfs, a loopback block device
2518 file is created in /var/lib/machines.raw with a btrfs file
2519 system inside. It is then mounted to /var/lib/machines to
2520 enable btrfs features for container management. The loopback
2521 file and btrfs file system is grown as needed when container
2522 images are imported via systemd-importd.
2524 * systemd-machined/systemd-importd gained support for btrfs
2525 quota, to enforce container disk space limits on disk. This
2526 is exposed in "machinectl set-limit".
2528 * systemd-importd now can import containers from local .tar,
2529 .raw and .qcow2 images, and export them to .tar and .raw. It
2530 can also import dkr v2 images now from the network (on top
2533 * systemd-importd gained support for verifying downloaded
2534 images with gpg2 (previously only gpg1 was supported).
2536 * systemd-machined, systemd-logind, systemd: most bus calls
2537 are now accessible to unprivileged processes via
2538 PolicyKit. Also, systemd-logind will now allow users to kill
2539 their own sessions without further privileges or
2542 * systemd-shutdownd has been removed. This service was
2543 previously responsible for implementing scheduled shutdowns
2544 as exposed in /usr/bin/shutdown's time parameter. This
2545 functionality has now been moved into systemd-logind and is
2546 accessible via a bus interface.
2548 * "systemctl reboot" gained a new switch --firmware-setup that
2549 can be used to reboot into the EFI firmware setup, if that
2550 is available. systemd-logind now exposes an API on the bus
2551 to trigger such reboots, in case graphical desktop UIs want
2552 to cover this functionality.
2554 * "systemctl enable", "systemctl disable" and "systemctl mask"
2555 now support a new "--now" switch. If specified the units
2556 that are enabled will also be started, and the ones
2557 disabled/masked also stopped.
2559 * The Gummiboot EFI boot loader tool has been merged into
2560 systemd, and renamed to "systemd-boot". The bootctl tool has been
2561 updated to support systemd-boot.
2563 * An EFI kernel stub has been added that may be used to create
2564 kernel EFI binaries that contain not only the actual kernel,
2565 but also an initrd, boot splash, command line and OS release
2566 information. This combined binary can then be signed as a
2567 single image, so that the firmware can verify it all in one
2568 step. systemd-boot has special support for EFI binaries created
2569 like this and can extract OS release information from them
2570 and show them in the boot menu. This functionality is useful
2571 to implement cryptographically verified boot schemes.
2573 * Optional support has been added to systemd-fsck to pass
2574 fsck's progress report to an AF_UNIX socket in the file
2577 * udev will no longer create device symlinks for all block
2578 devices by default. A blacklist for excluding special block
2579 devices from this logic has been turned into a whitelist
2580 that requires picking block devices explicitly that require
2583 * A new (currently still internal) API sd-device.h has been
2584 added to libsystemd. This modernized API is supposed to
2585 replace libudev eventually. In fact, already much of libudev
2586 is now just a wrapper around sd-device.h.
2588 * A new hwdb database for storing metadata about pointing
2589 stick devices has been added.
2591 * systemd-tmpfiles gained support for setting file attributes
2592 similar to the "chattr" tool with new 'h' and 'H' lines.
2594 * systemd-journald will no longer unconditionally set the
2595 btrfs NOCOW flag on new journal files. This is instead done
2596 with tmpfiles snippet using the new 'h' line type. This
2597 allows easy disabling of this logic, by masking the
2598 journal-nocow.conf tmpfiles file.
2600 * systemd-journald will now translate audit message types to
2601 human readable identifiers when writing them to the
2602 journal. This should improve readability of audit messages.
2604 * The LUKS logic gained support for the offset= and skip=
2605 options in /etc/crypttab, as previously implemented by
2608 * /usr/lib/os-release gained a new optional field VARIANT= for
2609 distributions that support multiple variants (such as a
2610 desktop edition, a server edition, ...)
2612 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
2613 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
2614 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
2615 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
2616 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
2617 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
2618 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
2619 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
2620 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
2621 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
2622 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
2623 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
2624 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
2625 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
2626 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
2627 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
2628 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
2629 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
2630 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
2631 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
2632 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
2633 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
2634 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
2635 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
2636 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
2637 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
2638 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
2640 — Berlin, 2015-05-22
2644 * Introduce a new API "sd-hwdb.h" for querying the hardware
2645 metadata database. With this minimal interface one can query
2646 and enumerate the udev hwdb, decoupled from the old libudev
2647 library. libudev's interface for this is now only a wrapper
2648 around sd-hwdb. A new tool systemd-hwdb has been added to
2649 interface with and update the database.
2651 * When any of systemd's tools copies files (for example due to
2652 tmpfiles' C lines) a btrfs reflink will attempted first,
2653 before bytewise copying is done.
2655 * systemd-nspawn gained a new --ephemeral switch. When
2656 specified a btrfs snapshot is taken of the container's root
2657 directory, and immediately removed when the container
2658 terminates again. Thus, a container can be started whose
2659 changes never alter the container's root directory, and are
2660 lost on container termination. This switch can also be used
2661 for starting a container off the root file system of the
2662 host without affecting the host OS. This switch is only
2663 available on btrfs file systems.
2665 * systemd-nspawn gained a new --template= switch. It takes the
2666 path to a container tree to use as template for the tree
2667 specified via --directory=, should that directory be
2668 missing. This allows instantiating containers dynamically,
2669 on first run. This switch is only available on btrfs file
2672 * When a .mount unit refers to a mount point on which multiple
2673 mounts are stacked, and the .mount unit is stopped all of
2674 the stacked mount points will now be unmounted until no
2675 mount point remains.
2677 * systemd now has an explicit notion of supported and
2678 unsupported unit types. Jobs enqueued for unsupported unit
2679 types will now fail with an "unsupported" error code. More
2680 specifically .swap, .automount and .device units are not
2681 supported in containers, .busname units are not supported on
2682 non-kdbus systems. .swap and .automount are also not
2683 supported if their respective kernel compile time options
2686 * machinectl gained support for two new "copy-from" and
2687 "copy-to" commands for copying files from a running
2688 container to the host or vice versa.
2690 * machinectl gained support for a new "bind" command to bind
2691 mount host directories into local containers. This is
2692 currently only supported for nspawn containers.
2694 * networkd gained support for configuring bridge forwarding
2695 database entries (fdb) from .network files.
2697 * A new tiny daemon "systemd-importd" has been added that can
2698 download container images in tar, raw, qcow2 or dkr formats,
2699 and make them available locally in /var/lib/machines, so
2700 that they can run as nspawn containers. The daemon can GPG
2701 verify the downloads (not supported for dkr, since it has no
2702 provisions for verifying downloads). It will transparently
2703 decompress bz2, xz, gzip compressed downloads if necessary,
2704 and restore sparse files on disk. The daemon uses privilege
2705 separation to ensure the actual download logic runs with
2706 fewer privileges than the daemon itself. machinectl has
2707 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
2708 make the functionality of importd available to the
2709 user. With this in place the Fedora and Ubuntu "Cloud"
2710 images can be downloaded and booted as containers unmodified
2711 (the Fedora images lack the appropriate GPG signature files
2712 currently, so they cannot be verified, but this will change
2713 soon, hopefully). Note that downloading images is currently
2714 only fully supported on btrfs.
2716 * machinectl is now able to list container images found in
2717 /var/lib/machines, along with some metadata about sizes of
2718 disk and similar. If the directory is located on btrfs and
2719 quota is enabled, this includes quota display. A new command
2720 "image-status" has been added that shows additional
2721 information about images.
2723 * machinectl is now able to clone container images
2724 efficiently, if the underlying file system (btrfs) supports
2725 it, with the new "machinectl clone" command. It also
2726 gained commands for renaming and removing images, as well as
2727 marking them read-only or read-write (supported also on
2728 legacy file systems).
2730 * networkd gained support for collecting LLDP network
2731 announcements, from hardware that supports this. This is
2732 shown in networkctl output.
2734 * systemd-run gained support for a new -t (--pty) switch for
2735 invoking a binary on a pty whose input and output is
2736 connected to the invoking terminal. This allows executing
2737 processes as system services while interactively
2738 communicating with them via the terminal. Most interestingly
2739 this is supported across container boundaries. Invoking
2740 "systemd-run -t /bin/bash" is an alternative to running a
2741 full login session, the difference being that the former
2742 will not register a session, nor go through the PAM session
2745 * tmpfiles gained support for a new "v" line type for creating
2746 btrfs subvolumes. If the underlying file system is a legacy
2747 file system, this automatically degrades to creating a
2748 normal directory. Among others /var/lib/machines is now
2749 created like this at boot, should it be missing.
2751 * The directory /var/lib/containers/ has been deprecated and
2752 been replaced by /var/lib/machines. The term "machines" has
2753 been used in the systemd context as generic term for both
2754 VMs and containers, and hence appears more appropriate for
2755 this, as the directory can also contain raw images bootable
2758 * systemd-nspawn when invoked with -M but without --directory=
2759 or --image= is now capable of searching for the container
2760 root directory, subvolume or disk image automatically, in
2761 /var/lib/machines. systemd-nspawn@.service has been updated
2762 to make use of this, thus allowing it to be used for raw
2765 * A new machines.target unit has been introduced that is
2766 supposed to group all containers/VMs invoked as services on
2767 the system. systemd-nspawn@.service has been updated to
2768 integrate with that.
2770 * machinectl gained a new "start" command, for invoking a
2771 container as a service. "machinectl start foo" is mostly
2772 equivalent to "systemctl start systemd-nspawn@foo.service",
2773 but handles escaping in a nicer way.
2775 * systemd-nspawn will now mount most of the cgroupfs tree
2776 read-only into each container, with the exception of the
2777 container's own subtree in the name=systemd hierarchy.
2779 * journald now sets the special FS_NOCOW file flag for its
2780 journal files. This should improve performance on btrfs, by
2781 avoiding heavy fragmentation when journald's write-pattern
2782 is used on COW file systems. It degrades btrfs' data
2783 integrity guarantees for the files to the same levels as for
2784 ext3/ext4 however. This should be OK though as journald does
2785 its own data integrity checks and all its objects are
2786 checksummed on disk. Also, journald should handle btrfs disk
2787 full events a lot more gracefully now, by processing SIGBUS
2788 errors, and not relying on fallocate() anymore.
2790 * When journald detects that journal files it is writing to
2791 have been deleted it will immediately start new journal
2794 * systemd now provides a way to store file descriptors
2795 per-service in PID 1. This is useful for daemons to ensure
2796 that fds they require are not lost during a daemon
2797 restart. The fds are passed to the daemon on the next
2798 invocation in the same way socket activation fds are
2799 passed. This is now used by journald to ensure that the
2800 various sockets connected to all the system's stdout/stderr
2801 are not lost when journald is restarted. File descriptors
2802 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
2803 an extension to sd_notify(). Note that a limit is enforced
2804 on the number of fds a service can store in PID 1, and it
2805 defaults to 0, so that no fds may be stored, unless this is
2806 explicitly turned on.
2808 * The default TERM variable to use for units connected to a
2809 terminal, when no other value is explicitly is set is now
2810 vt220 rather than vt102. This should be fairly safe still,
2811 but allows PgUp/PgDn work.
2813 * The /etc/crypttab option header= as known from Debian is now
2816 * "loginctl user-status" and "loginctl session-status" will
2817 now show the last 10 lines of log messages of the
2818 user/session following the status output. Similar,
2819 "machinectl status" will show the last 10 log lines
2820 associated with a virtual machine or container
2821 service. (Note that this is usually not the log messages
2822 done in the VM/container itself, but simply what the
2823 container manager logs. For nspawn this includes all console
2826 * "loginctl session-status" without further argument will now
2827 show the status of the session of the caller. Similar,
2828 "lock-session", "unlock-session", "activate",
2829 "enable-linger", "disable-linger" may now be called without
2830 session/user parameter in which case they apply to the
2831 caller's session/user.
2833 * An X11 session scriptlet is now shipped that uploads
2834 $DISPLAY and $XAUTHORITY into the environment of the systemd
2835 --user daemon if a session begins. This should improve
2836 compatibility with X11 enabled applications run as systemd
2839 * Generators are now subject to masking via /etc and /run, the
2840 same way as unit files.
2842 * networkd .network files gained support for configuring
2843 per-link IPv4/IPv6 packet forwarding as well as IPv4
2844 masquerading. This is by default turned on for veth links to
2845 containers, as registered by systemd-nspawn. This means that
2846 nspawn containers run with --network-veth will now get
2847 automatic routed access to the host's networks without any
2848 further configuration or setup, as long as networkd runs on
2851 * systemd-nspawn gained the --port= (-p) switch to expose TCP
2852 or UDP posts of a container on the host. With this in place
2853 it is possible to run containers with private veth links
2854 (--network-veth), and have their functionality exposed on
2855 the host as if their services were running directly on the
2858 * systemd-nspawn's --network-veth switch now gained a short
2859 version "-n", since with the changes above it is now truly
2860 useful out-of-the-box. The systemd-nspawn@.service has been
2861 updated to make use of it too by default.
2863 * systemd-nspawn will now maintain a per-image R/W lock, to
2864 ensure that the same image is not started more than once
2865 writable. (It's OK to run an image multiple times
2866 simultaneously in read-only mode.)
2868 * systemd-nspawn's --image= option is now capable of
2869 dissecting and booting MBR and GPT disk images that contain
2870 only a single active Linux partition. Previously it
2871 supported only GPT disk images with proper GPT type
2872 IDs. This allows running cloud images from major
2873 distributions directly with systemd-nspawn, without
2876 * In addition to collecting mouse dpi data in the udev
2877 hardware database, there's now support for collecting angle
2878 information for mouse scroll wheels. The database is
2879 supposed to guarantee similar scrolling behavior on mice
2880 that it knows about. There's also support for collecting
2881 information about Touchpad types.
2883 * udev's input_id built-in will now also collect touch screen
2884 dimension data and attach it to probed devices.
2886 * /etc/os-release gained support for a Distribution Privacy
2889 * networkd gained support for creating "ipvlan", "gretap",
2890 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
2892 * systemd-tmpfiles gained support for "a" lines for setting
2895 * systemd-nspawn will now mount /tmp in the container to
2896 tmpfs, automatically.
2898 * systemd now exposes the memory.usage_in_bytes cgroup
2899 attribute and shows it for each service in the "systemctl
2900 status" output, if available.
2902 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
2903 immediate reboot is triggered. This useful if shutdown is
2904 hung and is unable to complete, to expedite the
2905 operation. Note that this kind of reboot will still unmount
2906 all file systems, and hence should not result in fsck being
2909 * A .device unit for an optical block device will now be
2910 considered active only when a medium is in the drive. Also,
2911 mount units are now bound to their backing devices thus
2912 triggering automatic unmounting when devices become
2913 unavailable. With this in place systemd will now
2914 automatically unmount left-over mounts when a CD-ROM is
2915 ejected or an USB stick is yanked from the system.
2917 * networkd-wait-online now has support for waiting for
2918 specific interfaces only (with globbing), and for giving up
2919 after a configurable timeout.
2921 * networkd now exits when idle. It will be automatically
2922 restarted as soon as interfaces show up, are removed or
2923 change state. networkd will stay around as long as there is
2924 at least one DHCP state machine or similar around, that keep
2927 * networkd may now configure IPv6 link-local addressing in
2928 addition to IPv4 link-local addressing.
2930 * The IPv6 "token" for use in SLAAC may now be configured for
2931 each .network interface in networkd.
2933 * Routes configured with networkd may now be assigned a scope
2936 * networkd's [Match] sections now support globbing and lists
2937 of multiple space-separated matches per item.
2939 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
2940 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
2941 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
2942 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
2943 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
2944 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
2945 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
2946 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
2947 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
2948 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
2949 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
2950 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
2951 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
2952 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
2953 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
2954 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
2955 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
2956 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
2957 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
2958 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
2959 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
2960 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
2961 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
2962 Hoffmann, Zbigniew Jędrzejewski-Szmek
2964 — Berlin, 2015-02-16
2968 * When querying unit file enablement status (for example via
2969 "systemctl is-enabled"), a new state "indirect" is now known
2970 which indicates that a unit might not be enabled itself, but
2971 another unit listed in its Also= setting might be.
2973 * Similar to the various existing ConditionXYZ= settings for
2974 units, there are now matching AssertXYZ= settings. While
2975 failing conditions cause a unit to be skipped, but its job
2976 to succeed, failing assertions declared like this will cause
2977 a unit start operation and its job to fail.
2979 * hostnamed now knows a new chassis type "embedded".
2981 * systemctl gained a new "edit" command. When used on a unit
2982 file, this allows extending unit files with .d/ drop-in
2983 configuration snippets or editing the full file (after
2984 copying it from /usr/lib to /etc). This will invoke the
2985 user's editor (as configured with $EDITOR), and reload the
2986 modified configuration after editing.
2988 * "systemctl status" now shows the suggested enablement state
2989 for a unit, as declared in the (usually vendor-supplied)
2990 system preset files.
2992 * nss-myhostname will now resolve the single-label host name
2993 "gateway" to the locally configured default IP routing
2994 gateways, ordered by their metrics. This assigns a stable
2995 name to the used gateways, regardless which ones are
2996 currently configured. Note that the name will only be
2997 resolved after all other name sources (if nss-myhostname is
2998 configured properly) and should hence not negatively impact
2999 systems that use the single-label host name "gateway" in
3002 * systemd-inhibit now allows filtering by mode when listing
3005 * Scope and service units gained a new "Delegate" boolean
3006 property, which, when set, allows processes running inside the
3007 unit to further partition resources. This is primarily
3008 useful for systemd user instances as well as container
3011 * journald will now pick up audit messages directly from
3012 the kernel, and log them like any other log message. The
3013 audit fields are split up and fully indexed. This means that
3014 journalctl in many ways is now a (nicer!) alternative to
3015 ausearch, the traditional audit client. Note that this
3016 implements only a minimal audit client. If you want the
3017 special audit modes like reboot-on-log-overflow, please use
3018 the traditional auditd instead, which can be used in
3019 parallel to journald.
3021 * The ConditionSecurity= unit file option now understands the
3022 special string "audit" to check whether auditing is
3025 * journalctl gained two new commands --vacuum-size= and
3026 --vacuum-time= to delete old journal files until the
3027 remaining ones take up no more than the specified size on disk,
3028 or are not older than the specified time.
3030 * A new, native PPPoE library has been added to sd-network,
3031 systemd's library of light-weight networking protocols. This
3032 library will be used in a future version of networkd to
3033 enable PPPoE communication without an external pppd daemon.
3035 * The busctl tool now understands a new "capture" verb that
3036 works similar to "monitor", but writes a packet capture
3037 trace to STDOUT that can be redirected to a file which is
3038 compatible with libcap's capture file format. This can then
3039 be loaded in Wireshark and similar tools to inspect bus
3042 * The busctl tool now understands a new "tree" verb that shows
3043 the object trees of a specific service on the bus, or of all
3046 * The busctl tool now understands a new "introspect" verb that
3047 shows all interfaces and members of objects on the bus,
3048 including their signature and values. This is particularly
3049 useful to get more information about bus objects shown by
3050 the new "busctl tree" command.
3052 * The busctl tool now understands new verbs "call",
3053 "set-property" and "get-property" for invoking bus method
3054 calls, setting and getting bus object properties in a
3057 * busctl gained a new --augment-creds= argument that controls
3058 whether the tool shall augment credential information it
3059 gets from the bus with data from /proc, in a possibly
3062 * nspawn's --link-journal= switch gained two new values
3063 "try-guest" and "try-host" that work like "guest" and
3064 "host", but do not fail if the host has no persistent
3065 journalling enabled. -j is now equivalent to
3066 --link-journal=try-guest.
3068 * macvlan network devices created by nspawn will now have
3069 stable MAC addresses.
3071 * A new SmackProcessLabel= unit setting has been added, which
3072 controls the SMACK security label processes forked off by
3073 the respective unit shall use.
3075 * If compiled with --enable-xkbcommon, systemd-localed will
3076 verify x11 keymap settings by compiling the given keymap. It
3077 will spew out warnings if the compilation fails. This
3078 requires libxkbcommon to be installed.
3080 * When a coredump is collected, a larger number of metadata
3081 fields is now collected and included in the journal records
3082 created for it. More specifically, control group membership,
3083 environment variables, memory maps, working directory,
3084 chroot directory, /proc/$PID/status, and a list of open file
3085 descriptors is now stored in the log entry.
3087 * The udev hwdb now contains DPI information for mice. For
3090 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
3092 * All systemd programs that read standalone configuration
3093 files in /etc now also support a corresponding series of
3094 .conf.d configuration directories in /etc/, /run/,
3095 /usr/local/lib/, /usr/lib/, and (if configured with
3096 --enable-split-usr) /lib/. In particular, the following
3097 configuration files now have corresponding configuration
3098 directories: system.conf user.conf, logind.conf,
3099 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
3100 resolved.conf, timesyncd.conf, journal-remote.conf, and
3101 journal-upload.conf. Note that distributions should use the
3102 configuration directories in /usr/lib/; the directories in
3103 /etc/ are reserved for the system administrator.
3105 * systemd-rfkill will no longer take the rfkill device name
3106 into account when storing rfkill state on disk, as the name
3107 might be dynamically assigned and not stable. Instead, the
3108 ID_PATH udev variable combined with the rfkill type (wlan,
3109 bluetooth, ...) is used.
