1 systemd System and Service Manager
3 CHANGES WITH 239 in spe:
5 * NETWORK INTERFACE DEVICE NAMING CHANGES: elogind-udevd's "net_id"
6 builtin may name network interfaces differently than in previous
7 versions. SR-IOV virtual functions and NPAR partitions with PCI
8 function numbers of 8 and above will be named more predictably, and
9 udev may generate names based on the PCI slot number in some cases
10 where it previously did not.
12 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
13 elogind.service. Since v235, IPAddressDeny=any has been set to
14 the unit. So, it is expected that the default behavior of elogind
15 is not changed. However, if distribution packagers or administrators
16 disabled or modified IPAddressDeny= setting by a drop-in config file,
17 then it may be necessary to update the file to re-enable AF_INET and
18 AF_INET6 to support network user name services, e.g. NIS.
20 * When the RestrictNamespaces= unit property is specified multiple times,
21 then the specified types are merged now. Previously, only the last
22 assignment was used. So, if distribution packagers or administrators
23 modified the setting by a drop-in config file, then it may be necessary
26 * When OnFailure= is used in combination with Restart= on a service
27 unit, then the specified units will no longer be triggered on
28 failures that result in restarting. Previously, the specified units
29 would be activated each time the unit failed, even when the unit was
30 going to be restarted automatically. This behaviour contradicted the
31 documentation. With this release the code is adjusted to match the
34 * elogind-tmpfiles will now print a notice whenever it encounters
35 tmpfiles.d/ lines referencing the /var/run/ directory. It will
36 recommend reworking them to use the /run/ directory instead (for
37 which /var/run/ is simply a symlinked compatibility alias). This way
38 elogind-tmpfiles can properly detect line conflicts and merge lines
39 referencing the same file by two paths, without having to access
42 * systemctl disable/unmask/preset/preset-all cannot be used with
43 --runtime. Previously this was allowed, but resulted in unintuitive
44 behaviour that wasn't useful. systemctl disable/unmask will now
45 undo both runtime and persistent enablement/masking, i.e. it will
46 remove any relevant symlinks both in /run and /etc.
48 * sd-boot acquired new loader configuration settings to optionally turn
49 off Windows and MacOS boot partition discovery as well as
50 reboot-into-firmware menu items. It is also able to pick a better
51 screen resolution for HiDPI systems, and now provides loader
52 configuration settings to change the resolution explicitly.
54 * The elogind-resolve tool has been renamed to resolvectl (it also
55 remains available under the old name, for compatibility), and its
56 interface is now verb-based, similar in style to the other <xyz>ctl
57 tools, such as systemctl or loginctl.
59 * The resolvectl/elogind-resolve tool also provides 'resolveconf'
60 compatibility. It may be symlinked under the 'resolveconf' name, in
61 which case it will take arguments and input compatible with the
62 Debian and FreeBSD resolvconf tool.
64 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
65 where the system initially suspends, and after a time-out resumes and
68 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
69 set the client will only send a DUID as client identifier.
71 * The nss-elogind glibc NSS module will now enumerate dynamic users and
72 groups in effect. Previously, it could resolve UIDs/GIDs to user
73 names/groups and vice versa, but did not support enumeration.
75 * journald's Compress= configuration setting now optionally accepts a
76 byte threshold value. All journal objects larger than this threshold
77 will be compressed, smaller ones will not. Previously this threshold
78 was not configurable and set to 512.
80 * A new system.conf setting NoNewPrivileges= is now available which may
81 be used to turn off acquisition of new privileges system-wide
82 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
83 for all its children). Note that turning this option on means setuid
84 binaries and file system capabilities lose their special powers.
85 While turning on this option is a big step towards a more secure
86 system, doing so is likely to break numerous pre-existing UNIX tools,
87 in particular su and sudo.
89 * A new service elogind-time-sync-wait.service has been added. If
90 enabled it will delay the time-sync.target unit at boot until time
91 synchronization has been received from the network. This
92 functionality is useful on systems lacking a local RTC or where it is
93 acceptable that the boot process shall be delayed by external network
96 * When hibernating, elogind will now inform the kernel of the image
97 write offset, on kernels new enough to support this. This means swap
98 files should work for hibernation now.
100 * When loading unit files, elogind will now look for drop-in unit files
101 extensions in additional places. Previously, for a unit file name
102 "foo-bar-baz.service" it would look for dropin files in
103 "foo-bar-baz.service.d/*.conf". Now, it will also look in
104 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
105 service name truncated after all inner dashes. This scheme allows
106 writing drop-ins easily that apply to a whole set of unit files at
107 once. It's particularly useful for mount and slice units (as their
108 naming is prefix based), but is also useful for service and other
109 units, for packages that install multiple unit files at once,
110 following a strict naming regime of beginning the unit file name with
111 the package's name. Two new specifiers are now supported in unit
112 files to match this: %j and %J are replaced by the part of the unit
113 name following the last dash.
115 * Unit files and other configuration files that support specifier
116 expansion now understand another two new specifiers: %T and %V will
117 resolve to /tmp and /var/tmp respectively, or whatever temporary
118 directory has been set for the calling user.
120 * The ExecStart= lines of unit files are no longer required to
121 reference absolute paths. If non-absolute paths are specified the
122 specified binary name is searched within the service manager's
123 built-in $PATH, which may be queried with 'elogind-path
124 search-binaries-default'. It's generally recommended to continue to
125 use absolute paths for all binaries specified in unit files.
127 * Units gained a new load state "bad-setting", which is used when a
128 unit file was loaded, but contained fatal errors which prevent it
129 from being started (for example, an ExecStart= path which references
130 a non-existent executable).
132 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
133 support alternative debuggers, for example lldb. The old name
134 continues to be available however, for compatibility reasons. Use the
135 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
136 to pick an alternative debugger instead of the default gdb.
138 * systemctl and the other tools will now output escape sequences that
139 generate proper clickable hyperlinks in various terminal emulators
140 where useful (for example, in the "systemctl status" output you can
141 now click on the unit file name to quickly open it in the
142 editor/viewer of your choice). Note that not all terminal emulators
143 support this functionality yet, but many do. Unfortunately, the
144 "less" pager doesn't support this yet, hence this functionality is
145 currently automatically turned off when a pager is started (which
146 happens quite often due to auto-paging). We hope to remove this
147 limitation as soon as "less" learns these escape sequences. This new
148 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
149 environment variable. For details on these escape sequences see:
150 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
152 * networkd's .network files now support a new IPv6MTUBytes= option for
153 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
154 option in the [Route] section to configure the MTU to use for
155 specific routes. It also gained support for configuration of the DHCP
156 "UserClass" option through the new UserClass= setting. It gained
157 three new options in the new [CAN] section for configuring CAN
158 networks. The MULTICAST and ALLMULTI interface flags may now be
159 controlled explicitly with the new Multicast= and AllMulticast=
162 * networkd will now automatically make use of the kernel's route
163 expiration feature, if it is available.
165 * udevd's .link files now support setting the number of receive and
166 transmit channels, using the RxChannels=, TxChannels=,
167 OtherChannels=, CombinedChannels= settings.
169 * Support for UDPSegmentationOffload= has been removed, given its
170 limited support in hardware, and waning software support.
172 * networkd's .netdev files now support creating "netdevsim" interfaces.
174 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
175 to query the unit belonging to a specific kernel control group.
177 * elogind-analyze gained a new verb "cat-config", which may be used to
178 dump the contents of any configuration file, with all its matching
179 drop-in files added in, and honouring the usual search and masking
180 logic applied to elogind configuration files. For example use
181 "elogind-analyze cat-config elogind/system.conf" to get the complete
182 system configuration file of elogind how it would be loaded by PID 1
183 itself. Similar to this, various tools such as elogind-tmpfiles or
184 elogind-sysusers, gained a new option "--cat-config", which does the
185 corresponding operation for their own configuration settings. For
186 example, "elogind-tmpfiles --cat-config" will now output the full
187 list of tmpfiles.d/ lines in place.
189 * timedatectl gained two new verbs "timesync-status" (to show the
190 current NTP synchronization state of elogind-timesyncd) and
191 "show-timesync" (to show bus properties of elogind-timesyncd).
193 * elogind-timesyncd gained a bus interface on which it exposes details
196 * elogind-nspawn gained a new --rlimit= switch for setting initial
197 resource limits for the container payload. There's a new switch
198 --hostname= to explicitly override the container's hostname. A new
199 --no-new-privileges= switch may be used to control the
200 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
201 --oom-score-adjust= switch controls the OOM scoring adjustment value
202 for the payload. The new --cpu-affinity= switch controls the CPU
203 affinity of the container payload. The new --resolv-conf= switch
204 allows more detailed control of /etc/resolv.conf handling of the
205 container. Similarly, the new --timezone= switch allows more detailed
206 control of /etc/localtime handling of the container.
208 * elogind-detect-virt gained a new --list switch, which will print a
209 list of all currently known VM and container environments.
211 * Support for "Portable Services" has been added, see
212 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
213 experimental, but this is expected to change soon. Reflecting this
214 experimental state, the "portablectl" binary is not installed into
215 /usr/bin yet. The binary has to be called with the full path
216 /usr/lib/elogind/portablectl instead.
218 * journalctl's and systemctl's -o switch now knows a new log output
219 mode "with-unit". The output it generates is very similar to the
220 regular "short" mode, but displays the unit name instead of the
221 syslog tag for each log line. Also, the date is shown with timezone
222 information. This mode is probably more useful than the classic
223 "short" output mode for most purposes, except where pixel-perfect
224 compatibility with classic /var/log/messages formatting is required.
226 * A new --dump-bus-properties switch has been added to the elogind
227 binary, which may be used to dump all supported D-Bus properties.
228 (Options which are still supported, but are deprecated, are *not*
231 * sd-bus gained a set of new calls:
232 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
233 enable/disable the "floating" state of a bus slot object,
234 i.e. whether the slot object pins the bus it is allocated for into
235 memory or if the bus slot object gets disconnected when the bus goes
236 away. sd_bus_open_with_description(),
237 sd_bus_open_user_with_description(),
238 sd_bus_open_system_with_description() may be used to allocate bus
239 objects and set their description string already during allocation.
241 * sd-event gained support for watching inotify events from the event
242 loop, in an efficient way, sharing inotify handles between multiple
243 users. For this a new function sd_event_add_inotify() has been added.
245 * sd-event and sd-bus gained support for calling special user-supplied
246 destructor functions for userdata pointers associated with
247 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
248 functions sd_bus_slot_set_destroy_callback,
249 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
250 sd_bus_track_get_destroy_callback,
251 sd_event_source_set_destroy_callback,
252 sd_event_source_get_destroy_callback have been added.
254 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
256 * PID 1 will now automatically reschedule .timer units whenever the
257 local timezone changes. (They previously got rescheduled
258 automatically when the system clock changed.)
260 * New documentation has been added to document cgroups delegation,
261 portable services and the various code quality tools we have set up:
263 https://github.com/systemd/systemd/blob/master/doc/CGROUP_DELEGATION.md
264 https://github.com/systemd/systemd/blob/master/doc/PORTABLE_SERVICES.md
265 https://github.com/systemd/systemd/blob/master/doc/CODE_QUALITY.md
267 * pam_elogind will now look for PAM userdata fields elogind.memory_max,
268 elogind.tasks_max, elogind.cpu_weight, elogind.io_weight set by
269 earlier PAM modules. The data in these fields is used to initialize
270 the session scope's resource properties. Thus external PAM modules
271 may now configure per-session limits, for example sourced from
272 external user databases.
274 * socket units with Accept=yes will now maintain a "refused" counter in
275 addition to the existing "accepted" counter, counting connections
276 refused due to the enforced limits.
278 * The "elogind-path search-binaries-default" command may now be use to
279 query the default, built-in $PATH PID 1 will pass to the services it
282 * A new unit file setting PrivateMounts= has been added. It's a boolean
283 option. If enabled the unit's processes are invoked in their own file
284 system namespace. Note that this behaviour is also implied if any
285 other file system namespacing options (such as PrivateTmp=,
286 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
287 primarily useful for services that do not use any of the other file
288 system namespacing options. One such service is elogind-udevd.service
289 wher this is now used by default.
291 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
292 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Antique, Arnaud
293 Rebillout, Brian J. Murrell, Bruno Vernay, Chris Lesiak, Christian
294 Brauner, Christian Hesse, Daniel Dao, Daniel Lin, Danylo Korostil,
295 Davide Cavalca, David Tardon, Dimitri John Ledkov, Dmitriy Geels,
296 Douglas Christman, Elia Geretto, emelenas, Evegeny Vereshchagin, Evgeny
297 Vereshchagin, Felipe Sateler, Feng Sun, Filipe Brandenburger, Franck
298 Bui, futpib, Giuseppe Scrivano, Guillem Jover, guixxx, Hans de Goede,
299 Henrique Dante de Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko,
300 Ivan Shapovalov, James Cowgill, Jan Janssen, Jan Synacek, Jared
301 Kazimir, João Paulo Rechi Vita, Joost Heitbrink, juergbi, Jui-Chi Ricky
302 Liang, Kai-Heng Feng, Karol Augustin, Krzysztof Nowicki, Lauri
303 Tirkkonen, Lennart Poettering, Leonard, Long Li, Luca Boccassi, Lucas
304 Werkmeister, Marcel Hoppe, Marc Kleine-Budde, Mario Limonciello, Martin
305 Jansa, Martin Wilck, Mathieu Malaterre, Matteo F. Vescovi, Michael
306 Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal Sekletar,
307 Mike Gilbert, Mikhail Kasimov, Milan, Milan Broz, mourikwa, Muhammet
308 Kara, Nicolas Boichat, Omer Katz, Paride Legovini, Paul Menzel, Paul
309 Milliken, Peter A. Bigot, Peter Hutterer, Peter Jones, Philip Sequeira,
310 Philip Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de
311 Araujo, Rosen Penev, rubensa, Ryan Gonzalez, Salvo 'LtWorf' Tomaselli,
312 Sebastian Reichel, Sergio Lindo Mansilla, Stefan Schweter, Stephen
313 Hemminger, Stuart Hayes, Susant Sahani, Sylvain Plantefève, Thomas
314 H. P. Andersen, Tobias Jungel, Tomasz Torcz, Vito Caputo, Will Dietz,
315 Will Thompson, xginn8, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
321 * The MemoryAccounting= unit property now defaults to on. After
322 discussions with the upstream control group maintainers we learnt
323 that the negative impact of cgroup memory accounting on current
324 kernels is finally relatively minimal, so that it should be safe to
325 enable this by default without affecting system performance. Besides
326 memory accounting only task accounting is turned on by default, all
327 other forms of resource accounting (CPU, IO, IP) remain off for now,
328 because it's not clear yet that their impact is small enough to move
329 from opt-in to opt-out. We recommend downstreams to leave memory
330 accounting on by default if kernel 4.14 or higher is primarily
331 used. On very resource constrained systems or when support for old
332 kernels is a necessity, -Dmemory-accounting-default=false can be used
333 to revert this change.
335 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
336 %udev_rules_update) and the journal catalog (%journal_catalog_update)
337 from the upgrade scriptlets of individual packages now do nothing.
338 Transfiletriggers have been added which will perform those updates
339 once at the end of the transaction.
341 Similar transfiletriggers have been added to execute any sysctl.d
342 and binfmt.d rules. Thus, it should be unnecessary to provide any
343 scriptlets to execute this configuration from package installation
346 * elogind-sysusers gained a mode where the configuration to execute is
347 specified on the command line, but this configuration is not executed
348 directly, but instead it is merged with the configuration on disk,
349 and the result is executed. This is useful for package installation
350 scripts which want to create the user before installing any files on
351 disk (in case some of those files are owned by that user), while
352 still allowing local admin overrides.
354 This functionality is exposed to rpm scriptlets through a new
355 %sysusers_create_package macro. Old %sysusers_create and
356 %sysusers_create_inline macros are deprecated.
358 A transfiletrigger for sysusers.d configuration is now installed,
359 which means that it should be unnecessary to call elogind-sysusers from
360 package installation scripts, unless the package installs any files
361 owned by those newly-created users, in which case
362 %sysusers_create_package should be used.
364 * Analogous change has been done for elogind-tmpfiles: it gained a mode
365 where the command-line configuration is merged with the configuration
366 on disk. This is exposed as the new %tmpfiles_create_package macro,
367 and %tmpfiles_create is deprecated. A transfiletrigger is installed
368 for tmpfiles.d, hence it should be unnecessary to call elogind-tmpfiles
369 from package installation scripts.
371 * sysusers.d configuration for a user may now also specify the group
372 number, in addition to the user number ("u username 123:456"), or
373 without the user number ("u username -:456").
375 * Configution items for elogind-sysusers can now be specified as
376 positional arguments when the new --inline switch is used.
378 * The login shell of users created through sysusers.d may now be
379 specified (previously, it was always /bin/sh for root and
380 /sbin/nologin for other users).
382 * elogind-analyze gained a new --global switch to look at global user
383 configuration. It also gained a unit-paths verb to list the unit load
384 paths that are compiled into elogind (which can be used with
385 --elogind, --user, or --global).
387 * udevadm trigger gained a new --settle/-w option to wait for any
388 triggered events to finish (but just those, and not any other events
389 which are triggered meanwhile).
391 * The action that elogind takes when the lid is closed and the
392 machine is connected to external power can now be configured using
393 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
394 was determined by HandleLidSwitch=, and, for backwards compatibility,
395 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
397 * journalctl will periodically call sd_journal_process() to make it
398 resilient against inotify queue overruns when journal files are
399 rotated very quickly.
401 * Two new functions in libelogind — sd_bus_get_n_queued_read and
402 sd_bus_get_n_queued_write — may be used to check the number of
403 pending bus messages.
405 * elogind gained a new
406 org.freedesktop.elogind1.Manager.AttachProcessesToUnit dbus call
407 which can be used to migrate foreign processes to scope and service
408 units. The primary user for this new API is elogind itself: the
409 elogind --user instance uses this call of the elogind --system
410 instance to migrate processes if it itself gets the request to
411 migrate processes and the kernel refuses this due to access
412 restrictions. Thanks to this "elogind-run --scope --user …" works
413 again in pure cgroups v2 environments when invoked from the user
416 * A new TemporaryFileSystem= setting can be used to mask out part of
417 the real file system tree with tmpfs mounts. This may be combined
418 with BindPaths= and BindReadOnlyPaths= to hide files or directories
419 not relevant to the unit, while still allowing some paths lower in
420 the tree to be accessed.
422 ProtectHome=tmpfs may now be used to hide user home and runtime
423 directories from units, in a way that is mostly equivalent to
424 "TemporaryFileSystem=/home /run/user /root".
426 * Non-service units are now started with KeyringMode=shared by default.
427 This means that mount and swapon and other mount tools have access
428 to keys in the main keyring.
430 * /sys/fs/bpf is now mounted automatically.
432 * QNX virtualization is now detected by elogind-detect-virt and may
433 be used in ConditionVirtualization=.
435 * IPAccounting= may now be enabled also for slice units.
437 * A new -Dsplit-bin= build configuration switch may be used to specify
438 whether bin and sbin directories are merged, or if they should be
439 included separately in $PATH and various listings of executable
440 directories. The build configuration scripts will try to autodetect
441 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
442 system, but distributions are encouraged to configure this
445 * A new -Dok-color= build configuration switch may be used to change
446 the colour of "OK" status messages.
448 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
449 PrivateNetwork=yes was buggy in previous versions of elogind. This
450 means that after the upgrade and daemon-reexec, any such units must
453 * INCOMPATIBILITY: as announced in the NEWS for 237, elogind-tmpfiles
454 will not exclude read-only files owned by root from cleanup.
456 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
457 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
458 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
459 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
460 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
461 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
462 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
463 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
464 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
465 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
466 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
467 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
468 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
469 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
470 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
471 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
477 * Some keyboards come with a zoom see-saw or rocker which until now got
478 mapped to the Linux "zoomin/out" keys in hwdb. However, these
479 keycodes are not recognized by any major desktop. They now produce
480 Up/Down key events so that they can be used for scrolling.
482 * INCOMPATIBILITY: elogind-tmpfiles' "f" lines changed behaviour
483 slightly: previously, if an argument was specified for lines of this
484 type (i.e. the right-most column was set) this string was appended to
485 existing files each time elogind-tmpfiles was run. This behaviour was
486 different from what the documentation said, and not particularly
487 useful, as repeated elogind-tmpfiles invocations would not be
488 idempotent and grow such files without bounds. With this release
489 behaviour has been altered slightly, to match what the documentation
490 says: lines of this type only have an effect if the indicated files
491 don't exist yet, and only then the argument string is written to the
494 * FUTURE INCOMPATIBILITY: In elogind v238 we intend to slightly change
495 elogind-tmpfiles behaviour: previously, read-only files owned by root
496 were always excluded from the file "aging" algorithm (i.e. the
497 automatic clean-up of directories like /tmp based on
498 atime/mtime/ctime). We intend to drop this restriction, and age files
499 by default even when owned by root and read-only. This behaviour was
500 inherited from older tools, but there have been requests to remove
501 it, and it's not obvious why this restriction was made in the first
502 place. Please speak up now, if you are aware of software that reqires
503 this behaviour, otherwise we'll remove the restriction in v238.
505 * A new environment variable $SYSTEMD_OFFLINE is now understood by
506 systemctl. It takes a boolean argument. If on, systemctl assumes it
507 operates on an "offline" OS tree, and will not attempt to talk to the
508 service manager. Previously, this mode was implicitly enabled if a
509 chroot() environment was detected, and this new environment variable
510 now provides explicit control.
512 * .path and .socket units may now be created transiently, too.
513 Previously only service, mount, automount and timer units were
514 supported as transient units. The elogind-run tool has been updated
515 to expose this new functionality, you may hence use it now to bind
516 arbitrary commands to path or socket activation on-the-fly from the
517 command line. Moreover, almost all properties are now exposed for the
518 unit types that already supported transient operation.
520 * The elogind-mount command gained support for a new --owner= parameter
521 which takes a user name, which is then resolved and included in uid=
522 and gid= mount options string of the file system to mount.
524 * A new unit condition ConditionControlGroupController= has been added
525 that checks whether a specific cgroup controller is available.
527 * Unit files, udev's .link files, and elogind-networkd's .netdev and
528 .network files all gained support for a new condition
529 ConditionKernelVersion= for checking against specific kernel
532 * In elogind-networkd, the [IPVLAN] section in .netdev files gained
533 support for configuring device flags in the Flags= setting. In the
534 same files, the [Tunnel] section gained support for configuring
535 AllowLocalRemote=. The [Route] section in .network files gained
536 support for configuring InitialCongestionWindow=,
537 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
538 understands RapidCommit=.
540 * elogind-networkd's DHCPv6 support gained support for Prefix
543 * sd-bus gained support for a new "watch-bind" feature. When this
544 feature is enabled, an sd_bus connection may be set up to connect to
545 an AF_UNIX socket in the file system as soon as it is created. This
546 functionality is useful for writing early-boot services that
547 automatically connect to the system bus as soon as it is started,
548 without ugly time-based polling. elogind-networkd and
549 elogind-resolved have been updated to make use of this
550 functionality. busctl exposes this functionality in a new
551 --watch-bind= command line switch.
553 * sd-bus will now optionally synthesize a local "Connected" signal as
554 soon as a D-Bus connection is set up fully. This message mirrors the
555 already existing "Disconnected" signal which is synthesized when the
556 connection is terminated. This signal is generally useful but
557 particularly handy in combination with the "watch-bind" feature
558 described above. Synthesizing of this message has to be requested
559 explicitly through the new API call sd_bus_set_connected_signal(). In
560 addition a new call sd_bus_is_ready() has been added that checks
561 whether a connection is fully set up (i.e. between the "Connected" and
562 "Disconnected" signals).
564 * sd-bus gained two new calls sd_bus_request_name_async() and
565 sd_bus_release_name_async() for asynchronously registering bus
566 names. Similar, there is now sd_bus_add_match_async() for installing
567 a signal match asynchronously. All of elogind's own services have
568 been updated to make use of these calls. Doing these operations
569 asynchronously has two benefits: it reduces the risk of deadlocks in
570 case of cyclic dependencies between bus services, and it speeds up
571 service initialization since synchronization points for bus
572 round-trips are removed.
574 * sd-bus gained two new calls sd_bus_match_signal() and
575 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
576 and sd_bus_add_match_async() but instead of taking a D-Bus match
577 string take match fields as normal function parameters.