3111 * A new service systemd-machine-id-commit.service has been
3112 added. When used on systems where /etc is read-only during
3113 boot, and /etc/machine-id is not initialized (but an empty
3114 file), this service will copy the temporary machine ID
3115 created as replacement into /etc after the system is fully
3116 booted up. This is useful for systems that are freshly
3117 installed with a non-initialized machine ID, but should get
3118 a fixed machine ID for subsequent boots.
3120 * networkd's .netdev files now provide a large set of
3121 configuration parameters for VXLAN devices. Similarly, the
3122 bridge port cost parameter is now configurable in .network
3123 files. There's also new support for configuring IP source
3124 routing. networkd .link files gained support for a new
3125 OriginalName= match that is useful to match against the
3126 original interface name the kernel assigned. .network files
3127 may include MTU= and MACAddress= fields for altering the MTU
3128 and MAC address while being connected to a specific network
3131 * The LUKS logic gained supported for configuring
3132 UUID-specific key files. There's also new support for naming
3133 LUKS device from the kernel command line, using the new
3134 luks.name= argument.
3136 * Timer units may now be transiently created via the bus API
3137 (this was previously already available for scope and service
3138 units). In addition it is now possible to create multiple
3139 transient units at the same time with a single bus call. The
3140 "systemd-run" tool has been updated to make use of this for
3141 running commands on a specified time, in at(1)-style.
3143 * tmpfiles gained support for "t" lines, for assigning
3144 extended attributes to files. Among other uses this may be
3145 used to assign SMACK labels to files.
3147 Contributions from: Alin Rauta, Alison Chaiken, Andrej
3148 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
3149 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
3150 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
3151 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
3152 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
3153 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
3154 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
3155 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
3156 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
3157 Michael Biebl, Michael Chapman, Michael Marineau, Michal
3158 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
3159 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
3160 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
3161 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
3162 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3163 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
3164 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
3166 — Berlin, 2014-12-10
3170 * journalctl gained the new options -t/--identifier= to match
3171 on the syslog identifier (aka "tag"), as well as --utc to
3172 show log timestamps in the UTC timezone. journalctl now also
3173 accepts -n/--lines=all to disable line capping in a pager.
3175 * journalctl gained a new switch, --flush, that synchronously
3176 flushes logs from /run/log/journal to /var/log/journal if
3177 persistent storage is enabled. systemd-journal-flush.service
3178 now waits until the operation is complete.
3180 * Services can notify the manager before they start a reload
3181 (by sending RELOADING=1) or shutdown (by sending
3182 STOPPING=1). This allows the manager to track and show the
3183 internal state of daemons and closes a race condition when
3184 the process is still running but has closed its D-Bus
3187 * Services with Type=oneshot do not have to have any ExecStart
3190 * User units are now loaded also from
3191 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
3192 /run/systemd/user directory that was already previously
3193 supported, but is under the control of the user.
3195 * Job timeouts (i.e. time-outs on the time a job that is
3196 queued stays in the run queue) can now optionally result in
3197 immediate reboot or power-off actions (JobTimeoutAction= and
3198 JobTimeoutRebootArgument=). This is useful on ".target"
3199 units, to limit the maximum time a target remains
3200 undispatched in the run queue, and to trigger an emergency
3201 operation in such a case. This is now used by default to
3202 turn off the system if boot-up (as defined by everything in
3203 basic.target) hangs and does not complete for at least
3204 15min. Also, if power-off or reboot hang for at least 30min
3205 an immediate power-off/reboot operation is triggered. This
3206 functionality is particularly useful to increase reliability
3207 on embedded devices, but also on laptops which might
3208 accidentally get powered on when carried in a backpack and
3209 whose boot stays stuck in a hard disk encryption passphrase
3212 * systemd-logind can be configured to also handle lid switch
3213 events even when the machine is docked or multiple displays
3214 are attached (HandleLidSwitchDocked= option).
3216 * A helper binary and a service have been added which can be
3217 used to resume from hibernation in the initramfs. A
3218 generator will parse the resume= option on the kernel
3219 command line to trigger resume.
3221 * A user console daemon systemd-consoled has been
3222 added. Currently, it is a preview, and will so far open a
3223 single terminal on each session of the user marked as
3224 Desktop=systemd-console.
3226 * Route metrics can be specified for DHCP routes added by
3229 * The SELinux context of socket-activated services can be set
3230 from the information provided by the networking stack
3231 (SELinuxContextFromNet= option).
3233 * Userspace firmware loading support has been removed and
3234 the minimum supported kernel version is thus bumped to 3.7.
3236 * Timeout for udev workers has been increased from 1 to 3
3237 minutes, but a warning will be printed after 1 minute to
3238 help diagnose kernel modules that take a long time to load.
3240 * Udev rules can now remove tags on devices with TAG-="foobar".
3242 * systemd's readahead implementation has been removed. In many
3243 circumstances it didn't give expected benefits even for
3244 rotational disk drives and was becoming less relevant in the
3245 age of SSDs. As none of the developers has been using
3246 rotating media anymore, and nobody stepped up to actively
3247 maintain this component of systemd it has now been removed.
3249 * Swap units can use Options= to specify discard options.
3250 Discard options specified for swaps in /etc/fstab are now
3253 * Docker containers are now detected as a separate type of
3256 * The Password Agent protocol gained support for queries where
3257 the user input is shown, useful e.g. for user names.
3258 systemd-ask-password gained a new --echo option to turn that
3261 * The default sysctl.d/ snippets will now set:
3263 net.core.default_qdisc = fq_codel
3265 This selects Fair Queuing Controlled Delay as the default
3266 queuing discipline for network interfaces. fq_codel helps
3267 fight the network bufferbloat problem. It is believed to be
3268 a good default with no tuning required for most workloads.
3269 Downstream distributions may override this choice. On 10Gbit
3270 servers that do not do forwarding, "fq" may perform better.
3271 Systems without a good clocksource should use "pfifo_fast".
3273 * If kdbus is enabled during build a new option BusPolicy= is
3274 available for service units, that allows locking all service
3275 processes into a stricter bus policy, in order to limit
3276 access to various bus services, or even hide most of them
3277 from the service's view entirely.
3279 * networkctl will now show the .network and .link file
3280 networkd has applied to a specific interface.
3282 * sd-login gained a new API call sd_session_get_desktop() to
3283 query which desktop environment has been selected for a
3286 * UNIX utmp support is now compile-time optional to support
3287 legacy-free systems.
3289 * systemctl gained two new commands "add-wants" and
3290 "add-requires" for pulling in units from specific targets
3293 * If the word "rescue" is specified on the kernel command line
3294 the system will now boot into rescue mode (aka
3295 rescue.target), which was previously available only by
3296 specifying "1" or "systemd.unit=rescue.target" on the kernel
3297 command line. This new kernel command line option nicely
3298 mirrors the already existing "emergency" kernel command line
3301 * New kernel command line options mount.usr=, mount.usrflags=,
3302 mount.usrfstype= have been added that match root=, rootflags=,
3303 rootfstype= but allow mounting a specific file system to
3306 * The $NOTIFY_SOCKET is now also passed to control processes of
3307 services, not only the main process.
3309 * This version reenables support for fsck's -l switch. This
3310 means at least version v2.25 of util-linux is required for
3311 operation, otherwise dead-locks on device nodes may
3312 occur. Again: you need to update util-linux to at least
3313 v2.25 when updating systemd to v217.
3315 * The "multi-seat-x" tool has been removed from systemd, as
3316 its functionality has been integrated into X servers 1.16,
3317 and the tool is hence redundant. It is recommended to update
3318 display managers invoking this tool to simply invoke X
3319 directly from now on, again.
3321 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
3322 message flag has been added for all of systemd's PolicyKit
3323 authenticated method calls has been added. In particular
3324 this now allows optional interactive authorization via
3325 PolicyKit for many of PID1's privileged operations such as
3326 unit file enabling and disabling.
3328 * "udevadm hwdb --update" learnt a new switch "--usr" for
3329 placing the rebuilt hardware database in /usr instead of
3330 /etc. When used only hardware database entries stored in
3331 /usr will be used, and any user database entries in /etc are
3332 ignored. This functionality is useful for vendors to ship a
3333 pre-built database on systems where local configuration is
3334 unnecessary or unlikely.
3336 * Calendar time specifications in .timer units now also
3337 understand the strings "semi-annually", "quarterly" and
3338 "minutely" as shortcuts (in addition to the preexisting
3339 "anually", "hourly", ...).
3341 * systemd-tmpfiles will now correctly create files in /dev
3342 at boot which are marked for creation only at boot. It is
3343 recommended to always create static device nodes with 'c!'
3344 and 'b!', so that they are created only at boot and not
3345 overwritten at runtime.
3347 * When the watchdog logic is used for a service (WatchdogSec=)
3348 and the watchdog timeout is hit the service will now be
3349 terminated with SIGABRT (instead of just SIGTERM), in order
3350 to make sure a proper coredump and backtrace is
3351 generated. This ensures that hanging services will result in
3352 similar coredump/backtrace behaviour as services that hit a
3355 Contributions from: Andreas Henriksson, Andrei Borzenkov,
3356 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
3357 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
3358 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
3359 Herrmann, David Sommerseth, David Strauss, Emil Renner
3360 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
3361 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
3362 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
3363 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
3364 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
3365 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
3366 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
3367 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
3368 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
3369 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
3370 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
3371 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
3372 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
3373 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
3374 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
3375 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
3378 — Berlin, 2014-10-28
3382 * timedated no longer reads NTP implementation unit names from
3383 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
3384 implementations should add a
3386 Conflicts=systemd-timesyncd.service
3388 to their unit files to take over and replace systemd's NTP
3389 default functionality.
3391 * systemd-sysusers gained a new line type "r" for configuring
3392 which UID/GID ranges to allocate system users/groups
3393 from. Lines of type "u" may now add an additional column
3394 that specifies the home directory for the system user to be
3395 created. Also, systemd-sysusers may now optionally read user
3396 information from STDIN instead of a file. This is useful for
3397 invoking it from RPM preinst scriptlets that need to create
3398 users before the first RPM file is installed since these
3399 files might need to be owned by them. A new
3400 %sysusers_create_inline RPM macro has been introduced to do
3401 just that. systemd-sysusers now updates the shadow files as
3402 well as the user/group databases, which should enhance
3403 compatibility with certain tools like grpck.
3405 * A number of bus APIs of PID 1 now optionally consult
3406 PolicyKit to permit access for otherwise unprivileged
3407 clients under certain conditions. Note that this currently
3408 doesn't support interactive authentication yet, but this is
3409 expected to be added eventually, too.
3411 * /etc/machine-info now has new fields for configuring the
3412 deployment environment of the machine, as well as the
3413 location of the machine. hostnamectl has been updated with
3414 new command to update these fields.
3416 * systemd-timesyncd has been updated to automatically acquire
3417 NTP server information from systemd-networkd, which might
3418 have been discovered via DHCP.
3420 * systemd-resolved now includes a caching DNS stub resolver
3421 and a complete LLMNR name resolution implementation. A new
3422 NSS module "nss-resolve" has been added which can be used
3423 instead of glibc's own "nss-dns" to resolve hostnames via
3424 systemd-resolved. Hostnames, addresses and arbitrary RRs may
3425 be resolved via systemd-resolved D-Bus APIs. In contrast to
3426 the glibc internal resolver systemd-resolved is aware of
3427 multi-homed system, and keeps DNS server and caches separate
3428 and per-interface. Queries are sent simultaneously on all
3429 interfaces that have DNS servers configured, in order to
3430 properly handle VPNs and local LANs which might resolve
3431 separate sets of domain names. systemd-resolved may acquire
3432 DNS server information from systemd-networkd automatically,
3433 which in turn might have discovered them via DHCP. A tool
3434 "systemd-resolve-host" has been added that may be used to
3435 query the DNS logic in resolved. systemd-resolved implements
3436 IDNA and automatically uses IDNA or UTF-8 encoding depending
3437 on whether classic DNS or LLMNR is used as transport. In the
3438 next releases we intend to add a DNSSEC and mDNS/DNS-SD
3439 implementation to systemd-resolved.
3441 * A new NSS module nss-mymachines has been added, that
3442 automatically resolves the names of all local registered
3443 containers to their respective IP addresses.
3445 * A new client tool "networkctl" for systemd-networkd has been
3446 added. It currently is entirely passive and will query
3447 networking configuration from udev, rtnetlink and networkd,
3448 and present it to the user in a very friendly
3449 way. Eventually, we hope to extend it to become a full
3450 control utility for networkd.
3452 * .socket units gained a new DeferAcceptSec= setting that
3453 controls the kernels' TCP_DEFER_ACCEPT sockopt for
3454 TCP. Similarly, support for controlling TCP keep-alive
3455 settings has been added (KeepAliveTimeSec=,
3456 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
3457 turning off Nagle's algorithm on TCP has been added
3460 * logind learned a new session type "web", for use in projects
3461 like Cockpit which register web clients as PAM sessions.
3463 * timer units with at least one OnCalendar= setting will now
3464 be started only after time-sync.target has been
3465 reached. This way they will not elapse before the system
3466 clock has been corrected by a local NTP client or
3467 similar. This is particular useful on RTC-less embedded
3468 machines, that come up with an invalid system clock.
3470 * systemd-nspawn's --network-veth= switch should now result in
3471 stable MAC addresses for both the outer and the inner side
3474 * systemd-nspawn gained a new --volatile= switch for running
3475 container instances with /etc or /var unpopulated.
3477 * The kdbus client code has been updated to use the new Linux
3478 3.17 memfd subsystem instead of the old kdbus-specific one.
3480 * systemd-networkd's DHCP client and server now support
3481 FORCERENEW. There are also new configuration options to
3482 configure the vendor client identifier and broadcast mode
3485 * systemd will no longer inform the kernel about the current
3486 timezone, as this is necessarily incorrect and racy as the
3487 kernel has no understanding of DST and similar
3488 concepts. This hence means FAT timestamps will be always
3489 considered UTC, similar to what Android is already
3490 doing. Also, when the RTC is configured to the local time
3491 (rather than UTC) systemd will never synchronize back to it,
3492 as this might confuse Windows at a later boot.
3494 * systemd-analyze gained a new command "verify" for offline
3495 validation of unit files.
3497 * systemd-networkd gained support for a couple of additional
3498 settings for bonding networking setups. Also, the metric for
3499 statically configured routes may now be configured. For
3500 network interfaces where this is appropriate the peer IP
3501 address may now be configured.
3503 * systemd-networkd's DHCP client will no longer request
3504 broadcasting by default, as this tripped up some networks.
3505 For hardware where broadcast is required the feature should
3506 be switched back on using RequestBroadcast=yes.
3508 * systemd-networkd will now set up IPv4LL addresses (when
3509 enabled) even if DHCP is configured successfully.
3511 * udev will now default to respect network device names given
3512 by the kernel when the kernel indicates that these are
3513 predictable. This behavior can be tweaked by changing
3514 NamePolicy= in the relevant .link file.
3516 * A new library systemd-terminal has been added that
3517 implements full TTY stream parsing and rendering. This
3518 library is supposed to be used later on for implementing a
3519 full userspace VT subsystem, replacing the current kernel
3522 * A new tool systemd-journal-upload has been added to push
3523 journal data to a remote system running
3524 systemd-journal-remote.
3526 * journald will no longer forward all local data to another
3527 running syslog daemon. This change has been made because
3528 rsyslog (which appears to be the most commonly used syslog
3529 implementation these days) no longer makes use of this, and
3530 instead pulls the data out of the journal on its own. Since
3531 forwarding the messages to a non-existent syslog server is
3532 more expensive than we assumed we have now turned this
3533 off. If you run a syslog server that is not a recent rsyslog
3534 version, you have to turn this option on again
3535 (ForwardToSyslog= in journald.conf).
3537 * journald now optionally supports the LZ4 compressor for
3538 larger journal fields. This compressor should perform much
3539 better than XZ which was the previous default.
3541 * machinectl now shows the IP addresses of local containers,
3542 if it knows them, plus the interface name of the container.
3544 * A new tool "systemd-escape" has been added that makes it
3545 easy to escape strings to build unit names and similar.
3547 * sd_notify() messages may now include a new ERRNO= field
3548 which is parsed and collected by systemd and shown among the
3549 "systemctl status" output for a service.
3551 * A new component "systemd-firstboot" has been added that
3552 queries the most basic systemd information (timezone,
3553 hostname, root password) interactively on first
3554 boot. Alternatively it may also be used to provision these
3555 things offline on OS images installed into directories.
3557 * The default sysctl.d/ snippets will now set
3559 net.ipv4.conf.default.promote_secondaries=1
3561 This has the benefit of no flushing secondary IP addresses
3562 when primary addresses are removed.
3564 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
3565 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
3566 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
3567 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
3568 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
3569 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
3570 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
3571 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
3572 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
3573 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
3574 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
3575 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
3576 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
3577 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
3578 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
3580 — Berlin, 2014-08-19
3584 * A new tool systemd-sysusers has been added. This tool
3585 creates system users and groups in /etc/passwd and
3586 /etc/group, based on static declarative system user/group
3587 definitions in /usr/lib/sysusers.d/. This is useful to
3588 enable factory resets and volatile systems that boot up with
3589 an empty /etc directory, and thus need system users and
3590 groups created during early boot. systemd now also ships
3591 with two default sysusers.d/ files for the most basic
3592 users and groups systemd and the core operating system
3595 * A new tmpfiles snippet has been added that rebuilds the
3596 essential files in /etc on boot, should they be missing.
3598 * A directive for ensuring automatic clean-up of
3599 /var/cache/man/ has been removed from the default
3600 configuration. This line should now be shipped by the man
3601 implementation. The necessary change has been made to the
3602 man-db implementation. Note that you need to update your man
3603 implementation to one that ships this line, otherwise no
3604 automatic clean-up of /var/cache/man will take place.
3606 * A new condition ConditionNeedsUpdate= has been added that
3607 may conditionalize services to only run when /etc or /var
3608 are "older" than the vendor operating system resources in
3609 /usr. This is useful for reconstructing or updating /etc
3610 after an offline update of /usr or a factory reset, on the
3611 next reboot. Services that want to run once after such an
3612 update or reset should use this condition and order
3613 themselves before the new systemd-update-done.service, which
3614 will mark the two directories as fully updated. A number of
3615 service files have been added making use of this, to rebuild
3616 the udev hardware database, the journald message catalog and
3617 dynamic loader cache (ldconfig). The systemd-sysusers tool
3618 described above also makes use of this now. With this in
3619 place it is now possible to start up a minimal operating
3620 system with /etc empty cleanly. For more information on the
3621 concepts involved see this recent blog story:
3623 http://0pointer.de/blog/projects/stateless.html
3625 * A new system group "input" has been introduced, and all
3626 input device nodes get this group assigned. This is useful
3627 for system-level software to get access to input devices. It
3628 complements what is already done for "audio" and "video".
3630 * systemd-networkd learnt minimal DHCPv4 server support in
3631 addition to the existing DHCPv4 client support. It also
3632 learnt DHCPv6 client and IPv6 Router Solicitation client
3633 support. The DHCPv4 client gained support for static routes
3634 passed in from the server. Note that the [DHCPv4] section
3635 known in older systemd-networkd versions has been renamed to
3636 [DHCP] and is now also used by the DHCPv6 client. Existing
3637 .network files using settings of this section should be
3638 updated, though compatibility is maintained. Optionally, the
3639 client hostname may now be sent to the DHCP server.
3641 * networkd gained support for vxlan virtual networks as well
3642 as tun/tap and dummy devices.
3644 * networkd gained support for automatic allocation of address
3645 ranges for interfaces from a system-wide pool of
3646 addresses. This is useful for dynamically managing a large
3647 number of interfaces with a single network configuration
3648 file. In particular this is useful to easily assign
3649 appropriate IP addresses to the veth links of a large number
3650 of nspawn instances.
3652 * RPM macros for processing sysusers, sysctl and binfmt
3653 drop-in snippets at package installation time have been
3656 * The /etc/os-release file should now be placed in
3657 /usr/lib/os-release. The old location is automatically
3658 created as symlink. /usr/lib is the more appropriate
3659 location of this file, since it shall actually describe the
3660 vendor operating system shipped in /usr, and not the
3661 configuration stored in /etc.
3663 * .mount units gained a new boolean SloppyOptions= setting
3664 that maps to mount(8)'s -s option which enables permissive
3665 parsing of unknown mount options.
3667 * tmpfiles learnt a new "L+" directive which creates a symlink
3668 but (unlike "L") deletes a pre-existing file first, should
3669 it already exist and not already be the correct
3670 symlink. Similarly, "b+", "c+" and "p+" directives have been
3671 added as well, which create block and character devices, as
3672 well as fifos in the filesystem, possibly removing any
3673 pre-existing files of different types.
3675 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
3676 'argument' field (which so far specified the source to
3677 symlink/copy the files from) is now optional. If omitted the
3678 same file os copied from /usr/share/factory/ suffixed by the
3679 full destination path. This is useful for populating /etc
3680 with essential files, by copying them from vendor defaults
3681 shipped in /usr/share/factory/etc.
3683 * A new command "systemctl preset-all" has been added that
3684 applies the service preset settings to all installed unit
3685 files. A new switch --preset-mode= has been added that
3686 controls whether only enable or only disable operations
3689 * A new command "systemctl is-system-running" has been added
3690 that allows checking the overall state of the system, for
3691 example whether it is fully up and running.
3693 * When the system boots up with an empty /etc, the equivalent
3694 to "systemctl preset-all" is executed during early boot, to
3695 make sure all default services are enabled after a factory
3698 * systemd now contains a minimal preset file that enables the
3699 most basic services systemd ships by default.
3701 * Unit files' [Install] section gained a new DefaultInstance=
3702 field for defining the default instance to create if a
3703 template unit is enabled with no instance specified.
3705 * A new passive target cryptsetup-pre.target has been added
3706 that may be used by services that need to make they run and
3707 finish before the first LUKS cryptographic device is set up.
3709 * The /dev/loop-control and /dev/btrfs-control device nodes
3710 are now owned by the "disk" group by default, opening up
3711 access to this group.
3713 * systemd-coredump will now automatically generate a
3714 stack trace of all core dumps taking place on the system,
3715 based on elfutils' libdw library. This stack trace is logged
3718 * systemd-coredump may now optionally store coredumps directly
3719 on disk (in /var/lib/systemd/coredump, possibly compressed),
3720 instead of storing them unconditionally in the journal. This
3721 mode is the new default. A new configuration file
3722 /etc/systemd/coredump.conf has been added to configure this
3723 and other parameters of systemd-coredump.
3725 * coredumpctl gained a new "info" verb to show details about a
3726 specific coredump. A new switch "-1" has also been added
3727 that makes sure to only show information about the most
3728 recent entry instead of all entries. Also, as the tool is
3729 generally useful now the "systemd-" prefix of the binary
3730 name has been removed. Distributions that want to maintain
3731 compatibility with the old name should add a symlink from
3732 the old name to the new name.
3734 * journald's SplitMode= now defaults to "uid". This makes sure
3735 that unprivileged users can access their own coredumps with
3736 coredumpctl without restrictions.
3738 * New kernel command line options "systemd.wants=" (for
3739 pulling an additional unit during boot), "systemd.mask="
3740 (for masking a specific unit for the boot), and
3741 "systemd.debug-shell" (for enabling the debug shell on tty9)
3742 have been added. This is implemented in the new generator
3743 "systemd-debug-generator".
3745 * systemd-nspawn will now by default filter a couple of
3746 syscalls for containers, among them those required for
3747 kernel module loading, direct x86 IO port access, swap
3748 management, and kexec. Most importantly though
3749 open_by_handle_at() is now prohibited for containers,
3750 closing a hole similar to a recently discussed vulnerability
3751 in docker regarding access to files on file hierarchies the
3752 container should normally not have access to. Note that, for
3753 nspawn, we generally make no security claims anyway (and
3754 this is explicitly documented in the man page), so this is
3755 just a fix for one of the most obvious problems.
3757 * A new man page file-hierarchy(7) has been added that
3758 contains a minimized, modernized version of the file system
3759 layout systemd expects, similar in style to the FHS
3760 specification or hier(5). A new tool systemd-path(1) has
3761 been added to query many of these paths for the local
3764 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
3765 longer done. Since the directory now has a per-user size
3766 limit, and is cleaned on logout this appears unnecessary,
3767 in particular since this now brings the lifecycle of this
3768 directory closer in line with how IPC objects are handled.
3770 * systemd.pc now exports a number of additional directories,
3771 including $libdir (which is useful to identify the library
3772 path for the primary architecture of the system), and a
3773 couple of drop-in directories.
3775 * udev's predictable network interface names now use the dev_port
3776 sysfs attribute, introduced in linux 3.15 instead of dev_id to
3777 distinguish between ports of the same PCI function. dev_id should
3778 only be used for ports using the same HW address, hence the need
3781 * machined has been updated to export the OS version of a
3782 container (read from /etc/os-release and
3783 /usr/lib/os-release) on the bus. This is now shown in
3784 "machinectl status" for a machine.
3786 * A new service setting RestartForceExitStatus= has been
3787 added. If configured to a set of exit signals or process
3788 return values, the service will be restarted when the main
3789 daemon process exits with any of them, regardless of the
3792 * systemctl's -H switch for connecting to remote systemd
3793 machines has been extended so that it may be used to
3794 directly connect to a specific container on the
3795 host. "systemctl -H root@foobar:waldi" will now connect as
3796 user "root" to host "foobar", and then proceed directly to
3797 the container named "waldi". Note that currently you have to
3798 authenticate as user "root" for this to work, as entering
3799 containers is a privileged operation.
3801 Contributions from: Andreas Henriksson, Benjamin Steinwender,
3802 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
3803 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
3804 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
3805 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
3806 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
3807 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
3808 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
3809 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
3810 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
3811 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
3812 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
3814 — Berlin, 2014-07-03
3818 * As an experimental feature, udev now tries to lock the
3819 disk device node (flock(LOCK_SH|LOCK_NB)) while it
3820 executes events for the disk or any of its partitions.
3821 Applications like partitioning programs can lock the
3822 disk device node (flock(LOCK_EX)) and claim temporary
3823 device ownership that way; udev will entirely skip all event
3824 handling for this disk and its partitions. If the disk
3825 was opened for writing, the close will trigger a partition
3826 table rescan in udev's "watch" facility, and if needed
3827 synthesize "change" events for the disk and all its partitions.
3828 This is now unconditionally enabled, and if it turns out to
3829 cause major problems, we might turn it on only for specific
3830 devices, or might need to disable it entirely. Device Mapper
3831 devices are excluded from this logic.
3833 * We temporarily dropped the "-l" switch for fsck invocations,
3834 since they collide with the flock() logic above. util-linux
3835 upstream has been changed already to avoid this conflict,
3836 and we will readd "-l" as soon as util-linux with this
3837 change has been released.
3839 * The dependency on libattr has been removed. Since a long
3840 time, the extended attribute calls have moved to glibc, and
3841 libattr is thus unnecessary.
3843 * Virtualization detection works without privileges now. This
3844 means the systemd-detect-virt binary no longer requires
3845 CAP_SYS_PTRACE file capabilities, and our daemons can run
3846 with fewer privileges.
3848 * systemd-networkd now runs under its own "systemd-network"
3849 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
3850 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
3851 loses the ability to write to files owned by root this way.
3853 * Similarly, systemd-resolved now runs under its own
3854 "systemd-resolve" user with no capabilities remaining.
3856 * Similarly, systemd-bus-proxyd now runs under its own
3857 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
3859 * systemd-networkd gained support for setting up "veth"
3860 virtual Ethernet devices for container connectivity, as well
3861 as GRE and VTI tunnels.
3863 * systemd-networkd will no longer automatically attempt to
3864 manually load kernel modules necessary for certain tunnel
3865 transports. Instead, it is assumed the kernel loads them
3866 automatically when required. This only works correctly on
3867 very new kernels. On older kernels, please consider adding
3868 the kernel modules to /etc/modules-load.d/ as a work-around.
3870 * The resolv.conf file systemd-resolved generates has been
3871 moved to /run/systemd/resolve/. If you have a symlink from
3872 /etc/resolv.conf, it might be necessary to correct it.
3874 * Two new service settings, ProtectHome= and ProtectSystem=,
3875 have been added. When enabled, they will make the user data
3876 (such as /home) inaccessible or read-only and the system
3877 (such as /usr) read-only, for specific services. This allows
3878 very light-weight per-service sandboxing to avoid
3879 modifications of user data or system files from
3880 services. These two new switches have been enabled for all
3881 of systemd's long-running services, where appropriate.
3883 * Socket units gained new SocketUser= and SocketGroup=
3884 settings to set the owner user and group of AF_UNIX sockets
3885 and FIFOs in the file system.
3887 * Socket units gained a new RemoveOnStop= setting. If enabled,
3888 all FIFOS and sockets in the file system will be removed
3889 when the specific socket unit is stopped.
3891 * Socket units gained a new Symlinks= setting. It takes a list
3892 of symlinks to create to file system sockets or FIFOs
3893 created by the specific Unix sockets. This is useful to
3894 manage symlinks to socket nodes with the same life-cycle as
3897 * The /dev/log socket and /dev/initctl FIFO have been moved to
3898 /run, and have been replaced by symlinks. This allows
3899 connecting to these facilities even if PrivateDevices=yes is
3900 used for a service (which makes /dev/log itself unavailable,
3901 but /run is left). This also has the benefit of ensuring
3902 that /dev only contains device nodes, directories and
3903 symlinks, and nothing else.
3905 * sd-daemon gained two new calls sd_pid_notify() and
3906 sd_pid_notifyf(). They are similar to sd_notify() and
3907 sd_notifyf(), but allow overriding of the source PID of
3908 notification messages if permissions permit this. This is
3909 useful to send notify messages on behalf of a different
3910 process (for example, the parent process). The
3911 systemd-notify tool has been updated to make use of this
3912 when sending messages (so that notification messages now
3913 originate from the shell script invoking systemd-notify and
3914 not the systemd-notify process itself. This should minimize
3915 a race where systemd fails to associate notification
3916 messages to services when the originating process already
3919 * A new "on-abnormal" setting for Restart= has been added. If
3920 set, it will result in automatic restarts on all "abnormal"
3921 reasons for a process to exit, which includes unclean
3922 signals, core dumps, timeouts and watchdog timeouts, but
3923 does not include clean and unclean exit codes or clean
3924 signals. Restart=on-abnormal is an alternative for
3925 Restart=on-failure for services that shall be able to
3926 terminate and avoid restarts on certain errors, by
3927 indicating so with an unclean exit code. Restart=on-failure
3928 or Restart=on-abnormal is now the recommended setting for
3929 all long-running services.
3931 * If the InaccessibleDirectories= service setting points to a
3932 mount point (or if there are any submounts contained within
3933 it), it is now attempted to completely unmount it, to make
3934 the file systems truly unavailable for the respective
3937 * The ReadOnlyDirectories= service setting and
3938 systemd-nspawn's --read-only parameter are now recursively
3939 applied to all submounts, too.
3941 * Mount units may now be created transiently via the bus APIs.
3943 * The support for SysV and LSB init scripts has been removed
3944 from the systemd daemon itself. Instead, it is now
3945 implemented as a generator that creates native systemd units
3946 from these scripts when needed. This enables us to remove a
3947 substantial amount of legacy code from PID 1, following the
3948 fact that many distributions only ship a very small number
3949 of LSB/SysV init scripts nowadays.
3951 * Privileged Xen (dom0) domains are not considered
3952 virtualization anymore by the virtualization detection
3953 logic. After all, they generally have unrestricted access to
3954 the hardware and usually are used to manage the unprivileged
3957 * systemd-tmpfiles gained a new "C" line type, for copying
3958 files or entire directories.
3960 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
3961 lines. So far, they have been non-globbing versions of the
3962 latter, and have thus been redundant. In future, it is
3963 recommended to only use "z". "m" has hence been removed
3964 from the documentation, even though it stays supported.
3966 * A tmpfiles snippet to recreate the most basic structure in
3967 /var has been added. This is enough to create the /var/run →
3968 /run symlink and create a couple of structural
3969 directories. This allows systems to boot up with an empty or
3970 volatile /var. Of course, while with this change, the core OS
3971 now is capable with dealing with a volatile /var, not all
3972 user services are ready for it. However, we hope that sooner
3973 or later, many service daemons will be changed upstream so
3974 that they are able to automatically create their necessary
3975 directories in /var at boot, should they be missing. This is
3976 the first step to allow state-less systems that only require
3977 the vendor image for /usr to boot.
3979 * systemd-nspawn has gained a new --tmpfs= switch to mount an
3980 empty tmpfs instance to a specific directory. This is
3981 particularly useful for making use of the automatic
3982 reconstruction of /var (see above), by passing --tmpfs=/var.
3984 * Access modes specified in tmpfiles snippets may now be
3985 prefixed with "~", which indicates that they shall be masked
3986 by whether the existing file or directory is currently
3987 writable, readable or executable at all. Also, if specified,
3988 the sgid/suid/sticky bits will be masked for all
3991 * A new passive target unit "network-pre.target" has been
3992 added which is useful for services that shall run before any
3993 network is configured, for example firewall scripts.
3995 * The "floppy" group that previously owned the /dev/fd*
3996 devices is no longer used. The "disk" group is now used
3997 instead. Distributions should probably deprecate usage of
4000 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
4001 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
4002 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
4003 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
4004 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
4005 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
4008 — Berlin, 2014-06-11
4012 * A new "systemd-timesyncd" daemon has been added for
4013 synchronizing the system clock across the network. It
4014 implements an SNTP client. In contrast to NTP
4015 implementations such as chrony or the NTP reference server,
4016 this only implements a client side, and does not bother with
4017 the full NTP complexity, focusing only on querying time from
4018 one remote server and synchronizing the local clock to
4019 it. Unless you intend to serve NTP to networked clients or
4020 want to connect to local hardware clocks, this simple NTP
4021 client should be more than appropriate for most
4022 installations. The daemon runs with minimal privileges, and
4023 has been hooked up with networkd to only operate when
4024 network connectivity is available. The daemon saves the
4025 current clock to disk every time a new NTP sync has been
4026 acquired, and uses this to possibly correct the system clock
4027 early at bootup, in order to accommodate for systems that
4028 lack an RTC such as the Raspberry Pi and embedded devices,
4029 and to make sure that time monotonically progresses on these
4030 systems, even if it is not always correct. To make use of
4031 this daemon, a new system user and group "systemd-timesync"
4032 needs to be created on installation of systemd.
4034 * The queue "seqnum" interface of libudev has been disabled, as
4035 it was generally incompatible with device namespacing as
4036 sequence numbers of devices go "missing" if the devices are
4037 part of a different namespace.
4039 * "systemctl list-timers" and "systemctl list-sockets" gained
4040 a --recursive switch for showing units of these types also
4041 for all local containers, similar in style to the already
4042 supported --recursive switch for "systemctl list-units".
4044 * A new RebootArgument= setting has been added for service
4045 units, which may be used to specify a kernel reboot argument
4046 to use when triggering reboots with StartLimitAction=.
4048 * A new FailureAction= setting has been added for service
4049 units which may be used to specify an operation to trigger
4050 when a service fails. This works similarly to
4051 StartLimitAction=, but unlike it, controls what is done
4052 immediately rather than only after several attempts to
4053 restart the service in question.
4055 * hostnamed got updated to also expose the kernel name,
4056 release, and version on the bus. This is useful for
4057 executing commands like hostnamectl with the -H switch.
4058 systemd-analyze makes use of this to properly display
4059 details when running non-locally.
4061 * The bootchart tool can now show cgroup information in the
4062 graphs it generates.
4064 * The CFS CPU quota cgroup attribute is now exposed for
4065 services. The new CPUQuota= switch has been added for this
4066 which takes a percentage value. Setting this will have the
4067 result that a service may never get more CPU time than the
4068 specified percentage, even if the machine is otherwise idle.
4070 * systemd-networkd learned IPIP and SIT tunnel support.
4072 * LSB init scripts exposing a dependency on $network will now
4073 get a dependency on network-online.target rather than simply
4074 network.target. This should bring LSB handling closer to
4075 what it was on SysV systems.
4077 * A new fsck.repair= kernel option has been added to control
4078 how fsck shall deal with unclean file systems at boot.
4080 * The (.ini) configuration file parser will now silently
4081 ignore sections whose name begins with "X-". This may be
4082 used to maintain application-specific extension sections in unit
4085 * machined gained a new API to query the IP addresses of
4086 registered containers. "machinectl status" has been updated
4087 to show these addresses in its output.
4089 * A new call sd_uid_get_display() has been added to the
4090 sd-login APIs for querying the "primary" session of a
4091 user. The "primary" session of the user is elected from the
4092 user's sessions and generally a graphical session is
4093 preferred over a text one.
4095 * A minimal systemd-resolved daemon has been added. It
4096 currently simply acts as a companion to systemd-networkd and
4097 manages resolv.conf based on per-interface DNS
4098 configuration, possibly supplied via DHCP. In the long run
4099 we hope to extend this into a local DNSSEC enabled DNS and
4102 * The systemd-networkd-wait-online tool is now enabled by
4103 default. It will delay network-online.target until a network
4104 connection has been configured. The tool primarily integrates
4105 with networkd, but will also make a best effort to make sense
4106 of network configuration performed in some other way.
4108 * Two new service options StartupCPUShares= and
4109 StartupBlockIOWeight= have been added that work similarly to
4110 CPUShares= and BlockIOWeight= however only apply during
4111 system startup. This is useful to prioritize certain services
4112 differently during bootup than during normal runtime.
4114 * hostnamed has been changed to prefer the statically
4115 configured hostname in /etc/hostname (unless set to
4116 'localhost' or empty) over any dynamic one supplied by
4117 dhcp. With this change, the rules for picking the hostname
4118 match more closely the rules of other configuration settings
4119 where the local administrator's configuration in /etc always
4120 overrides any other settings.
4122 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
4123 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
4124 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
4125 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
4126 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
4127 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
4128 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
4129 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
4130 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
4131 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
4132 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
4133 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
4134 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
4135 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
4136 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
4137 Lindskog, WaLyong Cho, Will Woods, Zbigniew
4140 — Beijing, 2014-05-28
4144 * When restoring the screen brightness at boot, stay away from
4145 the darkest setting or from the lowest 5% of the available
4146 range, depending on which is the larger value of both. This
4147 should effectively protect the user from rebooting into a
4148 black screen, should the brightness have been set to minimum
4151 * sd-login gained a new sd_machine_get_class() call to
4152 determine the class ("vm" or "container") of a machine
4153 registered with machined.
4155 * sd-login gained new calls
4156 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
4157 to query the identity of the peer of a local AF_UNIX
4158 connection. They operate similarly to their sd_pid_get_xyz()
4161 * PID 1 will now maintain a system-wide system state engine
4162 with the states "starting", "running", "degraded",
4163 "maintenance", "stopping". These states are bound to system
4164 startup, normal runtime, runtime with at least one failed
4165 service, rescue/emergency mode and system shutdown. This
4166 state is shown in the "systemctl status" output when no unit
4167 name is passed. It is useful to determine system state, in
4168 particularly when doing so for many systems or containers at
4171 * A new command "list-machines" has been added to "systemctl"
4172 that lists all local OS containers and shows their system
4173 state (see above), if systemd runs inside of them.
4175 * systemctl gained a new "-r" switch to recursively enumerate
4176 units on all local containers, when used with the
4177 "list-unit" command (which is the default one that is
4178 executed when no parameters are specified).
4180 * The GPT automatic partition discovery logic will now honour
4181 two GPT partition flags: one may be set on a partition to
4182 cause it to be mounted read-only, and the other may be set
4183 on a partition to ignore it during automatic discovery.
4185 * Two new GPT type UUIDs have been added for automatic root
4186 partition discovery, for 32-bit and 64-bit ARM. This is not
4187 particularly useful for discovering the root directory on
4188 these architectures during bare-metal boots (since UEFI is
4189 not common there), but still very useful to allow booting of
4190 ARM disk images in nspawn with the -i option.
4192 * MAC addresses of interfaces created with nspawn's
4193 --network-interface= switch will now be generated from the
4194 machine name, and thus be stable between multiple invocations
4197 * logind will now automatically remove all IPC objects owned
4198 by a user if she or he fully logs out. This makes sure that
4199 users who are logged out cannot continue to consume IPC
4200 resources. This covers SysV memory, semaphores and message
4201 queues as well as POSIX shared memory and message
4202 queues. Traditionally, SysV and POSIX IPC had no life-cycle
4203 limits. With this functionality, that is corrected. This may
4204 be turned off by using the RemoveIPC= switch of logind.conf.
4206 * The systemd-machine-id-setup and tmpfiles tools gained a
4207 --root= switch to operate on a specific root directory,
4210 * journald can now forward logged messages to the TTYs of all
4211 logged in users ("wall"). This is the default for all
4212 emergency messages now.
4214 * A new tool systemd-journal-remote has been added to stream
4215 journal log messages across the network.
4217 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
4218 controller trees are mounted into it. Note that the
4219 directories mounted beneath it are not read-only. This is a
4220 security measure and is particularly useful because glibc
4221 actually includes a search logic to pick any tmpfs it can
4222 find to implement shm_open() if /dev/shm is not available
4223 (which it might very well be in namespaced setups).
4225 * machinectl gained a new "poweroff" command to cleanly power
4226 down a local OS container.
4228 * The PrivateDevices= unit file setting will now also drop the
4229 CAP_MKNOD capability from the capability bound set, and
4230 imply DevicePolicy=closed.
4232 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
4233 comprehensively on all long-running systemd services where
4234 this is appropriate.
4236 * systemd-udevd will now run in a disassociated mount
4237 namespace. To mount directories from udev rules, make sure to
4238 pull in mount units via SYSTEMD_WANTS properties.
4240 * The kdbus support gained support for uploading policy into
4241 the kernel. sd-bus gained support for creating "monitoring"
4242 connections that can eavesdrop into all bus communication
4243 for debugging purposes.