579 * sd-bus gained two new calls sd_bus_set_sender() and
580 sd_bus_message_set_sender() for setting the sender name of outgoing
581 messages (either for all outgoing messages or for just one specific
582 one). These calls are only useful in direct connections as on
583 brokered connections the broker fills in the sender anyway,
584 overwriting whatever the client filled in.
586 * sd-event gained a new pseudo-handle that may be specified on all API
587 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
588 used this refers to the default event loop object of the calling
589 thread. Note however that this does not implicitly allocate one —
590 which has to be done prior by using sd_event_default(). Similarly
591 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
592 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
593 to the default bus of the specified type of the calling thread. Here
594 too this does not implicitly allocate bus connection objects, this
595 has to be done prior with sd_bus_default() and friends.
597 * sd-event gained a new call pair
598 sd_event_source_{get|set}_io_fd_own(). This may be used to request
599 automatic closure of the file descriptor an IO event source watches
600 when the event source is destroyed.
602 * elogind-networkd gained support for natively configuring WireGuard
605 * In previous versions elogind synthesized user records both for the
606 "nobody" (UID 65534) and "root" (UID 0) users in nss-elogind and
607 internally. In order to simplify distribution-wide renames of the
608 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
609 new transitional flag file has been added: if
610 /etc/elogind/dont-synthesize-nobody exists synthesizing of the 65534
611 user and group record within the elogind codebase is disabled.
613 * elogind-notify gained a new --uid= option for selecting the source
614 user/UID to use for notification messages sent to the service
617 * journalctl gained a new --grep= option to list only entries in which
618 the message matches a certain pattern. By default matching is case
619 insensitive if the pattern is lowercase, and case sensitive
620 otherwise. Option --case-sensitive=yes|no can be used to override
621 this an specify case sensitivity or case insensitivity.
623 * There's now a "elogind-analyze service-watchdogs" command for printing
624 the current state of the service runtime watchdog, and optionally
625 enabling or disabling the per-service watchdogs system-wide if given a
626 boolean argument (i.e. the concept you configure in WatchdogSec=), for
627 debugging purposes. There's also a kernel command line option
628 elogind.service_watchdogs= for controlling the same.
630 * Two new "log-level" and "log-target" options for elogind-analyze were
631 added that merge the now deprecated get-log-level, set-log-level and
632 get-log-target, set-log-target pairs. The deprecated options are still
633 understood for backwards compatibility. The two new options print the
634 current value when no arguments are given, and set them when a
635 level/target is given as an argument.
637 * sysusers.d's "u" lines now optionally accept both a UID and a GID
638 specification, separated by a ":" character, in order to create users
639 where UID and GID do not match.
641 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
642 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
643 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
644 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
645 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
646 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
647 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
648 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
649 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
650 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
651 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
652 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
653 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
654 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
655 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
656 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
657 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
658 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
659 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
660 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
667 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
668 in v235 has been extended to also set the dummy.ko module option
669 numdummies=0, preventing the kernel from automatically creating
670 dummy0. All dummy interfaces must now be explicitly created.
672 * Unknown '%' specifiers in configuration files are now rejected. This
673 applies to units and tmpfiles.d configuration. Any percent characters
674 that are followed by a letter or digit that are not supposed to be
675 interpreted as the beginning of a specifier should be escaped by
676 doubling ("%%"). (So "size=5%" is still accepted, as well as
677 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
678 valid specifiers today.)
680 * systemd-resolved now maintains a new dynamic
681 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
682 recommended to make /etc/resolv.conf a symlink to it. This file
683 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
684 includes dynamically acquired search domains, achieving more correct
685 DNS resolution by software that bypasses local DNS APIs such as NSS.
687 * The "uaccess" udev tag has been dropped from /dev/kvm and
688 /dev/dri/renderD*. These devices now have the 0666 permissions by
689 default (but this may be changed at build-time). /dev/dri/renderD*
690 will now be owned by the "render" group along with /dev/kfd.
692 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
693 systemd-journal-gatewayd.service and
694 systemd-journal-upload.service. This means "nss-systemd" must be
695 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
696 services are resolved properly.
698 * In /etc/fstab two new mount options are now understood:
699 x-systemd.makefs and x-systemd.growfs. The former has the effect that
700 the configured file system is formatted before it is mounted, the
701 latter that the file system is resized to the full block device size
702 after it is mounted (i.e. if the file system is smaller than the
703 partition it resides on, it's grown). This is similar to the fsck
704 logic in /etc/fstab, and pulls in systemd-makefs@.service and
705 systemd-growfs@.service as necessary, similar to
706 systemd-fsck@.service. Resizing is currently only supported on ext4
709 * In systemd-networkd, the IPv6 RA logic now optionally may announce
710 DNS server and domain information.
712 * Support for the LUKS2 on-disk format for encrypted partitions has
713 been added. This requires libcryptsetup2 during compilation and
716 * The systemd --user instance will now signal "readiness" when its
717 basic.target unit has been reached, instead of when the run queue ran
718 empty for the first time.
720 * Tmpfiles.d with user configuration are now also supported.
721 systemd-tmpfiles gained a new --user switch, and snippets placed in
722 ~/.config/user-tmpfiles.d/ and corresponding directories will be
723 executed by systemd-tmpfiles --user running in the new
724 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
725 running in the user session.
727 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
728 %S resolves to the top-level state directory (/var/lib for the system
729 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
730 top-level cache directory (/var/cache for the system instance,
731 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
732 logs directory (/var/log for the system instance,
733 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
734 existing %t specifier, that resolves to the top-level runtime
735 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
738 * journalctl learnt a new parameter --output-fields= for limiting the
739 set of journal fields to output in verbose and JSON output modes.
741 * systemd-timesyncd's configuration file gained a new option
742 RootDistanceMaxSec= for setting the maximum root distance of servers
743 it'll use, as well as the new options PollIntervalMinSec= and
744 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
746 * bootctl gained a new command "list" for listing all available boot
747 menu items on systems that follow the boot loader specification.
749 * systemctl gained a new --dry-run switch that shows what would be done
750 instead of doing it, and is currently supported by the shutdown and
753 * ConditionSecurity= can now detect the TOMOYO security module.
755 * Unit file [Install] sections are now also respected in unit drop-in
756 files. This is intended to be used by drop-ins under /usr/lib/.
758 * systemd-firstboot may now also set the initial keyboard mapping.
760 * Udev "changed" events for devices which are exposed as systemd
761 .device units are now propagated to units specified in
762 ReloadPropagatedFrom= as reload requests.
764 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
765 unit template name (i.e. a name in the form of 'foobar@.service',
766 without the instance component between the '@' and - the '.'), then
767 the escaped sysfs path of the device is automatically used as the
770 * SystemCallFilter= in unit files has been extended so that an "errno"
771 can be specified individually for each system call. Example:
772 SystemCallFilter=~uname:EILSEQ.
774 * The cgroup delegation logic has been substantially updated. Delegate=
775 now optionally takes a list of controllers (instead of a boolean, as
776 before), which lists the controllers to delegate at least.
778 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
780 * A new LogLevelMax= setting configures the maximum log level any
781 process of the service may log at (i.e. anything with a lesser
782 priority than what is specified is automatically dropped). A new
783 LogExtraFields= setting allows configuration of additional journal
784 fields to attach to all log records generated by any of the unit's
787 * New StandardInputData= and StandardInputText= settings along with the
788 new option StandardInput=data may be used to configure textual or
789 binary data that shall be passed to the executed service process via
790 standard input, encoded in-line in the unit file.
792 * StandardInput=, StandardOutput= and StandardError= may now be used to
793 connect stdin/stdout/stderr of executed processes directly with a
794 file or AF_UNIX socket in the file system, using the new "file:" option.
796 * A new unit file option CollectMode= has been added, that allows
797 tweaking the garbage collection logic for units. It may be used to
798 tell systemd to garbage collect units that have failed automatically
799 (normally it only GCs units that exited successfully). systemd-run
800 and systemd-mount expose this new functionality with a new -G option.
802 * "machinectl bind" may now be used to bind mount non-directories
803 (i.e. regularfiles, devices, fifos, sockets).
805 * systemd-analyze gained a new verb "calendar" for validating and
806 testing calendar time specifications to use for OnCalendar= in timer
807 units. Besides validating the expression it will calculate the next
808 time the specified expression would elapse.
810 * In addition to the pre-existing FailureAction= unit file setting
811 there's now SuccessAction=, for configuring a shutdown action to
812 execute when a unit completes successfully. This is useful in
813 particular inside containers that shall terminate after some workload
814 has been completed. Also, both options are now supported for all unit
815 types, not just services.
817 * networkds's IP rule support gained two new options
818 IncomingInterface= and OutgoingInterface= for configuring the incoming
819 and outgoing interfaces of configured rules. systemd-networkd also
820 gained support for "vxcan" network devices.
822 * networkd gained a new setting RequiredForOnline=, taking a
823 boolean. If set, systemd-wait-online will take it into consideration
824 when determining that the system is up, otherwise it will ignore the
825 interface for this purpose.
827 * The sd_notify() protocol gained support for a new operation: with
828 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
829 store again, ahead of POLLHUP or POLLERR when they are removed
832 * A new document doc/UIDS-GIDS.md has been added to the source tree,
833 that documents the UID/GID range and assignment assumptions and
834 requirements of systemd.
836 * The watchdog device PID 1 will ping may now be configured through the
837 WatchdogDevice= configuration file setting, or by setting the
838 systemd.watchdog_service= kernel commandline option.
840 * systemd-resolved's gained support for registering DNS-SD services on
841 the local network using MulticastDNS. Services may either be
842 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
843 the same dir below /run, /usr/lib), or through its D-Bus API.
845 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
846 extend the effective start, runtime, and stop time. The service must
847 continue to send EXTEND_TIMEOUT_USEC within the period specified to
848 prevent the service manager from making the service as timedout.
850 * elogind-resolved's DNSSEC support gained support for RFC 8080
851 (Ed25519 keys and signatures).
853 * The elogind-resolve command line tool gained a new set of options
854 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
855 --set-nta= and --revert to configure per-interface DNS configuration
856 dynamically during runtime. It's useful for pushing DNS information
857 into elogind-resolved from DNS hook scripts that various interface
858 managing software supports (such as pppd).
860 * elogind-nspawn gained a new --network-namespace-path= command line
861 option, which may be used to make a container join an existing
862 network namespace, by specifying a path to a "netns" file.
864 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
865 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
866 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
867 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
868 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
869 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
870 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
871 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
872 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
873 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
874 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
875 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
876 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
877 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
878 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
879 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
880 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
881 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
882 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
883 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
884 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
885 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
886 Jędrzejewski-Szmek, Zeal Jagannatha
892 * INCOMPATIBILITY: systemd-logind.service and other long-running
893 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
894 communication with the outside. This generally improves security of
895 the system, and is in almost all cases a safe and good choice, as
896 these services do not and should not provide any network-facing
897 functionality. However, systemd-logind uses the glibc NSS API to
898 query the user database. This creates problems on systems where NSS
899 is set up to directly consult network services for user database
900 lookups. In particular, this creates incompatibilities with the
901 "nss-nis" module, which attempts to directly contact the NIS/YP
902 network servers it is configured for, and will now consistently
903 fail. In such cases, it is possible to turn off IP sandboxing for
904 systemd-logind.service (set IPAddressDeny= in its [Service] section
905 to the empty string, via a .d/ unit file drop-in). Downstream
906 distributions might want to update their nss-nis packaging to include
907 such a drop-in snippet, accordingly, to hide this incompatibility
908 from the user. Another option is to make use of glibc's nscd service
909 to proxy such network requests through a privilege-separated, minimal
910 local caching daemon, or to switch to more modern technologies such
911 sssd, whose NSS hook-ups generally do not involve direct network
912 access. In general, we think it's definitely time to question the
913 implementation choices of nss-nis, i.e. whether it's a good idea
914 today to embed a network-facing loadable module into all local
915 processes that need to query the user database, including the most
916 trivial and benign ones, such as "ls". For more details about
917 IPAddressDeny= see below.
919 * A new modprobe.d drop-in is now shipped by default that sets the
920 bonding module option max_bonds=0. This overrides the kernel default,
921 to avoid conflicts and ambiguity as to whether or not bond0 should be
922 managed by systemd-networkd or not. This resolves multiple issues
923 with bond0 properties not being applied, when bond0 is configured
924 with systemd-networkd. Distributors may choose to not package this,
925 however in that case users will be prevented from correctly managing
926 bond0 interface using systemd-networkd.
928 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
929 which print the logging level and target of the system manager. They
930 complement the existing "set-log-level" and "set-log-target" verbs
931 used to change those values.
933 * journald.conf gained a new boolean setting ReadKMsg= which defaults
934 to on. If turned off kernel log messages will not be read by
935 systemd-journald or included in the logs. It also gained a new
936 setting LineMax= for configuring the maximum line length in
937 STDOUT/STDERR log streams. The new default for this value is 48K, up
938 from the previous hardcoded 2048.
940 * A new unit setting RuntimeDirectoryPreserve= has been added, which
941 allows more detailed control of what to do with a runtime directory
942 configured with RuntimeDirectory= (i.e. a directory below /run or
943 $XDG_RUNTIME_DIR) after a unit is stopped.
945 * The RuntimeDirectory= setting for units gained support for creating
946 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
947 one top-level directory.
949 * Units gained new options StateDirectory=, CacheDirectory=,
950 LogsDirectory= and ConfigurationDirectory= which are closely related
951 to RuntimeDirectory= but manage per-service directories below
952 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
953 possible to write unit files which when activated automatically gain
954 properly owned service specific directories in these locations, thus
955 making unit files self-contained and increasing compatibility with
956 stateless systems and factory reset where /etc or /var are
957 unpopulated at boot. Matching these new settings there's also
958 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
959 ConfigurationDirectoryMode= for configuring the access mode of these
960 directories. These settings are particularly useful in combination
961 with DynamicUser=yes as they provide secure, properly-owned,
962 writable, and stateful locations for storage, excluded from the
963 sandbox that such services live in otherwise.
965 * Automake support has been removed from this release. systemd is now
968 * systemd-journald will now aggressively cache client metadata during
969 runtime, speeding up log write performance under pressure. This comes
970 at a small price though: as much of the metadata is read
971 asynchronously from /proc/ (and isn't implicitly attached to log
972 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
973 metadata stored alongside a log entry might be slightly
974 out-of-date. Previously it could only be slightly newer than the log
975 message. The time window is small however, and given that the kernel
976 is unlikely to be improved anytime soon in this regard, this appears
979 * nss-myhostname/systemd-resolved will now by default synthesize an
980 A/AAAA resource record for the "_gateway" hostname, pointing to the
981 current default IP gateway. Previously it did that for the "gateway"
982 name, hampering adoption, as some distributions wanted to leave that
983 host name open for local use. The old behaviour may still be
984 requested at build time.
986 * systemd-networkd's [Address] section in .network files gained a new
987 Scope= setting for configuring the IP address scope. The [Network]
988 section gained a new boolean setting ConfigureWithoutCarrier= that
989 tells systemd-networkd to ignore link sensing when configuring the
990 device. The [DHCP] section gained a new Anonymize= boolean option for
991 turning on a number of options suggested in RFC 7844. A new
992 [RoutingPolicyRule] section has been added for configuring the IP
993 routing policy. The [Route] section has gained support for a new
994 Type= setting which permits configuring
995 blackhole/unreachable/prohibit routes.
997 * The [VRF] section in .netdev files gained a new Table= setting for
998 configuring the routing table to use. The [Tunnel] section gained a
999 new Independent= boolean field for configuring tunnels independent of
1000 an underlying network interface. The [Bridge] section gained a new
1001 GroupForwardMask= option for configuration of propagation of link
1002 local frames between bridge ports.
1004 * The WakeOnLan= setting in .link files gained support for a number of
1005 new modes. A new TCP6SegmentationOffload= setting has been added for
1006 configuring TCP/IPv6 hardware segmentation offload.
1008 * The IPv6 RA sender implementation may now optionally send out RDNSS
1009 and RDNSSL records to supply DNS configuration to peers.
1011 * systemd-nspawn gained support for a new --system-call-filter= command
1012 line option for adding and removing entries in the default system
1013 call filter it applies. Moreover systemd-nspawn has been changed to
1014 implement a system call whitelist instead of a blacklist.
1016 * systemd-run gained support for a new --pipe command line option. If
1017 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
1018 are directly passed on to the activated transient service
1019 executable. This allows invoking arbitrary processes as systemd
1020 services (for example to take benefit of dependency management,
1021 accounting management, resource management or log management that is
1022 done automatically for services) — while still allowing them to be
1023 integrated in a classic UNIX shell pipeline.
1025 * When a service sends RELOAD=1 via sd_notify() and reload propagation
1026 using ReloadPropagationTo= is configured, a reload is now propagated
1027 to configured units. (Previously this was only done on explicitly
1028 requested reloads, using "systemctl reload" or an equivalent
1031 * For each service unit a restart counter is now kept: it is increased
1032 each time the service is restarted due to Restart=, and may be
1033 queried using "systemctl show -p NRestarts …".
1035 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
1036 @signal and @timer have been added, for usage with SystemCallFilter=
1037 in unit files and the new --system-call-filter= command line option
1038 of systemd-nspawn (see above).
1040 * ExecStart= lines in unit files gained two new modifiers: when a
1041 command line is prefixed with "!" the command will be executed as
1042 configured, except for the credentials applied by
1043 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
1044 "+", but does still apply namespacing options unlike "+". There's
1045 also "!!" now, which is mostly identical, but becomes a NOP on
1046 systems that support ambient capabilities. This is useful to write
1047 unit files that work with ambient capabilities where possible but
1048 automatically fall back to traditional privilege dropping mechanisms
1049 on systems where this is not supported.
1051 * ListenNetlink= settings in socket units now support RDMA netlink
1054 * A new unit file setting LockPersonality= has been added which permits
1055 locking down the chosen execution domain ("personality") of a service
1058 * A new special target "getty-pre.target" has been added, which is
1059 ordered before all text logins, and may be used to order services
1060 before textual logins acquire access to the console.
1062 * systemd will now attempt to load the virtio-rng.ko kernel module very
1063 early on if a VM environment supporting this is detected. This should
1064 improve entropy during early boot in virtualized environments.
1066 * A _netdev option is now supported in /etc/crypttab that operates in a
1067 similar way as the same option in /etc/fstab: it permits configuring
1068 encrypted devices that need to be ordered after the network is up.
1069 Following this logic, two new special targets
1070 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
1071 added that are to cryptsetup.target what remote-fs.target and
1072 remote-fs-pre.target are to local-fs.target.
1074 * Service units gained a new UnsetEnvironment= setting which permits
1075 unsetting specific environment variables for services that are
1076 normally passed to it (for example in order to mask out locale
1077 settings for specific services that can't deal with it).
1079 * Units acquired a new boolean option IPAccounting=. When turned on, IP
1080 traffic accounting (packet count as well as byte count) is done for
1081 the service, and shown as part of "systemctl status" or "systemd-run
1084 * Service units acquired two new options IPAddressAllow= and
1085 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
1086 for configuring a simple IP access control list for all sockets of
1087 the unit. These options are available also on .slice and .socket
1088 units, permitting flexible access list configuration for individual
1089 services as well as groups of services (as defined by a slice unit),
1090 including system-wide. Note that IP ACLs configured this way are
1091 enforced on every single IPv4 and IPv6 socket created by any process
1092 of the service unit, and apply to ingress as well as egress traffic.
1094 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
1095 structured log message is generated each time the unit is stopped,
1096 containing information about the consumed resources of this
1099 * A new setting KeyringMode= has been added to unit files, which may be
1100 used to control how the kernel keyring is set up for executed
1103 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
1104 "systemctl kexec" and "systemctl exit" are now always asynchronous in
1105 behaviour (that is: these commands return immediately after the
1106 operation was enqueued instead of waiting for the operation to
1107 complete). Previously, "systemctl poweroff" and "systemctl reboot"
1108 were asynchronous on systems using systemd-logind (i.e. almost
1109 always, and like they were on sysvinit), and the other three commands
1110 were unconditionally synchronous. With this release this is cleaned
1111 up, and callers will see the same asynchronous behaviour on all
1112 systems for all five operations.
1114 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
1117 * .timer units now accept calendar specifications in other timezones
1118 than UTC or the local timezone.
1120 * The tmpfiles snippet var.conf has been changed to create
1121 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
1122 the "utmp" group already, and it appears to be generally understood
1123 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
1124 databases. Previously this was implemented correctly for all these
1125 databases excepts btmp, which has been opened up like this now
1126 too. Note that while the other databases are world-readable
1127 (i.e. 0644), btmp is not and remains more restrictive.
1129 * The systemd-resolve tool gained a new --reset-server-features
1130 switch. When invoked like this systemd-resolved will forget
1131 everything it learnt about the features supported by the configured
1132 upstream DNS servers, and restarts the feature probing logic on the
1133 next resolver look-up for them at the highest feature level
1136 * The status dump systemd-resolved sends to the logs upon receiving
1137 SIGUSR1 now also includes information about all DNS servers it is
1138 configured to use, and the features levels it probed for them.
1140 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
1141 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
1142 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
1143 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
1144 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
1145 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
1146 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
1147 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
1148 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
1149 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
1150 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
1151 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
1152 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
1153 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
1154 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
1155 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
1156 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
1157 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
1158 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
1159 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1161 — Berlin, 2017-10-06
1165 * Meson is now supported as build system in addition to Automake. It is
1166 our plan to remove Automake in one of our next releases, so that
1167 Meson becomes our exclusive build system. Hence, please start using
1168 the Meson build system in your downstream packaging. There's plenty
1169 of documentation around how to use Meson, the extremely brief
1172 ./autogen.sh && ./configure && make && sudo make install
1176 meson build && ninja -C build && sudo ninja -C build install
1178 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
1179 which permits configuring a timeout on the time a job is
1180 running. This is particularly useful for setting timeouts on jobs for
1183 * Unit files gained two new options ConditionUser= and ConditionGroup=
1184 for conditionalizing units based on the identity of the user/group
1185 running a systemd user instance.
1187 * systemd-networkd now understands a new FlowLabel= setting in the
1188 [VXLAN] section of .network files, as well as a Priority= in
1189 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
1190 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
1191 gained support for configuration of GENEVE links, and IPv6 address
1192 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
1194 * .link files now understand a new Port= setting.
1196 * systemd-networkd's DHCP support gained support for DHCP option 119
1197 (domain search list).
1199 * systemd-networkd gained support for serving IPv6 address ranges using
1200 the Router Advertisement protocol. The new .network configuration
1201 section [IPv6Prefix] may be used to configure the ranges to
1202 serve. This is implemented based on a new, minimal, native server
1203 implementation of RA.
1205 * journalctl's --output= switch gained support for a new parameter
1206 "short-iso-precise" for a mode where timestamps are shown as precise
1209 * systemd-udevd's "net_id" builtin may now generate stable network
1210 interface names from IBM PowerVM VIO devices as well as ACPI platform
1213 * MulticastDNS support in systemd-resolved may now be explicitly
1214 enabled/disabled using the new MulticastDNS= configuration file
1217 * systemd-resolved may now optionally use libidn2 instead of the libidn
1218 for processing internationalized domain names. Support for libidn2
1219 should be considered experimental and should not be enabled by
1222 * "machinectl pull-tar" and related call may now do verification of
1223 downloaded images using SUSE-style .sha256 checksum files in addition
1224 to the already existing support for validating using Ubuntu-style
1227 * sd-bus gained support for a new sd_bus_message_appendv() call which
1228 is va_list equivalent of sd_bus_message_append().
1230 * sd-boot gained support for validating images using SHIM/MOK.
1232 * The SMACK code learnt support for "onlycap".
1234 * systemd-mount --umount is now much smarter in figuring out how to
1235 properly unmount a device given its mount or device path.
1237 * The code to call libnss_dns as a fallback from libnss_resolve when
1238 the communication with systemd-resolved fails was removed. This
1239 fallback was redundant and interfered with the [!UNAVAIL=return]
1240 suffix. See nss-resolve(8) for the recommended configuration.
1242 * systemd-logind may now be restarted without losing state. It stores
1243 the file descriptors for devices it manages in the system manager
1244 using the FDSTORE= mechanism. Please note that further changes in
1245 other components may be required to make use of this (for example
1246 Xorg has code to listen for stops of systemd-logind and terminate
1247 itself when logind is stopped or restarted, in order to avoid using
1248 stale file descriptors for graphical devices, which is now
1249 counterproductive and must be reverted in order for restarts of
1250 systemd-logind to be safe. See
1251 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
1253 * All kernel install plugins are called with the environment variable
1254 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
1255 /etc/machine-id. If the file is missing or empty, the variable is
1256 empty and BOOT_DIR_ABS is the path of a temporary directory which is
1257 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
1258 is empty, all plugins should not put anything in BOOT_DIR_ABS.