4245 * Timestamps may now be specified in seconds since the UNIX
4246 epoch Jan 1st, 1970 by specifying "@" followed by the value
4249 * Native tcpwrap support in systemd has been removed. tcpwrap
4250 is old code, not really maintained anymore and has serious
4251 shortcomings, and better options such as firewalls
4252 exist. For setups that require tcpwrap usage, please
4253 consider invoking your socket-activated service via tcpd,
4254 like on traditional inetd.
4256 * A new system.conf configuration option
4257 DefaultTimerAccuracySec= has been added that controls the
4258 default AccuracySec= setting of .timer units.
4260 * Timer units gained a new WakeSystem= switch. If enabled,
4261 timers configured this way will cause the system to resume
4262 from system suspend (if the system supports that, which most
4265 * Timer units gained a new Persistent= switch. If enabled,
4266 timers configured this way will save to disk when they have
4267 been last triggered. This information is then used on next
4268 reboot to possible execute overdue timer events, that
4269 could not take place because the system was powered off.
4270 This enables simple anacron-like behaviour for timer units.
4272 * systemctl's "list-timers" will now also list the time a
4273 timer unit was last triggered in addition to the next time
4274 it will be triggered.
4276 * systemd-networkd will now assign predictable IPv4LL
4277 addresses to its local interfaces.
4279 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
4280 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
4281 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
4282 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
4283 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
4284 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
4285 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
4286 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
4289 — Berlin, 2014-03-25
4293 * A new unit file setting RestrictAddressFamilies= has been
4294 added to restrict which socket address families unit
4295 processes gain access to. This takes address family names
4296 like "AF_INET" or "AF_UNIX", and is useful to minimize the
4297 attack surface of services via exotic protocol stacks. This
4298 is built on seccomp system call filters.
4300 * Two new unit file settings RuntimeDirectory= and
4301 RuntimeDirectoryMode= have been added that may be used to
4302 manage a per-daemon runtime directories below /run. This is
4303 an alternative for setting up directory permissions with
4304 tmpfiles snippets, and has the advantage that the runtime
4305 directory's lifetime is bound to the daemon runtime and that
4306 the daemon starts up with an empty directory each time. This
4307 is particularly useful when writing services that drop
4308 privileges using the User= or Group= setting.
4310 * The DeviceAllow= unit setting now supports globbing for
4311 matching against device group names.
4313 * The systemd configuration file system.conf gained new
4314 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
4315 DefaultMemoryAccounting= to globally turn on/off accounting
4316 for specific resources (cgroups) for all units. These
4317 settings may still be overridden individually in each unit
4320 * systemd-gpt-auto-generator is now able to discover /srv and
4321 root partitions in addition to /home and swap partitions. It
4322 also supports LUKS-encrypted partitions now. With this in
4323 place, automatic discovery of partitions to mount following
4324 the Discoverable Partitions Specification
4325 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
4326 is now a lot more complete. This allows booting without
4327 /etc/fstab and without root= on the kernel command line on
4328 systems prepared appropriately.
4330 * systemd-nspawn gained a new --image= switch which allows
4331 booting up disk images and Linux installations on any block
4332 device that follow the Discoverable Partitions Specification
4333 (see above). This means that installations made with
4334 appropriately updated installers may now be started and
4335 deployed using container managers, completely
4336 unmodified. (We hope that libvirt-lxc will add support for
4337 this feature soon, too.)
4339 * systemd-nspawn gained a new --network-macvlan= setting to
4340 set up a private macvlan interface for the
4341 container. Similarly, systemd-networkd gained a new
4342 Kind=macvlan setting in .netdev files.
4344 * systemd-networkd now supports configuring local addresses
4347 * A new tool systemd-network-wait-online has been added to
4348 synchronously wait for network connectivity using
4351 * The sd-bus.h bus API gained a new sd_bus_track object for
4352 tracking the life-cycle of bus peers. Note that sd-bus.h is
4353 still not a public API though (unless you specify
4354 --enable-kdbus on the configure command line, which however
4355 voids your warranty and you get no API stability guarantee).
4357 * The $XDG_RUNTIME_DIR runtime directories for each user are
4358 now individual tmpfs instances, which has the benefit of
4359 introducing separate pools for each user, with individual
4360 size limits, and thus making sure that unprivileged clients
4361 can no longer negatively impact the system or other users by
4362 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
4363 RuntimeDirectorySize= has been introduced that allows
4364 controlling the default size limit for all users. It
4365 defaults to 10% of the available physical memory. This is no
4366 replacement for quotas on tmpfs though (which the kernel
4367 still does not support), as /dev/shm and /tmp are still
4368 shared resources used by both the system and unprivileged
4371 * logind will now automatically turn off automatic suspending
4372 on laptop lid close when more than one display is
4373 connected. This was previously expected to be implemented
4374 individually in desktop environments (such as GNOME),
4375 however has been added to logind now, in order to fix a
4376 boot-time race where a desktop environment might not have
4377 been started yet and thus not been able to take an inhibitor
4378 lock at the time where logind already suspends the system
4379 due to a closed lid.
4381 * logind will now wait at least 30s after each system
4382 suspend/resume cycle, and 3min after system boot before
4383 suspending the system due to a closed laptop lid. This
4384 should give USB docking stations and similar enough time to
4385 be probed and configured after system resume and boot in
4386 order to then act as suspend blocker.
4388 * systemd-run gained a new --property= setting which allows
4389 initialization of resource control properties (and others)
4390 for the created scope or service unit. Example: "systemd-run
4391 --property=BlockIOWeight=10 updatedb" may be used to run
4392 updatedb at a low block IO scheduling weight.
4394 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
4395 now also work in --scope mode.
4397 * When systemd is compiled with kdbus support, basic support
4398 for enforced policies is now in place. (Note that enabling
4399 kdbus still voids your warranty and no API compatibility
4402 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
4403 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4404 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
4405 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
4406 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
4407 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
4408 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
4409 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
4410 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
4411 Zbigniew Jędrzejewski-Szmek
4413 — Berlin, 2014-03-12
4417 * systemd will now relabel /dev after loading the SMACK policy
4418 according to SMACK rules.
4420 * A new unit file option AppArmorProfile= has been added to
4421 set the AppArmor profile for the processes of a unit.
4423 * A new condition check ConditionArchitecture= has been added
4424 to conditionalize units based on the system architecture, as
4425 reported by uname()'s "machine" field.
4427 * systemd-networkd now supports matching on the system
4428 virtualization, architecture, kernel command line, host name
4431 * logind is now a lot more aggressive when suspending the
4432 machine due to a closed laptop lid. Instead of acting only
4433 on the lid close action, it will continuously watch the lid
4434 status and act on it. This is useful for laptops where the
4435 power button is on the outside of the chassis so that it can
4436 be reached without opening the lid (such as the Lenovo
4437 Yoga). On those machines, logind will now immediately
4438 re-suspend the machine if the power button has been
4439 accidentally pressed while the laptop was suspended and in a
4440 backpack or similar.
4442 * logind will now watch SW_DOCK switches and inhibit reaction
4443 to the lid switch if it is pressed. This means that logind
4444 will not suspend the machine anymore if the lid is closed
4445 and the system is docked, if the laptop supports SW_DOCK
4446 notifications via the input layer. Note that ACPI docking
4447 stations do not generate this currently. Also note that this
4448 logic is usually not fully sufficient and Desktop
4449 Environments should take a lid switch inhibitor lock when an
4450 external display is connected, as systemd will not watch
4453 * nspawn will now make use of the devices cgroup controller by
4454 default, and only permit creation of and access to the usual
4455 API device nodes like /dev/null or /dev/random, as well as
4456 access to (but not creation of) the pty devices.
4458 * We will now ship a default .network file for
4459 systemd-networkd that automatically configures DHCP for
4460 network interfaces created by nspawn's --network-veth or
4461 --network-bridge= switches.
4463 * systemd will now understand the usual M, K, G, T suffixes
4464 according to SI conventions (i.e. to the base 1000) when
4465 referring to throughput and hardware metrics. It will stay
4466 with IEC conventions (i.e. to the base 1024) for software
4467 metrics, according to what is customary according to
4468 Wikipedia. We explicitly document which base applies for
4469 each configuration option.
4471 * The DeviceAllow= setting in unit files now supports a syntax
4472 to whitelist an entire group of devices node majors at once,
4473 based on the /proc/devices listing. For example, with the
4474 string "char-pts", it is now possible to whitelist all
4475 current and future pseudo-TTYs at once.
4477 * sd-event learned a new "post" event source. Event sources of
4478 this type are triggered by the dispatching of any event
4479 source of a type that is not "post". This is useful for
4480 implementing clean-up and check event sources that are
4481 triggered by other work being done in the program.
4483 * systemd-networkd is no longer statically enabled, but uses
4484 the usual [Install] sections so that it can be
4485 enabled/disabled using systemctl. It still is enabled by
4488 * When creating a veth interface pair with systemd-nspawn, the
4489 host side will now be prefixed with "vb-" if
4490 --network-bridge= is used, and with "ve-" if --network-veth
4491 is used. This way, it is easy to distinguish these cases on
4492 the host, for example to apply different configuration to
4493 them with systemd-networkd.
4495 * The compatibility libraries for libsystemd-journal.so,
4496 libsystem-id128.so, libsystemd-login.so and
4497 libsystemd-daemon.so do not make use of IFUNC
4498 anymore. Instead, we now build libsystemd.so multiple times
4499 under these alternative names. This means that the footprint
4500 is drastically increased, but given that these are
4501 transitional compatibility libraries, this should not matter
4502 much. This change has been made necessary to support the ARM
4503 platform for these compatibility libraries, as the ARM
4504 toolchain is not really at the same level as the toolchain
4505 for other architectures like x86 and does not support
4506 IFUNC. Please make sure to use --enable-compat-libs only
4507 during a transitional period!
4509 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
4510 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4511 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
4512 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
4513 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
4514 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
4515 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
4516 Zbigniew Jędrzejewski-Szmek
4518 — Berlin, 2014-02-24
4522 * A new component "systemd-networkd" has been added that can
4523 be used to configure local network interfaces statically or
4524 via DHCP. It is capable of bringing up bridges, VLANs, and
4525 bonding. Currently, no hook-ups for interactive network
4526 configuration are provided. Use this for your initrd,
4527 container, embedded, or server setup if you need a simple,
4528 yet powerful, network configuration solution. This
4529 configuration subsystem is quite nifty, as it allows wildcard
4530 hotplug matching in interfaces. For example, with a single
4531 configuration snippet, you can configure that all Ethernet
4532 interfaces showing up are automatically added to a bridge,
4533 or similar. It supports link-sensing and more.
4535 * A new tool "systemd-socket-proxyd" has been added which can
4536 act as a bidirectional proxy for TCP sockets. This is
4537 useful for adding socket activation support to services that
4538 do not actually support socket activation, including virtual
4539 machines and the like.
4541 * Add a new tool to save/restore rfkill state on
4544 * Save/restore state of keyboard backlights in addition to
4545 display backlights on shutdown/boot.
4547 * udev learned a new SECLABEL{} construct to label device
4548 nodes with a specific security label when they appear. For
4549 now, only SECLABEL{selinux} is supported, but the syntax is
4550 prepared for additional security frameworks.
4552 * udev gained a new scheme to configure link-level attributes
4553 from files in /etc/systemd/network/*.link. These files can
4554 match against MAC address, device path, driver name and type,
4555 and will apply attributes like the naming policy, link speed,
4556 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
4557 address assignment policy (randomized, ...).
4559 * The configuration of network interface naming rules for
4560 "permanent interface names" has changed: a new NamePolicy=
4561 setting in the [Link] section of .link files determines the
4562 priority of possible naming schemes (onboard, slot, MAC,
4563 path). The default value of this setting is determined by
4564 /usr/lib/net/links/99-default.link. Old
4565 80-net-name-slot.rules udev configuration file has been
4566 removed, so local configuration overriding this file should
4567 be adapted to override 99-default.link instead.
4569 * When the User= switch is used in a unit file, also
4570 initialize $SHELL= based on the user database entry.
4572 * systemd no longer depends on libdbus. All communication is
4573 now done with sd-bus, systemd's low-level bus library
4576 * kdbus support has been added to PID 1 itself. When kdbus is
4577 enabled, this causes PID 1 to set up the system bus and
4578 enable support for a new ".busname" unit type that
4579 encapsulates bus name activation on kdbus. It works a little
4580 bit like ".socket" units, except for bus names. A new
4581 generator has been added that converts classic dbus1 service
4582 activation files automatically into native systemd .busname
4585 * sd-bus: add a light-weight vtable implementation that allows
4586 defining objects on the bus with a simple static const
4587 vtable array of its methods, signals and properties.
4589 * systemd will not generate or install static dbus
4590 introspection data anymore to /usr/share/dbus-1/interfaces,
4591 as the precise format of these files is unclear, and
4592 nothing makes use of it.
4594 * A proxy daemon is now provided to proxy clients connecting
4595 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
4596 compatibility with classic D-Bus.
4598 * A bus driver implementation has been added that supports the
4599 classic D-Bus bus driver calls on kdbus, also for
4600 compatibility purposes.
4602 * A new API "sd-event.h" has been added that implements a
4603 minimal event loop API built around epoll. It provides a
4604 couple of features that direct epoll usage is lacking:
4605 prioritization of events, scales to large numbers of timer
4606 events, per-event timer slack (accuracy), system-wide
4607 coalescing of timer events, exit handlers, watchdog
4608 supervision support using systemd's sd_notify() API, child
4611 * A new API "sd-rntl.h" has been added that provides an API
4612 around the route netlink interface of the kernel, similar in
4613 style to "sd-bus.h".
4615 * A new API "sd-dhcp-client.h" has been added that provides a
4616 small DHCPv4 client-side implementation. This is used by
4619 * There is a new kernel command line option
4620 "systemd.restore_state=0|1". When set to "0", none of the
4621 systemd tools will restore saved runtime state to hardware
4622 devices. More specifically, the rfkill and backlight states
4625 * The FsckPassNo= compatibility option in mount/service units
4626 has been removed. The fstab generator will now add the
4627 necessary dependencies automatically, and does not require
4628 PID1's support for that anymore.
4630 * journalctl gained a new switch, --list-boots, that lists
4631 recent boots with their times and boot IDs.
4633 * The various tools like systemctl, loginctl, timedatectl,
4634 busctl, systemd-run, ... have gained a new switch "-M" to
4635 connect to a specific, local OS container (as direct
4636 connection, without requiring SSH). This works on any
4637 container that is registered with machined, such as those
4638 created by libvirt-lxc or nspawn.
4640 * systemd-run and systemd-analyze also gained support for "-H"
4641 to connect to remote hosts via SSH. This is particularly
4642 useful for systemd-run because it enables queuing of jobs
4643 onto remote systems.
4645 * machinectl gained a new command "login" to open a getty
4646 login in any local container. This works with any container
4647 that is registered with machined (such as those created by
4648 libvirt-lxc or nspawn), and which runs systemd inside.
4650 * machinectl gained a new "reboot" command that may be used to
4651 trigger a reboot on a specific container that is registered
4652 with machined. This works on any container that runs an init
4653 system of some kind.
4655 * systemctl gained a new "list-timers" command to print a nice
4656 listing of installed timer units with the times they elapse
4659 * Alternative reboot() parameters may now be specified on the
4660 "systemctl reboot" command line and are passed to the
4661 reboot() system call.
4663 * systemctl gained a new --job-mode= switch to configure the
4664 mode to queue a job with. This is a more generic version of
4665 --fail, --irreversible, and --ignore-dependencies, which are
4666 still available but not advertised anymore.
4668 * /etc/systemd/system.conf gained new settings to configure
4669 various default timeouts of units, as well as the default
4670 start limit interval and burst. These may still be overridden
4673 * PID1 will now export on the bus profile data of the security
4674 policy upload process (such as the SELinux policy upload to
4677 * journald: when forwarding logs to the console, include
4678 timestamps (following the setting in
4679 /sys/module/printk/parameters/time).
4681 * OnCalendar= in timer units now understands the special
4682 strings "yearly" and "annually". (Both are equivalent)
4684 * The accuracy of timer units is now configurable with the new
4685 AccuracySec= setting. It defaults to 1min.
4687 * A new dependency type JoinsNamespaceOf= has been added that
4688 allows running two services within the same /tmp and network
4689 namespace, if PrivateNetwork= or PrivateTmp= are used.
4691 * A new command "cat" has been added to systemctl. It outputs
4692 the original unit file of a unit, and concatenates the
4693 contents of additional "drop-in" unit file snippets, so that
4694 the full configuration is shown.
4696 * systemctl now supports globbing on the various "list-xyz"
4697 commands, like "list-units" or "list-sockets", as well as on
4698 those commands which take multiple unit names.
4700 * journalctl's --unit= switch gained support for globbing.
4702 * All systemd daemons now make use of the watchdog logic so
4703 that systemd automatically notices when they hang.
4705 * If the $container_ttys environment variable is set,
4706 getty-generator will automatically spawn a getty for each
4707 listed tty. This is useful for container managers to request
4708 login gettys to be spawned on as many ttys as needed.
4710 * %h, %s, %U specifier support is not available anymore when
4711 used in unit files for PID 1. This is because NSS calls are
4712 not safe from PID 1. They stay available for --user
4713 instances of systemd, and as special case for the root user.
4715 * loginctl gained a new "--no-legend" switch to turn off output
4718 * The "sd-login.h" API gained three new calls:
4719 sd_session_is_remote(), sd_session_get_remote_user(),
4720 sd_session_get_remote_host() to query information about
4723 * The udev hardware database now also carries vendor/product
4724 information of SDIO devices.
4726 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
4727 determine whether watchdog notifications are requested by
4730 * Socket-activated per-connection services now include a
4731 short description of the connection parameters in the
4734 * tmpfiles gained a new "--boot" option. When this is not used,
4735 only lines where the command character is not suffixed with
4736 "!" are executed. When this option is specified, those
4737 options are executed too. This partitions tmpfiles
4738 directives into those that can be safely executed at any
4739 time, and those which should be run only at boot (for
4740 example, a line that creates /run/nologin).
4742 * A new API "sd-resolve.h" has been added which provides a simple
4743 asynchronous wrapper around glibc NSS host name resolution
4744 calls, such as getaddrinfo(). In contrast to glibc's
4745 getaddrinfo_a(), it does not use signals. In contrast to most
4746 other asynchronous name resolution libraries, this one does
4747 not reimplement DNS, but reuses NSS, so that alternate
4748 host name resolution systems continue to work, such as mDNS,
4749 LDAP, etc. This API is based on libasyncns, but it has been
4750 cleaned up for inclusion in systemd.
4752 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
4753 "sd-daemon.h" are no longer found in individual libraries
4754 libsystemd-journal.so, libsystemd-login.so,
4755 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
4756 merged them into a single library, libsystemd.so, which
4757 provides all symbols. The reason for this is cyclic
4758 dependencies, as these libraries tend to use each other's
4759 symbols. So far, we have managed to workaround that by linking
4760 a copy of a good part of our code into each of these
4761 libraries again and again, which, however, makes certain
4762 things hard to do, like sharing static variables. Also, it
4763 substantially increases footprint. With this change, there
4764 is only one library for the basic APIs systemd
4765 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
4766 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
4767 library as well, however are subject to the --enable-kdbus
4768 switch (see below). Note that "sd-dhcp-client.h" is not part
4769 of this library (this is because it only consumes, never
4770 provides, services of/to other APIs). To make the transition
4771 easy from the separate libraries to the unified one, we
4772 provide the --enable-compat-libs compile-time switch which
4773 will generate stub libraries that are compatible with the
4774 old ones but redirect all calls to the new one.
4776 * All of the kdbus logic and the new APIs "sd-bus.h",
4777 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
4778 and "sd-utf8.h" are compile-time optional via the
4779 "--enable-kdbus" switch, and they are not compiled in by
4780 default. To make use of kdbus, you have to explicitly enable
4781 the switch. Note however, that neither the kernel nor the
4782 userspace API for all of this is considered stable yet. We
4783 want to maintain the freedom to still change the APIs for
4784 now. By specifying this build-time switch, you acknowledge
4785 that you are aware of the instability of the current
4788 * Also, note that while kdbus is pretty much complete,
4789 it lacks one thing: proper policy support. This means you
4790 can build a fully working system with all features; however,
4791 it will be highly insecure. Policy support will be added in
4792 one of the next releases, at the same time that we will
4793 declare the APIs stable.
4795 * When the kernel command line argument "kdbus" is specified,
4796 systemd will automatically load the kdbus.ko kernel module. At
4797 this stage of development, it is only useful for testing kdbus
4798 and should not be used in production. Note: if "--enable-kdbus"
4799 is specified, and the kdbus.ko kernel module is available, and
4800 "kdbus" is added to the kernel command line, the entire system
4801 runs with kdbus instead of dbus-daemon, with the above mentioned
4802 problem of missing the system policy enforcement. Also a future
4803 version of kdbus.ko or a newer systemd will not be compatible with
4804 each other, and will unlikely be able to boot the machine if only
4805 one of them is updated.
4807 * systemctl gained a new "import-environment" command which
4808 uploads the caller's environment (or parts thereof) into the
4809 service manager so that it is inherited by services started
4810 by the manager. This is useful to upload variables like
4811 $DISPLAY into the user service manager.