1260 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
1261 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
1262 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
1263 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
1264 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
1265 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
1266 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
1267 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
1268 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
1269 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
1270 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
1271 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
1272 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
1273 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
1274 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
1275 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
1276 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
1277 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
1278 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
1279 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
1280 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
1281 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
1282 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
1283 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
1284 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
1285 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
1286 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
1287 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
1288 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
1291 — Berlin, 2017-07-12
1295 * The "hybrid" control group mode has been modified to improve
1296 compatibility with "legacy" cgroups-v1 setups. Specifically, the
1297 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
1298 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
1299 cgroups-v1 hierarchy), the only externally visible change being that
1300 the cgroups-v2 hierarchy is also mounted, to
1301 /sys/fs/cgroup/unified. This should provide a large degree of
1302 compatibility with "legacy" cgroups-v1, while taking benefit of the
1303 better management capabilities of cgroups-v2.
1305 * The default control group setup mode may be selected both a boot-time
1306 via a set of kernel command line parameters (specifically:
1307 systemd.unified_cgroup_hierarchy= and
1308 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
1309 default selected on the configure command line
1310 (--with-default-hierarchy=). The upstream default is "hybrid"
1311 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
1312 this will change in a future systemd version to be "unified" (pure
1313 cgroups-v2 mode). The third option for the compile time option is
1314 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
1315 distributions to default to "hybrid" mode for release distributions,
1316 starting with v233. We recommend "unified" for development
1317 distributions (specifically: distributions such as Fedora's rawhide)
1318 as that's where things are headed in the long run. Use "legacy" for
1319 greatest stability and compatibility only.
1321 * Note one current limitation of "unified" and "hybrid" control group
1322 setup modes: the kernel currently does not permit the systemd --user
1323 instance (i.e. unprivileged code) to migrate processes between two
1324 disconnected cgroup subtrees, even if both are managed and owned by
1325 the user. This effectively means "systemd-run --user --scope" doesn't
1326 work when invoked from outside of any "systemd --user" service or
1327 scope. Specifically, it is not supported from session scopes. We are
1328 working on fixing this in a future systemd version. (See #3388 for
1329 further details about this.)
1331 * DBus policy files are now installed into /usr rather than /etc. Make
1332 sure your system has dbus >= 1.9.18 running before upgrading to this
1333 version, or override the install path with --with-dbuspolicydir= .
1335 * All python scripts shipped with systemd (specifically: the various
1336 tests written in Python) now require Python 3.
1338 * systemd unit tests can now run standalone (without the source or
1339 build directories), and can be installed into /usr/lib/systemd/tests/
1340 with 'make install-tests'.
1342 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
1343 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
1346 * Support for the %c, %r, %R specifiers in unit files has been
1347 removed. Specifiers are not supposed to be dependent on configuration
1348 in the unit file itself (so that they resolve the same regardless
1349 where used in the unit files), but these specifiers were influenced
1350 by the Slice= option.
1352 * The shell invoked by debug-shell.service now defaults to /bin/sh in
1353 all cases. If distributions want to use a different shell for this
1354 purpose (for example Fedora's /sbin/sushell) they need to specify
1355 this explicitly at configure time using --with-debug-shell=.
1357 * The confirmation spawn prompt has been reworked to offer the
1360 (c)ontinue, proceed without asking anymore
1361 (D)ump, show the state of the unit
1362 (f)ail, don't execute the command and pretend it failed
1364 (i)nfo, show a short summary of the unit
1365 (j)obs, show jobs that are in progress
1366 (s)kip, don't execute the command and pretend it succeeded
1367 (y)es, execute the command
1369 The 'n' choice for the confirmation spawn prompt has been removed,
1370 because its meaning was confusing.
1372 The prompt may now also be redirected to an alternative console by
1373 specifying the console as parameter to systemd.confirm_spawn=.
1375 * Services of Type=notify require a READY=1 notification to be sent
1376 during startup. If no such message is sent, the service now fails,
1377 even if the main process exited with a successful exit code.
1379 * Services that fail to start up correctly now always have their
1380 ExecStopPost= commands executed. Previously, they'd enter "failed"
1381 state directly, without executing these commands.
1383 * The option MulticastDNS= of network configuration files has acquired
1384 an actual implementation. With MulticastDNS=yes a host can resolve
1385 names of remote hosts and reply to mDNS A and AAAA requests.
1387 * When units are about to be started an additional check is now done to
1388 ensure that all dependencies of type BindsTo= (when used in
1389 combination with After=) have been started.
1391 * systemd-analyze gained a new verb "syscall-filter" which shows which
1392 system call groups are defined for the SystemCallFilter= unit file
1393 setting, and which system calls they contain.
1395 * A new system call filter group "@filesystem" has been added,
1396 consisting of various file system related system calls. Group
1397 "@reboot" has been added, covering reboot, kexec and shutdown related
1398 calls. Finally, group "@swap" has been added covering swap
1399 configuration related calls.
1401 * A new unit file option RestrictNamespaces= has been added that may be
1402 used to restrict access to the various process namespace types the
1403 Linux kernel provides. Specifically, it may be used to take away the
1404 right for a service unit to create additional file system, network,
1405 user, and other namespaces. This sandboxing option is particularly
1406 relevant due to the high amount of recently discovered namespacing
1407 related vulnerabilities in the kernel.
1409 * systemd-udev's .link files gained support for a new AutoNegotiation=
1410 setting for configuring Ethernet auto-negotiation.
1412 * systemd-networkd's .network files gained support for a new
1413 ListenPort= setting in the [DHCP] section to explicitly configure the
1414 UDP client port the DHCP client shall listen on.
1416 * .network files gained a new Unmanaged= boolean setting for explicitly
1417 excluding one or more interfaces from management by systemd-networkd.
1419 * The systemd-networkd ProxyARP= option has been renamed to
1420 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
1421 renamed to ReduceARPProxy=. The old names continue to be available
1424 * systemd-networkd gained support for configuring IPv6 Proxy NDP
1425 addresses via the new IPv6ProxyNDPAddress= .network file setting.
1427 * systemd-networkd's bonding device support gained support for two new
1428 configuration options ActiveSlave= and PrimarySlave=.
1430 * The various options in the [Match] section of .network files gained
1431 support for negative matching.
1433 * New systemd-specific mount options are now understood in /etc/fstab:
1435 x-systemd.mount-timeout= may be used to configure the maximum
1436 permitted runtime of the mount command.
1438 x-systemd.device-bound may be set to bind a mount point to its
1439 backing device unit, in order to automatically remove a mount point
1440 if its backing device is unplugged. This option may also be
1441 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
1442 on the block device, which is now automatically set for all CDROM
1443 drives, so that mounted CDs are automatically unmounted when they are
1444 removed from the drive.
1446 x-systemd.after= and x-systemd.before= may be used to explicitly
1447 order a mount after or before another unit or mount point.
1449 * Enqueued start jobs for device units are now automatically garbage
1450 collected if there are no jobs waiting for them anymore.
1452 * systemctl list-jobs gained two new switches: with --after, for every
1453 queued job the jobs it's waiting for are shown; with --before the
1454 jobs which it's blocking are shown.
1456 * systemd-nspawn gained support for ephemeral boots from disk images
1457 (or in other words: --ephemeral and --image= may now be
1458 combined). Moreover, ephemeral boots are now supported for normal
1459 directories, even if the backing file system is not btrfs. Of course,
1460 if the file system does not support file system snapshots or
1461 reflinks, the initial copy operation will be relatively expensive, but
1462 this should still be suitable for many use cases.
1464 * Calendar time specifications in .timer units now support
1465 specifications relative to the end of a month by using "~" instead of
1466 "-" as separator between month and day. For example, "*-02~03" means
1467 "the third last day in February". In addition a new syntax for
1468 repeated events has been added using the "/" character. For example,
1469 "9..17/2:00" means "every two hours from 9am to 5pm".
1471 * systemd-socket-proxyd gained a new parameter --connections-max= for
1472 configuring the maximum number of concurrent connections.
1474 * sd-id128 gained a new API for generating unique IDs for the host in a
1475 way that does not leak the machine ID. Specifically,
1476 sd_id128_get_machine_app_specific() derives an ID based on the
1477 machine ID a in well-defined, non-reversible, stable way. This is
1478 useful whenever an identifier for the host is needed but where the
1479 identifier shall not be useful to identify the system beyond the
1480 scope of the application itself. (Internally this uses HMAC-SHA256 as
1481 keyed hash function using the machine ID as input.)
1483 * NotifyAccess= gained a new supported value "exec". When set
1484 notifications are accepted from all processes systemd itself invoked,
1485 including all control processes.
1487 * .nspawn files gained support for defining overlay mounts using the
1488 Overlay= and OverlayReadOnly= options. Previously this functionality
1489 was only available on the systemd-nspawn command line.
1491 * systemd-nspawn's --bind= and --overlay= options gained support for
1492 bind/overlay mounts whose source lies within the container tree by
1493 prefixing the source path with "+".
1495 * systemd-nspawn's --bind= and --overlay= options gained support for
1496 automatically allocating a temporary source directory in /var/tmp
1497 that is removed when the container dies. Specifically, if the source
1498 directory is specified as empty string this mechanism is selected. An
1499 example usage is --overlay=+/var::/var, which creates an overlay
1500 mount based on the original /var contained in the image, overlayed
1501 with a temporary directory in the host's /var/tmp. This way changes
1502 to /var are automatically flushed when the container shuts down.
1504 * systemd-nspawn --image= option does now permit raw file system block
1505 devices (in addition to images containing partition tables, as
1508 * The disk image dissection logic in systemd-nspawn gained support for
1509 automatically setting up LUKS encrypted as well as Verity protected
1510 partitions. When a container is booted from an encrypted image the
1511 passphrase is queried at start-up time. When a container with Verity
1512 data is started, the root hash is search in a ".roothash" file
1513 accompanying the disk image (alternatively, pass the root hash via
1514 the new --root-hash= command line option).
1516 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
1517 be used to dissect disk images the same way as systemd-nspawn does
1518 it, following the Bootable Partition Specification. It may even be
1519 used to mount disk images with complex partition setups (including
1520 LUKS and Verity partitions) to a local host directory, in order to
1521 inspect them. This tool is not considered public API (yet), and is
1522 thus not installed into /usr/bin. Please do not rely on its
1523 existence, since it might go away or be changed in later systemd
1526 * A new generator "systemd-verity-generator" has been added, similar in
1527 style to "systemd-cryptsetup-generator", permitting automatic setup of
1528 Verity root partitions when systemd boots up. In order to make use of
1529 this your partition setup should follow the Discoverable Partitions
1530 Specification, and the GPT partition ID of the root file system
1531 partition should be identical to the upper 128bit of the Verity root
1532 hash. The GPT partition ID of the Verity partition protecting it
1533 should be the lower 128bit of the Verity root hash. If the partition
1534 image follows this model it is sufficient to specify a single
1535 "roothash=" kernel command line argument to both configure which root
1536 image and verity partition to use as well as the root hash for
1537 it. Note that systemd-nspawn's Verity support follows the same
1538 semantics, meaning that disk images with proper Verity data in place
1539 may be booted in containers with systemd-nspawn as well as on
1540 physical systems via the verity generator. Also note that the "mkosi"
1541 tool available at https://github.com/systemd/mkosi has been updated
1542 to generate Verity protected disk images following this scheme. In
1543 fact, it has been updated to generate disk images that optionally
1544 implement a complete UEFI SecureBoot trust chain, involving a signed
1545 kernel and initrd image that incorporates such a root hash as well as
1546 a Verity-enabled root partition.
1548 * The hardware database (hwdb) udev supports has been updated to carry
1549 accelerometer quirks.
1551 * All system services are now run with a fresh kernel keyring set up
1552 for them. The invocation ID is stored by default in it, thus
1553 providing a safe, non-overridable way to determine the invocation
1556 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
1557 options for bind mounting arbitrary paths in a service-specific
1558 way. When these options are used, arbitrary host or service files and
1559 directories may be mounted to arbitrary locations in the service's
1562 * Documentation has been added that lists all of systemd's low-level
1563 environment variables:
1565 https://github.com/systemd/systemd/blob/master/doc/ENVIRONMENT.md
1567 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
1568 whether a specific socket file descriptor matches a specified socket
1571 * systemd-firstboot has been updated to check for the
1572 systemd.firstboot= kernel command line option. It accepts a boolean
1573 and when set to false the first boot questions are skipped.
1575 * systemd-fstab-generator has been updated to check for the
1576 systemd.volatile= kernel command line option, which either takes an
1577 optional boolean parameter or the special value "state". If used the
1578 system may be booted in a "volatile" boot mode. Specifically,
1579 "systemd.volatile" is used, the root directory will be mounted as
1580 tmpfs, and only /usr is mounted from the actual root file system. If
1581 "systemd.volatile=state" is used, the root directory will be mounted
1582 as usual, but /var is mounted as tmpfs. This concept provides similar
1583 functionality as systemd-nspawn's --volatile= option, but provides it
1584 on physical boots. Use this option for implementing stateless
1585 systems, or testing systems with all state and/or configuration reset
1586 to the defaults. (Note though that many distributions are not
1587 prepared to boot up without a populated /etc or /var, though.)
1589 * systemd-gpt-auto-generator gained support for LUKS encrypted root
1590 partitions. Previously it only supported LUKS encrypted partitions
1591 for all other uses, except for the root partition itself.
1593 * Socket units gained support for listening on AF_VSOCK sockets for
1594 communication in virtualized QEMU environments.
1596 * The "configure" script gained a new option --with-fallback-hostname=
1597 for specifying the fallback hostname to use if none is configured in
1598 /etc/hostname. For example, by specifying
1599 --with-fallback-hostname=fedora it is possible to default to a
1600 hostname of "fedora" on pristine installations.
1602 * systemd-cgls gained support for a new --unit= switch for listing only
1603 the control groups of a specific unit. Similar --user-unit= has been
1604 added for listing only the control groups of a specific user unit.
1606 * systemd-mount gained a new --umount switch for unmounting a mount or
1607 automount point (and all mount/automount points below it).
1609 * systemd will now refuse full configuration reloads (via systemctl
1610 daemon-reload and related calls) unless at least 16MiB of free space
1611 are available in /run. This is a safety precaution in order to ensure
1612 that generators can safely operate after the reload completed.
1614 * A new unit file option RootImage= has been added, which has a similar
1615 effect as RootDirectory= but mounts the service's root directory from
1616 a disk image instead of plain directory. This logic reuses the same
1617 image dissection and mount logic that systemd-nspawn already uses,
1618 and hence supports any disk images systemd-nspawn supports, including
1619 those following the Discoverable Partition Specification, as well as
1620 Verity enabled images. This option enables systemd to run system
1621 services directly off disk images acting as resource bundles,
1622 possibly even including full integrity data.
1624 * A new MountAPIVFS= unit file option has been added, taking a boolean
1625 argument. If enabled /proc, /sys and /dev (collectively called the
1626 "API VFS") will be mounted for the service. This is only relevant if
1627 RootDirectory= or RootImage= is used for the service, as these mounts
1628 are of course in place in the host mount namespace anyway.
1630 * systemd-nspawn gained support for a new --pivot-root= switch. If
1631 specified the root directory within the container image is pivoted to
1632 the specified mount point, while the original root disk is moved to a
1633 different place. This option enables booting of ostree images
1634 directly with systemd-nspawn.
1636 * The systemd build scripts will no longer complain if the NTP server
1637 addresses are not changed from the defaults. Google now supports
1638 these NTP servers officially. We still recommend downstreams to
1639 properly register an NTP pool with the NTP pool project though.
1641 * coredumpctl gained a new "--reverse" option for printing the list
1642 of coredumps in reverse order.
1644 * coredumpctl will now show additional information about truncated and
1645 inaccessible coredumps, as well as coredumps that are still being
1646 processed. It also gained a new --quiet switch for suppressing
1647 additional informational message in its output.
1649 * coredumpctl gained support for only showing coredumps newer and/or
1650 older than specific timestamps, using the new --since= and --until=
1651 options, reminiscent of journalctl's options by the same name.
1653 * The systemd-coredump logic has been improved so that it may be reused
1654 to collect backtraces in non-compiled languages, for example in
1655 scripting languages such as Python.
1657 * machinectl will now show the UID shift of local containers, if user
1658 namespacing is enabled for them.
1660 * systemd will now optionally run "environment generator" binaries at
1661 configuration load time. They may be used to add environment
1662 variables to the environment block passed to services invoked. One
1663 user environment generator is shipped by default that sets up
1664 environment variables based on files dropped into /etc/environment.d
1665 and ~/.config/environment.d/.
1667 * systemd-resolved now includes the new, recently published 2017 DNSSEC
1670 * hostnamed has been updated to report a new chassis type of
1671 "convertible" to cover "foldable" laptops that can both act as a
1672 tablet and as a laptop, such as various Lenovo Yoga devices.
1674 Contributions from: Adrián López, Alexander Galanin, Alexander
1675 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
1676 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
1677 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
1678 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
1679 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
1680 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
1681 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
1682 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
1683 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
1684 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
1685 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
1686 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
1687 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
1688 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
1689 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
1690 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
1691 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
1692 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
1693 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
1694 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
1695 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
1696 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
1697 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
1698 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
1699 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
1700 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
1701 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
1704 — Berlin, 2017-03-01
1708 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
1709 RestrictAddressFamilies= enabled. These sandboxing options should
1710 generally be compatible with the various external udev call-out
1711 binaries we are aware of, however there may be exceptions, in
1712 particular when exotic languages for these call-outs are used. In
1713 this case, consider turning off these settings locally.
1715 * The new RemoveIPC= option can be used to remove IPC objects owned by
1716 the user or group of a service when that service exits.
1718 * The new ProtectKernelModules= option can be used to disable explicit
1719 load and unload operations of kernel modules by a service. In
1720 addition access to /usr/lib/modules is removed if this option is set.
1722 * ProtectSystem= option gained a new value "strict", which causes the
1723 whole file system tree with the exception of /dev, /proc, and /sys,
1724 to be remounted read-only for a service.
1726 * The new ProtectKernelTunables= option can be used to disable
1727 modification of configuration files in /sys and /proc by a service.
1728 Various directories and files are remounted read-only, so access is
1729 restricted even if the file permissions would allow it.
1731 * The new ProtectControlGroups= option can be used to disable write
1732 access by a service to /sys/fs/cgroup.
1734 * Various systemd services have been hardened with
1735 ProtectKernelTunables=yes, ProtectControlGroups=yes,
1736 RestrictAddressFamilies=.
1738 * Support for dynamically creating users for the lifetime of a service
1739 has been added. If DynamicUser=yes is specified, user and group IDs
1740 will be allocated from the range 61184..65519 for the lifetime of the
1741 service. They can be resolved using the new nss-systemd.so NSS
1742 module. The module must be enabled in /etc/nsswitch.conf. Services
1743 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
1744 any resources allocated by the service will be cleaned up when the
1745 service exits. They also have ProtectHome=read-only and
1746 ProtectSystem=strict enabled, so they are not able to make any
1747 permanent modifications to the system.
1749 * The nss-systemd module also always resolves root and nobody, making
1750 it possible to have no /etc/passwd or /etc/group files in minimal
1751 container or chroot environments.
1753 * Services may be started with their own user namespace using the new
1754 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
1755 under which the service is running are mapped. All other users are
1758 * Support for the cgroup namespace has been added to systemd-nspawn. If
1759 supported by kernel, the container system started by systemd-nspawn
1760 will have its own view of the cgroup hierarchy. This new behaviour
1761 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
1763 * The new MemorySwapMax= option can be used to limit the maximum swap
1764 usage under the unified cgroup hierarchy.
1766 * Support for the CPU controller in the unified cgroup hierarchy has
1767 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
1768 options. This controller requires out-of-tree patches for the kernel
1769 and the support is provisional.
1771 * Mount and automount units may now be created transiently
1772 (i.e. dynamically at runtime via the bus API, instead of requiring
1773 unit files in the file system).
1775 * systemd-mount is a new tool which may mount file systems – much like
1776 mount(8), optionally pulling in additional dependencies through
1777 transient .mount and .automount units. For example, this tool
1778 automatically runs fsck on a backing block device before mounting,
1779 and allows the automount logic to be used dynamically from the
1780 command line for establishing mount points. This tool is particularly
1781 useful when dealing with removable media, as it will ensure fsck is
1782 run – if necessary – before the first access and that the file system
1783 is quickly unmounted after each access by utilizing the automount
1784 logic. This maximizes the chance that the file system on the
1785 removable media stays in a clean state, and if it isn't in a clean
1786 state is fixed automatically.
1788 * LazyUnmount=yes option for mount units has been added to expose the
1789 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
1792 * /efi will be used as the mount point of the EFI boot partition, if
1793 the directory is present, and the mount point was not configured
1794 through other means (e.g. fstab). If /efi directory does not exist,
1795 /boot will be used as before. This makes it easier to automatically
1796 mount the EFI partition on systems where /boot is used for something
1799 * When operating on GPT disk images for containers, systemd-nspawn will
1800 now mount the ESP to /boot or /efi according to the same rules as PID
1801 1 running on a host. This allows tools like "bootctl" to operate
1802 correctly within such containers, in order to make container images
1803 bootable on physical systems.
1805 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
1807 * Two new user session targets have been added to support running
1808 graphical sessions under the systemd --user instance:
1809 graphical-session.target and graphical-session-pre.target. See
1810 systemd.special(7) for a description of how those targets should be
1813 * The vconsole initialization code has been significantly reworked to
1814 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
1815 support unicode keymaps. Font and keymap configuration will now be
1816 copied to all allocated virtual consoles.
1818 * FreeBSD's bhyve virtualization is now detected.
1820 * Information recorded in the journal for core dumps now includes the
1821 contents of /proc/mountinfo and the command line of the process at
1822 the top of the process hierarchy (which is usually the init process
1825 * systemd-journal-gatewayd learned the --directory= option to serve
1826 files from the specified location.
1828 * journalctl --root=… can be used to peruse the journal in the
1829 /var/log/ directories inside of a container tree. This is similar to
1830 the existing --machine= option, but does not require the container to
1833 * The hardware database has been extended to support
1834 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
1837 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
1838 specify the click rate for mice which include a horizontal wheel with
1839 a click rate that is different than the one for the vertical wheel.
1841 * systemd-run gained a new --wait option that makes service execution
1842 synchronous. (Specifically, the command will not return until the
1843 specified service binary exited.)
1845 * systemctl gained a new --wait option that causes the start command to
1846 wait until the units being started have terminated again.
1848 * A new journal output mode "short-full" has been added which displays
1849 timestamps with abbreviated English day names and adds a timezone
1850 suffix. Those timestamps include more information than the default
1851 "short" output mode, and can be passed directly to journalctl's
1852 --since= and --until= options.
1854 * /etc/resolv.conf will be bind-mounted into containers started by
1855 systemd-nspawn, if possible, so any changes to resolv.conf contents
1856 are automatically propagated to the container.
1858 * The number of instances for socket-activated services originating
1859 from a single IP address can be limited with
1860 MaxConnectionsPerSource=, extending the existing setting of
1863 * systemd-networkd gained support for vcan ("Virtual CAN") interface
1866 * .netdev and .network configuration can now be extended through
1869 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
1870 Segmentation Offload, Generic Receive Offload, Large Receive Offload
1871 can be enabled and disabled using the new UDPSegmentationOffload=,
1872 TCPSegmentationOffload=, GenericSegmentationOffload=,
1873 GenericReceiveOffload=, LargeReceiveOffload= options in the
1874 [Link] section of .link files.
1876 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
1877 Port VLAN ID can be configured for bridge devices using the new STP=,
1878 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
1879 section of .netdev files.
1881 * The route table to which routes received over DHCP or RA should be
1882 added can be configured with the new RouteTable= option in the [DHCP]
1883 and [IPv6AcceptRA] sections of .network files.
1885 * The Address Resolution Protocol can be disabled on links managed by
1886 systemd-networkd using the ARP=no setting in the [Link] section of
1889 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
1890 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
1891 encode information about the result and exit codes of the current
1892 service runtime cycle.
1894 * systemd-sysctl will now configure kernel parameters in the order
1895 they occur in the configuration files. This matches what sysctl
1896 has been traditionally doing.