4813 * A new PrivateDevices= switch has been added to service units
4814 which allows running a service with a namespaced /dev
4815 directory that does not contain any device nodes for
4816 physical devices. More specifically, it only includes devices
4817 such as /dev/null, /dev/urandom, and /dev/zero which are API
4820 * logind has been extended to support behaviour like VT
4821 switching on seats that do not support a VT. This makes
4822 multi-session available on seats that are not the first seat
4823 (seat0), and on systems where kernel support for VTs has
4824 been disabled at compile-time.
4826 * If a process holds a delay lock for system sleep or shutdown
4827 and fails to release it in time, we will now log its
4828 identity. This makes it easier to identify processes that
4829 cause slow suspends or power-offs.
4831 * When parsing /etc/crypttab, support for a new key-slot=
4832 option as supported by Debian is added. It allows indicating
4833 which LUKS slot to use on disk, speeding up key loading.
4835 * The sd_journald_sendv() API call has been checked and
4836 officially declared to be async-signal-safe so that it may
4837 be invoked from signal handlers for logging purposes.
4839 * Boot-time status output is now enabled automatically after a
4840 short timeout if boot does not progress, in order to give
4841 the user an indication what she or he is waiting for.
4843 * The boot-time output has been improved to show how much time
4844 remains until jobs expire.
4846 * The KillMode= switch in service units gained a new possible
4847 value "mixed". If set, and the unit is shut down, then the
4848 initial SIGTERM signal is sent only to the main daemon
4849 process, while the following SIGKILL signal is sent to
4850 all remaining processes of the service.
4852 * When a scope unit is registered, a new property "Controller"
4853 may be set. If set to a valid bus name, systemd will send a
4854 RequestStop() signal to this name when it would like to shut
4855 down the scope. This may be used to hook manager logic into
4856 the shutdown logic of scope units. Also, scope units may now
4857 be put in a special "abandoned" state, in which case the
4858 manager process which created them takes no further
4859 responsibilities for it.
4861 * When reading unit files, systemd will now verify
4862 the access mode of these files, and warn about certain
4863 suspicious combinations. This has been added to make it
4864 easier to track down packaging bugs where unit files are
4865 marked executable or world-writable.
4867 * systemd-nspawn gained a new "--setenv=" switch to set
4868 container-wide environment variables. The similar option in
4869 systemd-activate was renamed from "--environment=" to
4870 "--setenv=" for consistency.
4872 * systemd-nspawn has been updated to create a new kdbus domain
4873 for each container that is invoked, thus allowing each
4874 container to have its own set of system and user buses,
4875 independent of the host.
4877 * systemd-nspawn gained a new --drop-capability= switch to run
4878 the container with less capabilities than the default. Both
4879 --drop-capability= and --capability= now take the special
4880 string "all" for dropping or keeping all capabilities.
4882 * systemd-nspawn gained new switches for executing containers
4883 with specific SELinux labels set.
4885 * systemd-nspawn gained a new --quiet switch to not generate
4886 any additional output but the container's own console
4889 * systemd-nspawn gained a new --share-system switch to run a
4890 container without PID namespacing enabled.
4892 * systemd-nspawn gained a new --register= switch to control
4893 whether the container is registered with systemd-machined or
4894 not. This is useful for containers that do not run full
4895 OS images, but only specific apps.
4897 * systemd-nspawn gained a new --keep-unit which may be used
4898 when invoked as the only program from a service unit, and
4899 results in registration of the unit service itself in
4900 systemd-machined, instead of a newly opened scope unit.
4902 * systemd-nspawn gained a new --network-interface= switch for
4903 moving arbitrary interfaces to the container. The new
4904 --network-veth switch creates a virtual Ethernet connection
4905 between host and container. The new --network-bridge=
4906 switch then allows assigning the host side of this virtual
4907 Ethernet connection to a bridge device.
4909 * systemd-nspawn gained a new --personality= switch for
4910 setting the kernel personality for the container. This is
4911 useful when running a 32-bit container on a 64-bit host. A
4912 similar option Personality= is now also available for service
4915 * logind will now also track a "Desktop" identifier for each
4916 session which encodes the desktop environment of it. This is
4917 useful for desktop environments that want to identify
4918 multiple running sessions of itself easily.
4920 * A new SELinuxContext= setting for service units has been
4921 added that allows setting a specific SELinux execution
4922 context for a service.
4924 * Most systemd client tools will now honour $SYSTEMD_LESS for
4925 settings of the "less" pager. By default, these tools will
4926 override $LESS to allow certain operations to work, such as
4927 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
4928 influence this logic.
4930 * systemd's "seccomp" hook-up has been changed to make use of
4931 the libseccomp library instead of using its own
4932 implementation. This has benefits for portability among
4935 * For usage together with SystemCallFilter=, a new
4936 SystemCallErrorNumber= setting has been introduced that
4937 allows configuration of a system error number to be returned
4938 on filtered system calls, instead of immediately killing the
4939 process. Also, SystemCallArchitectures= has been added to
4940 limit access to system calls of a particular architecture
4941 (in order to turn off support for unused secondary
4942 architectures). There is also a global
4943 SystemCallArchitectures= setting in system.conf now to turn
4944 off support for non-native system calls system-wide.
4946 * systemd requires a kernel with a working name_to_handle_at(),
4947 please see the kernel config requirements in the README file.
4949 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
4950 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
4951 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
4952 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
4953 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
4954 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
4955 Elia Pinto, Florian Weimer, George McCollister, Goffredo
4956 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
4957 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
4958 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
4959 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
4960 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
4961 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
4962 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
4963 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
4964 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
4965 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
4966 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
4967 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
4968 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
4969 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
4970 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
4971 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
4972 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
4974 — Berlin, 2014-02-20
4978 * logind has gained support for facilitating privileged input
4979 and drm device access for unprivileged clients. This work is
4980 useful to allow Wayland display servers (and similar
4981 programs, such as kmscon) to run under the user's ID and
4982 access input and drm devices which are normally
4983 protected. When this is used (and the kernel is new enough)
4984 logind will "mute" IO on the file descriptors passed to
4985 Wayland as long as it is in the background and "unmute" it
4986 if it returns into the foreground. This allows secure
4987 session switching without allowing background sessions to
4988 eavesdrop on input and display data. This also introduces
4989 session switching support if VT support is turned off in the
4990 kernel, and on seats that are not seat0.
4992 * A new kernel command line option luks.options= is understood
4993 now which allows specifying LUKS options for usage for LUKS
4994 encrypted partitions specified with luks.uuid=.
4996 * tmpfiles.d(5) snippets may now use specifier expansion in
4997 path names. More specifically %m, %b, %H, %v, are now
4998 replaced by the local machine id, boot id, hostname, and
4999 kernel version number.
5001 * A new tmpfiles.d(5) command "m" has been introduced which
5002 may be used to change the owner/group/access mode of a file
5003 or directory if it exists, but do nothing if it does not.
5005 * This release removes high-level support for the
5006 MemorySoftLimit= cgroup setting. The underlying kernel
5007 cgroup attribute memory.soft_limit= is currently badly
5008 designed and likely to be removed from the kernel API in its
5009 current form, hence we should not expose it for now.
5011 * The memory.use_hierarchy cgroup attribute is now enabled for
5012 all cgroups systemd creates in the memory cgroup
5013 hierarchy. This option is likely to be come the built-in
5014 default in the kernel anyway, and the non-hierarchical mode
5015 never made much sense in the intrinsically hierarchical
5018 * A new field _SYSTEMD_SLICE= is logged along with all journal
5019 messages containing the slice a message was generated
5020 from. This is useful to allow easy per-customer filtering of
5021 logs among other things.
5023 * systemd-journald will no longer adjust the group of journal
5024 files it creates to the "systemd-journal" group. Instead we
5025 rely on the journal directory to be owned by the
5026 "systemd-journal" group, and its setgid bit set, so that the
5027 kernel file system layer will automatically enforce that
5028 journal files inherit this group assignment. The reason for
5029 this change is that we cannot allow NSS look-ups from
5030 journald which would be necessary to resolve
5031 "systemd-journal" to a numeric GID, because this might
5032 create deadlocks if NSS involves synchronous queries to
5033 other daemons (such as nscd, or sssd) which in turn are
5034 logging clients of journald and might block on it, which
5035 would then dead lock. A tmpfiles.d(5) snippet included in
5036 systemd will make sure the setgid bit and group are
5037 properly set on the journal directory if it exists on every
5038 boot. However, we recommend adjusting it manually after
5039 upgrades too (or from RPM scriptlets), so that the change is
5040 not delayed until next reboot.
5042 * Backlight and random seed files in /var/lib/ have moved into
5043 the /var/lib/systemd/ directory, in order to centralize all
5044 systemd generated files in one directory.
5046 * Boot time performance measurements (as displayed by
5047 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
5048 performance information if that's available to determine how
5049 much time BIOS and boot loader initialization required. With
5050 a sufficiently new BIOS you hence no longer need to boot
5051 with Gummiboot to get access to such information.
5053 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
5054 Cristian Rodríguez, Dave Reisner, David Herrmann, David
5055 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
5056 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
5057 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
5058 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
5059 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5061 — Berlin, 2013-10-02
5065 * The Restart= option for services now understands a new
5066 on-watchdog setting, which will restart the service
5067 automatically if the service stops sending out watchdog keep
5068 alive messages (as configured with WatchdogSec=).
5070 * The getty generator (which is responsible for bringing up a
5071 getty on configured serial consoles) will no longer only
5072 start a getty on the primary kernel console but on all
5073 others, too. This makes the order in which console= is
5074 specified on the kernel command line less important.
5076 * libsystemd-logind gained a new sd_session_get_vt() call to
5077 retrieve the VT number of a session.
5079 * If the option "tries=0" is set for an entry of /etc/crypttab
5080 its passphrase is queried indefinitely instead of any
5081 maximum number of tries.
5083 * If a service with a configure PID file terminates its PID
5084 file will now be removed automatically if it still exists
5085 afterwards. This should put an end to stale PID files.
5087 * systemd-run will now also take relative binary path names
5088 for execution and no longer insists on absolute paths.
5090 * InaccessibleDirectories= and ReadOnlyDirectories= now take
5091 paths that are optionally prefixed with "-" to indicate that
5092 it should not be considered a failure if they do not exist.
5094 * journalctl -o (and similar commands) now understands a new
5095 output mode "short-precise", it is similar to "short" but
5096 shows timestamps with usec accuracy.
5098 * The option "discard" (as known from Debian) is now
5099 synonymous to "allow-discards" in /etc/crypttab. In fact,
5100 "discard" is preferred now (since it is easier to remember
5103 * Some licensing clean-ups were made, so that more code is now
5104 LGPL-2.1 licensed than before.
5106 * A minimal tool to save/restore the display backlight
5107 brightness across reboots has been added. It will store the
5108 backlight setting as late as possible at shutdown, and
5109 restore it as early as possible during reboot.
5111 * A logic to automatically discover and enable home and swap
5112 partitions on GPT disks has been added. With this in place
5113 /etc/fstab becomes optional for many setups as systemd can
5114 discover certain partitions located on the root disk
5115 automatically. Home partitions are recognized under their
5116 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
5117 partitions are recognized under their GPT type ID
5118 0657fd6da4ab43c484e50933c84b4f4f.
5120 * systemd will no longer pass any environment from the kernel
5121 or initrd to system services. If you want to set an
5122 environment for all services, do so via the kernel command
5123 line systemd.setenv= assignment.
5125 * The systemd-sysctl tool no longer natively reads the file
5126 /etc/sysctl.conf. If desired, the file should be symlinked
5127 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
5128 legacy support by a symlink rather than built-in code, it
5129 also makes the otherwise hidden order of application of the
5130 different files visible. (Note that this partly reverts to a
5131 pre-198 application order of sysctl knobs!)
5133 * The "systemctl set-log-level" and "systemctl dump" commands
5134 have been moved to systemd-analyze.
5136 * systemd-run learned the new --remain-after-exit switch,
5137 which causes the scope unit not to be cleaned up
5138 automatically after the process terminated.
5140 * tmpfiles learned a new --exclude-prefix= switch to exclude
5141 certain paths from operation.
5143 * journald will now automatically flush all messages to disk
5144 as soon as a message at the log level CRIT, ALERT or EMERG
5147 Contributions from: Andrew Cook, Brandon Philips, Christian
5148 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
5149 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
5150 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
5151 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
5152 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
5153 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
5154 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
5155 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
5156 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
5157 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
5158 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
5159 William Giokas, Zbigniew Jędrzejewski-Szmek
5161 — Berlin, 2013-09-13
5165 * The documentation has been updated to cover the various new
5166 concepts introduced with 205.
5168 * Unit files now understand the new %v specifier which
5169 resolves to the kernel version string as returned by "uname
5172 * systemctl now supports filtering the unit list output by
5173 load state, active state and sub state, using the new
5176 * "systemctl status" will now show the results of the
5177 condition checks (like ConditionPathExists= and similar) of
5178 the last start attempts of the unit. They are also logged to
5181 * "journalctl -b" may now be used to look for boot output of a
5182 specific boot. Try "journalctl -b -1" for the previous boot,
5183 but the syntax is substantially more powerful.
5185 * "journalctl --show-cursor" has been added which prints the
5186 cursor string the last shown log line. This may then be used
5187 with the new "journalctl --after-cursor=" switch to continue
5188 browsing logs from that point on.
5190 * "journalctl --force" may now be used to force regeneration
5193 * Creation of "dead" device nodes has been moved from udev
5194 into kmod and tmpfiles. Previously, udev would read the kmod
5195 databases to pre-generate dead device nodes based on meta
5196 information contained in kernel modules, so that these would
5197 be auto-loaded on access rather then at boot. As this
5198 does not really have much to do with the exposing actual
5199 kernel devices to userspace this has always been slightly
5200 alien in the udev codebase. Following the new scheme kmod
5201 will now generate a runtime snippet for tmpfiles from the
5202 module meta information and it now is tmpfiles' job to the
5203 create the nodes. This also allows overriding access and
5204 other parameters for the nodes using the usual tmpfiles
5205 facilities. As side effect this allows us to remove the
5206 CAP_SYS_MKNOD capability bit from udevd entirely.
5208 * logind's device ACLs may now be applied to these "dead"
5209 devices nodes too, thus finally allowing managed access to
5210 devices such as /dev/snd/sequencer without loading the
5211 backing module right-away.
5213 * A new RPM macro has been added that may be used to apply
5214 tmpfiles configuration during package installation.
5216 * systemd-detect-virt and ConditionVirtualization= now can
5217 detect User-Mode-Linux machines (UML).
5219 * journald will now implicitly log the effective capabilities
5220 set of processes in the message metadata.
5222 * systemd-cryptsetup has gained support for TrueCrypt volumes.
5224 * The initrd interface has been simplified (more specifically,
5225 support for passing performance data via environment
5226 variables and fsck results via files in /run has been
5227 removed). These features were non-essential, and are
5228 nowadays available in a much nicer way by having systemd in
5229 the initrd serialize its state and have the hosts systemd
5230 deserialize it again.
5232 * The udev "keymap" data files and tools to apply keyboard
5233 specific mappings of scan to key codes, and force-release
5234 scan code lists have been entirely replaced by a udev
5235 "keyboard" builtin and a hwdb data file.
5237 * systemd will now honour the kernel's "quiet" command line
5238 argument also during late shutdown, resulting in a
5239 completely silent shutdown when used.
5241 * There's now an option to control the SO_REUSEPORT socket
5242 option in .socket units.
5244 * Instance units will now automatically get a per-template
5245 subslice of system.slice unless something else is explicitly
5246 configured. For example, instances of sshd@.service will now
5247 implicitly be placed in system-sshd.slice rather than
5248 system.slice as before.
5250 * Test coverage support may now be enabled at build time.
5252 Contributions from: Dave Reisner, Frederic Crozat, Harald
5253 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
5254 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
5255 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
5256 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
5257 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
5258 Giokas, Zbigniew Jędrzejewski-Szmek
5260 — Berlin, 2013-07-23
5264 * Two new unit types have been introduced:
5266 Scope units are very similar to service units, however, are
5267 created out of pre-existing processes — instead of PID 1
5268 forking off the processes. By using scope units it is
5269 possible for system services and applications to group their
5270 own child processes (worker processes) in a powerful way
5271 which then maybe used to organize them, or kill them
5272 together, or apply resource limits on them.
5274 Slice units may be used to partition system resources in an
5275 hierarchical fashion and then assign other units to them. By
5276 default there are now three slices: system.slice (for all
5277 system services), user.slice (for all user sessions),
5278 machine.slice (for VMs and containers).
5280 Slices and scopes have been introduced primarily in
5281 context of the work to move cgroup handling to a
5282 single-writer scheme, where only PID 1
5283 creates/removes/manages cgroups.
5285 * There's a new concept of "transient" units. In contrast to
5286 normal units these units are created via an API at runtime,
5287 not from configuration from disk. More specifically this
5288 means it is now possible to run arbitrary programs as
5289 independent services, with all execution parameters passed
5290 in via bus APIs rather than read from disk. Transient units
5291 make systemd substantially more dynamic then it ever was,
5292 and useful as a general batch manager.
5294 * logind has been updated to make use of scope and slice units
5295 for managing user sessions. As a user logs in he will get
5296 his own private slice unit, to which all sessions are added
5297 as scope units. We also added support for automatically
5298 adding an instance of user@.service for the user into the
5299 slice. Effectively logind will no longer create cgroup
5300 hierarchies on its own now, it will defer entirely to PID 1
5301 for this by means of scope, service and slice units. Since
5302 user sessions this way become entities managed by PID 1
5303 the output of "systemctl" is now a lot more comprehensive.
5305 * A new mini-daemon "systemd-machined" has been added which
5306 may be used by virtualization managers to register local
5307 VMs/containers. nspawn has been updated accordingly, and
5308 libvirt will be updated shortly. machined will collect a bit
5309 of meta information about the VMs/containers, and assign
5310 them their own scope unit (see above). The collected
5311 meta-data is then made available via the "machinectl" tool,
5312 and exposed in "ps" and similar tools. machined/machinectl
5313 is compile-time optional.
5315 * As discussed earlier, the low-level cgroup configuration
5316 options ControlGroup=, ControlGroupModify=,
5317 ControlGroupPersistent=, ControlGroupAttribute= have been
5318 removed. Please use high-level attribute settings instead as
5319 well as slice units.
5321 * A new bus call SetUnitProperties() has been added to alter
5322 various runtime parameters of a unit. This is primarily
5323 useful to alter cgroup parameters dynamically in a nice way,
5324 but will be extended later on to make more properties
5325 modifiable at runtime. systemctl gained a new set-properties
5326 command that wraps this call.
5328 * A new tool "systemd-run" has been added which can be used to
5329 run arbitrary command lines as transient services or scopes,
5330 while configuring a number of settings via the command
5331 line. This tool is currently very basic, however already
5332 very useful. We plan to extend this tool to even allow
5333 queuing of execution jobs with time triggers from the
5334 command line, similar in fashion to "at".
5336 * nspawn will now inform the user explicitly that kernels with
5337 audit enabled break containers, and suggest the user to turn
5340 * Support for detecting the IMA and AppArmor security
5341 frameworks with ConditionSecurity= has been added.
5343 * journalctl gained a new "-k" switch for showing only kernel
5344 messages, mimicking dmesg output; in addition to "--user"
5345 and "--system" switches for showing only user's own logs
5348 * systemd-delta can now show information about drop-in
5349 snippets extending unit files.
5351 * libsystemd-bus has been substantially updated but is still
5352 not available as public API.
5354 * systemd will now look for the "debug" argument on the kernel
5355 command line and enable debug logging, similar to what
5356 "systemd.log_level=debug" already did before.
5358 * "systemctl set-default", "systemctl get-default" has been
5359 added to configure the default.target symlink, which
5360 controls what to boot into by default.
5362 * "systemctl set-log-level" has been added as a convenient
5363 way to raise and lower systemd logging threshold.
5365 * "systemd-analyze plot" will now show the time the various
5366 generators needed for execution, as well as information
5367 about the unit file loading.
5369 * libsystemd-journal gained a new sd_journal_open_files() call
5370 for opening specific journal files. journactl also gained a
5371 new switch to expose this new functionality. Previously we
5372 only supported opening all files from a directory, or all
5373 files from the system, as opening individual files only is
5374 racy due to journal file rotation.
5376 * systemd gained the new DefaultEnvironment= setting in
5377 /etc/systemd/system.conf to set environment variables for
5380 * If a privileged process logs a journal message with the
5381 OBJECT_PID= field set, then journald will automatically
5382 augment this with additional OBJECT_UID=, OBJECT_GID=,
5383 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
5384 system services want to log events about specific client
5385 processes. journactl/systemctl has been updated to make use
5386 of this information if all log messages regarding a specific
5389 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
5390 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
5391 Reisner, David Coppa, David King, David Strauss, Eelco
5392 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
5393 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
5394 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
5395 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
5396 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
5397 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
5398 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
5399 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
5400 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
5401 Łukasz Stelmach, 장동준
5405 * The Python bindings gained some minimal support for the APIs
5406 exposed by libsystemd-logind.
5408 * ConditionSecurity= gained support for detecting SMACK. Since
5409 this condition already supports SELinux and AppArmor we only
5410 miss IMA for this. Patches welcome!