1898 * kernel-install "plugins" that are executed to perform various
1899 tasks after a new kernel is added and before an old one is removed
1900 can now return a special value to terminate the procedure and
1901 prevent any later plugins from running.
1903 * Journald's SplitMode=login setting has been deprecated. It has been
1904 removed from documentation, and its use is discouraged. In a future
1905 release it will be completely removed, and made equivalent to current
1906 default of SplitMode=uid.
1908 * Storage=both option setting in /etc/systemd/coredump.conf has been
1909 removed. With fast LZ4 compression storing the core dump twice is not
1912 * The --share-system systemd-nspawn option has been replaced with an
1913 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
1914 this functionality is discouraged. In addition the variables
1915 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
1916 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
1917 individual namespaces.
1919 * "machinectl list" now shows the IP address of running containers in
1920 the output, as well as OS release information.
1922 * "loginctl list" now shows the TTY of each session in the output.
1924 * sd-bus gained new API calls sd_bus_track_set_recursive(),
1925 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
1926 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
1927 tracking objects in a "recursive" mode, where a single client can be
1928 counted multiple times, if it takes multiple references.
1930 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
1931 sd_bus_get_exit_on_disconnect(). They may be used to make a
1932 process using sd-bus automatically exit if the bus connection is
1935 * Bus clients of the service manager may now "pin" loaded units into
1936 memory, by taking an explicit reference on them. This is useful to
1937 ensure the client can retrieve runtime data about the service even
1938 after the service completed execution. Taking such a reference is
1939 available only for privileged clients and should be helpful to watch
1940 running services in a race-free manner, and in particular collect
1941 information about exit statuses and results.
1943 * The nss-resolve module has been changed to strictly return UNAVAIL
1944 when communication via D-Bus with resolved failed, and NOTFOUND when
1945 a lookup completed but was negative. This means it is now possible to
1946 neatly configure fallbacks using nsswitch.conf result checking
1947 expressions. Taking benefit of this, the new recommended
1948 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
1950 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
1952 * A new setting CtrlAltDelBurstAction= has been added to
1953 /etc/systemd/system.conf which may be used to configure the precise
1954 behaviour if the user on the console presses Ctrl-Alt-Del more often
1955 than 7 times in 2s. Previously this would unconditionally result in
1956 an expedited, immediate reboot. With this new setting the precise
1957 operation may be configured in more detail, and also turned off
1960 * In .netdev files two new settings RemoteChecksumTx= and
1961 RemoteChecksumRx= are now understood that permit configuring the
1962 remote checksumming logic for VXLAN networks.
1964 * The service manager learnt a new "invocation ID" concept for invoked
1965 services. Each runtime cycle of a service will get a new invocation
1966 ID (a 128bit random UUID) assigned that identifies the current
1967 run of the service uniquely and globally. A new invocation ID
1968 is generated each time a service starts up. The journal will store
1969 the invocation ID of a service along with any logged messages, thus
1970 making the invocation ID useful for matching the online runtime of a
1971 service with the offline log data it generated in a safe way without
1972 relying on synchronized timestamps. In many ways this new service
1973 invocation ID concept is similar to the kernel's boot ID concept that
1974 uniquely and globally identifies the runtime of each boot. The
1975 invocation ID of a service is passed to the service itself via an
1976 environment variable ($INVOCATION_ID). A new bus call
1977 GetUnitByInvocationID() has been added that is similar to GetUnit()
1978 but instead of retrieving the bus path for a unit by its name
1979 retrieves it by its invocation ID. The returned path is valid only as
1980 long as the passed invocation ID is current.
1982 * systemd-resolved gained a new "DNSStubListener" setting in
1983 resolved.conf. It either takes a boolean value or the special values
1984 "udp" and "tcp", and configures whether to enable the stub DNS
1985 listener on 127.0.0.53:53.
1987 * IP addresses configured via networkd may now carry additional
1988 configuration settings supported by the kernel. New options include:
1989 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
1990 PrefixRoute=, AutoJoin=.
1992 * The PAM configuration fragment file for "user@.service" shipped with
1993 systemd (i.e. the --user instance of systemd) has been stripped to
1994 the minimum necessary to make the system boot. Previously, it
1995 contained Fedora-specific stanzas that did not apply to other
1996 distributions. It is expected that downstream distributions add
1997 additional configuration lines, matching their needs to this file,
1998 using it only as rough template of what systemd itself needs. Note
1999 that this reduced fragment does not even include an invocation of
2000 pam_limits which most distributions probably want to add, even though
2001 systemd itself does not need it. (There's also the new build time
2002 option --with-pamconfdir=no to disable installation of the PAM
2005 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
2006 capability is now also dropped from its set (in addition to
2007 CAP_SYS_MKNOD as before).
2009 * In service unit files it is now possible to connect a specific named
2010 file descriptor with stdin/stdout/stdout of an executed service. The
2011 name may be specified in matching .socket units using the
2012 FileDescriptorName= setting.
2014 * A number of journal settings may now be configured on the kernel
2015 command line. Specifically, the following options are now understood:
2016 systemd.journald.max_level_console=,
2017 systemd.journald.max_level_store=,
2018 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
2019 systemd.journald.max_level_wall=.
2021 * "systemctl is-enabled --full" will now show by which symlinks a unit
2022 file is enabled in the unit dependency tree.
2024 * Support for VeraCrypt encrypted partitions has been added to the
2025 "cryptsetup" logic and /etc/crypttab.
2027 * systemd-detect-virt gained support for a new --private-users switch
2028 that checks whether the invoking processes are running inside a user
2029 namespace. Similar, a new special value "private-users" for the
2030 existing ConditionVirtualization= setting has been added, permitting
2031 skipping of specific units in user namespace environments.
2033 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
2034 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
2035 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
2036 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
2037 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
2038 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
2039 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
2040 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
2041 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
2042 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
2043 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
2044 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
2045 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
2046 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
2047 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
2048 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
2049 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
2050 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
2051 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
2052 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
2053 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
2054 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
2055 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
2056 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
2057 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
2058 Jędrzejewski-Szmek, Zeal Jagannatha
2060 — Santa Fe, 2016-11-03
2064 * In service units the various ExecXYZ= settings have been extended
2065 with an additional special character as first argument of the
2066 assigned value: if the character '+' is used the specified command
2067 line it will be run with full privileges, regardless of User=,
2068 Group=, CapabilityBoundingSet= and similar options. The effect is
2069 similar to the existing PermissionsStartOnly= option, but allows
2070 configuration of this concept for each executed command line
2073 * Services may now alter the service watchdog timeout at runtime by
2074 sending a WATCHDOG_USEC= message via sd_notify().
2076 * MemoryLimit= and related unit settings now optionally take percentage
2077 specifications. The percentage is taken relative to the amount of
2078 physical memory in the system (or in case of containers, the assigned
2079 amount of memory). This allows scaling service resources neatly with
2080 the amount of RAM available on the system. Similarly, systemd-logind's
2081 RuntimeDirectorySize= option now also optionally takes percentage
2084 * In similar fashion TasksMax= takes percentage values now, too. The
2085 value is taken relative to the configured maximum number of processes
2086 on the system. The per-service task maximum has been changed to 15%
2087 using this functionality. (Effectively this is an increase of 512 →
2088 4915 for service units, given the kernel's default pid_max setting.)
2090 * Calendar time specifications in .timer units now understand a ".."
2091 syntax for time ranges. Example: "4..7:10" may now be used for
2092 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
2095 * The InaccessableDirectories=, ReadOnlyDirectories= and
2096 ReadWriteDirectories= unit file settings have been renamed to
2097 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
2098 applied to all kinds of file nodes, and not just directories, with
2099 the exception of symlinks. Specifically these settings may now be
2100 used on block and character device nodes, UNIX sockets and FIFOS as
2101 well as regular files. The old names of these settings remain
2102 available for compatibility.
2104 * systemd will now log about all service processes it kills forcibly
2105 (using SIGKILL) because they remained after the clean shutdown phase
2106 of the service completed. This should help identifying services that
2107 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
2108 systemd-logind's configuration a similar log message is generated for
2109 processes killed at the end of each session due to this setting.
2111 * systemd will now set the $JOURNAL_STREAM environment variable for all
2112 services whose stdout/stderr are connected to the Journal (which
2113 effectively means by default: all services). The variable contains
2114 the device and inode number of the file descriptor used for
2115 stdout/stderr. This may be used by invoked programs to detect whether
2116 their stdout/stderr is connected to the Journal, in which case they
2117 can switch over to direct Journal communication, thus being able to
2118 pass extended, structured metadata along with their log messages. As
2119 one example, this is now used by glib's logging primitives.
2121 * When using systemd's default tmp.mount unit for /tmp, the mount point
2122 will now be established with the "nosuid" and "nodev" options. This
2123 avoids privilege escalation attacks that put traps and exploits into
2124 /tmp. However, this might cause problems if you e. g. put container
2125 images or overlays into /tmp; if you need this, override tmp.mount's
2126 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
2129 * systemd now supports the "memory" cgroup controller also on
2132 * The systemd-cgtop tool now optionally takes a control group path as
2133 command line argument. If specified, the control group list shown is
2134 limited to subgroups of that group.
2136 * The SystemCallFilter= unit file setting gained support for
2137 pre-defined, named system call filter sets. For example
2138 SystemCallFilter=@clock is now an effective way to make all clock
2139 changing-related system calls unavailable to a service. A number of
2140 similar pre-defined groups are defined. Writing system call filters
2141 for system services is simplified substantially with this new
2142 concept. Accordingly, all of systemd's own, long-running services now
2143 enable system call filtering based on this, by default.
2145 * A new service setting MemoryDenyWriteExecute= has been added, taking
2146 a boolean value. If turned on, a service may no longer create memory
2147 mappings that are writable and executable at the same time. This
2148 enhances security for services where this is enabled as it becomes
2149 harder to dynamically write and then execute memory in exploited
2150 service processes. This option has been enabled for all of systemd's
2151 own long-running services.
2153 * A new RestrictRealtime= service setting has been added, taking a
2154 boolean argument. If set the service's processes may no longer
2155 acquire realtime scheduling. This improves security as realtime
2156 scheduling may otherwise be used to easily freeze the system.
2158 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
2159 value. This may be used for requesting that the system manager inside
2160 of the container reports start-up completion to nspawn which then
2161 propagates this notification further to the service manager
2162 supervising nspawn itself. A related option NotifyReady= in .nspawn
2163 files has been added too. This functionality allows ordering of the
2164 start-up of multiple containers using the usual systemd ordering
2167 * machinectl gained a new command "stop" that is an alias for
2170 * systemd-resolved gained support for contacting DNS servers on
2171 link-local IPv6 addresses.
2173 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
2174 its caches. A method call for requesting the same operation has been
2175 added to the bus API too, and is made available via "systemd-resolve
2178 * systemd-resolve gained a new --status switch. If passed a brief
2179 summary of the used DNS configuration with per-interface information
2182 * resolved.conf gained a new Cache= boolean option, defaulting to
2183 on. If turned off local DNS caching is disabled. This comes with a
2184 performance penalty in particular when DNSSEC is enabled. Note that
2185 resolved disables its internal caching implicitly anyway, when the
2186 configured DNS server is on a host-local IP address such as ::1 or
2187 127.0.0.1, thus automatically avoiding double local caching.
2189 * systemd-resolved now listens on the local IP address 127.0.0.53:53
2190 for DNS requests. This improves compatibility with local programs
2191 that do not use the libc NSS or systemd-resolved's bus APIs for name
2192 resolution. This minimal DNS service is only available to local
2193 programs and does not implement the full DNS protocol, but enough to
2194 cover local DNS clients. A new, static resolv.conf file, listing just
2195 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
2196 now recommended to make /etc/resolv.conf a symlink to this file in
2197 order to route all DNS lookups to systemd-resolved, regardless if
2198 done via NSS, the bus API or raw DNS packets. Note that this local
2199 DNS service is not as fully featured as the libc NSS or
2200 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
2201 used to deliver link-local address information (as this implies
2202 sending a local interface index along), LLMNR/mDNS support via this
2203 interface is severely restricted. It is thus strongly recommended for
2204 all applications to use the libc NSS API or native systemd-resolved
2207 * systemd-networkd's bridge support learned a new setting
2208 VLANFiltering= for controlling VLAN filtering. Moreover a new section
2209 in .network files has been added for configuring VLAN bridging in
2210 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
2212 * systemd-networkd's IPv6 Router Advertisement code now makes use of
2213 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
2214 now be acquired without relying on DHCPv6. Two new options
2215 UseDomains= and UseDNS= have been added to configure this behaviour.
2217 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
2218 renamed IPv6AcceptRA=, without altering its behaviour. The old
2219 setting name remains available for compatibility reasons.
2221 * The systemd-networkd VTI/VTI6 tunneling support gained new options
2222 Key=, InputKey= and OutputKey=.
2224 * systemd-networkd gained support for VRF ("Virtual Routing Function")
2225 interface configuration.
2227 * "systemctl edit" may now be used to create new unit files by
2228 specifying the --force switch.
2230 * sd-event gained a new function sd_event_get_iteration() for
2231 requesting the current iteration counter of the event loop. It starts
2232 at zero and is increased by one with each event loop iteration.
2234 * A new rpm macro %systemd_ordering is provided by the macros.systemd
2235 file. It can be used in lieu of %systemd_requires in packages which
2236 don't use any systemd functionality and are intended to be installed
2237 in minimal containers without systemd present. This macro provides
2238 ordering dependencies to ensure that if the package is installed in
2239 the same rpm transaction as systemd, systemd will be installed before
2240 the scriptlets for the package are executed, allowing unit presets
2243 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
2244 been added to simplify packaging of generators.
2246 * The os-release file gained VERSION_CODENAME field for the
2247 distribution nickname (e.g. VERSION_CODENAME=woody).
2249 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
2250 can be set to disable parsing of metadata and the creation
2251 of persistent symlinks for that device.
2253 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
2254 to make them available to logged-in users has been reverted.
2256 * Much of the common code of the various systemd components is now
2257 built into an internal shared library libsystemd-shared-231.so
2258 (incorporating the systemd version number in the name, to be updated
2259 with future releases) that the components link to. This should
2260 decrease systemd footprint both in memory during runtime and on
2261 disk. Note that the shared library is not for public use, and is
2262 neither API nor ABI stable, but is likely to change with every new
2263 released update. Packagers need to make sure that binaries
2264 linking to libsystemd-shared.so are updated in step with the
2267 * Configuration for "mkosi" is now part of the systemd
2268 repository. mkosi is a tool to easily build legacy-free OS images,
2269 and is available on github: https://github.com/systemd/mkosi. If
2270 "mkosi" is invoked in the build tree a new raw OS image is generated
2271 incorporating the systemd sources currently being worked on and a
2272 clean, fresh distribution installation. The generated OS image may be
2273 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
2274 UEFI PC. This functionality is particularly useful to easily test
2275 local changes made to systemd in a pristine, defined environment. See
2276 doc/HACKING for details.
2278 * configure learned the --with-support-url= option to specify the
2279 distribution's bugtracker.
2281 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
2282 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
2283 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
2284 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
2285 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
2286 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
2287 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
2288 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
2289 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
2290 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
2291 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
2292 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
2293 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
2294 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
2295 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
2296 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
2297 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
2298 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
2299 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
2301 — Berlin, 2016-07-25
2305 * DNSSEC is now turned on by default in systemd-resolved (in
2306 "allow-downgrade" mode), but may be turned off during compile time by
2307 passing "--with-default-dnssec=no" to "configure" (and of course,
2308 during runtime with DNSSEC= in resolved.conf). We recommend
2309 downstreams to leave this on at least during development cycles and
2310 report any issues with the DNSSEC logic upstream. We are very
2311 interested in collecting feedback about the DNSSEC validator and its
2312 limitations in the wild. Note however, that DNSSEC support is
2313 probably nothing downstreams should turn on in stable distros just
2314 yet, as it might create incompatibilities with a few DNS servers and
2315 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
2316 automatically whenever we detect such incompatible setups, but there
2317 might be systems we do not cover yet. Hence: please help us testing
2318 the DNSSEC code, leave this on where you can, report back, but then
2319 again don't consider turning this on in your stable, LTS or
2320 production release just yet. (Note that you have to enable
2321 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
2322 and its DNSSEC mode for host name resolution from local
2325 * systemd-resolve conveniently resolves DANE records with the --tlsa
2326 option and OPENPGPKEY records with the --openpgp option. It also
2327 supports dumping raw DNS record data via the new --raw= switch.
2329 * systemd-logind will now by default terminate user processes that are
2330 part of the user session scope unit (session-XX.scope) when the user
2331 logs out. This behavior is controlled by the KillUserProcesses=
2332 setting in logind.conf, and the previous default of "no" is now
2333 changed to "yes". This means that user sessions will be properly
2334 cleaned up after, but additional steps are necessary to allow
2335 intentionally long-running processes to survive logout.
2337 While the user is logged in at least once, user@.service is running,
2338 and any service that should survive the end of any individual login
2339 session can be started at a user service or scope using systemd-run.
2340 systemd-run(1) man page has been extended with an example which shows
2341 how to run screen in a scope unit underneath user@.service. The same
2342 command works for tmux.
2344 After the user logs out of all sessions, user@.service will be
2345 terminated too, by default, unless the user has "lingering" enabled.
2346 To effectively allow users to run long-term tasks even if they are
2347 logged out, lingering must be enabled for them. See loginctl(1) for
2348 details. The default polkit policy was modified to allow users to
2349 set lingering for themselves without authentication.
2351 Previous defaults can be restored at compile time by the
2352 --without-kill-user-processes option to "configure".
2354 * systemd-logind gained new configuration settings SessionsMax= and
2355 InhibitorsMax=, both with a default of 8192. It will not register new
2356 user sessions or inhibitors above this limit.
2358 * systemd-logind will now reload configuration on SIGHUP.
2360 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
2361 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
2362 enable. Also, support for the "io" cgroup controller in the unified
2363 hierarchy has been added, so that the "memory", "pids" and "io" are
2364 now the controllers that are supported on the unified hierarchy.
2366 WARNING: it is not possible to use previous systemd versions with
2367 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
2368 is necessary to also update systemd in the initramfs if using the
2369 unified hierarchy. An updated SELinux policy is also required.
2371 * LLDP support has been extended, and both passive (receive-only) and
2372 active (sender) modes are supported. Passive mode ("routers-only") is
2373 enabled by default in systemd-networkd. Active LLDP mode is enabled
2374 by default for containers on the internal network. The "networkctl
2375 lldp" command may be used to list information gathered. "networkctl
2376 status" will also show basic LLDP information on connected peers now.
2378 * The IAID and DUID unique identifier sent in DHCP requests may now be
2379 configured for the system and each .network file managed by
2380 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
2382 * systemd-networkd gained support for configuring proxy ARP support for
2383 each interface, via the ProxyArp= setting in .network files. It also
2384 gained support for configuring the multicast querier feature of
2385 bridge devices, via the new MulticastQuerier= setting in .netdev
2386 files. Similarly, snooping on the IGMP traffic can be controlled
2387 via the new setting MulticastSnooping=.
2389 A new setting PreferredLifetime= has been added for addresses
2390 configured in .network file to configure the lifetime intended for an
2393 The systemd-networkd DHCP server gained the option EmitRouter=, which
2394 defaults to yes, to configure whether the DHCP Option 3 (Router)
2397 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
2398 systemd-socket-activate and installed into /usr/bin. It is now fully
2401 * systemd-journald now uses separate threads to flush changes to disk
2402 when closing journal files, thus reducing impact of slow disk I/O on
2403 logging performance.
2405 * The sd-journal API gained two new calls
2406 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
2407 can be used to open journal files using file descriptors instead of
2408 file or directory paths. sd_journal_open_container() has been
2409 deprecated, sd_journal_open_directory_fd() should be used instead
2410 with the flag SD_JOURNAL_OS_ROOT.
2412 * journalctl learned a new output mode "-o short-unix" that outputs log
2413 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
2414 UTC). It also gained support for a new --no-hostname setting to
2415 suppress the hostname column in the family of "short" output modes.
2417 * systemd-ask-password now optionally skips printing of the password to
2418 stdout with --no-output which can be useful in scripts.
2420 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
2421 (devices tagged with ID_MAKER_TOOL) are now tagged with
2422 "uaccess" and are available to logged in users.
2424 * The DeviceAllow= unit setting now supports specifiers (with "%").
2426 * "systemctl show" gained a new --value switch, which allows print a
2427 only the contents of a specific unit property, without also printing
2428 the property's name. Similar support was added to "show*" verbs
2429 of loginctl and machinectl that output "key=value" lists.
2431 * A new unit type "generated" was added for files dynamically generated
2432 by generator tools. Similarly, a new unit type "transient" is used
2433 for unit files created using the runtime API. "systemctl enable" will
2434 refuse to operate on such files.
2436 * A new command "systemctl revert" has been added that may be used to
2437 revert to the vendor version of a unit file, in case local changes
2438 have been made by adding drop-ins or overriding the unit file.
2440 * "machinectl clean" gained a new verb to automatically remove all or
2441 just hidden container images.
2443 * systemd-tmpfiles gained support for a new line type "e" for emptying
2444 directories, if they exist, without creating them if they don't.
2446 * systemd-nspawn gained support for automatically patching the UID/GIDs
2447 of the owners and the ACLs of all files and directories in a
2448 container tree to match the UID/GID user namespacing range selected
2449 for the container invocation. This mode is enabled via the new
2450 --private-users-chown switch. It also gained support for
2451 automatically choosing a free, previously unused UID/GID range when
2452 starting a container, via the new --private-users=pick setting (which
2453 implies --private-users-chown). Together, these options for the first
2454 time make user namespacing for nspawn containers fully automatic and
2455 thus deployable. The systemd-nspawn@.service template unit file has
2456 been changed to use this functionality by default.
2458 * systemd-nspawn gained a new --network-zone= switch, that allows
2459 creating ad-hoc virtual Ethernet links between multiple containers,
2460 that only exist as long as at least one container referencing them is
2461 running. This allows easy connecting of multiple containers with a
2462 common link that implements an Ethernet broadcast domain. Each of
2463 these network "zones" may be named relatively freely by the user, and
2464 may be referenced by any number of containers, but each container may
2465 only reference one of these "zones". On the lower level, this is
2466 implemented by an automatically managed bridge network interface for
2467 each zone, that is created when the first container referencing its
2468 zone is created and removed when the last one referencing its zone
2471 * The default start timeout may now be configured on the kernel command
2472 line via systemd.default_timeout_start_sec=. It was already
2473 configurable via the DefaultTimeoutStartSec= option in
2474 /etc/systemd/system.conf.
2476 * Socket units gained a new TriggerLimitIntervalSec= and
2477 TriggerLimitBurst= setting to configure a limit on the activation
2478 rate of the socket unit.
2480 * The LimitNICE= setting now optionally takes normal UNIX nice values
2481 in addition to the raw integer limit value. If the specified
2482 parameter is prefixed with "+" or "-" and is in the range -20..19 the
2483 value is understood as UNIX nice value. If not prefixed like this it
2484 is understood as raw RLIMIT_NICE limit.
2486 * Note that the effect of the PrivateDevices= unit file setting changed
2487 slightly with this release: the per-device /dev file system will be
2488 mounted read-only from this version on, and will have "noexec"
2489 set. This (minor) change of behavior might cause some (exceptional)
2490 legacy software to break, when PrivateDevices=yes is set for its
2491 service. Please leave PrivateDevices= off if you run into problems
2494 * systemd-bootchart has been split out to a separate repository:
2495 https://github.com/systemd/systemd-bootchart
2497 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
2498 merged into the kernel in its current form.
2500 * The compatibility libraries libsystemd-daemon.so,
2501 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
2502 which have been deprecated since systemd-209 have been removed along
2503 with the corresponding pkg-config files. All symbols provided by
2504 those libraries are provided by libsystemd.so.
2506 * The Capabilities= unit file setting has been removed (it is ignored
2507 for backwards compatibility). AmbientCapabilities= and
2508 CapabilityBoundingSet= should be used instead.
2510 * A new special target has been added, initrd-root-device.target,
2511 which creates a synchronization point for dependencies of the root
2512 device in early userspace. Initramfs builders must ensure that this
2513 target is now included in early userspace.
2515 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
2516 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
2517 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
2518 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
2519 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
2520 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
2521 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
2522 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
2523 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
2524 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
2525 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
2526 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
2527 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
2528 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
2529 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
2530 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
2531 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
2532 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
2533 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
2534 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
2535 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
2536 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
2537 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
2538 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
2539 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
2542 — Fairfax, 2016-05-21
2546 * The systemd-resolved DNS resolver service has gained a substantial
2547 set of new features, most prominently it may now act as a DNSSEC
2548 validating stub resolver. DNSSEC mode is currently turned off by
2549 default, but is expected to be turned on by default in one of the
2550 next releases. For now, we invite everybody to test the DNSSEC logic
2551 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
2552 service also gained a full set of D-Bus interfaces, including calls
2553 to configure DNS and DNSSEC settings per link (for use by external
2554 network management software). systemd-resolved and systemd-networkd
2555 now distinguish between "search" and "routing" domains. The former
2556 are used to qualify single-label names, the latter are used purely
2557 for routing lookups within certain domains to specific links.
2558 resolved now also synthesizes RRs for all entries from /etc/hosts.
2560 * The systemd-resolve tool (which is a client utility for
2561 systemd-resolved) has been improved considerably and is now fully
2562 supported and documented. Hence it has moved from /usr/lib/systemd to
2565 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
2568 * The coredump collection logic has been reworked: when a coredump is
2569 collected it is now written to disk, compressed and processed
2570 (including stacktrace extraction) from a new instantiated service
2571 systemd-coredump@.service, instead of directly from the
2572 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
2573 processing large coredumps can take up a substantial amount of
2574 resources and time, and this previously happened entirely outside of
2575 systemd's service supervision. With the new logic the core_pattern
2576 hook only does minimal metadata collection before passing off control
2577 to the new instantiated service, which is configured with a time
2578 limit, a nice level and other settings to minimize negative impact on
2579 the rest of the system. Also note that the new logic will honour the
2580 RLIMIT_CORE setting of the crashed process, which now allows users
2581 and processes to turn off coredumping for their processes by setting
2584 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
2585 and all forked processes by default. Previously, PID 1 would leave
2586 the setting at "0" for all processes, as set by the kernel. Note that
2587 the resource limit traditionally has no effect on the generated
2588 coredumps on the system if the /proc/sys/kernel/core_pattern hook
2589 logic is used. Since the limit is now honoured (see above) its
2590 default has been changed so that the coredumping logic is enabled by
2591 default for all processes, while allowing specific opt-out.
2593 * When the stacktrace is extracted from processes of system users, this
2594 is now done as "systemd-coredump" user, in order to sandbox this
2595 potentially security sensitive parsing operation. (Note that when
2596 processing coredumps of normal users this is done under the user ID
2597 of process that crashed, as before.) Packagers should take notice
2598 that it is now necessary to create the "systemd-coredump" system user
2599 and group at package installation time.
2601 * The systemd-activate socket activation testing tool gained support
2602 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
2603 and --seqpacket switches. It also has been extended to support both
2604 new-style and inetd-style file descriptor passing. Use the new
2605 --inetd switch to request inetd-style file descriptor passing.
2607 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
2608 variable, which takes a boolean value. If set to false, ANSI color
2609 output is disabled in the tools even when run on a terminal that
2612 * The VXLAN support in networkd now supports two new settings
2613 DestinationPort= and PortRange=.
2615 * A new systemd.machine_id= kernel command line switch has been added,
2616 that may be used to set the machine ID in /etc/machine-id if it is
2617 not initialized yet. This command line option has no effect if the
2618 file is already initialized.
2620 * systemd-nspawn gained a new --as-pid2 switch that invokes any
2621 specified command line as PID 2 rather than PID 1 in the
2622 container. In this mode PID 1 is a minimal stub init process that
2623 implements the special POSIX and Linux semantics of PID 1 regarding
2624 signal and child process management. Note that this stub init process
2625 is implemented in nspawn itself and requires no support from the
2626 container image. This new logic is useful to support running
2627 arbitrary commands in the container, as normal processes are
2628 generally not prepared to run as PID 1.
2630 * systemd-nspawn gained a new --chdir= switch for setting the current
2631 working directory for the process started in the container.
2633 * "journalctl /dev/sda" will now output all kernel log messages for
2634 specified device from the current boot, in addition to all devices
2635 that are parents of it. This should make log output about devices
2636 pretty useful, as long as kernel drivers attach enough metadata to
2637 the log messages. (The usual SATA drivers do.)
2639 * The sd-journal API gained two new calls
2640 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
2641 that report whether log data from /run or /var has been found.
2643 * journalctl gained a new switch "--fields" that prints all journal
2644 record field names currently in use in the journal. This is backed
2645 by two new sd-journal API calls sd_journal_enumerate_fields() and
2646 sd_journal_restart_fields().
2648 * Most configurable timeouts in systemd now expect an argument of
2649 "infinity" to turn them off, instead of "0" as before. The semantics
2650 from now on is that a timeout of "0" means "now", and "infinity"
2651 means "never". To maintain backwards compatibility, "0" continues to
2652 turn off previously existing timeout settings.
2654 * "systemctl reload-or-try-restart" has been renamed to "systemctl
2655 try-reload-or-restart" to clarify what it actually does: the "try"
2656 logic applies to both reloading and restarting, not just restarting.
2657 The old name continues to be accepted for compatibility.
2659 * On boot-up, when PID 1 detects that the system clock is behind the
2660 release date of the systemd version in use, the clock is now set
2661 to the latter. Previously, this was already done in timesyncd, in order
2662 to avoid running with clocks set to the various clock epochs such as
2663 1902, 1938 or 1970. With this change the logic is now done in PID 1
2664 in addition to timesyncd during early boot-up, so that it is enforced
2665 before the first process is spawned by systemd. Note that the logic
2666 in timesyncd remains, as it is more comprehensive and ensures
2667 clock monotonicity by maintaining a persistent timestamp file in
2668 /var. Since /var is generally not available in earliest boot or the
2669 initrd, this part of the logic remains in timesyncd, and is not done
2672 * Support for tweaking details in net_cls.class_id through the
2673 NetClass= configuration directive has been removed, as the kernel
2674 people have decided to deprecate that controller in cgroup v2.
2675 Userspace tools such as nftables are moving over to setting rules
2676 that are specific to the full cgroup path of a task, which obsoletes
2677 these controllers anyway. The NetClass= directive is kept around for
2678 legacy compatibility reasons. For a more in-depth description of the
2679 kernel change, please refer to the respective upstream commit:
2681 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
2683 * A new service setting RuntimeMaxSec= has been added that may be used
2684 to specify a maximum runtime for a service. If the timeout is hit, the
2685 service is terminated and put into a failure state.
2687 * A new service setting AmbientCapabilities= has been added. It allows
2688 configuration of additional Linux process capabilities that are
2689 passed to the activated processes. This is only available on very
2692 * The process resource limit settings in service units may now be used
2693 to configure hard and soft limits individually.
2695 * The various libsystemd APIs such as sd-bus or sd-event now publicly
2696 expose support for gcc's __attribute__((cleanup())) C extension.
2697 Specifically, for many object destructor functions alternative
2698 versions have been added that have names suffixed with "p" and take a
2699 pointer to a pointer to the object to destroy, instead of just a
2700 pointer to the object itself. This is useful because these destructor
2701 functions may be used directly as parameters to the cleanup
2702 construct. Internally, systemd has been a heavy user of this GCC
2703 extension for a long time, and with this change similar support is
2704 now available to consumers of the library outside of systemd. Note
2705 that by using this extension in your sources compatibility with old
2706 and strictly ANSI compatible C compilers is lost. However, all gcc or
2707 LLVM versions of recent years support this extension.
2709 * Timer units gained support for a new setting RandomizedDelaySec= that
2710 allows configuring some additional randomized delay to the configured
2711 time. This is useful to spread out timer events to avoid load peaks in
2712 clusters or larger setups.
2714 * Calendar time specifications now support sub-second accuracy.
2716 * Socket units now support listening on SCTP and UDP-lite protocol
2719 * The sd-event API now comes with a full set of man pages.
2721 * Older versions of systemd contained experimental support for
2722 compressing journal files and coredumps with the LZ4 compressor that
2723 was not compatible with the lz4 binary (due to API limitations of the
2724 lz4 library). This support has been removed; only support for files
2725 compatible with the lz4 binary remains. This LZ4 logic is now
2726 officially supported and no longer considered experimental.
2728 * The dkr image import logic has been removed again from importd. dkr's
2729 micro-services focus doesn't fit into the machine image focus of
2730 importd, and quickly got out of date with the upstream dkr API.
2732 * Creation of the /run/lock/lockdev/ directory was dropped from
2733 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
2734 been available for many years. If you still need this, you need to
2735 create your own tmpfiles.d config file with:
2737 d /run/lock/lockdev 0775 root lock -
2739 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
2740 and RebootArgument= have been moved from the [Service] section of
2741 unit files to [Unit], and they are now supported on all unit types,
2742 not just service units. Of course, elogind will continue to
2743 understand these settings also at the old location, in order to
2744 maintain compatibility.
2746 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
2747 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
2748 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
2749 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
2750 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
2751 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
2752 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
2753 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
2754 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
2755 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
2756 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
2757 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
2758 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
2759 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
2760 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
2761 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
2762 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
2763 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
2764 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2766 — Berlin, 2016-02-11
2770 * A number of properties previously only settable in unit
2771 files are now also available as properties to set when
2772 creating transient units programmatically via the bus, as it
2773 is exposed with systemd-run's --property=
2774 setting. Specifically, these are: SyslogIdentifier=,
2775 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
2776 EnvironmentFile=, ReadWriteDirectories=,
2777 ReadOnlyDirectories=, InaccessibleDirectories=,
2778 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
2780 * When creating transient services via the bus API it is now
2781 possible to pass in a set of file descriptors to use as
2782 STDIN/STDOUT/STDERR for the invoked process.
2784 * Slice units may now be created transiently via the bus APIs,
2785 similar to the way service and scope units may already be
2786 created transiently.
2788 * Wherever systemd expects a calendar timestamp specification
2789 (like in journalctl's --since= and --until= switches) UTC
2790 timestamps are now supported. Timestamps suffixed with "UTC"
2791 are now considered to be in Universal Time Coordinated
2792 instead of the local timezone. Also, timestamps may now
2793 optionally be specified with sub-second accuracy. Both of
2794 these additions also apply to recurring calendar event
2795 specification, such as OnCalendar= in timer units.
2797 * journalctl gained a new "--sync" switch that asks the
2798 journal daemon to write all so far unwritten log messages to
2799 disk and sync the files, before returning.
2801 * systemd-tmpfiles learned two new line types "q" and "Q" that
2802 operate like "v", but also set up a basic btrfs quota
2803 hierarchy when used on a btrfs file system with quota
2806 * tmpfiles' "v", "q" and "Q" will now create a plain directory
2807 instead of a subvolume (even on a btrfs file system) if the
2808 root directory is a plain directory, and not a
2809 subvolume. This should simplify things with certain chroot()
2810 environments which are not aware of the concept of btrfs
2813 * systemd-detect-virt gained a new --chroot switch to detect
2814 whether execution takes place in a chroot() environment.
2816 * CPUAffinity= now takes CPU index ranges in addition to
2819 * The various memory-related resource limit settings (such as
2820 LimitAS=) now understand the usual K, M, G, ... suffixes to
2821 the base of 1024 (IEC). Similar, the time-related resource
2822 limit settings understand the usual min, h, day, ...
2825 * There's a new system.conf setting DefaultTasksMax= to
2826 control the default TasksMax= setting for services and
2827 scopes running on the system. (TasksMax= is the primary
2828 setting that exposes the "pids" cgroup controller on systemd
2829 and was introduced in the previous systemd release.) The
2830 setting now defaults to 512, which means services that are
2831 not explicitly configured otherwise will only be able to
2832 create 512 processes or threads at maximum, from this
2833 version on. Note that this means that thread- or
2834 process-heavy services might need to be reconfigured to set
2835 TasksMax= to a higher value. It is sufficient to set
2836 TasksMax= in these specific unit files to a higher value, or
2837 even "infinity". Similar, there's now a logind.conf setting
2838 UserTasksMax= that defaults to 4096 and limits the total
2839 number of processes or tasks each user may own
2840 concurrently. nspawn containers also have the TasksMax=
2841 value set by default now, to 8192. Note that all of this
2842 only has an effect if the "pids" cgroup controller is
2843 enabled in the kernel. The general benefit of these changes
2844 should be a more robust and safer system, that provides a
2845 certain amount of per-service fork() bomb protection.
2847 * systemd-nspawn gained the new --network-veth-extra= switch
2848 to define additional and arbitrarily-named virtual Ethernet
2849 links between the host and the container.
2851 * A new service execution setting PassEnvironment= has been
2852 added that allows importing select environment variables
2853 from PID1's environment block into the environment block of
2856 * Timer units gained support for a new RemainAfterElapse=
2857 setting which takes a boolean argument. It defaults to on,
2858 exposing behaviour unchanged to previous releases. If set to
2859 off, timer units are unloaded after they elapsed if they
2860 cannot elapse again. This is particularly useful for
2861 transient timer units, which shall not stay around longer
2862 than until they first elapse.
2864 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
2865 default now (the kernel default is 16). This is beneficial
2866 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
2867 allows substantially larger numbers of queued
2868 datagrams. This should increase the capability of systemd to
2869 parallelize boot-up, as logging and sd_notify() are unlikely
2870 to stall execution anymore. If you need to change the value
2871 from the new defaults, use the usual sysctl.d/ snippets.
2873 * The compression framing format used by the journal or
2874 coredump processing has changed to be in line with what the
2875 official LZ4 tools generate. LZ4 compression support in
2876 systemd was considered unsupported previously, as the format
2877 was not compatible with the normal tools. With this release
2878 this has changed now, and it is hence safe for downstream
2879 distributions to turn it on. While not compressing as well
2880 as the XZ, LZ4 is substantially faster, which makes
2881 it a good default choice for the compression logic in the
2882 journal and in coredump handling.
2884 * Any reference to /etc/mtab has been dropped from
2885 systemd. The file has been obsolete since a while, but
2886 systemd refused to work on systems where it was incorrectly
2887 set up (it should be a symlink or non-existent). Please make
2888 sure to update to util-linux 2.27.1 or newer in conjunction
2889 with this systemd release, which also drops any reference to
2890 /etc/mtab. If you maintain a distribution make sure that no
2891 software you package still references it, as this is a
2892 likely source of bugs. There's also a glibc bug pending,
2893 asking for removal of any reference to this obsolete file:
2895 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
2897 Note that only util-linux versions built with
2898 --enable-libmount-force-mountinfo are supported.
2900 * Support for the ".snapshot" unit type has been removed. This
2901 feature turned out to be little useful and little used, and
2902 has now been removed from the core and from systemctl.
2904 * The dependency types RequiresOverridable= and
2905 RequisiteOverridable= have been removed from systemd. They
2906 have been used only very sparingly to our knowledge and
2907 other options that provide a similar effect (such as
2908 systemctl --mode=ignore-dependencies) are much more useful
2909 and commonly used. Moreover, they were only half-way
2910 implemented as the option to control behaviour regarding
2911 these dependencies was never added to systemctl. By removing
2912 these dependency types the execution engine becomes a bit
2913 simpler. Unit files that use these dependencies should be
2914 changed to use the non-Overridable dependency types
2915 instead. In fact, when parsing unit files with these
2916 options, that's what systemd will automatically convert them
2917 too, but it will also warn, asking users to fix the unit
2918 files accordingly. Removal of these dependency types should
2919 only affect a negligible number of unit files in the wild.
2921 * Behaviour of networkd's IPForward= option changed
2922 (again). It will no longer maintain a per-interface setting,
2923 but propagate one way from interfaces where this is enabled
2924 to the global kernel setting. The global setting will be
2925 enabled when requested by a network that is set up, but
2926 never be disabled again. This change was made to make sure
2927 IPv4 and IPv6 behaviour regarding packet forwarding is
2928 similar (as the Linux IPv6 stack does not support
2929 per-interface control of this setting) and to minimize
2932 * In unit files the behaviour of %u, %U, %h, %s has
2933 changed. These specifiers will now unconditionally resolve
2934 to the various user database fields of the user that the
2935 systemd instance is running as, instead of the user
2936 configured in the specific unit via User=. Note that this
2937 effectively doesn't change much, as resolving of these
2938 specifiers was already turned off in the --system instance
2939 of systemd, as we cannot do NSS lookups from PID 1. In the
2940 --user instance of systemd these specifiers where correctly
2941 resolved, but hardly made any sense, since the user instance
2942 lacks privileges to do user switches anyway, and User= is
2943 hence useless. Moreover, even in the --user instance of
2944 systemd behaviour was awkward as it would only take settings
2945 from User= assignment placed before the specifier into
2946 account. In order to unify and simplify the logic around
2947 this the specifiers will now always resolve to the
2948 credentials of the user invoking the manager (which in case
2949 of PID 1 is the root user).
2951 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
2952 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
2953 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
2954 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
2955 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
2956 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
2957 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
2958 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
2959 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
2960 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
2961 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
2962 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
2963 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
2964 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
2967 — Berlin, 2015-11-18
2971 * systemd now depends on util-linux v2.27. More specifically,
2972 the newly added mount monitor feature in libmount now
2973 replaces systemd's former own implementation.
2975 * libmount mandates /etc/mtab not to be regular file, and
2976 systemd now enforces this condition at early boot.
2977 /etc/mtab has been deprecated and warned about for a very
2978 long time, so systems running systemd should already have
2979 stopped having this file around as anything else than a
2980 symlink to /proc/self/mounts.
2982 * Support for the "pids" cgroup controller has been added. It
2983 allows accounting the number of tasks in a cgroup and
2984 enforcing limits on it. This adds two new setting
2985 TasksAccounting= and TasksMax= to each unit, as well as a
2986 global option DefaultTasksAccounting=.
2988 * Support for the "net_cls" cgroup controller has been added.
2989 It allows assigning a net class ID to each task in the
2990 cgroup, which can then be used in firewall rules and traffic
2991 shaping configurations. Note that the kernel netfilter net
2992 class code does not currently work reliably for ingress
2993 packets on unestablished sockets.
2995 This adds a new config directive called NetClass= to CGroup
2996 enabled units. Allowed values are positive numbers for fixed
2997 assignments and "auto" for picking a free value
3000 * 'systemctl is-system-running' now returns 'offline' if the
3001 system is not booted with systemd. This command can now be
3002 used as a substitute for 'systemd-notify --booted'.
3004 * Watchdog timeouts have been increased to 3 minutes for all
3005 in-tree service files. Apparently, disk IO issues are more
3006 frequent than we hoped, and user reported >1 minute waiting
3009 * 'machine-id-commit' functionality has been merged into
3010 'machine-id-setup --commit'. The separate binary has been
3013 * The WorkingDirectory= directive in unit files may now be set
3014 to the special value '~'. In this case, the working
3015 directory is set to the home directory of the user
3016 configured in User=.
3018 * "machinectl shell" will now open the shell in the home
3019 directory of the selected user by default.
3021 * The CrashChVT= configuration file setting is renamed to
3022 CrashChangeVT=, following our usual logic of not
3023 abbreviating unnecessarily. The old directive is still
3024 supported for compat reasons. Also, this directive now takes
3025 an integer value between 1 and 63, or a boolean value. The
3026 formerly supported '-1' value for disabling stays around for
3029 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
3030 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
3031 RootDirectory= properties can now be set for transient
3034 * The systemd-analyze tool gained a new "set-log-target" verb
3035 to change the logging target the system manager logs to
3036 dynamically during runtime. This is similar to how
3037 "systemd-analyze set-log-level" already changes the log
3040 * In nspawn /sys is now mounted as tmpfs, with only a selected
3041 set of subdirectories mounted in from the real sysfs. This
3042 enhances security slightly, and is useful for ensuring user
3043 namespaces work correctly.
3045 * Support for USB FunctionFS activation has been added. This
3046 allows implementation of USB gadget services that are
3047 activated as soon as they are requested, so that they don't
3048 have to run continuously, similar to classic socket
3051 * The "systemctl exit" command now optionally takes an
3052 additional parameter that sets the exit code to return from
3053 the systemd manager when exiting. This is only relevant when
3054 running the systemd user instance, or when running the
3055 system instance in a container.
3057 * sd-bus gained the new API calls sd_bus_path_encode_many()
3058 and sd_bus_path_decode_many() that allow easy encoding and
3059 decoding of multiple identifier strings inside a D-Bus
3060 object path. Another new call sd_bus_default_flush_close()
3061 has been added to flush and close per-thread default
3064 * systemd-cgtop gained support for a -M/--machine= switch to
3065 show the control groups within a certain container only.
3067 * "systemctl kill" gained support for an optional --fail
3068 switch. If specified the requested operation will fail of no
3069 processes have been killed, because the unit had no
3070 processes attached, or similar.
3072 * A new systemd.crash_reboot=1 kernel command line option has
3073 been added that triggers a reboot after crashing. This can
3074 also be set through CrashReboot= in systemd.conf.
3076 * The RuntimeDirectory= setting now understands unit
3077 specifiers like %i or %f.
3079 * A new (still internal) library API sd-ipv4acd has been added,
3080 that implements address conflict detection for IPv4. It's
3081 based on code from sd-ipv4ll, and will be useful for
3082 detecting DHCP address conflicts.
3084 * File descriptors passed during socket activation may now be
3085 named. A new API sd_listen_fds_with_names() is added to
3086 access the names. The default names may be overridden,
3087 either in the .socket file using the FileDescriptorName=
3088 parameter, or by passing FDNAME= when storing the file
3089 descriptors using sd_notify().
3091 * systemd-networkd gained support for:
3093 - Setting the IPv6 Router Advertisement settings via
3094 IPv6AcceptRouterAdvertisements= in .network files.
3096 - Configuring the HelloTimeSec=, MaxAgeSec= and
3097 ForwardDelaySec= bridge parameters in .netdev files.
3099 - Configuring PreferredSource= for static routes in
3102 * The "ask-password" framework used to query for LUKS harddisk
3103 passwords or SSL passwords during boot gained support for
3104 caching passwords in the kernel keyring, if it is
3105 available. This makes sure that the user only has to type in
3106 a passphrase once if there are multiple objects to unlock
3107 with the same one. Previously, such password caching was
3108 available only when Plymouth was used; this moves the
3109 caching logic into the systemd codebase itself. The
3110 "systemd-ask-password" utility gained a new --keyname=
3111 switch to control which kernel keyring key to use for
3112 caching a password in. This functionality is also useful for
3113 enabling display managers such as gdm to automatically
3114 unlock the user's GNOME keyring if its passphrase, the
3115 user's password and the harddisk password are the same, if
3116 gdm-autologin is used.
3118 * When downloading tar or raw images using "machinectl
3119 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
3120 file is now also downloaded, if it is available and stored
3121 next to the image file.
3123 * Units of type ".socket" gained a new boolean setting
3124 Writable= which is only useful in conjunction with
3125 ListenSpecial=. If true, enables opening the specified
3126 special file in O_RDWR mode rather than O_RDONLY mode.
3128 * systemd-rfkill has been reworked to become a singleton
3129 service that is activated through /dev/rfkill on each rfkill
3130 state change and saves the settings to disk. This way,
3131 systemd-rfkill is now compatible with devices that exist
3132 only intermittendly, and even restores state if the previous
3133 system shutdown was abrupt rather than clean.
3135 * The journal daemon gained support for vacuuming old journal
3136 files controlled by the number of files that shall remain,
3137 in addition to the already existing control by size and by
3138 date. This is useful as journal interleaving performance
3139 degrades with too many separate journal files, and allows
3140 putting an effective limit on them. The new setting defaults
3141 to 100, but this may be changed by setting SystemMaxFiles=
3142 and RuntimeMaxFiles= in journald.conf. Also, the
3143 "journalctl" tool gained the new --vacuum-files= switch to
3144 manually vacuum journal files to leave only the specified
3145 number of files in place.
3147 * udev will now create /dev/disk/by-path links for ATA devices
3148 on kernels where that is supported.
3150 * Galician, Serbian, Turkish and Korean translations were added.
3152 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
3153 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
3154 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
3155 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
3156 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
3157 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
3158 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
3159 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
3160 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
3161 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
3162 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
3163 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
3164 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
3165 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
3166 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
3167 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
3168 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
3169 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
3171 — Berlin, 2015-10-07
3175 * The DHCP implementation of systemd-networkd gained a set of
3178 - The DHCP server now supports emitting DNS and NTP
3179 information. It may be enabled and configured via
3180 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
3181 and NTP information is enabled, but no servers are
3182 configured, the corresponding uplink information (if there
3183 is any) is propagated.