5412 Contributions from: Karol Lewandowski, Lennart Poettering,
5413 Zbigniew Jędrzejewski-Szmek
5417 * systemd-nspawn will now create /etc/resolv.conf if
5418 necessary, before bind-mounting the host's file onto it.
5420 * systemd-nspawn will now store meta information about a
5421 container on the container's cgroup as extended attribute
5422 fields, including the root directory.
5424 * The cgroup hierarchy has been reworked in many ways. All
5425 objects any of the components systemd creates in the cgroup
5426 tree are now suffixed. More specifically, user sessions are
5427 now placed in cgroups suffixed with ".session", users in
5428 cgroups suffixed with ".user", and nspawn containers in
5429 cgroups suffixed with ".nspawn". Furthermore, all cgroup
5430 names are now escaped in a simple scheme to avoid collision
5431 of userspace object names with kernel filenames. This work
5432 is preparation for making these objects relocatable in the
5433 cgroup tree, in order to allow easy resource partitioning of
5434 these objects without causing naming conflicts.
5436 * systemctl list-dependencies gained the new switches
5437 --plain, --reverse, --after and --before.
5439 * systemd-inhibit now shows the process name of processes that
5440 have taken an inhibitor lock.
5442 * nss-myhostname will now also resolve "localhost"
5443 implicitly. This makes /etc/hosts an optional file and
5444 nicely handles that on IPv6 ::1 maps to both "localhost" and
5447 * libsystemd-logind.so gained a new call
5448 sd_get_machine_names() to enumerate running containers and
5449 VMs (currently only supported by very new libvirt and
5450 nspawn). sd_login_monitor can now be used to watch
5451 VMs/containers coming and going.
5453 * .include is not allowed recursively anymore, and only in
5454 unit files. Usually it is better to use drop-in snippets in
5455 .d/*.conf anyway, as introduced with systemd 198.
5457 * systemd-analyze gained a new "critical-chain" command that
5458 determines the slowest chain of units run during system
5459 boot-up. It is very useful for tracking down where
5460 optimizing boot time is the most beneficial.
5462 * systemd will no longer allow manipulating service paths in
5463 the name=systemd:/system cgroup tree using ControlGroup= in
5464 units. (But is still fine with it in all other dirs.)
5466 * There's a new systemd-nspawn@.service service file that may
5467 be used to easily run nspawn containers as system
5468 services. With the container's root directory in
5469 /var/lib/container/foobar it is now sufficient to run
5470 "systemctl start systemd-nspawn@foobar.service" to boot it.
5472 * systemd-cgls gained a new parameter "--machine" to list only
5473 the processes within a certain container.
5475 * ConditionSecurity= now can check for "apparmor". We still
5476 are lacking checks for SMACK and IMA for this condition
5477 check though. Patches welcome!
5479 * A new configuration file /etc/systemd/sleep.conf has been
5480 added that may be used to configure which kernel operation
5481 systemd is supposed to execute when "suspend", "hibernate"
5482 or "hybrid-sleep" is requested. This makes the new kernel
5483 "freeze" state accessible to the user.
5485 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
5486 the passed argument if applicable.
5488 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
5489 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
5490 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
5491 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
5492 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
5493 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
5494 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
5499 * The output of 'systemctl list-jobs' got some polishing. The
5500 '--type=' argument may now be passed more than once. A new
5501 command 'systemctl list-sockets' has been added which shows
5502 a list of kernel sockets systemd is listening on with the
5503 socket units they belong to, plus the units these socket
5506 * The experimental libsystemd-bus library got substantial
5507 updates to work in conjunction with the (also experimental)
5508 kdbus kernel project. It works well enough to exchange
5509 messages with some sophistication. Note that kdbus is not
5510 ready yet, and the library is mostly an elaborate test case
5511 for now, and not installable.
5513 * systemd gained a new unit 'systemd-static-nodes.service'
5514 that generates static device nodes earlier during boot, and
5515 can run in conjunction with udev.
5517 * libsystemd-login gained a new call sd_pid_get_user_unit()
5518 to retrieve the user systemd unit a process is running
5519 in. This is useful for systems where systemd is used as
5522 * systemd-nspawn now places all containers in the new /machine
5523 top-level cgroup directory in the name=systemd
5524 hierarchy. libvirt will soon do the same, so that we get a
5525 uniform separation of /system, /user and /machine for system
5526 services, user processes and containers/virtual
5527 machines. This new cgroup hierarchy is also useful to stick
5528 stable names to specific container instances, which can be
5529 recognized later this way (this name may be controlled
5530 via systemd-nspawn's new -M switch). libsystemd-login also
5531 gained a new call sd_pid_get_machine_name() to retrieve the
5532 name of the container/VM a specific process belongs to.
5534 * bootchart can now store its data in the journal.
5536 * libsystemd-journal gained a new call
5537 sd_journal_add_conjunction() for AND expressions to the
5538 matching logic. This can be used to express more complex
5539 logical expressions.
5541 * journactl can now take multiple --unit= and --user-unit=
5544 * The cryptsetup logic now understands the "luks.key=" kernel
5545 command line switch for specifying a file to read the
5546 decryption key from. Also, if a configured key file is not
5547 found the tool will now automatically fall back to prompting
5550 * Python systemd.journal module was updated to wrap recently
5551 added functions from libsystemd-journal. The interface was
5552 changed to bring the low level interface in s.j._Reader
5553 closer to the C API, and the high level interface in
5554 s.j.Reader was updated to wrap and convert all data about
5557 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
5558 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
5559 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
5560 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
5561 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
5562 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5566 * journalctl --update-catalog now understands a new --root=
5567 option to operate on catalogs found in a different root
5570 * During shutdown after systemd has terminated all running
5571 services a final killing loop kills all remaining left-over
5572 processes. We will now print the name of these processes
5573 when we send SIGKILL to them, since this usually indicates a
5576 * If /etc/crypttab refers to password files stored on
5577 configured mount points automatic dependencies will now be
5578 generated to ensure the specific mount is established first
5579 before the key file is attempted to be read.
5581 * 'systemctl status' will now show information about the
5582 network sockets a socket unit is listening on.
5584 * 'systemctl status' will also shown information about any
5585 drop-in configuration file for units. (Drop-In configuration
5586 files in this context are files such as
5587 /etc/systemd/systemd/foobar.service.d/*.conf)
5589 * systemd-cgtop now optionally shows summed up CPU times of
5590 cgroups. Press '%' while running cgtop to switch between
5591 percentage and absolute mode. This is useful to determine
5592 which cgroups use up the most CPU time over the entire
5593 runtime of the system. systemd-cgtop has also been updated
5594 to be 'pipeable' for processing with further shell tools.
5596 * 'hostnamectl set-hostname' will now allow setting of FQDN
5599 * The formatting and parsing of time span values has been
5600 changed. The parser now understands fractional expressions
5601 such as "5.5h". The formatter will now output fractional
5602 expressions for all time spans under 1min, i.e. "5.123456s"
5603 rather than "5s 123ms 456us". For time spans under 1s
5604 millisecond values are shown, for those under 1ms
5605 microsecond values are shown. This should greatly improve
5606 all time-related output of systemd.
5608 * libsystemd-login and libsystemd-journal gained new
5609 functions for querying the poll() events mask and poll()
5610 timeout value for integration into arbitrary event
5613 * localectl gained the ability to list available X11 keymaps
5614 (models, layouts, variants, options).
5616 * 'systemd-analyze dot' gained the ability to filter for
5617 specific units via shell-style globs, to create smaller,
5618 more useful graphs. I.e. it is now possible to create simple
5619 graphs of all the dependencies between only target units, or
5620 of all units that Avahi has dependencies with.
5622 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
5623 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
5624 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
5625 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
5626 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
5627 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
5628 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
5632 * The boot-time readahead implementation for rotating media
5633 will now read the read-ahead data in multiple passes which
5634 consist of all read requests made in equidistant time
5635 intervals. This means instead of strictly reading read-ahead
5636 data in its physical order on disk we now try to find a
5637 middle ground between physical and access time order.
5639 * /etc/os-release files gained a new BUILD_ID= field for usage
5640 on operating systems that provide continuous builds of OS
5643 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
5644 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
5645 William Douglas, Zbigniew Jędrzejewski-Szmek
5649 * systemd-python gained an API exposing libsystemd-daemon.
5651 * The SMACK setup logic gained support for uploading CIPSO
5654 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
5655 ReadOnlyDirectories= and InaccessibleDirectories= has
5656 changed. The private /tmp and /var/tmp directories are now
5657 shared by all processes of a service (which means
5658 ExecStartPre= may now leave data in /tmp that ExecStart= of
5659 the same service can still access). When a service is
5660 stopped its temporary directories are immediately deleted
5661 (normal clean-up with tmpfiles is still done in addition to
5664 * By default, systemd will now set a couple of sysctl
5665 variables in the kernel: the safe sysrq options are turned
5666 on, IP route verification is turned on, and source routing
5667 disabled. The recently added hardlink and softlink
5668 protection of the kernel is turned on. These settings should
5669 be reasonably safe, and good defaults for all new systems.
5671 * The predictable network naming logic may now be turned off
5672 with a new kernel command line switch: net.ifnames=0.
5674 * A new libsystemd-bus module has been added that implements a
5675 pretty complete D-Bus client library. For details see:
5677 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
5679 * journald will now explicitly flush the journal files to disk
5680 at the latest 5min after each write. The file will then also
5681 be marked offline until the next write. This should increase
5682 reliability in case of a crash. The synchronization delay
5683 can be configured via SyncIntervalSec= in journald.conf.
5685 * There's a new remote-fs-setup.target unit that can be used
5686 to pull in specific services when at least one remote file
5687 system is to be mounted.
5689 * There are new targets timers.target and paths.target as
5690 canonical targets to pull user timer and path units in
5691 from. This complements sockets.target with a similar
5692 purpose for socket units.
5694 * libudev gained a new call udev_device_set_attribute_value()
5695 to set sysfs attributes of a device.
5697 * The udev daemon now sets the default number of worker
5698 processes executed in parallel based on the number of available
5699 CPUs instead of the amount of available RAM. This is supposed
5700 to provide a more reliable default and limit a too aggressive
5701 parallelism for setups with 1000s of devices connected.
5703 Contributions from: Auke Kok, Colin Walters, Cristian
5704 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
5705 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
5706 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
5707 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
5708 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
5709 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
5710 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
5711 Zbigniew Jędrzejewski-Szmek
5715 * Configuration of unit files may now be extended via drop-in
5716 files without having to edit/override the unit files
5717 themselves. More specifically, if the administrator wants to
5718 change one value for a service file foobar.service he can
5719 now do so by dropping in a configuration snippet into
5720 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
5721 will load all these snippets and apply them on top of the
5722 main unit configuration file, possibly extending or
5723 overriding its settings. Using these drop-in snippets is
5724 generally nicer than the two earlier options for changing
5725 unit files locally: copying the files from
5726 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
5727 them there; or creating a new file in /etc/systemd/system/
5728 that incorporates the original one via ".include". Drop-in
5729 snippets into these .d/ directories can be placed in any
5730 directory systemd looks for units in, and the usual
5731 overriding semantics between /usr/lib, /etc and /run apply
5734 * Most unit file settings which take lists of items can now be
5735 reset by assigning the empty string to them. For example,
5736 normally, settings such as Environment=FOO=BAR append a new
5737 environment variable assignment to the environment block,
5738 each time they are used. By assigning Environment= the empty
5739 string the environment block can be reset to empty. This is
5740 particularly useful with the .d/*.conf drop-in snippets
5741 mentioned above, since this adds the ability to reset list
5742 settings from vendor unit files via these drop-ins.
5744 * systemctl gained a new "list-dependencies" command for
5745 listing the dependencies of a unit recursively.
5747 * Inhibitors are now honored and listed by "systemctl
5748 suspend", "systemctl poweroff" (and similar) too, not only
5749 GNOME. These commands will also list active sessions by
5752 * Resource limits (as exposed by the various control group
5753 controllers) can now be controlled dynamically at runtime
5754 for all units. More specifically, you can now use a command
5755 like "systemctl set-cgroup-attr foobar.service cpu.shares
5756 2000" to alter the CPU shares a specific service gets. These
5757 settings are stored persistently on disk, and thus allow the
5758 administrator to easily adjust the resource usage of
5759 services with a few simple commands. This dynamic resource
5760 management logic is also available to other programs via the
5761 bus. Almost any kernel cgroup attribute and controller is
5764 * systemd-vconsole-setup will now copy all font settings to
5765 all allocated VTs, where it previously applied them only to
5768 * libsystemd-login gained the new sd_session_get_tty() API
5771 * This release drops support for a few legacy or
5772 distribution-specific LSB facility names when parsing init
5773 scripts: $x-display-manager, $mail-transfer-agent,
5774 $mail-transport-agent, $mail-transfer-agent, $smtp,
5775 $null. Also, the mail-transfer-agent.target unit backing
5776 this has been removed. Distributions which want to retain
5777 compatibility with this should carry the burden for
5778 supporting this themselves and patch support for these back
5779 in, if they really need to. Also, the facilities $syslog and
5780 $local_fs are now ignored, since systemd does not support
5781 early-boot LSB init scripts anymore, and these facilities
5782 are implied anyway for normal services. syslog.target has
5785 * There are new bus calls on PID1's Manager object for
5786 cancelling jobs, and removing snapshot units. Previously,
5787 both calls were only available on the Job and Snapshot
5790 * systemd-journal-gatewayd gained SSL support.
5792 * The various "environment" files, such as /etc/locale.conf
5793 now support continuation lines with a backslash ("\") as
5794 last character in the line, similarly in style (but different)
5795 to how this is supported in shells.
5797 * For normal user processes the _SYSTEMD_USER_UNIT= field is
5798 now implicitly appended to every log entry logged. systemctl
5799 has been updated to filter by this field when operating on a
5800 user systemd instance.
5802 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
5803 CAP_AUDIT_CONTROL capabilities to the capabilities set for
5804 the container. This makes it easier to boot unmodified
5805 Fedora systems in a container, which however still requires
5806 audit=0 to be passed on the kernel command line. Auditing in
5807 kernel and userspace is unfortunately still too broken in
5808 context of containers, hence we recommend compiling it out
5809 of the kernel or using audit=0. Hopefully this will be fixed
5810 one day for good in the kernel.
5812 * nspawn gained the new --bind= and --bind-ro= parameters to
5813 bind mount specific directories from the host into the
5816 * nspawn will now mount its own devpts file system instance
5817 into the container, in order not to leak pty devices from
5818 the host into the container.
5820 * systemd will now read the firmware boot time performance
5821 information from the EFI variables, if the used boot loader
5822 supports this, and takes it into account for boot performance
5823 analysis via "systemd-analyze". This is currently supported
5824 only in conjunction with Gummiboot, but could be supported
5825 by other boot loaders too. For details see:
5827 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
5829 * A new generator has been added that automatically mounts the
5830 EFI System Partition (ESP) to /boot, if that directory
5831 exists, is empty, and no other file system has been
5832 configured to be mounted there.
5834 * logind will now send out PrepareForSleep(false) out
5835 unconditionally, after coming back from suspend. This may be
5836 used by applications as asynchronous notification for
5837 system resume events.
5839 * "systemctl unlock-sessions" has been added, that allows
5840 unlocking the screens of all user sessions at once, similar
5841 to how "systemctl lock-sessions" already locked all users
5842 sessions. This is backed by a new D-Bus call UnlockSessions().
5844 * "loginctl seat-status" will now show the master device of a
5845 seat. (i.e. the device of a seat that needs to be around for
5846 the seat to be considered available, usually the graphics
5849 * tmpfiles gained a new "X" line type, that allows
5850 configuration of files and directories (with wildcards) that
5851 shall be excluded from automatic cleanup ("aging").
5853 * udev default rules set the device node permissions now only
5854 at "add" events, and do not change them any longer with a
5855 later "change" event.
5857 * The log messages for lid events and power/sleep keypresses
5858 now carry a message ID.
5860 * We now have a substantially larger unit test suite, but this
5861 continues to be work in progress.
5863 * udevadm hwdb gained a new --root= parameter to change the
5864 root directory to operate relative to.
5866 * logind will now issue a background sync() request to the kernel
5867 early at shutdown, so that dirty buffers are flushed to disk early
5868 instead of at the last moment, in order to optimize shutdown
5871 * A new bootctl tool has been added that is an interface for
5872 certain boot loader operations. This is currently a preview
5873 and is likely to be extended into a small mechanism daemon
5874 like timedated, localed, hostnamed, and can be used by
5875 graphical UIs to enumerate available boot options, and
5876 request boot into firmware operations.
5878 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
5879 the rest of the package. It also has been updated to work
5880 correctly in initrds.
5882 * Policykit previously has been runtime optional, and is now
5883 also compile time optional via a configure switch.
5885 * systemd-analyze has been reimplemented in C. Also "systemctl
5886 dot" has moved into systemd-analyze.
5888 * "systemctl status" with no further parameters will now print
5889 the status of all active or failed units.
5891 * Operations such as "systemctl start" can now be executed
5892 with a new mode "--irreversible" which may be used to queue
5893 operations that cannot accidentally be reversed by a later
5894 job queuing. This is by default used to make shutdown
5895 requests more robust.
5897 * The Python API of systemd now gained a new module for
5898 reading journal files.
5900 * A new tool kernel-install has been added that can install
5901 kernel images according to the Boot Loader Specification:
5903 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
5905 * Boot time console output has been improved to provide
5906 animated boot time output for hanging jobs.
5908 * A new tool systemd-activate has been added which can be used
5909 to test socket activation with, directly from the command
5910 line. This should make it much easier to test and debug
5911 socket activation in daemons.
5913 * journalctl gained a new "--reverse" (or -r) option to show
5914 journal output in reverse order (i.e. newest line first).
5916 * journalctl gained a new "--pager-end" (or -e) option to jump
5917 to immediately jump to the end of the journal in the
5918 pager. This is only supported in conjunction with "less".
5920 * journalctl gained a new "--user-unit=" option, that works
5921 similarly to "--unit=" but filters for user units rather than
5924 * A number of unit files to ease adoption of systemd in
5925 initrds has been added. This moves some minimal logic from
5926 the various initrd implementations into systemd proper.
5928 * The journal files are now owned by a new group
5929 "systemd-journal", which exists specifically to allow access
5930 to the journal, and nothing else. Previously, we used the
5931 "adm" group for that, which however possibly covers more
5932 than just journal/log file access. This new group is now
5933 already used by systemd-journal-gatewayd to ensure this
5934 daemon gets access to the journal files and as little else
5935 as possible. Note that "make install" will also set FS ACLs
5936 up for /var/log/journal to give "adm" and "wheel" read
5937 access to it, in addition to "systemd-journal" which owns
5938 the journal files. We recommend that packaging scripts also
5939 add read access to "adm" + "wheel" to /var/log/journal, and
5940 all existing/future journal files. To normal users and
5941 administrators little changes, however packagers need to
5942 ensure to create the "systemd-journal" system group at
5943 package installation time.
5945 * The systemd-journal-gatewayd now runs as unprivileged user
5946 systemd-journal-gateway:systemd-journal-gateway. Packaging
5947 scripts need to create these system user/group at
5950 * timedated now exposes a new boolean property CanNTP that
5951 indicates whether a local NTP service is available or not.
5953 * systemd-detect-virt will now also detect xen PVs
5955 * The pstore file system is now mounted by default, if it is
5958 * In addition to the SELinux and IMA policies we will now also
5959 load SMACK policies at early boot.
5961 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
5962 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
5963 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
5964 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
5965 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
5966 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
5967 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
5968 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
5969 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
5970 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
5971 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
5972 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
5973 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
5974 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
5978 * Timer units now support calendar time events in addition to
5979 monotonic time events. That means you can now trigger a unit
5980 based on a calendar time specification such as "Thu,Fri
5981 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
5982 or fifth day of any month of the year 2013, given that it is
5983 a thursday or friday. This brings timer event support
5984 considerably closer to cron's capabilities. For details on
5985 the supported calendar time specification language see
5988 * udev now supports a number of different naming policies for
5989 network interfaces for predictable names, and a combination
5990 of these policies is now the default. Please see this wiki
5991 document for details:
5993 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
5995 * Auke Kok's bootchart implementation has been added to the
5996 systemd tree. It is an optional component that can graph the
5997 boot in quite some detail. It is one of the best bootchart
5998 implementations around and minimal in its code and
6001 * nss-myhostname has been integrated into the systemd source
6002 tree. nss-myhostname guarantees that the local hostname
6003 always stays resolvable via NSS. It has been a weak
6004 requirement of systemd-hostnamed since a long time, and
6005 since its code is actually trivial we decided to just
6006 include it in systemd's source tree. It can be turned off
6007 with a configure switch.
6009 * The read-ahead logic is now capable of properly detecting
6010 whether a btrfs file system is on SSD or rotating media, in
6011 order to optimize the read-ahead scheme. Previously, it was
6012 only capable of detecting this on traditional file systems
6015 * In udev, additional device properties are now read from the
6016 IAB in addition to the OUI database. Also, Bluetooth company
6017 identities are attached to the devices as well.