3185 - Server and client now support transmission and reception
3186 of timezone information. It can be configured via the
3187 newly introduced network options UseTimezone=,
3188 EmitTimezone=, and Timezone=. Transmission of timezone
3189 information is enabled between host and containers by
3190 default now: the container will change its local timezone
3191 to what the host has set.
3193 - Lease timeouts can now be configured via
3194 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
3196 - The DHCP server improved on the stability of
3197 leases. Clients are more likely to get the same lease
3198 information back, even if the server loses state.
3200 - The DHCP server supports two new configuration options to
3201 control the lease address pool metrics, PoolOffset= and
3204 * The encapsulation limit of tunnels in systemd-networkd may
3205 now be configured via 'EncapsulationLimit='. It allows
3206 modifying the maximum additional levels of encapsulation
3207 that are permitted to be prepended to a packet.
3209 * systemd now supports the concept of user buses replacing
3210 session buses, if used with dbus-1.10 (and enabled via dbus
3211 --enable-user-session). It previously only supported this on
3212 kdbus-enabled systems, and this release expands this to
3213 'dbus-daemon' systems.
3215 * systemd-networkd now supports predictable interface names
3218 * systemd now optionally supports the new Linux kernel
3219 "unified" control group hierarchy. If enabled via the kernel
3220 command-line option 'systemd.unified_cgroup_hierarchy=1',
3221 systemd will try to mount the unified cgroup hierarchy
3222 directly on /sys/fs/cgroup. If not enabled, or not
3223 available, systemd will fall back to the legacy cgroup
3224 hierarchy setup, as before. Host system and containers can
3225 mix and match legacy and unified hierarchies as they
3226 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
3227 environment variable to individually select the hierarchy to
3228 use for executed containers. By default, nspawn will use the
3229 unified hierarchy for the containers if the host uses the
3230 unified hierarchy, and the legacy hierarchy otherwise.
3231 Please note that at this point the unified hierarchy is an
3232 experimental kernel feature and is likely to change in one
3233 of the next kernel releases. Therefore, it should not be
3234 enabled by default in downstream distributions yet. The
3235 minimum required kernel version for the unified hierarchy to
3236 work is 4.2. Note that when the unified hierarchy is used
3237 for the first time delegated access to controllers is
3238 safe. Because of this systemd-nspawn containers will get
3239 access to controllers now, as will systemd user
3240 sessions. This means containers and user sessions may now
3241 manage their own resources, partitioning up what the system
3244 * A new special scope unit "init.scope" has been introduced
3245 that encapsulates PID 1 of the system. It may be used to
3246 determine resource usage and enforce resource limits on PID
3247 1 itself. PID 1 hence moved out of the root of the control
3250 * The cgtop tool gained support for filtering out kernel
3251 threads when counting tasks in a control group. Also, the
3252 count of processes is now recursively summed up by
3253 default. Two options -k and --recursive= have been added to
3254 revert to old behaviour. The tool has also been updated to
3255 work correctly in containers now.
3257 * systemd-nspawn's --bind= and --bind-ro= options have been
3258 extended to allow creation of non-recursive bind mounts.
3260 * libsystemd gained two new calls sd_pid_get_cgroup() and
3261 sd_peer_get_cgroup() which return the control group path of
3262 a process or peer of a connected AF_UNIX socket. This
3263 function call is particularly useful when implementing
3264 delegated subtrees support in the control group hierarchy.
3266 * The "sd-event" event loop API of libsystemd now supports
3267 correct dequeuing of real-time signals, without losing
3270 * When systemd requests a PolicyKit decision when managing
3271 units it will now add additional fields to the request,
3272 including unit name and desired operation. This enables more
3273 powerful PolicyKit policies, that make decisions depending
3274 on these parameters.
3276 * nspawn learnt support for .nspawn settings files, that may
3277 accompany the image files or directories of containers, and
3278 may contain additional settings for the container. This is
3279 an alternative to configuring container parameters via the
3280 nspawn command line.
3282 Contributions from: Cristian Rodríguez, Daniel Mack, David
3283 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
3284 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
3285 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
3286 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
3287 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
3288 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
3289 Andersen, Tom Gundersen, Torstein Husebø
3291 — Berlin, 2015-09-08
3295 * machinectl gained a new verb 'shell' which opens a fresh
3296 shell on the target container or the host. It is similar to
3297 the existing 'login' command of machinectl, but spawns the
3298 shell directly without prompting for username or
3299 password. The pseudo machine '.host' now refers to the local
3300 host and is used by default. Hence, 'machinectl shell' can
3301 be used as replacement for 'su -' which spawns a session as
3302 a fresh systemd unit in a way that is fully isolated from
3303 the originating session.
3305 * systemd-networkd learned to cope with private-zone DHCP
3306 options and allows other programs to query the values.
3308 * SELinux access control when enabling/disabling units is no
3309 longer enforced with this release. The previous
3310 implementation was incorrect, and a new corrected
3311 implementation is not yet available. As unit file operations
3312 are still protected via PolicyKit and D-Bus policy this is
3313 not a security problem. Yet, distributions which care about
3314 optimal SELinux support should probably not stabilize on
3317 * sd-bus gained support for matches of type "arg0has=", that
3318 test for membership of strings in string arrays sent in bus
3321 * systemd-resolved now dumps the contents of its DNS and LLMNR
3322 caches to the logs on reception of the SIGUSR1 signal. This
3323 is useful to debug DNS behaviour.
3325 * The coredumpctl tool gained a new --directory= option to
3326 operate on journal files in a specific directory.
3328 * "systemctl reboot" and related commands gained a new
3329 "--message=" option which may be used to set a free-text
3330 wall message when shutting down or rebooting the
3331 system. This message is also logged, which is useful for
3332 figuring out the reason for a reboot or shutdown a
3335 * The "systemd-resolve-host" tool's -i switch now takes
3336 network interface numbers as alternative to interface names.
3338 * A new unit file setting for services has been introduced:
3339 UtmpMode= allows configuration of how precisely systemd
3340 handles utmp and wtmp entries for the service if this is
3341 enabled. This allows writing services that appear similar to
3342 user sessions in the output of the "w", "who", "last" and
3345 * systemd-resolved will now locally synthesize DNS resource
3346 records for the "localhost" and "gateway" domains as well as
3347 the local hostname. This should ensure that clients querying
3348 RRs via resolved will get similar results as those going via
3349 NSS, if nss-myhostname is enabled.
3351 Contributions from: Alastair Hughes, Alex Crawford, Daniel
3352 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
3353 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
3354 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
3355 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
3356 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
3357 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
3358 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
3359 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
3360 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
3361 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
3362 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3364 — Berlin, 2015-08-27
3368 * The systemd-efi-boot-generator functionality was merged into
3369 systemd-gpt-auto-generator.
3371 * systemd-networkd now supports Group Policy for vxlan
3372 devices. It can be enabled via the new boolean configuration
3373 option called 'GroupPolicyExtension='.
3375 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
3376 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
3377 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
3379 — Berlin, 2015-07-31
3383 * The python-systemd code has been removed from the systemd repository.
3384 A new repository has been created which accommodates the code from
3385 now on, and we kindly ask distributions to create a separate package
3386 for this: https://github.com/systemd/python-systemd
3388 * The systemd daemon will now reload its main configuration
3389 (/etc/systemd/system.conf) on daemon-reload.
3391 * sd-dhcp now exposes vendor specific extensions via
3392 sd_dhcp_lease_get_vendor_specific().
3394 * systemd-networkd gained a number of new configuration options.
3396 - A new boolean configuration option for TAP devices called
3397 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
3398 device, thus allowing to send and receive GSO packets.
3400 - A new tunnel configuration option called 'CopyDSCP='.
3401 If enabled, the DSCP field of ip6 tunnels is copied into the
3402 decapsulated packet.
3404 - A set of boolean bridge configuration options were added.
3405 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
3406 and 'UnicastFlood=' are now parsed by networkd and applied to the
3407 respective bridge link device via the respective IFLA_BRPORT_*
3410 - A new string configuration option to override the hostname sent
3411 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
3412 is true, networkd will use the configured hostname instead of the
3413 system hostname when sending DHCP requests.
3415 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
3416 networkd will configure the IPv6 flow-label of the tunnel device
3417 according to RFC2460.
3419 - The 'macvtap' virtual network devices are now supported, similar to
3420 the already supported 'macvlan' devices.
3422 * systemd-resolved now implements RFC5452 to improve resilience against
3423 cache poisoning. Additionally, source port randomization is enabled
3424 by default to further protect against DNS spoofing attacks.
3426 * nss-mymachines now supports translating UIDs and GIDs of running
3427 containers with user-namespaces enabled. If a container 'foo'
3428 translates a host uid 'UID' to the container uid 'TUID', then
3429 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
3430 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
3431 mapped as 'vg-foo-TGID'.
3433 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
3434 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
3435 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
3436 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
3437 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
3438 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
3439 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
3440 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
3441 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
3442 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
3444 — Berlin, 2015-07-29
3448 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
3449 There are no known issues with current sysfs, and udev does not need
3450 or should be used to work around such bugs.
3452 * udev does no longer enable USB HID power management. Several reports
3453 indicate, that some devices cannot handle that setting.
3455 * The udev accelerometer helper was removed. The functionality
3456 is now fully included in iio-sensor-proxy. But this means,
3457 older iio-sensor-proxy versions will no longer provide
3458 accelerometer/orientation data with this systemd version.
3459 Please upgrade iio-sensor-proxy to version 1.0.
3461 * networkd gained a new configuration option IPv6PrivacyExtensions=
3462 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
3463 for Stateless Address") on selected networks.
3465 * For the sake of fewer build-time dependencies and less code in the
3466 main repository, the python bindings are about to be removed in the
3467 next release. A new repository has been created which accommodates
3468 the code from now on, and we kindly ask distributions to create a
3469 separate package for this. The removal will take place in v223.
3471 https://github.com/systemd/python-systemd
3473 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
3474 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
3475 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
3476 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
3477 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
3478 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
3479 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
3480 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
3481 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3482 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3484 — Berlin, 2015-07-07
3488 * The sd-bus.h and sd-event.h APIs have now been declared
3489 stable and have been added to the official interface of
3490 libsystemd.so. sd-bus implements an alternative D-Bus client
3491 library, that is relatively easy to use, very efficient and
3492 supports both classic D-Bus as well as kdbus as transport
3493 backend. sd-event is a generic event loop abstraction that
3494 is built around Linux epoll, but adds features such as event
3495 prioritization or efficient timer handling. Both APIs are good
3496 choices for C programs looking for a bus and/or event loop
3497 implementation that is minimal and does not have to be
3498 portable to other kernels.
3500 * kdbus support is no longer compile-time optional. It is now
3501 always built-in. However, it can still be disabled at
3502 runtime using the kdbus=0 kernel command line setting, and
3503 that setting may be changed to default to off, by specifying
3504 --disable-kdbus at build-time. Note though that the kernel
3505 command line setting has no effect if the kdbus.ko kernel
3506 module is not installed, in which case kdbus is (obviously)
3507 also disabled. We encourage all downstream distributions to
3508 begin testing kdbus by adding it to the kernel images in the
3509 development distributions, and leaving kdbus support in
3512 * The minimal required util-linux version has been bumped to
3515 * Support for chkconfig (--enable-chkconfig) was removed in
3516 favor of calling an abstraction tool
3517 /lib/systemd/systemd-sysv-install. This needs to be
3518 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
3519 in README for details.
3521 * If there's a systemd unit and a SysV init script for the
3522 same service name, and the user executes "systemctl enable"
3523 for it (or a related call), then this will now enable both
3524 (or execute the related operation on both), not just the
3527 * The libudev API documentation has been converted from gtkdoc
3530 * gudev has been removed from the systemd tree, it is now an
3533 * The systemd-cgtop tool learnt a new --raw switch to generate
3534 "raw" (machine parsable) output.
3536 * networkd's IPForwarding= .network file setting learnt the
3537 new setting "kernel", which ensures that networkd does not
3538 change the IP forwarding sysctl from the default kernel
3541 * The systemd-logind bus API now exposes a new boolean
3542 property "Docked" that reports whether logind considers the
3543 system "docked", i.e. connected to a docking station or not.
3545 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
3546 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
3547 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
3548 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
3549 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
3550 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
3551 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
3552 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
3553 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
3554 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
3555 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
3556 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
3557 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
3558 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3559 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
3560 Fink, Zbigniew Jędrzejewski-Szmek
3562 — Berlin, 2015-06-19
3566 * The gudev library has been extracted into a separate repository
3567 available at: https://git.gnome.org/browse/libgudev/
3568 It is now managed as part of the Gnome project. Distributions
3569 are recommended to pass --disable-gudev to systemd and use
3570 gudev from the Gnome project instead. gudev is still included
3571 in systemd, for now. It will be removed soon, though. Please
3572 also see the announcement-thread on systemd-devel:
3573 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
3575 * systemd now exposes a CPUUsageNSec= property for each
3576 service unit on the bus, that contains the overall consumed
3577 CPU time of a service (the sum of what each process of the
3578 service consumed). This value is only available if
3579 CPUAccounting= is turned on for a service, and is then shown
3580 in the "systemctl status" output.
3582 * Support for configuring alternative mappings of the old SysV
3583 runlevels to systemd targets has been removed. They are now
3584 hardcoded in a way that runlevels 2, 3, 4 all map to
3585 multi-user.target and 5 to graphical.target (which
3586 previously was already the default behaviour).
3588 * The auto-mounter logic gained support for mount point
3589 expiry, using a new TimeoutIdleSec= setting in .automount
3590 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
3592 * The EFI System Partition (ESP) as mounted to /boot by
3593 systemd-efi-boot-generator will now be unmounted
3594 automatically after 2 minutes of not being used. This should
3595 minimize the risk of ESP corruptions.
3597 * New /etc/fstab options x-systemd.requires= and
3598 x-systemd.requires-mounts-for= are now supported to express
3599 additional dependencies for mounts. This is useful for
3600 journalling file systems that support external journal
3601 devices or overlay file systems that require underlying file
3602 systems to be mounted.
3604 * systemd does not support direct live-upgrades (via systemctl
3605 daemon-reexec) from versions older than v44 anymore. As no
3606 distribution we are aware of shipped such old versions in a
3607 stable release this should not be problematic.
3609 * When systemd forks off a new per-connection service instance
3610 it will now set the $REMOTE_ADDR environment variable to the
3611 remote IP address, and $REMOTE_PORT environment variable to
3612 the remote IP port. This behaviour is similar to the
3613 corresponding environment variables defined by CGI.
3615 * systemd-networkd gained support for uplink failure
3616 detection. The BindCarrier= option allows binding interface
3617 configuration dynamically to the link sense of other
3618 interfaces. This is useful to achieve behaviour like in
3621 * systemd-networkd gained support for configuring the DHCP
3622 client identifier to use when requesting leases.
3624 * systemd-networkd now has a per-network UseNTP= option to
3625 configure whether NTP server information acquired via DHCP
3626 is passed on to services like systemd-timesyncd.
3628 * systemd-networkd gained support for vti6 tunnels.
3630 * Note that systemd-networkd manages the sysctl variable
3631 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
3632 it is configured for since v219. The variable controls IP
3633 forwarding, and is a per-interface alternative to the global
3634 /proc/sys/net/ipv[46]/ip_forward. This setting is
3635 configurable in the IPForward= option, which defaults to
3636 "no". This means if networkd is used for an interface it is
3637 no longer sufficient to set the global sysctl option to turn
3638 on IP forwarding! Instead, the .network file option
3639 IPForward= needs to be turned on! Note that the
3640 implementation of this behaviour was broken in v219 and has
3643 * Many bonding and vxlan options are now configurable in
3646 * systemd-nspawn gained a new --property= setting to set unit
3647 properties for the container scope. This is useful for
3648 setting resource parameters (e.g. "CPUShares=500") on
3649 containers started from the command line.
3651 * systemd-nspawn gained a new --private-users= switch to make
3652 use of user namespacing available on recent Linux kernels.
3654 * systemd-nspawn may now be called as part of a shell pipeline
3655 in which case the pipes used for stdin and stdout are passed
3656 directly to the process invoked in the container, without
3657 indirection via a pseudo tty.
3659 * systemd-nspawn gained a new switch to control the UNIX
3660 signal to use when killing the init process of the container
3663 * systemd-nspawn gained a new --overlay= switch for mounting
3664 overlay file systems into the container using the new kernel
3667 * When a container image is imported via systemd-importd and
3668 the host file system is not btrfs, a loopback block device
3669 file is created in /var/lib/machines.raw with a btrfs file
3670 system inside. It is then mounted to /var/lib/machines to
3671 enable btrfs features for container management. The loopback
3672 file and btrfs file system is grown as needed when container
3673 images are imported via systemd-importd.
3675 * systemd-machined/systemd-importd gained support for btrfs
3676 quota, to enforce container disk space limits on disk. This
3677 is exposed in "machinectl set-limit".
3679 * systemd-importd now can import containers from local .tar,
3680 .raw and .qcow2 images, and export them to .tar and .raw. It
3681 can also import dkr v2 images now from the network (on top
3684 * systemd-importd gained support for verifying downloaded
3685 images with gpg2 (previously only gpg1 was supported).
3687 * systemd-machined, systemd-logind, systemd: most bus calls
3688 are now accessible to unprivileged processes via
3689 PolicyKit. Also, systemd-logind will now allow users to kill
3690 their own sessions without further privileges or
3693 * systemd-shutdownd has been removed. This service was
3694 previously responsible for implementing scheduled shutdowns
3695 as exposed in /usr/bin/shutdown's time parameter. This
3696 functionality has now been moved into systemd-logind and is
3697 accessible via a bus interface.
3699 * "systemctl reboot" gained a new switch --firmware-setup that
3700 can be used to reboot into the EFI firmware setup, if that
3701 is available. systemd-logind now exposes an API on the bus
3702 to trigger such reboots, in case graphical desktop UIs want
3703 to cover this functionality.
3705 * "systemctl enable", "systemctl disable" and "systemctl mask"
3706 now support a new "--now" switch. If specified the units
3707 that are enabled will also be started, and the ones
3708 disabled/masked also stopped.
3710 * The Gummiboot EFI boot loader tool has been merged into
3711 systemd, and renamed to "systemd-boot". The bootctl tool has been
3712 updated to support systemd-boot.
3714 * An EFI kernel stub has been added that may be used to create
3715 kernel EFI binaries that contain not only the actual kernel,
3716 but also an initrd, boot splash, command line and OS release
3717 information. This combined binary can then be signed as a
3718 single image, so that the firmware can verify it all in one
3719 step. systemd-boot has special support for EFI binaries created
3720 like this and can extract OS release information from them
3721 and show them in the boot menu. This functionality is useful
3722 to implement cryptographically verified boot schemes.
3724 * Optional support has been added to systemd-fsck to pass
3725 fsck's progress report to an AF_UNIX socket in the file
3728 * udev will no longer create device symlinks for all block
3729 devices by default. A blacklist for excluding special block
3730 devices from this logic has been turned into a whitelist
3731 that requires picking block devices explicitly that require
3734 * A new (currently still internal) API sd-device.h has been
3735 added to libsystemd. This modernized API is supposed to
3736 replace libudev eventually. In fact, already much of libudev
3737 is now just a wrapper around sd-device.h.
3739 * A new hwdb database for storing metadata about pointing
3740 stick devices has been added.
3742 * systemd-tmpfiles gained support for setting file attributes
3743 similar to the "chattr" tool with new 'h' and 'H' lines.
3745 * systemd-journald will no longer unconditionally set the
3746 btrfs NOCOW flag on new journal files. This is instead done
3747 with tmpfiles snippet using the new 'h' line type. This
3748 allows easy disabling of this logic, by masking the
3749 journal-nocow.conf tmpfiles file.
3751 * systemd-journald will now translate audit message types to
3752 human readable identifiers when writing them to the
3753 journal. This should improve readability of audit messages.
3755 * The LUKS logic gained support for the offset= and skip=
3756 options in /etc/crypttab, as previously implemented by
3759 * /usr/lib/os-release gained a new optional field VARIANT= for
3760 distributions that support multiple variants (such as a
3761 desktop edition, a server edition, ...)
3763 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
3764 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
3765 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
3766 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
3767 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
3768 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
3769 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
3770 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
3771 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
3772 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
3773 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
3774 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
3775 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
3776 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
3777 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
3778 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
3779 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
3780 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
3781 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
3782 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
3783 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
3784 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
3785 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
3786 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
3787 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
3788 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
3789 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
3791 — Berlin, 2015-05-22
3795 * Introduce a new API "sd-hwdb.h" for querying the hardware
3796 metadata database. With this minimal interface one can query
3797 and enumerate the udev hwdb, decoupled from the old libudev
3798 library. libudev's interface for this is now only a wrapper
3799 around sd-hwdb. A new tool systemd-hwdb has been added to
3800 interface with and update the database.
3802 * When any of systemd's tools copies files (for example due to
3803 tmpfiles' C lines) a btrfs reflink will attempted first,
3804 before bytewise copying is done.
3806 * systemd-nspawn gained a new --ephemeral switch. When
3807 specified a btrfs snapshot is taken of the container's root
3808 directory, and immediately removed when the container
3809 terminates again. Thus, a container can be started whose
3810 changes never alter the container's root directory, and are
3811 lost on container termination. This switch can also be used
3812 for starting a container off the root file system of the
3813 host without affecting the host OS. This switch is only
3814 available on btrfs file systems.
3816 * systemd-nspawn gained a new --template= switch. It takes the
3817 path to a container tree to use as template for the tree
3818 specified via --directory=, should that directory be
3819 missing. This allows instantiating containers dynamically,
3820 on first run. This switch is only available on btrfs file
3823 * When a .mount unit refers to a mount point on which multiple
3824 mounts are stacked, and the .mount unit is stopped all of
3825 the stacked mount points will now be unmounted until no
3826 mount point remains.
3828 * systemd now has an explicit notion of supported and
3829 unsupported unit types. Jobs enqueued for unsupported unit
3830 types will now fail with an "unsupported" error code. More
3831 specifically .swap, .automount and .device units are not
3832 supported in containers, .busname units are not supported on
3833 non-kdbus systems. .swap and .automount are also not
3834 supported if their respective kernel compile time options
3837 * machinectl gained support for two new "copy-from" and
3838 "copy-to" commands for copying files from a running
3839 container to the host or vice versa.
3841 * machinectl gained support for a new "bind" command to bind
3842 mount host directories into local containers. This is
3843 currently only supported for nspawn containers.
3845 * networkd gained support for configuring bridge forwarding
3846 database entries (fdb) from .network files.
3848 * A new tiny daemon "systemd-importd" has been added that can
3849 download container images in tar, raw, qcow2 or dkr formats,
3850 and make them available locally in /var/lib/machines, so
3851 that they can run as nspawn containers. The daemon can GPG
3852 verify the downloads (not supported for dkr, since it has no
3853 provisions for verifying downloads). It will transparently
3854 decompress bz2, xz, gzip compressed downloads if necessary,
3855 and restore sparse files on disk. The daemon uses privilege
3856 separation to ensure the actual download logic runs with
3857 fewer privileges than the daemon itself. machinectl has
3858 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
3859 make the functionality of importd available to the
3860 user. With this in place the Fedora and Ubuntu "Cloud"
3861 images can be downloaded and booted as containers unmodified
3862 (the Fedora images lack the appropriate GPG signature files
3863 currently, so they cannot be verified, but this will change
3864 soon, hopefully). Note that downloading images is currently
3865 only fully supported on btrfs.
3867 * machinectl is now able to list container images found in
3868 /var/lib/machines, along with some metadata about sizes of
3869 disk and similar. If the directory is located on btrfs and
3870 quota is enabled, this includes quota display. A new command
3871 "image-status" has been added that shows additional
3872 information about images.
3874 * machinectl is now able to clone container images
3875 efficiently, if the underlying file system (btrfs) supports
3876 it, with the new "machinectl clone" command. It also
3877 gained commands for renaming and removing images, as well as
3878 marking them read-only or read-write (supported also on
3879 legacy file systems).
3881 * networkd gained support for collecting LLDP network
3882 announcements, from hardware that supports this. This is
3883 shown in networkctl output.
3885 * systemd-run gained support for a new -t (--pty) switch for
3886 invoking a binary on a pty whose input and output is
3887 connected to the invoking terminal. This allows executing
3888 processes as system services while interactively
3889 communicating with them via the terminal. Most interestingly
3890 this is supported across container boundaries. Invoking
3891 "systemd-run -t /bin/bash" is an alternative to running a
3892 full login session, the difference being that the former
3893 will not register a session, nor go through the PAM session
3896 * tmpfiles gained support for a new "v" line type for creating
3897 btrfs subvolumes. If the underlying file system is a legacy
3898 file system, this automatically degrades to creating a
3899 normal directory. Among others /var/lib/machines is now
3900 created like this at boot, should it be missing.
3902 * The directory /var/lib/containers/ has been deprecated and
3903 been replaced by /var/lib/machines. The term "machines" has
3904 been used in the systemd context as generic term for both
3905 VMs and containers, and hence appears more appropriate for
3906 this, as the directory can also contain raw images bootable
3909 * systemd-nspawn when invoked with -M but without --directory=
3910 or --image= is now capable of searching for the container
3911 root directory, subvolume or disk image automatically, in
3912 /var/lib/machines. systemd-nspawn@.service has been updated
3913 to make use of this, thus allowing it to be used for raw
3916 * A new machines.target unit has been introduced that is
3917 supposed to group all containers/VMs invoked as services on
3918 the system. systemd-nspawn@.service has been updated to
3919 integrate with that.
3921 * machinectl gained a new "start" command, for invoking a
3922 container as a service. "machinectl start foo" is mostly
3923 equivalent to "systemctl start systemd-nspawn@foo.service",
3924 but handles escaping in a nicer way.
3926 * systemd-nspawn will now mount most of the cgroupfs tree
3927 read-only into each container, with the exception of the
3928 container's own subtree in the name=systemd hierarchy.
3930 * journald now sets the special FS_NOCOW file flag for its
3931 journal files. This should improve performance on btrfs, by
3932 avoiding heavy fragmentation when journald's write-pattern
3933 is used on COW file systems. It degrades btrfs' data
3934 integrity guarantees for the files to the same levels as for
3935 ext3/ext4 however. This should be OK though as journald does
3936 its own data integrity checks and all its objects are
3937 checksummed on disk. Also, journald should handle btrfs disk
3938 full events a lot more gracefully now, by processing SIGBUS
3939 errors, and not relying on fallocate() anymore.
3941 * When journald detects that journal files it is writing to
3942 have been deleted it will immediately start new journal
3945 * systemd now provides a way to store file descriptors
3946 per-service in PID 1. This is useful for daemons to ensure
3947 that fds they require are not lost during a daemon
3948 restart. The fds are passed to the daemon on the next
3949 invocation in the same way socket activation fds are
3950 passed. This is now used by journald to ensure that the
3951 various sockets connected to all the system's stdout/stderr
3952 are not lost when journald is restarted. File descriptors
3953 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
3954 an extension to sd_notify(). Note that a limit is enforced
3955 on the number of fds a service can store in PID 1, and it
3956 defaults to 0, so that no fds may be stored, unless this is
3957 explicitly turned on.
3959 * The default TERM variable to use for units connected to a
3960 terminal, when no other value is explicitly is set is now
3961 vt220 rather than vt102. This should be fairly safe still,
3962 but allows PgUp/PgDn work.
3964 * The /etc/crypttab option header= as known from Debian is now
3967 * "loginctl user-status" and "loginctl session-status" will
3968 now show the last 10 lines of log messages of the
3969 user/session following the status output. Similar,
3970 "machinectl status" will show the last 10 log lines
3971 associated with a virtual machine or container
3972 service. (Note that this is usually not the log messages
3973 done in the VM/container itself, but simply what the
3974 container manager logs. For nspawn this includes all console
3977 * "loginctl session-status" without further argument will now
3978 show the status of the session of the caller. Similar,
3979 "lock-session", "unlock-session", "activate",
3980 "enable-linger", "disable-linger" may now be called without
3981 session/user parameter in which case they apply to the
3982 caller's session/user.
3984 * An X11 session scriptlet is now shipped that uploads
3985 $DISPLAY and $XAUTHORITY into the environment of the systemd
3986 --user daemon if a session begins. This should improve
3987 compatibility with X11 enabled applications run as systemd
3990 * Generators are now subject to masking via /etc and /run, the
3991 same way as unit files.
3993 * networkd .network files gained support for configuring
3994 per-link IPv4/IPv6 packet forwarding as well as IPv4
3995 masquerading. This is by default turned on for veth links to
3996 containers, as registered by systemd-nspawn. This means that
3997 nspawn containers run with --network-veth will now get
3998 automatic routed access to the host's networks without any
3999 further configuration or setup, as long as networkd runs on
4002 * systemd-nspawn gained the --port= (-p) switch to expose TCP
4003 or UDP posts of a container on the host. With this in place
4004 it is possible to run containers with private veth links
4005 (--network-veth), and have their functionality exposed on
4006 the host as if their services were running directly on the
4009 * systemd-nspawn's --network-veth switch now gained a short
4010 version "-n", since with the changes above it is now truly
4011 useful out-of-the-box. The systemd-nspawn@.service has been
4012 updated to make use of it too by default.
4014 * systemd-nspawn will now maintain a per-image R/W lock, to
4015 ensure that the same image is not started more than once
4016 writable. (It's OK to run an image multiple times
4017 simultaneously in read-only mode.)
4019 * systemd-nspawn's --image= option is now capable of
4020 dissecting and booting MBR and GPT disk images that contain
4021 only a single active Linux partition. Previously it
4022 supported only GPT disk images with proper GPT type
4023 IDs. This allows running cloud images from major
4024 distributions directly with systemd-nspawn, without
4027 * In addition to collecting mouse dpi data in the udev
4028 hardware database, there's now support for collecting angle
4029 information for mouse scroll wheels. The database is
4030 supposed to guarantee similar scrolling behavior on mice
4031 that it knows about. There's also support for collecting
4032 information about Touchpad types.
4034 * udev's input_id built-in will now also collect touch screen
4035 dimension data and attach it to probed devices.
4037 * /etc/os-release gained support for a Distribution Privacy
4040 * networkd gained support for creating "ipvlan", "gretap",
4041 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
4043 * systemd-tmpfiles gained support for "a" lines for setting
4046 * systemd-nspawn will now mount /tmp in the container to
4047 tmpfs, automatically.
4049 * systemd now exposes the memory.usage_in_bytes cgroup
4050 attribute and shows it for each service in the "systemctl
4051 status" output, if available.
4053 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
4054 immediate reboot is triggered. This useful if shutdown is
4055 hung and is unable to complete, to expedite the
4056 operation. Note that this kind of reboot will still unmount
4057 all file systems, and hence should not result in fsck being
4060 * A .device unit for an optical block device will now be
4061 considered active only when a medium is in the drive. Also,
4062 mount units are now bound to their backing devices thus
4063 triggering automatic unmounting when devices become
4064 unavailable. With this in place systemd will now
4065 automatically unmount left-over mounts when a CD-ROM is
4066 ejected or an USB stick is yanked from the system.
4068 * networkd-wait-online now has support for waiting for
4069 specific interfaces only (with globbing), and for giving up
4070 after a configurable timeout.
4072 * networkd now exits when idle. It will be automatically
4073 restarted as soon as interfaces show up, are removed or
4074 change state. networkd will stay around as long as there is
4075 at least one DHCP state machine or similar around, that keep
4078 * networkd may now configure IPv6 link-local addressing in
4079 addition to IPv4 link-local addressing.
4081 * The IPv6 "token" for use in SLAAC may now be configured for
4082 each .network interface in networkd.
4084 * Routes configured with networkd may now be assigned a scope
4087 * networkd's [Match] sections now support globbing and lists
4088 of multiple space-separated matches per item.
4090 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
4091 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
4092 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
4093 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
4094 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
4095 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
4096 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
4097 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
4098 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
4099 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
4100 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
4101 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
4102 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
4103 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
4104 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
4105 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
4106 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
4107 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
4108 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
4109 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
4110 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
4111 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
4112 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
4113 Hoffmann, Zbigniew Jędrzejewski-Szmek
4115 — Berlin, 2015-02-16
4119 * When querying unit file enablement status (for example via
4120 "systemctl is-enabled"), a new state "indirect" is now known
4121 which indicates that a unit might not be enabled itself, but
4122 another unit listed in its Also= setting might be.
4124 * Similar to the various existing ConditionXYZ= settings for
4125 units, there are now matching AssertXYZ= settings. While
4126 failing conditions cause a unit to be skipped, but its job
4127 to succeed, failing assertions declared like this will cause
4128 a unit start operation and its job to fail.
4130 * hostnamed now knows a new chassis type "embedded".
4132 * systemctl gained a new "edit" command. When used on a unit
4133 file, this allows extending unit files with .d/ drop-in
4134 configuration snippets or editing the full file (after
4135 copying it from /usr/lib to /etc). This will invoke the
4136 user's editor (as configured with $EDITOR), and reload the
4137 modified configuration after editing.
4139 * "systemctl status" now shows the suggested enablement state
4140 for a unit, as declared in the (usually vendor-supplied)
4141 system preset files.
4143 * nss-myhostname will now resolve the single-label host name
4144 "gateway" to the locally configured default IP routing
4145 gateways, ordered by their metrics. This assigns a stable
4146 name to the used gateways, regardless which ones are
4147 currently configured. Note that the name will only be
4148 resolved after all other name sources (if nss-myhostname is
4149 configured properly) and should hence not negatively impact
4150 systems that use the single-label host name "gateway" in
4153 * systemd-inhibit now allows filtering by mode when listing
4156 * Scope and service units gained a new "Delegate" boolean
4157 property, which, when set, allows processes running inside the
4158 unit to further partition resources. This is primarily
4159 useful for systemd user instances as well as container
4162 * journald will now pick up audit messages directly from
4163 the kernel, and log them like any other log message. The
4164 audit fields are split up and fully indexed. This means that
4165 journalctl in many ways is now a (nicer!) alternative to
4166 ausearch, the traditional audit client. Note that this
4167 implements only a minimal audit client. If you want the
4168 special audit modes like reboot-on-log-overflow, please use
4169 the traditional auditd instead, which can be used in
4170 parallel to journald.
4172 * The ConditionSecurity= unit file option now understands the
4173 special string "audit" to check whether auditing is
4176 * journalctl gained two new commands --vacuum-size= and
4177 --vacuum-time= to delete old journal files until the
4178 remaining ones take up no more than the specified size on disk,
4179 or are not older than the specified time.
4181 * A new, native PPPoE library has been added to sd-network,
4182 systemd's library of light-weight networking protocols. This
4183 library will be used in a future version of networkd to
4184 enable PPPoE communication without an external pppd daemon.
4186 * The busctl tool now understands a new "capture" verb that
4187 works similar to "monitor", but writes a packet capture
4188 trace to STDOUT that can be redirected to a file which is
4189 compatible with libcap's capture file format. This can then
4190 be loaded in Wireshark and similar tools to inspect bus
4193 * The busctl tool now understands a new "tree" verb that shows
4194 the object trees of a specific service on the bus, or of all
4197 * The busctl tool now understands a new "introspect" verb that
4198 shows all interfaces and members of objects on the bus,
4199 including their signature and values. This is particularly
4200 useful to get more information about bus objects shown by
4201 the new "busctl tree" command.
4203 * The busctl tool now understands new verbs "call",
4204 "set-property" and "get-property" for invoking bus method
4205 calls, setting and getting bus object properties in a
4208 * busctl gained a new --augment-creds= argument that controls
4209 whether the tool shall augment credential information it
4210 gets from the bus with data from /proc, in a possibly
4213 * nspawn's --link-journal= switch gained two new values
4214 "try-guest" and "try-host" that work like "guest" and
4215 "host", but do not fail if the host has no persistent
4216 journalling enabled. -j is now equivalent to
4217 --link-journal=try-guest.
4219 * macvlan network devices created by nspawn will now have
4220 stable MAC addresses.
4222 * A new SmackProcessLabel= unit setting has been added, which
4223 controls the SMACK security label processes forked off by
4224 the respective unit shall use.
4226 * If compiled with --enable-xkbcommon, systemd-localed will
4227 verify x11 keymap settings by compiling the given keymap. It
4228 will spew out warnings if the compilation fails. This
4229 requires libxkbcommon to be installed.
4231 * When a coredump is collected, a larger number of metadata
4232 fields is now collected and included in the journal records
4233 created for it. More specifically, control group membership,
4234 environment variables, memory maps, working directory,
4235 chroot directory, /proc/$PID/status, and a list of open file
4236 descriptors is now stored in the log entry.
4238 * The udev hwdb now contains DPI information for mice. For
4241 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
4243 * All systemd programs that read standalone configuration
4244 files in /etc now also support a corresponding series of
4245 .conf.d configuration directories in /etc/, /run/,
4246 /usr/local/lib/, /usr/lib/, and (if configured with
4247 --enable-split-usr) /lib/. In particular, the following
4248 configuration files now have corresponding configuration
4249 directories: system.conf user.conf, logind.conf,
4250 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
4251 resolved.conf, timesyncd.conf, journal-remote.conf, and
4252 journal-upload.conf. Note that distributions should use the
4253 configuration directories in /usr/lib/; the directories in
4254 /etc/ are reserved for the system administrator.
4256 * systemd-rfkill will no longer take the rfkill device name
4257 into account when storing rfkill state on disk, as the name
4258 might be dynamically assigned and not stable. Instead, the
4259 ID_PATH udev variable combined with the rfkill type (wlan,
4260 bluetooth, ...) is used.
4262 * A new service systemd-machine-id-commit.service has been
4263 added. When used on systems where /etc is read-only during
4264 boot, and /etc/machine-id is not initialized (but an empty
4265 file), this service will copy the temporary machine ID
4266 created as replacement into /etc after the system is fully
4267 booted up. This is useful for systems that are freshly
4268 installed with a non-initialized machine ID, but should get
4269 a fixed machine ID for subsequent boots.
4271 * networkd's .netdev files now provide a large set of
4272 configuration parameters for VXLAN devices. Similarly, the
4273 bridge port cost parameter is now configurable in .network
4274 files. There's also new support for configuring IP source
4275 routing. networkd .link files gained support for a new
4276 OriginalName= match that is useful to match against the
4277 original interface name the kernel assigned. .network files
4278 may include MTU= and MACAddress= fields for altering the MTU
4279 and MAC address while being connected to a specific network
4282 * The LUKS logic gained supported for configuring
4283 UUID-specific key files. There's also new support for naming
4284 LUKS device from the kernel command line, using the new
4285 luks.name= argument.
4287 * Timer units may now be transiently created via the bus API
4288 (this was previously already available for scope and service
4289 units). In addition it is now possible to create multiple
4290 transient units at the same time with a single bus call. The
4291 "systemd-run" tool has been updated to make use of this for
4292 running commands on a specified time, in at(1)-style.
4294 * tmpfiles gained support for "t" lines, for assigning
4295 extended attributes to files. Among other uses this may be
4296 used to assign SMACK labels to files.
4298 Contributions from: Alin Rauta, Alison Chaiken, Andrej
4299 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
4300 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
4301 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
4302 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
4303 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
4304 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
4305 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
4306 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
4307 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
4308 Michael Biebl, Michael Chapman, Michael Marineau, Michal
4309 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
4310 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
4311 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
4312 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
4313 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4314 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
4315 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
4317 — Berlin, 2014-12-10
4321 * journalctl gained the new options -t/--identifier= to match
4322 on the syslog identifier (aka "tag"), as well as --utc to
4323 show log timestamps in the UTC timezone. journalctl now also
4324 accepts -n/--lines=all to disable line capping in a pager.
4326 * journalctl gained a new switch, --flush, that synchronously
4327 flushes logs from /run/log/journal to /var/log/journal if
4328 persistent storage is enabled. systemd-journal-flush.service
4329 now waits until the operation is complete.
4331 * Services can notify the manager before they start a reload
4332 (by sending RELOADING=1) or shutdown (by sending
4333 STOPPING=1). This allows the manager to track and show the
4334 internal state of daemons and closes a race condition when
4335 the process is still running but has closed its D-Bus
4338 * Services with Type=oneshot do not have to have any ExecStart
4341 * User units are now loaded also from
4342 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
4343 /run/systemd/user directory that was already previously
4344 supported, but is under the control of the user.
4346 * Job timeouts (i.e. time-outs on the time a job that is
4347 queued stays in the run queue) can now optionally result in
4348 immediate reboot or power-off actions (JobTimeoutAction= and
4349 JobTimeoutRebootArgument=). This is useful on ".target"
4350 units, to limit the maximum time a target remains
4351 undispatched in the run queue, and to trigger an emergency
4352 operation in such a case. This is now used by default to
4353 turn off the system if boot-up (as defined by everything in
4354 basic.target) hangs and does not complete for at least
4355 15min. Also, if power-off or reboot hang for at least 30min
4356 an immediate power-off/reboot operation is triggered. This
4357 functionality is particularly useful to increase reliability
4358 on embedded devices, but also on laptops which might
4359 accidentally get powered on when carried in a backpack and
4360 whose boot stays stuck in a hard disk encryption passphrase
4363 * systemd-logind can be configured to also handle lid switch
4364 events even when the machine is docked or multiple displays
4365 are attached (HandleLidSwitchDocked= option).
4367 * A helper binary and a service have been added which can be
4368 used to resume from hibernation in the initramfs. A
4369 generator will parse the resume= option on the kernel
4370 command line to trigger resume.
4372 * A user console daemon systemd-consoled has been
4373 added. Currently, it is a preview, and will so far open a
4374 single terminal on each session of the user marked as
4375 Desktop=systemd-console.
4377 * Route metrics can be specified for DHCP routes added by
4380 * The SELinux context of socket-activated services can be set
4381 from the information provided by the networking stack
4382 (SELinuxContextFromNet= option).
4384 * Userspace firmware loading support has been removed and
4385 the minimum supported kernel version is thus bumped to 3.7.
4387 * Timeout for udev workers has been increased from 1 to 3
4388 minutes, but a warning will be printed after 1 minute to
4389 help diagnose kernel modules that take a long time to load.
4391 * Udev rules can now remove tags on devices with TAG-="foobar".
4393 * systemd's readahead implementation has been removed. In many
4394 circumstances it didn't give expected benefits even for
4395 rotational disk drives and was becoming less relevant in the
4396 age of SSDs. As none of the developers has been using
4397 rotating media anymore, and nobody stepped up to actively
4398 maintain this component of systemd it has now been removed.
4400 * Swap units can use Options= to specify discard options.
4401 Discard options specified for swaps in /etc/fstab are now
4404 * Docker containers are now detected as a separate type of
4407 * The Password Agent protocol gained support for queries where
4408 the user input is shown, useful e.g. for user names.
4409 systemd-ask-password gained a new --echo option to turn that
4412 * The default sysctl.d/ snippets will now set:
4414 net.core.default_qdisc = fq_codel
4416 This selects Fair Queuing Controlled Delay as the default
4417 queuing discipline for network interfaces. fq_codel helps
4418 fight the network bufferbloat problem. It is believed to be
4419 a good default with no tuning required for most workloads.
4420 Downstream distributions may override this choice. On 10Gbit
4421 servers that do not do forwarding, "fq" may perform better.
4422 Systems without a good clocksource should use "pfifo_fast".
4424 * If kdbus is enabled during build a new option BusPolicy= is
4425 available for service units, that allows locking all service
4426 processes into a stricter bus policy, in order to limit
4427 access to various bus services, or even hide most of them
4428 from the service's view entirely.
4430 * networkctl will now show the .network and .link file
4431 networkd has applied to a specific interface.
4433 * sd-login gained a new API call sd_session_get_desktop() to
4434 query which desktop environment has been selected for a
4437 * UNIX utmp support is now compile-time optional to support
4438 legacy-free systems.
4440 * systemctl gained two new commands "add-wants" and
4441 "add-requires" for pulling in units from specific targets
4444 * If the word "rescue" is specified on the kernel command line
4445 the system will now boot into rescue mode (aka
4446 rescue.target), which was previously available only by
4447 specifying "1" or "systemd.unit=rescue.target" on the kernel
4448 command line. This new kernel command line option nicely
4449 mirrors the already existing "emergency" kernel command line
4452 * New kernel command line options mount.usr=, mount.usrflags=,
4453 mount.usrfstype= have been added that match root=, rootflags=,
4454 rootfstype= but allow mounting a specific file system to
4457 * The $NOTIFY_SOCKET is now also passed to control processes of
4458 services, not only the main process.
4460 * This version reenables support for fsck's -l switch. This
4461 means at least version v2.25 of util-linux is required for
4462 operation, otherwise dead-locks on device nodes may
4463 occur. Again: you need to update util-linux to at least
4464 v2.25 when updating systemd to v217.
4466 * The "multi-seat-x" tool has been removed from systemd, as
4467 its functionality has been integrated into X servers 1.16,
4468 and the tool is hence redundant. It is recommended to update
4469 display managers invoking this tool to simply invoke X
4470 directly from now on, again.
4472 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
4473 message flag has been added for all of systemd's PolicyKit
4474 authenticated method calls has been added. In particular
4475 this now allows optional interactive authorization via
4476 PolicyKit for many of PID1's privileged operations such as
4477 unit file enabling and disabling.
4479 * "udevadm hwdb --update" learnt a new switch "--usr" for
4480 placing the rebuilt hardware database in /usr instead of
4481 /etc. When used only hardware database entries stored in
4482 /usr will be used, and any user database entries in /etc are
4483 ignored. This functionality is useful for vendors to ship a
4484 pre-built database on systems where local configuration is
4485 unnecessary or unlikely.
4487 * Calendar time specifications in .timer units now also
4488 understand the strings "semi-annually", "quarterly" and
4489 "minutely" as shortcuts (in addition to the preexisting
4490 "anually", "hourly", ...).
4492 * systemd-tmpfiles will now correctly create files in /dev
4493 at boot which are marked for creation only at boot. It is
4494 recommended to always create static device nodes with 'c!'
4495 and 'b!', so that they are created only at boot and not
4496 overwritten at runtime.
4498 * When the watchdog logic is used for a service (WatchdogSec=)
4499 and the watchdog timeout is hit the service will now be
4500 terminated with SIGABRT (instead of just SIGTERM), in order
4501 to make sure a proper coredump and backtrace is
4502 generated. This ensures that hanging services will result in
4503 similar coredump/backtrace behaviour as services that hit a
4506 Contributions from: Andreas Henriksson, Andrei Borzenkov,
4507 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
4508 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
4509 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
4510 Herrmann, David Sommerseth, David Strauss, Emil Renner
4511 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
4512 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
4513 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
4514 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
4515 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
4516 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
4517 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
4518 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
4519 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
4520 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
4521 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
4522 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
4523 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
4524 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
4525 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
4526 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
4529 — Berlin, 2014-10-28
4533 * timedated no longer reads NTP implementation unit names from
4534 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
4535 implementations should add a
4537 Conflicts=systemd-timesyncd.service
4539 to their unit files to take over and replace systemd's NTP
4540 default functionality.
4542 * systemd-sysusers gained a new line type "r" for configuring
4543 which UID/GID ranges to allocate system users/groups
4544 from. Lines of type "u" may now add an additional column
4545 that specifies the home directory for the system user to be
4546 created. Also, systemd-sysusers may now optionally read user
4547 information from STDIN instead of a file. This is useful for
4548 invoking it from RPM preinst scriptlets that need to create
4549 users before the first RPM file is installed since these
4550 files might need to be owned by them. A new
4551 %sysusers_create_inline RPM macro has been introduced to do
4552 just that. systemd-sysusers now updates the shadow files as
4553 well as the user/group databases, which should enhance
4554 compatibility with certain tools like grpck.
4556 * A number of bus APIs of PID 1 now optionally consult
4557 PolicyKit to permit access for otherwise unprivileged
4558 clients under certain conditions. Note that this currently
4559 doesn't support interactive authentication yet, but this is
4560 expected to be added eventually, too.
4562 * /etc/machine-info now has new fields for configuring the
4563 deployment environment of the machine, as well as the
4564 location of the machine. hostnamectl has been updated with
4565 new command to update these fields.
4567 * systemd-timesyncd has been updated to automatically acquire
4568 NTP server information from systemd-networkd, which might
4569 have been discovered via DHCP.