6019 * In service files %U may be used as specifier that is
6020 replaced by the configured user name of the service.
6022 * nspawn may now be invoked without a controlling TTY. This
6023 makes it suitable for invocation as its own service. This
6024 may be used to set up a simple containerized server system
6025 using only core OS tools.
6027 * systemd and nspawn can now accept socket file descriptors
6028 when they are started for socket activation. This enables
6029 implementation of socket activated nspawn
6030 containers. i.e. think about autospawning an entire OS image
6031 when the first SSH or HTTP connection is received. We expect
6032 that similar functionality will also be added to libvirt-lxc
6035 * journalctl will now suppress ANSI color codes when
6036 presenting log data.
6038 * systemctl will no longer show control group information for
6039 a unit if the control group is empty anyway.
6041 * logind can now automatically suspend/hibernate/shutdown the
6044 * /etc/machine-info and hostnamed now also expose the chassis
6045 type of the system. This can be used to determine whether
6046 the local system is a laptop, desktop, handset or
6047 tablet. This information may either be configured by the
6048 user/vendor or is automatically determined from ACPI and DMI
6049 information if possible.
6051 * A number of PolicyKit actions are now bound together with
6052 "imply" rules. This should simplify creating UIs because
6053 many actions will now authenticate similar ones as well.
6055 * Unit files learnt a new condition ConditionACPower= which
6056 may be used to conditionalize a unit depending on whether an
6057 AC power source is connected or not, of whether the system
6058 is running on battery power.
6060 * systemctl gained a new "is-failed" verb that may be used in
6061 shell scripts and suchlike to check whether a specific unit
6062 is in the "failed" state.
6064 * The EnvironmentFile= setting in unit files now supports file
6065 globbing, and can hence be used to easily read a number of
6066 environment files at once.
6068 * systemd will no longer detect and recognize specific
6069 distributions. All distribution-specific #ifdeffery has been
6070 removed, systemd is now fully generic and
6071 distribution-agnostic. Effectively, not too much is lost as
6072 a lot of the code is still accessible via explicit configure
6073 switches. However, support for some distribution specific
6074 legacy configuration file formats has been dropped. We
6075 recommend distributions to simply adopt the configuration
6076 files everybody else uses now and convert the old
6077 configuration from packaging scripts. Most distributions
6078 already did that. If that's not possible or desirable,
6079 distributions are welcome to forward port the specific
6080 pieces of code locally from the git history.
6082 * When logging a message about a unit systemd will now always
6083 log the unit name in the message meta data.
6085 * localectl will now also discover system locale data that is
6086 not stored in locale archives, but directly unpacked.
6088 * logind will no longer unconditionally use framebuffer
6089 devices as seat masters, i.e. as devices that are required
6090 to be existing before a seat is considered preset. Instead,
6091 it will now look for all devices that are tagged as
6092 "seat-master" in udev. By default, framebuffer devices will
6093 be marked as such, but depending on local systems, other
6094 devices might be marked as well. This may be used to
6095 integrate graphics cards using closed source drivers (such
6096 as NVidia ones) more nicely into logind. Note however, that
6097 we recommend using the open source NVidia drivers instead,
6098 and no udev rules for the closed-source drivers will be
6099 shipped from us upstream.
6101 Contributions from: Adam Williamson, Alessandro Crismani, Auke
6102 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
6103 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
6104 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
6105 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
6106 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
6107 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
6108 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
6109 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
6110 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
6111 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
6112 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
6117 * udev gained support for loading additional device properties
6118 from an indexed database that is keyed by vendor/product IDs
6119 and similar device identifiers. For the beginning this
6120 "hwdb" is populated with data from the well-known PCI and
6121 USB database, but also includes PNP, ACPI and OID data. In
6122 the longer run this indexed database shall grow into
6123 becoming the one central database for non-essential
6124 userspace device metadata. Previously, data from the PCI/USB
6125 database was only attached to select devices, since the
6126 lookup was a relatively expensive operation due to O(n) time
6127 complexity (with n being the number of entries in the
6128 database). Since this is now O(1), we decided to add in this
6129 data for all devices where this is available, by
6130 default. Note that the indexed database needs to be rebuilt
6131 when new data files are installed. To achieve this you need
6132 to update your packaging scripts to invoke "udevadm hwdb
6133 --update" after installation of hwdb data files. For
6134 RPM-based distributions we introduced the new
6135 %udev_hwdb_update macro for this purpose.
6137 * The Journal gained support for the "Message Catalog", an
6138 indexed database to link up additional information with
6139 journal entries. For further details please check:
6141 https://www.freedesktop.org/wiki/Software/systemd/catalog
6143 The indexed message catalog database also needs to be
6144 rebuilt after installation of message catalog files. Use
6145 "journalctl --update-catalog" for this. For RPM-based
6146 distributions we introduced the %journal_catalog_update
6147 macro for this purpose.
6149 * The Python Journal bindings gained support for the standard
6150 Python logging framework.
6152 * The Journal API gained new functions for checking whether
6153 the underlying file system of a journal file is capable of
6154 properly reporting file change notifications, or whether
6155 applications that want to reflect journal changes "live"
6156 need to recheck journal files continuously in appropriate
6159 * It is now possible to set the "age" field for tmpfiles
6160 entries to 0, indicating that files matching this entry
6161 shall always be removed when the directories are cleaned up.
6163 * coredumpctl gained a new "gdb" verb which invokes gdb
6164 right-away on the selected coredump.
6166 * There's now support for "hybrid sleep" on kernels that
6167 support this, in addition to "suspend" and "hibernate". Use
6168 "systemctl hybrid-sleep" to make use of this.
6170 * logind's HandleSuspendKey= setting (and related settings)
6171 now gained support for a new "lock" setting to simply
6172 request the screen lock on all local sessions, instead of
6173 actually executing a suspend or hibernation.
6175 * systemd will now mount the EFI variables file system by
6178 * Socket units now gained support for configuration of the
6179 SMACK security label.
6181 * timedatectl will now output the time of the last and next
6182 daylight saving change.
6184 * We dropped support for various legacy and distro-specific
6185 concepts, such as insserv, early-boot SysV services
6186 (i.e. those for non-standard runlevels such as 'b' or 'S')
6187 or ArchLinux /etc/rc.conf support. We recommend the
6188 distributions who still need support this to either continue
6189 to maintain the necessary patches downstream, or find a
6190 different solution. (Talk to us if you have questions!)
6192 * Various systemd components will now bypass PolicyKit checks
6193 for root and otherwise handle properly if PolicyKit is not
6194 found to be around. This should fix most issues for
6195 PolicyKit-less systems. Quite frankly this should have been
6196 this way since day one. It is absolutely our intention to
6197 make systemd work fine on PolicyKit-less systems, and we
6198 consider it a bug if something does not work as it should if
6199 PolicyKit is not around.
6201 * For embedded systems it is now possible to build udev and
6202 systemd without blkid and/or kmod support.
6204 * "systemctl switch-root" is now capable of switching root
6205 more than once. I.e. in addition to transitions from the
6206 initrd to the host OS it is now possible to transition to
6207 further OS images from the host. This is useful to implement
6208 offline updating tools.
6210 * Various other additions have been made to the RPM macros
6211 shipped with systemd. Use %udev_rules_update() after
6212 installing new udev rules files. %_udevhwdbdir,
6213 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
6214 %_sysctldir are now available which resolve to the right
6215 directories for packages to place various data files in.
6217 * journalctl gained the new --full switch (in addition to
6218 --all, to disable ellipsation for long messages.
6220 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
6221 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
6222 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
6223 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
6224 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
6225 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
6226 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
6227 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
6228 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
6232 * journalctl gained new --since= and --until= switches to
6233 filter by time. It also now supports nice filtering for
6234 units via --unit=/-u.
6236 * Type=oneshot services may use ExecReload= and do the
6239 * The journal daemon now supports time-based rotation and
6240 vacuuming, in addition to the usual disk-space based
6243 * The journal will now index the available field values for
6244 each field name. This enables clients to show pretty drop
6245 downs of available match values when filtering. The bash
6246 completion of journalctl has been updated
6247 accordingly. journalctl gained a new switch -F to list all
6248 values a certain field takes in the journal database.
6250 * More service events are now written as structured messages
6251 to the journal, and made recognizable via message IDs.
6253 * The timedated, localed and hostnamed mini-services which
6254 previously only provided support for changing time, locale
6255 and hostname settings from graphical DEs such as GNOME now
6256 also have a minimal (but very useful) text-based client
6257 utility each. This is probably the nicest way to changing
6258 these settings from the command line now, especially since
6259 it lists available options and is fully integrated with bash
6262 * There's now a new tool "systemd-coredumpctl" to list and
6263 extract coredumps from the journal.
6265 * We now install a README each in /var/log/ and
6266 /etc/rc.d/init.d explaining where the system logs and init
6267 scripts went. This hopefully should help folks who go to
6268 that dirs and look into the otherwise now empty void and
6269 scratch their heads.
6271 * When user-services are invoked (by systemd --user) the
6272 $MANAGERPID env var is set to the PID of systemd.
6274 * SIGRTMIN+24 when sent to a --user instance will now result
6275 in immediate termination of systemd.
6277 * gatewayd received numerous feature additions such as a
6278 "follow" mode, for live syncing and filtering.
6280 * browse.html now allows filtering and showing detailed
6281 information on specific entries. Keyboard navigation and
6282 mouse screen support has been added.
6284 * gatewayd/journalctl now supports HTML5/JSON
6285 Server-Sent-Events as output.
6287 * The SysV init script compatibility logic will now
6288 heuristically determine whether a script supports the
6289 "reload" verb, and only then make this available as
6292 * "systemctl status --follow" has been removed, use "journalctl
6295 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
6296 have been removed since they are hardly useful to be
6299 * And I'd like to take the opportunity to specifically mention
6300 Zbigniew for his great contributions. Zbigniew, you rock!
6302 Contributions from: Andrew Eikum, Christian Hesse, Colin
6303 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
6304 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
6305 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
6306 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
6307 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
6308 Jędrzejewski-Szmek, Сковорода Никита Андреевич
6312 * If /etc/vconsole.conf is non-existent or empty we will no
6313 longer load any console font or key map at boot by
6314 default. Instead the kernel defaults will be left
6315 intact. This is definitely the right thing to do, as no
6316 configuration should mean no configuration, and hard-coding
6317 font names that are different on all archs is probably a bad
6318 idea. Also, the kernel default key map and font should be
6319 good enough for most cases anyway, and mostly identical to
6320 the userspace fonts/key maps we previously overloaded them
6321 with. If distributions want to continue to default to a
6322 non-kernel font or key map they should ship a default
6323 /etc/vconsole.conf with the appropriate contents.
6325 Contributions from: Colin Walters, Daniel J Walsh, Dave
6326 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
6327 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6331 * journalctl gained a new --cursor= switch to show entries
6332 starting from the specified location in the journal.
6334 * We now enforce a size limit on journal entry fields exported
6335 with "-o json" in journalctl. Fields larger than 4K will be
6336 assigned null. This can be turned off with --all.
6338 * An (optional) journal gateway daemon is now available as
6339 "systemd-journal-gatewayd.service". This service provides
6340 access to the journal via HTTP and JSON. This functionality
6341 will be used to implement live log synchronization in both
6342 pull and push modes, but has various other users too, such
6343 as easy log access for debugging of embedded devices. Right
6344 now it is already useful to retrieve the journal via HTTP:
6346 # systemctl start systemd-journal-gatewayd.service
6347 # wget http://localhost:19531/entries
6349 This will download the journal contents in a
6350 /var/log/messages compatible format. The same as JSON:
6352 # curl -H"Accept: application/json" http://localhost:19531/entries
6354 This service is also accessible via a web browser where a
6355 single static HTML5 app is served that uses the JSON logic
6356 to enable the user to do some basic browsing of the
6357 journal. This will be extended later on. Here's an example
6358 screenshot of this app in its current state:
6360 http://0pointer.de/public/journal-gatewayd
6362 Contributions from: Kay Sievers, Lennart Poettering, Robert
6363 Milasan, Tom Gundersen
6367 * The bash completion logic is now available for journalctl
6370 * We do not mount the "cpuset" controller anymore together with
6371 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
6372 started if no parameters are assigned to it. "cpuset" hence
6373 broke code that assumed it could create "cpu" groups and
6376 * journalctl -f will now subscribe to terminal size changes,
6377 and line break accordingly.
6379 Contributions from: Dave Reisner, Kay Sievers, Lennart
6380 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
6384 * nspawn will now create a symlink /etc/localtime in the
6385 container environment, copying the host's timezone
6386 setting. Previously this has been done via a bind mount, but
6387 since symlinks cannot be bind mounted this has now been
6388 changed to create/update the appropriate symlink.
6390 * journalctl -n's line number argument is now optional, and
6391 will default to 10 if omitted.
6393 * journald will now log the maximum size the journal files may
6394 take up on disk. This is particularly useful if the default
6395 built-in logic of determining this parameter from the file
6396 system size is used. Use "systemctl status
6397 systemd-journald.service" to see this information.
6399 * The multi-seat X wrapper tool has been stripped down. As X
6400 is now capable of enumerating graphics devices via udev in a
6401 seat-aware way the wrapper is not strictly necessary
6402 anymore. A stripped down temporary stop-gap is still shipped
6403 until the upstream display managers have been updated to
6404 fully support the new X logic. Expect this wrapper to be
6405 removed entirely in one of the next releases.
6407 * HandleSleepKey= in logind.conf has been split up into
6408 HandleSuspendKey= and HandleHibernateKey=. The old setting
6409 is not available anymore. X11 and the kernel are
6410 distinguishing between these keys and we should too. This
6411 also means the inhibition lock for these keys has been split
6414 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
6415 Poettering, Lukas Nykryn, Václav Pavlín
6419 * Whenever a unit changes state we will now log this to the
6420 journal and show along the unit's own log output in
6423 * ConditionPathIsMountPoint= can now properly detect bind
6424 mount points too. (Previously, a bind mount of one file
6425 system to another place in the same file system could not be
6426 detected as mount, since they shared struct stat's st_dev
6429 * We will now mount the cgroup controllers cpu, cpuacct,
6430 cpuset and the controllers net_cls, net_prio together by
6433 * nspawn containers will now have a virtualized boot
6434 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
6435 over with a randomized ID at container initialization). This
6436 has the effect of making "journalctl -b" do the right thing
6439 * The JSON output journal serialization has been updated not
6440 to generate "endless" list objects anymore, but rather one
6441 JSON object per line. This is more in line how most JSON
6442 parsers expect JSON objects. The new output mode
6443 "json-pretty" has been added to provide similar output, but
6444 neatly aligned for readability by humans.
6446 * We dropped all explicit sync() invocations in the shutdown
6447 code. The kernel does this implicitly anyway in the kernel
6448 reboot() syscall. halt(8)'s -n option is now a compatibility
6451 * We now support virtualized reboot() in containers, as
6452 supported by newer kernels. We will fall back to exit() if
6453 CAP_SYS_REBOOT is not available to the container. Also,
6454 nspawn makes use of this now and will actually reboot the
6455 container if the containerized OS asks for that.
6457 * journalctl will only show local log output by default
6458 now. Use --merge (-m) to show remote log output, too.
6460 * libsystemd-journal gained the new sd_journal_get_usage()
6461 call to determine the current disk usage of all journal
6462 files. This is exposed in the new "journalctl --disk-usage"
6465 * journald gained a new configuration setting SplitMode= in
6466 journald.conf which may be used to control how user journals
6467 are split off. See journald.conf(5) for details.
6469 * A new condition type ConditionFileNotEmpty= has been added.
6471 * tmpfiles' "w" lines now support file globbing, to write
6472 multiple files at once.
6474 * We added Python bindings for the journal submission
6475 APIs. More Python APIs for a number of selected APIs will
6476 likely follow. Note that we intend to add native bindings
6477 only for the Python language, as we consider it common
6478 enough to deserve bindings shipped within systemd. There are
6479 various projects outside of systemd that provide bindings
6480 for languages such as PHP or Lua.
6482 * Many conditions will now resolve specifiers such as %i. In
6483 addition, PathChanged= and related directives of .path units
6484 now support specifiers as well.
6486 * There's now a new RPM macro definition for the system preset
6489 * journald will now warn if it ca not forward a message to the
6490 syslog daemon because its socket is full.
6492 * timedated will no longer write or process /etc/timezone,
6493 except on Debian. As we do not support late mounted /usr
6494 anymore /etc/localtime always being a symlink is now safe,
6495 and hence the information in /etc/timezone is not necessary
6498 * logind will now always reserve one VT for a text getty (VT6
6499 by default). Previously if more than 6 X sessions where
6500 started they took up all the VTs with auto-spawned gettys,
6501 so that no text gettys were available anymore.
6503 * udev will now automatically inform the btrfs kernel logic
6504 about btrfs RAID components showing up. This should make
6505 simple hotplug based btrfs RAID assembly work.
6507 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
6508 (but not for its children which will stay at the kernel
6509 default). This should allow setups with a lot more listening
6512 * systemd will now always pass the configured timezone to the
6513 kernel at boot. timedated will do the same when the timezone
6516 * logind's inhibition logic has been updated. By default,
6517 logind will now handle the lid switch, the power and sleep
6518 keys all the time, even in graphical sessions. If DEs want
6519 to handle these events on their own they should take the new
6520 handle-power-key, handle-sleep-key and handle-lid-switch
6521 inhibitors during their runtime. A simple way to achieve
6522 that is to invoke the DE wrapped in an invocation of:
6524 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
6526 * Access to unit operations is now checked via SELinux taking
6527 the unit file label and client process label into account.
6529 * systemd will now notify the administrator in the journal
6530 when he over-mounts a non-empty directory.
6532 * There are new specifiers that are resolved in unit files,
6533 for the host name (%H), the machine ID (%m) and the boot ID
6536 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
6537 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
6538 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
6539 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
6540 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
6541 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
6542 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
6546 * Support for reading structured kernel messages from
6547 /dev/kmsg has now been added and is enabled by default.
6549 * Support for reading kernel messages from /proc/kmsg has now
6550 been removed. If you want kernel messages in the journal
6551 make sure to run a recent kernel (>= 3.5) that supports
6552 reading structured messages from /dev/kmsg (see
6553 above). /proc/kmsg is now exclusive property of classic
6554 syslog daemons again.
6556 * The libudev API gained the new
6557 udev_device_new_from_device_id() call.
6559 * The logic for file system namespace (ReadOnlyDirectory=,
6560 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
6561 require pivot_root() anymore. This means fewer temporary
6562 directories are created below /tmp for this feature.
6564 * nspawn containers will now see and receive all submounts
6565 made on the host OS below the root file system of the
6568 * Forward Secure Sealing is now supported for Journal files,
6569 which provide cryptographical sealing of journal files so
6570 that attackers cannot alter log history anymore without this
6571 being detectable. Lennart will soon post a blog story about
6572 this explaining it in more detail.
6574 * There are two new service settings RestartPreventExitStatus=
6575 and SuccessExitStatus= which allow configuration of exit
6576 status (exit code or signal) which will be excepted from the
6577 restart logic, resp. consider successful.
6579 * journalctl gained the new --verify switch that can be used
6580 to check the integrity of the structure of journal files and
6581 (if Forward Secure Sealing is enabled) the contents of
6584 * nspawn containers will now be run with /dev/stdin, /dev/fd/
6585 and similar symlinks pre-created. This makes running shells
6586 as container init process a lot more fun.
6588 * The fstab support can now handle PARTUUID= and PARTLABEL=
6591 * A new ConditionHost= condition has been added to match
6592 against the hostname (with globs) and machine ID. This is
6593 useful for clusters where a single OS image is used to
6594 provision a large number of hosts which shall run slightly
6595 different sets of services.
6597 * Services which hit the restart limit will now be placed in a
6600 Contributions from: Bertram Poettering, Dave Reisner, Huang
6601 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
6602 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
6606 * When running in --user mode systemd will now become a
6607 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
6608 tree a lot more organized.
6610 * A new PartOf= unit dependency type has been introduced that
6611 may be used to group services in a natural way.
6613 * "systemctl enable" may now be used to enable instances of
6616 * journalctl now prints error log levels in red, and
6617 warning/notice log levels in bright white. It also supports
6618 filtering by log level now.
6620 * cgtop gained a new -n switch (similar to top), to configure
6621 the maximum number of iterations to run for. It also gained
6622 -b, to run in batch mode (accepting no input).
6624 * The suffix ".service" may now be omitted on most systemctl
6625 command lines involving service unit names.
6627 * There's a new bus call in logind to lock all sessions, as
6628 well as a loginctl verb for it "lock-sessions".
6630 * libsystemd-logind.so gained a new call sd_journal_perror()
6631 that works similar to libc perror() but logs to the journal
6632 and encodes structured information about the error number.
6634 * /etc/crypttab entries now understand the new keyfile-size=
6637 * shutdown(8) now can send a (configurable) wall message when
6638 a shutdown is cancelled.
6640 * The mount propagation mode for the root file system will now
6641 default to "shared", which is useful to make containers work
6642 nicely out-of-the-box so that they receive new mounts from
6643 the host. This can be undone locally by running "mount
6644 --make-rprivate /" if needed.