4571 * systemd-resolved now includes a caching DNS stub resolver
4572 and a complete LLMNR name resolution implementation. A new
4573 NSS module "nss-resolve" has been added which can be used
4574 instead of glibc's own "nss-dns" to resolve hostnames via
4575 systemd-resolved. Hostnames, addresses and arbitrary RRs may
4576 be resolved via systemd-resolved D-Bus APIs. In contrast to
4577 the glibc internal resolver systemd-resolved is aware of
4578 multi-homed system, and keeps DNS server and caches separate
4579 and per-interface. Queries are sent simultaneously on all
4580 interfaces that have DNS servers configured, in order to
4581 properly handle VPNs and local LANs which might resolve
4582 separate sets of domain names. systemd-resolved may acquire
4583 DNS server information from systemd-networkd automatically,
4584 which in turn might have discovered them via DHCP. A tool
4585 "systemd-resolve-host" has been added that may be used to
4586 query the DNS logic in resolved. systemd-resolved implements
4587 IDNA and automatically uses IDNA or UTF-8 encoding depending
4588 on whether classic DNS or LLMNR is used as transport. In the
4589 next releases we intend to add a DNSSEC and mDNS/DNS-SD
4590 implementation to systemd-resolved.
4592 * A new NSS module nss-mymachines has been added, that
4593 automatically resolves the names of all local registered
4594 containers to their respective IP addresses.
4596 * A new client tool "networkctl" for systemd-networkd has been
4597 added. It currently is entirely passive and will query
4598 networking configuration from udev, rtnetlink and networkd,
4599 and present it to the user in a very friendly
4600 way. Eventually, we hope to extend it to become a full
4601 control utility for networkd.
4603 * .socket units gained a new DeferAcceptSec= setting that
4604 controls the kernels' TCP_DEFER_ACCEPT sockopt for
4605 TCP. Similarly, support for controlling TCP keep-alive
4606 settings has been added (KeepAliveTimeSec=,
4607 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
4608 turning off Nagle's algorithm on TCP has been added
4611 * logind learned a new session type "web", for use in projects
4612 like Cockpit which register web clients as PAM sessions.
4614 * timer units with at least one OnCalendar= setting will now
4615 be started only after time-sync.target has been
4616 reached. This way they will not elapse before the system
4617 clock has been corrected by a local NTP client or
4618 similar. This is particular useful on RTC-less embedded
4619 machines, that come up with an invalid system clock.
4621 * systemd-nspawn's --network-veth= switch should now result in
4622 stable MAC addresses for both the outer and the inner side
4625 * systemd-nspawn gained a new --volatile= switch for running
4626 container instances with /etc or /var unpopulated.
4628 * The kdbus client code has been updated to use the new Linux
4629 3.17 memfd subsystem instead of the old kdbus-specific one.
4631 * systemd-networkd's DHCP client and server now support
4632 FORCERENEW. There are also new configuration options to
4633 configure the vendor client identifier and broadcast mode
4636 * systemd will no longer inform the kernel about the current
4637 timezone, as this is necessarily incorrect and racy as the
4638 kernel has no understanding of DST and similar
4639 concepts. This hence means FAT timestamps will be always
4640 considered UTC, similar to what Android is already
4641 doing. Also, when the RTC is configured to the local time
4642 (rather than UTC) systemd will never synchronize back to it,
4643 as this might confuse Windows at a later boot.
4645 * systemd-analyze gained a new command "verify" for offline
4646 validation of unit files.
4648 * systemd-networkd gained support for a couple of additional
4649 settings for bonding networking setups. Also, the metric for
4650 statically configured routes may now be configured. For
4651 network interfaces where this is appropriate the peer IP
4652 address may now be configured.
4654 * systemd-networkd's DHCP client will no longer request
4655 broadcasting by default, as this tripped up some networks.
4656 For hardware where broadcast is required the feature should
4657 be switched back on using RequestBroadcast=yes.
4659 * systemd-networkd will now set up IPv4LL addresses (when
4660 enabled) even if DHCP is configured successfully.
4662 * udev will now default to respect network device names given
4663 by the kernel when the kernel indicates that these are
4664 predictable. This behavior can be tweaked by changing
4665 NamePolicy= in the relevant .link file.
4667 * A new library systemd-terminal has been added that
4668 implements full TTY stream parsing and rendering. This
4669 library is supposed to be used later on for implementing a
4670 full userspace VT subsystem, replacing the current kernel
4673 * A new tool systemd-journal-upload has been added to push
4674 journal data to a remote system running
4675 systemd-journal-remote.
4677 * journald will no longer forward all local data to another
4678 running syslog daemon. This change has been made because
4679 rsyslog (which appears to be the most commonly used syslog
4680 implementation these days) no longer makes use of this, and
4681 instead pulls the data out of the journal on its own. Since
4682 forwarding the messages to a non-existent syslog server is
4683 more expensive than we assumed we have now turned this
4684 off. If you run a syslog server that is not a recent rsyslog
4685 version, you have to turn this option on again
4686 (ForwardToSyslog= in journald.conf).
4688 * journald now optionally supports the LZ4 compressor for
4689 larger journal fields. This compressor should perform much
4690 better than XZ which was the previous default.
4692 * machinectl now shows the IP addresses of local containers,
4693 if it knows them, plus the interface name of the container.
4695 * A new tool "systemd-escape" has been added that makes it
4696 easy to escape strings to build unit names and similar.
4698 * sd_notify() messages may now include a new ERRNO= field
4699 which is parsed and collected by systemd and shown among the
4700 "systemctl status" output for a service.
4702 * A new component "systemd-firstboot" has been added that
4703 queries the most basic systemd information (timezone,
4704 hostname, root password) interactively on first
4705 boot. Alternatively it may also be used to provision these
4706 things offline on OS images installed into directories.
4708 * The default sysctl.d/ snippets will now set
4710 net.ipv4.conf.default.promote_secondaries=1
4712 This has the benefit of no flushing secondary IP addresses
4713 when primary addresses are removed.
4715 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
4716 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
4717 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
4718 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
4719 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
4720 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
4721 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
4722 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
4723 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
4724 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
4725 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
4726 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
4727 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
4728 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
4729 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
4731 — Berlin, 2014-08-19
4735 * A new tool systemd-sysusers has been added. This tool
4736 creates system users and groups in /etc/passwd and
4737 /etc/group, based on static declarative system user/group
4738 definitions in /usr/lib/sysusers.d/. This is useful to
4739 enable factory resets and volatile systems that boot up with
4740 an empty /etc directory, and thus need system users and
4741 groups created during early boot. systemd now also ships
4742 with two default sysusers.d/ files for the most basic
4743 users and groups systemd and the core operating system
4746 * A new tmpfiles snippet has been added that rebuilds the
4747 essential files in /etc on boot, should they be missing.
4749 * A directive for ensuring automatic clean-up of
4750 /var/cache/man/ has been removed from the default
4751 configuration. This line should now be shipped by the man
4752 implementation. The necessary change has been made to the
4753 man-db implementation. Note that you need to update your man
4754 implementation to one that ships this line, otherwise no
4755 automatic clean-up of /var/cache/man will take place.
4757 * A new condition ConditionNeedsUpdate= has been added that
4758 may conditionalize services to only run when /etc or /var
4759 are "older" than the vendor operating system resources in
4760 /usr. This is useful for reconstructing or updating /etc
4761 after an offline update of /usr or a factory reset, on the
4762 next reboot. Services that want to run once after such an
4763 update or reset should use this condition and order
4764 themselves before the new systemd-update-done.service, which
4765 will mark the two directories as fully updated. A number of
4766 service files have been added making use of this, to rebuild
4767 the udev hardware database, the journald message catalog and
4768 dynamic loader cache (ldconfig). The systemd-sysusers tool
4769 described above also makes use of this now. With this in
4770 place it is now possible to start up a minimal operating
4771 system with /etc empty cleanly. For more information on the
4772 concepts involved see this recent blog story:
4774 http://0pointer.de/blog/projects/stateless.html
4776 * A new system group "input" has been introduced, and all
4777 input device nodes get this group assigned. This is useful
4778 for system-level software to get access to input devices. It
4779 complements what is already done for "audio" and "video".
4781 * systemd-networkd learnt minimal DHCPv4 server support in
4782 addition to the existing DHCPv4 client support. It also
4783 learnt DHCPv6 client and IPv6 Router Solicitation client
4784 support. The DHCPv4 client gained support for static routes
4785 passed in from the server. Note that the [DHCPv4] section
4786 known in older systemd-networkd versions has been renamed to
4787 [DHCP] and is now also used by the DHCPv6 client. Existing
4788 .network files using settings of this section should be
4789 updated, though compatibility is maintained. Optionally, the
4790 client hostname may now be sent to the DHCP server.
4792 * networkd gained support for vxlan virtual networks as well
4793 as tun/tap and dummy devices.
4795 * networkd gained support for automatic allocation of address
4796 ranges for interfaces from a system-wide pool of
4797 addresses. This is useful for dynamically managing a large
4798 number of interfaces with a single network configuration
4799 file. In particular this is useful to easily assign
4800 appropriate IP addresses to the veth links of a large number
4801 of nspawn instances.
4803 * RPM macros for processing sysusers, sysctl and binfmt
4804 drop-in snippets at package installation time have been
4807 * The /etc/os-release file should now be placed in
4808 /usr/lib/os-release. The old location is automatically
4809 created as symlink. /usr/lib is the more appropriate
4810 location of this file, since it shall actually describe the
4811 vendor operating system shipped in /usr, and not the
4812 configuration stored in /etc.
4814 * .mount units gained a new boolean SloppyOptions= setting
4815 that maps to mount(8)'s -s option which enables permissive
4816 parsing of unknown mount options.
4818 * tmpfiles learnt a new "L+" directive which creates a symlink
4819 but (unlike "L") deletes a pre-existing file first, should
4820 it already exist and not already be the correct
4821 symlink. Similarly, "b+", "c+" and "p+" directives have been
4822 added as well, which create block and character devices, as
4823 well as fifos in the filesystem, possibly removing any
4824 pre-existing files of different types.
4826 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
4827 'argument' field (which so far specified the source to
4828 symlink/copy the files from) is now optional. If omitted the
4829 same file os copied from /usr/share/factory/ suffixed by the
4830 full destination path. This is useful for populating /etc
4831 with essential files, by copying them from vendor defaults
4832 shipped in /usr/share/factory/etc.
4834 * A new command "systemctl preset-all" has been added that
4835 applies the service preset settings to all installed unit
4836 files. A new switch --preset-mode= has been added that
4837 controls whether only enable or only disable operations
4840 * A new command "systemctl is-system-running" has been added
4841 that allows checking the overall state of the system, for
4842 example whether it is fully up and running.
4844 * When the system boots up with an empty /etc, the equivalent
4845 to "systemctl preset-all" is executed during early boot, to
4846 make sure all default services are enabled after a factory
4849 * systemd now contains a minimal preset file that enables the
4850 most basic services systemd ships by default.
4852 * Unit files' [Install] section gained a new DefaultInstance=
4853 field for defining the default instance to create if a
4854 template unit is enabled with no instance specified.
4856 * A new passive target cryptsetup-pre.target has been added
4857 that may be used by services that need to make they run and
4858 finish before the first LUKS cryptographic device is set up.
4860 * The /dev/loop-control and /dev/btrfs-control device nodes
4861 are now owned by the "disk" group by default, opening up
4862 access to this group.
4864 * systemd-coredump will now automatically generate a
4865 stack trace of all core dumps taking place on the system,
4866 based on elfutils' libdw library. This stack trace is logged
4869 * systemd-coredump may now optionally store coredumps directly
4870 on disk (in /var/lib/systemd/coredump, possibly compressed),
4871 instead of storing them unconditionally in the journal. This
4872 mode is the new default. A new configuration file
4873 /etc/systemd/coredump.conf has been added to configure this
4874 and other parameters of systemd-coredump.
4876 * coredumpctl gained a new "info" verb to show details about a
4877 specific coredump. A new switch "-1" has also been added
4878 that makes sure to only show information about the most
4879 recent entry instead of all entries. Also, as the tool is
4880 generally useful now the "systemd-" prefix of the binary
4881 name has been removed. Distributions that want to maintain
4882 compatibility with the old name should add a symlink from
4883 the old name to the new name.
4885 * journald's SplitMode= now defaults to "uid". This makes sure
4886 that unprivileged users can access their own coredumps with
4887 coredumpctl without restrictions.
4889 * New kernel command line options "systemd.wants=" (for
4890 pulling an additional unit during boot), "systemd.mask="
4891 (for masking a specific unit for the boot), and
4892 "systemd.debug-shell" (for enabling the debug shell on tty9)
4893 have been added. This is implemented in the new generator
4894 "systemd-debug-generator".
4896 * systemd-nspawn will now by default filter a couple of
4897 syscalls for containers, among them those required for
4898 kernel module loading, direct x86 IO port access, swap
4899 management, and kexec. Most importantly though
4900 open_by_handle_at() is now prohibited for containers,
4901 closing a hole similar to a recently discussed vulnerability
4902 in docker regarding access to files on file hierarchies the
4903 container should normally not have access to. Note that, for
4904 nspawn, we generally make no security claims anyway (and
4905 this is explicitly documented in the man page), so this is
4906 just a fix for one of the most obvious problems.
4908 * A new man page file-hierarchy(7) has been added that
4909 contains a minimized, modernized version of the file system
4910 layout systemd expects, similar in style to the FHS
4911 specification or hier(5). A new tool systemd-path(1) has
4912 been added to query many of these paths for the local
4915 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
4916 longer done. Since the directory now has a per-user size
4917 limit, and is cleaned on logout this appears unnecessary,
4918 in particular since this now brings the lifecycle of this
4919 directory closer in line with how IPC objects are handled.
4921 * systemd.pc now exports a number of additional directories,
4922 including $libdir (which is useful to identify the library
4923 path for the primary architecture of the system), and a
4924 couple of drop-in directories.
4926 * udev's predictable network interface names now use the dev_port
4927 sysfs attribute, introduced in linux 3.15 instead of dev_id to
4928 distinguish between ports of the same PCI function. dev_id should
4929 only be used for ports using the same HW address, hence the need
4932 * machined has been updated to export the OS version of a
4933 container (read from /etc/os-release and
4934 /usr/lib/os-release) on the bus. This is now shown in
4935 "machinectl status" for a machine.
4937 * A new service setting RestartForceExitStatus= has been
4938 added. If configured to a set of exit signals or process
4939 return values, the service will be restarted when the main
4940 daemon process exits with any of them, regardless of the
4943 * systemctl's -H switch for connecting to remote systemd
4944 machines has been extended so that it may be used to
4945 directly connect to a specific container on the
4946 host. "systemctl -H root@foobar:waldi" will now connect as
4947 user "root" to host "foobar", and then proceed directly to
4948 the container named "waldi". Note that currently you have to
4949 authenticate as user "root" for this to work, as entering
4950 containers is a privileged operation.
4952 Contributions from: Andreas Henriksson, Benjamin Steinwender,
4953 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
4954 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
4955 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
4956 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
4957 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
4958 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
4959 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
4960 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
4961 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
4962 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
4963 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
4965 — Berlin, 2014-07-03
4969 * As an experimental feature, udev now tries to lock the
4970 disk device node (flock(LOCK_SH|LOCK_NB)) while it
4971 executes events for the disk or any of its partitions.
4972 Applications like partitioning programs can lock the
4973 disk device node (flock(LOCK_EX)) and claim temporary
4974 device ownership that way; udev will entirely skip all event
4975 handling for this disk and its partitions. If the disk
4976 was opened for writing, the close will trigger a partition
4977 table rescan in udev's "watch" facility, and if needed
4978 synthesize "change" events for the disk and all its partitions.
4979 This is now unconditionally enabled, and if it turns out to
4980 cause major problems, we might turn it on only for specific
4981 devices, or might need to disable it entirely. Device Mapper
4982 devices are excluded from this logic.
4984 * We temporarily dropped the "-l" switch for fsck invocations,
4985 since they collide with the flock() logic above. util-linux
4986 upstream has been changed already to avoid this conflict,
4987 and we will readd "-l" as soon as util-linux with this
4988 change has been released.
4990 * The dependency on libattr has been removed. Since a long
4991 time, the extended attribute calls have moved to glibc, and
4992 libattr is thus unnecessary.
4994 * Virtualization detection works without privileges now. This
4995 means the systemd-detect-virt binary no longer requires
4996 CAP_SYS_PTRACE file capabilities, and our daemons can run
4997 with fewer privileges.
4999 * systemd-networkd now runs under its own "systemd-network"
5000 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
5001 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
5002 loses the ability to write to files owned by root this way.
5004 * Similarly, systemd-resolved now runs under its own
5005 "systemd-resolve" user with no capabilities remaining.
5007 * Similarly, systemd-bus-proxyd now runs under its own
5008 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
5010 * systemd-networkd gained support for setting up "veth"
5011 virtual Ethernet devices for container connectivity, as well
5012 as GRE and VTI tunnels.
5014 * systemd-networkd will no longer automatically attempt to
5015 manually load kernel modules necessary for certain tunnel
5016 transports. Instead, it is assumed the kernel loads them
5017 automatically when required. This only works correctly on
5018 very new kernels. On older kernels, please consider adding
5019 the kernel modules to /etc/modules-load.d/ as a work-around.
5021 * The resolv.conf file systemd-resolved generates has been
5022 moved to /run/systemd/resolve/. If you have a symlink from
5023 /etc/resolv.conf, it might be necessary to correct it.
5025 * Two new service settings, ProtectHome= and ProtectSystem=,
5026 have been added. When enabled, they will make the user data
5027 (such as /home) inaccessible or read-only and the system
5028 (such as /usr) read-only, for specific services. This allows
5029 very light-weight per-service sandboxing to avoid
5030 modifications of user data or system files from
5031 services. These two new switches have been enabled for all
5032 of systemd's long-running services, where appropriate.
5034 * Socket units gained new SocketUser= and SocketGroup=
5035 settings to set the owner user and group of AF_UNIX sockets
5036 and FIFOs in the file system.
5038 * Socket units gained a new RemoveOnStop= setting. If enabled,
5039 all FIFOS and sockets in the file system will be removed
5040 when the specific socket unit is stopped.
5042 * Socket units gained a new Symlinks= setting. It takes a list
5043 of symlinks to create to file system sockets or FIFOs
5044 created by the specific Unix sockets. This is useful to
5045 manage symlinks to socket nodes with the same life-cycle as
5048 * The /dev/log socket and /dev/initctl FIFO have been moved to
5049 /run, and have been replaced by symlinks. This allows
5050 connecting to these facilities even if PrivateDevices=yes is
5051 used for a service (which makes /dev/log itself unavailable,
5052 but /run is left). This also has the benefit of ensuring
5053 that /dev only contains device nodes, directories and
5054 symlinks, and nothing else.
5056 * sd-daemon gained two new calls sd_pid_notify() and
5057 sd_pid_notifyf(). They are similar to sd_notify() and
5058 sd_notifyf(), but allow overriding of the source PID of
5059 notification messages if permissions permit this. This is
5060 useful to send notify messages on behalf of a different
5061 process (for example, the parent process). The
5062 systemd-notify tool has been updated to make use of this
5063 when sending messages (so that notification messages now
5064 originate from the shell script invoking systemd-notify and
5065 not the systemd-notify process itself. This should minimize
5066 a race where systemd fails to associate notification
5067 messages to services when the originating process already
5070 * A new "on-abnormal" setting for Restart= has been added. If
5071 set, it will result in automatic restarts on all "abnormal"
5072 reasons for a process to exit, which includes unclean
5073 signals, core dumps, timeouts and watchdog timeouts, but
5074 does not include clean and unclean exit codes or clean
5075 signals. Restart=on-abnormal is an alternative for
5076 Restart=on-failure for services that shall be able to
5077 terminate and avoid restarts on certain errors, by
5078 indicating so with an unclean exit code. Restart=on-failure
5079 or Restart=on-abnormal is now the recommended setting for
5080 all long-running services.
5082 * If the InaccessibleDirectories= service setting points to a
5083 mount point (or if there are any submounts contained within
5084 it), it is now attempted to completely unmount it, to make
5085 the file systems truly unavailable for the respective
5088 * The ReadOnlyDirectories= service setting and
5089 systemd-nspawn's --read-only parameter are now recursively
5090 applied to all submounts, too.
5092 * Mount units may now be created transiently via the bus APIs.
5094 * The support for SysV and LSB init scripts has been removed
5095 from the systemd daemon itself. Instead, it is now
5096 implemented as a generator that creates native systemd units
5097 from these scripts when needed. This enables us to remove a
5098 substantial amount of legacy code from PID 1, following the
5099 fact that many distributions only ship a very small number
5100 of LSB/SysV init scripts nowadays.
5102 * Privileged Xen (dom0) domains are not considered
5103 virtualization anymore by the virtualization detection
5104 logic. After all, they generally have unrestricted access to
5105 the hardware and usually are used to manage the unprivileged
5108 * systemd-tmpfiles gained a new "C" line type, for copying
5109 files or entire directories.
5111 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
5112 lines. So far, they have been non-globbing versions of the
5113 latter, and have thus been redundant. In future, it is
5114 recommended to only use "z". "m" has hence been removed
5115 from the documentation, even though it stays supported.
5117 * A tmpfiles snippet to recreate the most basic structure in
5118 /var has been added. This is enough to create the /var/run →
5119 /run symlink and create a couple of structural
5120 directories. This allows systems to boot up with an empty or
5121 volatile /var. Of course, while with this change, the core OS
5122 now is capable with dealing with a volatile /var, not all
5123 user services are ready for it. However, we hope that sooner
5124 or later, many service daemons will be changed upstream so
5125 that they are able to automatically create their necessary
5126 directories in /var at boot, should they be missing. This is
5127 the first step to allow state-less systems that only require
5128 the vendor image for /usr to boot.
5130 * systemd-nspawn has gained a new --tmpfs= switch to mount an
5131 empty tmpfs instance to a specific directory. This is
5132 particularly useful for making use of the automatic
5133 reconstruction of /var (see above), by passing --tmpfs=/var.
5135 * Access modes specified in tmpfiles snippets may now be
5136 prefixed with "~", which indicates that they shall be masked
5137 by whether the existing file or directory is currently
5138 writable, readable or executable at all. Also, if specified,
5139 the sgid/suid/sticky bits will be masked for all
5142 * A new passive target unit "network-pre.target" has been
5143 added which is useful for services that shall run before any
5144 network is configured, for example firewall scripts.
5146 * The "floppy" group that previously owned the /dev/fd*
5147 devices is no longer used. The "disk" group is now used
5148 instead. Distributions should probably deprecate usage of
5151 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
5152 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
5153 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
5154 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
5155 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
5156 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
5159 — Berlin, 2014-06-11
5163 * A new "systemd-timesyncd" daemon has been added for
5164 synchronizing the system clock across the network. It
5165 implements an SNTP client. In contrast to NTP
5166 implementations such as chrony or the NTP reference server,
5167 this only implements a client side, and does not bother with
5168 the full NTP complexity, focusing only on querying time from
5169 one remote server and synchronizing the local clock to
5170 it. Unless you intend to serve NTP to networked clients or
5171 want to connect to local hardware clocks, this simple NTP
5172 client should be more than appropriate for most
5173 installations. The daemon runs with minimal privileges, and
5174 has been hooked up with networkd to only operate when
5175 network connectivity is available. The daemon saves the
5176 current clock to disk every time a new NTP sync has been
5177 acquired, and uses this to possibly correct the system clock
5178 early at bootup, in order to accommodate for systems that
5179 lack an RTC such as the Raspberry Pi and embedded devices,
5180 and to make sure that time monotonically progresses on these
5181 systems, even if it is not always correct. To make use of
5182 this daemon, a new system user and group "systemd-timesync"
5183 needs to be created on installation of systemd.
5185 * The queue "seqnum" interface of libudev has been disabled, as
5186 it was generally incompatible with device namespacing as
5187 sequence numbers of devices go "missing" if the devices are
5188 part of a different namespace.
5190 * "systemctl list-timers" and "systemctl list-sockets" gained
5191 a --recursive switch for showing units of these types also
5192 for all local containers, similar in style to the already
5193 supported --recursive switch for "systemctl list-units".
5195 * A new RebootArgument= setting has been added for service
5196 units, which may be used to specify a kernel reboot argument
5197 to use when triggering reboots with StartLimitAction=.
5199 * A new FailureAction= setting has been added for service
5200 units which may be used to specify an operation to trigger
5201 when a service fails. This works similarly to
5202 StartLimitAction=, but unlike it, controls what is done
5203 immediately rather than only after several attempts to
5204 restart the service in question.
5206 * hostnamed got updated to also expose the kernel name,
5207 release, and version on the bus. This is useful for
5208 executing commands like hostnamectl with the -H switch.
5209 systemd-analyze makes use of this to properly display
5210 details when running non-locally.