6646 * The prefdm.service file has been removed. Distributions
6647 should maintain this unit downstream if they intend to keep
6648 it around. However, we recommend writing normal unit files
6649 for display managers instead.
6651 * Since systemd is a crucial part of the OS we will now
6652 default to a number of compiler switches that improve
6653 security (hardening) such as read-only relocations, stack
6654 protection, and suchlike.
6656 * The TimeoutSec= setting for services is now split into
6657 TimeoutStartSec= and TimeoutStopSec= to allow configuration
6658 of individual time outs for the start and the stop phase of
6661 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
6662 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
6663 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
6664 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
6665 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
6666 Gundersen, Zbigniew Jędrzejewski-Szmek
6670 * The journal and id128 C APIs are now fully documented as man
6673 * Extra safety checks have been added when transitioning from
6674 the initial RAM disk to the main system to avoid accidental
6677 * /etc/crypttab entries now understand the new keyfile-offset=
6680 * systemctl -t can now be used to filter by unit load state.
6682 * The journal C API gained the new sd_journal_wait() call to
6683 make writing synchronous journal clients easier.
6685 * journalctl gained the new -D switch to show journals from a
6688 * journalctl now displays a special marker between log
6689 messages of two different boots.
6691 * The journal is now explicitly flushed to /var via a service
6692 systemd-journal-flush.service, rather than implicitly simply
6693 by seeing /var/log/journal to be writable.
6695 * journalctl (and the journal C APIs) can now match for much
6696 more complex expressions, with alternatives and
6699 * When transitioning from the initial RAM disk to the main
6700 system we will now kill all processes in a killing spree to
6701 ensure no processes stay around by accident.
6703 * Three new specifiers may be used in unit files: %u, %h, %s
6704 resolve to the user name, user home directory resp. user
6705 shell. This is useful for running systemd user instances.
6707 * We now automatically rotate journal files if their data
6708 object hash table gets a fill level > 75%. We also size the
6709 hash table based on the configured maximum file size. This
6710 together should lower hash collisions drastically and thus
6711 speed things up a bit.
6713 * journalctl gained the new "--header" switch to introspect
6714 header data of journal files.
6716 * A new setting SystemCallFilters= has been added to services
6717 which may be used to apply blacklists or whitelists to
6718 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
6720 * nspawn gained a new --link-journal= switch (and quicker: -j)
6721 to link the container journal with the host. This makes it
6722 very easy to centralize log viewing on the host for all
6723 guests while still keeping the journal files separated.
6725 * Many bugfixes and optimizations
6727 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
6728 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
6729 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
6734 * Several tools now understand kernel command line arguments,
6735 which are only read when run in an initial RAM disk. They
6736 usually follow closely their normal counterparts, but are
6739 * There's a new tool to analyze the readahead files that are
6740 automatically generated at boot. Use:
6742 /usr/lib/systemd/systemd-readahead analyze /.readahead
6744 * We now provide an early debug shell on tty9 if this enabled. Use:
6746 systemctl enable debug-shell.service
6748 * All plymouth related units have been moved into the Plymouth
6749 package. Please make sure to upgrade your Plymouth version
6752 * systemd-tmpfiles now supports getting passed the basename of
6753 a configuration file only, in which case it will look for it
6754 in all appropriate directories automatically.
6756 * udevadm info now takes a /dev or /sys path as argument, and
6757 does the right thing. Example:
6759 udevadm info /dev/sda
6760 udevadm info /sys/class/block/sda
6762 * systemctl now prints a warning if a unit is stopped but a
6763 unit that might trigger it continues to run. Example: a
6764 service is stopped but the socket that activates it is left
6767 * "systemctl status" will now mention if the log output was
6768 shortened due to rotation since a service has been started.
6770 * The journal API now exposes functions to determine the
6771 "cutoff" times due to rotation.
6773 * journald now understands SIGUSR1 and SIGUSR2 for triggering
6774 immediately flushing of runtime logs to /var if possible,
6775 resp. for triggering immediate rotation of the journal
6778 * It is now considered an error if a service is attempted to
6779 be stopped that is not loaded.
6781 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
6783 * systemd-analyze now supports Python 3
6785 * tmpfiles now supports cleaning up directories via aging
6786 where the first level dirs are always kept around but
6787 directories beneath it automatically aged. This is enabled
6788 by prefixing the age field with '~'.
6790 * Seat objects now expose CanGraphical, CanTTY properties
6791 which is required to deal with very fast bootups where the
6792 display manager might be running before the graphics drivers
6793 completed initialization.
6795 * Seat objects now expose a State property.
6797 * We now include RPM macros for service enabling/disabling
6798 based on the preset logic. We recommend RPM based
6799 distributions to make use of these macros if possible. This
6800 makes it simpler to reuse RPM spec files across
6803 * We now make sure that the collected systemd unit name is
6804 always valid when services log to the journal via
6807 * There's a new man page kernel-command-line(7) detailing all
6808 command line options we understand.
6810 * The fstab generator may now be disabled at boot by passing
6811 fstab=0 on the kernel command line.
6813 * A new kernel command line option modules-load= is now understood
6814 to load a specific kernel module statically, early at boot.
6816 * Unit names specified on the systemctl command line are now
6817 automatically escaped as needed. Also, if file system or
6818 device paths are specified they are automatically turned
6819 into the appropriate mount or device unit names. Example:
6821 systemctl status /home
6822 systemctl status /dev/sda
6824 * The SysVConsole= configuration option has been removed from
6825 system.conf parsing.
6827 * The SysV search path is no longer exported on the D-Bus
6830 * The Names= option has been removed from unit file parsing.
6832 * There's a new man page bootup(7) detailing the boot process.
6834 * Every unit and every generator we ship with systemd now
6835 comes with full documentation. The self-explanatory boot is
6838 * A couple of services gained "systemd-" prefixes in their
6839 name if they wrap systemd code, rather than only external
6840 code. Among them fsck@.service which is now
6841 systemd-fsck@.service.
6843 * The HaveWatchdog property has been removed from the D-Bus
6846 * systemd.confirm_spawn= on the kernel command line should now
6849 * There's a new man page crypttab(5) which details all options
6850 we actually understand.
6852 * systemd-nspawn gained a new --capability= switch to pass
6853 additional capabilities to the container.
6855 * timedated will now read known NTP implementation unit names
6856 from /usr/lib/systemd/ntp-units.d/*.list,
6857 systemd-timedated-ntp.target has been removed.
6859 * journalctl gained a new switch "-b" that lists log data of
6860 the current boot only.
6862 * The notify socket is in the abstract namespace again, in
6863 order to support daemons which chroot() at start-up.
6865 * There is a new Storage= configuration option for journald
6866 which allows configuration of where log data should go. This
6867 also provides a way to disable journal logging entirely, so
6868 that data collected is only forwarded to the console, the
6869 kernel log buffer or another syslog implementation.
6871 * Many bugfixes and optimizations
6873 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
6874 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
6875 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
6876 Shawn Landden, Tom Gundersen
6880 * "systemctl help <unit>" now shows the man page if one is
6883 * Several new man pages have been added.
6885 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
6886 MaxLevelConsole= can now be specified in
6887 journald.conf. These options allow reducing the amount of
6888 data stored on disk or forwarded by the log level.
6890 * TimerSlackNSec= can now be specified in system.conf for
6891 PID1. This allows system-wide power savings.
6893 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
6894 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
6899 * logind is now capable of (optionally) handling power and
6900 sleep keys as well as the lid switch.
6902 * journalctl now understands the syntax "journalctl
6903 /usr/bin/avahi-daemon" to get all log output of a specific
6906 * CapabilityBoundingSet= in system.conf now also influences
6907 the capability bound set of usermode helpers of the kernel.
6909 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
6910 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
6911 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
6912 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
6916 * Note that we skipped 139 releases here in order to set the
6917 new version to something that is greater than both udev's
6918 and systemd's most recent version number.
6920 * udev: all udev sources are merged into the systemd source tree now.
6921 All future udev development will happen in the systemd tree. It
6922 is still fully supported to use the udev daemon and tools without
6923 systemd running, like in initramfs or other init systems. Building
6924 udev though, will require the *build* of the systemd tree, but
6925 udev can be properly *run* without systemd.
6927 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
6928 should be used to create dead device nodes as workarounds for broken
6931 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
6932 no longer supported. udev_monitor_new_from_netlink() needs to be
6933 used to subscribe to events.
6935 * udev: when udevd is started by systemd, processes which are left
6936 behind by forking them off of udev rules, are unconditionally cleaned
6937 up and killed now after the event handling has finished. Services or
6938 daemons must be started as systemd services. Services can be
6939 pulled-in by udev to get started, but they can no longer be directly
6940 forked by udev rules.
6942 * udev: the daemon binary is called systemd-udevd now and installed
6943 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
6944 to adapt to that, create symlink, or rename the binary after building
6947 * libudev no longer provides these symbols:
6948 udev_monitor_from_socket()
6949 udev_queue_get_failed_list_entry()
6950 udev_get_{dev,sys,run}_path()
6951 The versions number was bumped and symbol versioning introduced.
6953 * systemd-loginctl and systemd-journalctl have been renamed
6954 to loginctl and journalctl to match systemctl.
6956 * The config files: /etc/systemd/systemd-logind.conf and
6957 /etc/systemd/systemd-journald.conf have been renamed to
6958 logind.conf and journald.conf. Package updates should rename
6959 the files to the new names on upgrade.
6961 * For almost all files the license is now LGPL2.1+, changed
6962 from the previous GPL2.0+. Exceptions are some minor stuff
6963 of udev (which will be changed to LGPL2.1 eventually, too),
6964 and the MIT licensed sd-daemon.[ch] library that is suitable
6965 to be used as drop-in files.
6967 * systemd and logind now handle system sleep states, in
6968 particular suspending and hibernating.
6970 * logind now implements a sleep/shutdown/idle inhibiting logic
6971 suitable for a variety of uses. Soonishly Lennart will blog
6972 about this in more detail.
6974 * var-run.mount and var-lock.mount are no longer provided
6975 (which previously bind mounted these directories to their new
6976 places). Distributions which have not converted these
6977 directories to symlinks should consider stealing these files
6978 from git history and add them downstream.
6980 * We introduced the Documentation= field for units and added
6981 this to all our shipped units. This is useful to make it
6982 easier to explore the boot and the purpose of the various
6985 * All smaller setup units (such as
6986 systemd-vconsole-setup.service) now detect properly if they
6987 are run in a container and are skipped when
6988 appropriate. This guarantees an entirely noise-free boot in
6989 Linux container environments such as systemd-nspawn.
6991 * A framework for implementing offline system updates is now
6992 integrated, for details see:
6993 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
6995 * A new service type Type=idle is available now which helps us
6996 avoiding ugly interleaving of getty output and boot status
6999 * There's now a system-wide CapabilityBoundingSet= option to
7000 globally reduce the set of capabilities for the
7001 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
7002 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
7003 even CAP_NET_ADMIN system-wide for secure systems.
7005 * There are now system-wide DefaultLimitXXX= options to
7006 globally change the defaults of the various resource limits
7007 for all units started by PID 1.
7009 * Harald Hoyer's systemd test suite has been integrated into
7010 systemd which allows easy testing of systemd builds in qemu
7011 and nspawn. (This is really awesome! Ask us for details!)
7013 * The fstab parser is now implemented as generator, not inside
7016 * systemctl will now warn you if .mount units generated from
7017 /etc/fstab are out of date due to changes in fstab that
7018 have not been read by systemd yet.
7020 * systemd is now suitable for usage in initrds. Dracut has
7021 already been updated to make use of this. With this in place
7022 initrds get a slight bit faster but primarily are much
7023 easier to introspect and debug since "systemctl status" in
7024 the host system can be used to introspect initrd services,
7025 and the journal from the initrd is kept around too.
7027 * systemd-delta has been added, a tool to explore differences
7028 between user/admin configuration and vendor defaults.
7030 * PrivateTmp= now affects both /tmp and /var/tmp.
7032 * Boot time status messages are now much prettier and feature
7033 proper english language. Booting up systemd has never been
7036 * Read-ahead pack files now include the inode number of all
7037 files to pre-cache. When the inode changes the pre-caching
7038 is not attempted. This should be nicer to deal with updated
7039 packages which might result in changes of read-ahead
7042 * We now temporaritly lower the kernel's read_ahead_kb variable
7043 when collecting read-ahead data to ensure the kernel's
7044 built-in read-ahead does not add noise to our measurements
7045 of necessary blocks to pre-cache.
7047 * There's now RequiresMountsFor= to add automatic dependencies
7048 for all mounts necessary for a specific file system path.
7050 * MountAuto= and SwapAuto= have been removed from
7051 system.conf. Mounting file systems at boot has to take place
7054 * nspawn now learned a new switch --uuid= to set the machine
7055 ID on the command line.
7057 * nspawn now learned the -b switch to automatically search
7060 * vt102 is now the default TERM for serial TTYs, upgraded from
7063 * systemd-logind now works on VT-less systems.
7065 * The build tree has been reorganized. The individual
7066 components now have directories of their own.
7068 * A new condition type ConditionPathIsReadWrite= is now available.
7070 * nspawn learned the new -C switch to create cgroups for the
7071 container in other hierarchies.
7073 * We now have support for hardware watchdogs, configurable in
7076 * The scheduled shutdown logic now has a public API.
7078 * We now mount /tmp as tmpfs by default, but this can be
7079 masked and /etc/fstab can override it.
7081 * Since udisks does not make use of /media anymore we are not
7082 mounting a tmpfs on it anymore.
7084 * journalctl gained a new --local switch to only interleave
7085 locally generated journal files.
7087 * We can now load the IMA policy at boot automatically.
7089 * The GTK tools have been split off into a systemd-ui.
7091 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
7092 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
7093 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
7094 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
7095 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
7096 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
7097 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
7098 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
7099 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
7104 * This is mostly a bugfix release
7106 * Support optional initialization of the machine ID from the
7107 KVM or container configured UUID.
7109 * Support immediate reboots with "systemctl reboot -ff"
7111 * Show /etc/os-release data in systemd-analyze output
7113 * Many bugfixes for the journal, including endianness fixes and
7114 ensuring that disk space enforcement works
7116 * sd-login.h is C++ compatible again
7118 * Extend the /etc/os-release format on request of the Debian
7121 * We now refuse non-UTF8 strings used in various configuration
7122 and unit files. This is done to ensure we do not pass invalid
7123 data over D-Bus or expose it elsewhere.
7125 * Register Mimo USB Screens as suitable for automatic seat
7128 * Read SELinux client context from journal clients in a race
7131 * Reorder configuration file lookup order. /etc now always
7132 overrides /run in order to allow the administrator to always
7133 and unconditionally override vendor-supplied or
7134 automatically generated data.
7136 * The various user visible bits of the journal now have man
7137 pages. We still lack man pages for the journal API calls
7140 * We now ship all man pages in HTML format again in the
7143 Contributions from: Dave Reisner, Dirk Eibach, Frederic
7144 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
7145 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
7150 * This is mostly a bugfix release
7152 * systems lacking /etc/os-release are no longer supported.
7154 * Various functionality updates to libsystemd-login.so
7156 * Track class of PAM logins to distinguish greeters from
7159 Contributions from: Kay Sievers, Lennart Poettering, Michael
7164 * This is an important bugfix release for v41.
7166 * Building man pages is now optional which should be useful
7167 for those building systemd from git but unwilling to install
7170 * Watchdog support for supervising services is now usable. In
7171 a future release support for hardware watchdogs
7172 (i.e. /dev/watchdog) will be added building on this.
7174 * Service start rate limiting is now configurable and can be
7175 turned off per service. When a start rate limit is hit a
7176 reboot can automatically be triggered.
7178 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
7180 Contributions from: Benjamin Franzke, Bill Nottingham,
7181 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
7182 Schmidt, Michał Górny, Piotr Drąg
7186 * The systemd binary is installed /usr/lib/systemd/systemd now;
7187 An existing /sbin/init symlink needs to be adapted with the
7190 * The code that loads kernel modules has been ported to invoke
7191 libkmod directly, instead of modprobe. This means we do not
7192 support systems with module-init-tools anymore.
7194 * Watchdog support is now already useful, but still not
7197 * A new kernel command line option systemd.setenv= is
7198 understood to set system wide environment variables
7199 dynamically at boot.
7201 * We now limit the set of capabilities of systemd-journald.
7203 * We now set SIGPIPE to ignore by default, since it only is
7204 useful in shell pipelines, and has little use in general
7205 code. This can be disabled with IgnoreSIPIPE=no in unit
7208 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
7209 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
7214 * This is mostly a bugfix release
7216 * We now expose the reason why a service failed in the
7217 "Result" D-Bus property.
7219 * Rudimentary service watchdog support (will be completed over
7220 the next few releases.)
7222 * When systemd forks off in order execute some service we will
7223 now immediately changes its argv[0] to reflect which process
7224 it will execute. This is useful to minimize the time window
7225 with a generic argv[0], which makes bootcharts more useful
7227 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
7228 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
7229 Mike Kazantsev, Ray Strode
7233 * This is mostly a test release, but incorporates many
7236 * New systemd-cgtop tool to show control groups by their
7239 * Linking against libacl for ACLs is optional again. If
7240 disabled, support tracking device access for active logins
7241 goes becomes unavailable, and so does access to the user
7242 journals by the respective users.
7244 * If a group "adm" exists, journal files are automatically
7245 owned by them, thus allow members of this group full access
7246 to the system journal as well as all user journals.
7248 * The journal now stores the SELinux context of the logging
7249 client for all entries.
7251 * Add C++ inclusion guards to all public headers
7253 * New output mode "cat" in the journal to print only text
7254 messages, without any meta data like date or time.
7256 * Include tiny X server wrapper as a temporary stop-gap to
7257 teach XOrg udev display enumeration. This is used by display
7258 managers such as gdm, and will go away as soon as XOrg
7259 learned native udev hotplugging for display devices.
7261 * Add new systemd-cat tool for executing arbitrary programs
7262 with STDERR/STDOUT connected to the journal. Can also act as
7263 BSD logger replacement, and does so by default.
7265 * Optionally store all locally generated coredumps in the
7266 journal along with meta data.
7268 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
7269 writing short strings to files (for usage for /sys), and for
7270 creating symlinks, character and block device nodes.
7272 * New unit file option ControlGroupPersistent= to make cgroups
7273 persistent, following the mechanisms outlined in
7274 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
7276 * Support multiple local RTCs in a sane way
7278 * No longer monopolize IO when replaying readahead data on
7279 rotating disks, since we might starve non-file-system IO to
7280 death, since fanotify() will not see accesses done by blkid,
7283 * Do not show kernel threads in systemd-cgls anymore, unless
7284 requested with new -k switch.
7286 Contributions from: Dan Horák, Kay Sievers, Lennart
7287 Poettering, Michal Schmidt
7291 * This is mostly a test release, but incorporates many
7294 * The git repository moved to:
7295 git://anongit.freedesktop.org/systemd/systemd
7296 ssh://git.freedesktop.org/git/systemd/systemd
7298 * First release with the journal
7299 http://0pointer.de/blog/projects/the-journal.html
7301 * The journal replaces both systemd-kmsg-syslogd and
7302 systemd-stdout-bridge.
7304 * New sd_pid_get_unit() API call in libsystemd-logind
7306 * Many systemadm clean-ups
7308 * Introduce remote-fs-pre.target which is ordered before all
7309 remote mounts and may be used to start services before all
7312 * Added Mageia support
7314 * Add bash completion for systemd-loginctl
7316 * Actively monitor PID file creation for daemons which exit in
7317 the parent process before having finished writing the PID
7318 file in the daemon process. Daemons which do this need to be
7319 fixed (i.e. PID file creation must have finished before the
7320 parent exits), but we now react a bit more gracefully to them.
7322 * Add colourful boot output, mimicking the well-known output
7323 of existing distributions.
7325 * New option PassCredentials= for socket units, for
7326 compatibility with a recent kernel ABI breakage.
7328 * /etc/rc.local is now hooked in via a generator binary, and
7329 thus will no longer act as synchronization point during
7332 * systemctl list-unit-files now supports --root=.
7334 * systemd-tmpfiles now understands two new commands: z, Z for
7335 relabelling files according to the SELinux database. This is
7336 useful to apply SELinux labels to specific files in /sys,
7339 * Output of SysV services is now forwarded to both the console
7340 and the journal by default, not only just the console.
7342 * New man pages for all APIs from libsystemd-login.
7344 * The build tree got reorganized and the build system is a
7345 lot more modular allowing embedded setups to specifically
7346 select the components of systemd they are interested in.
7348 * Support for Linux systems lacking the kernel VT subsystem is
7351 * configure's --with-rootdir= got renamed to
7352 --with-rootprefix= to follow the naming used by udev and
7355 * Unless specified otherwise we will now install to /usr instead
7356 of /usr/local by default.
7358 * Processes with '@' in argv[0][0] are now excluded from the
7359 final shut-down killing spree, following the logic explained
7361 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
7363 * All processes remaining in a service cgroup when we enter
7364 the START or START_PRE states are now killed with
7365 SIGKILL. That means it is no longer possible to spawn
7366 background processes from ExecStart= lines (which was never
7367 supported anyway, and bad style).
7369 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
7370 reloading of units together.
7372 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
7373 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
7374 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
7375 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
7376 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek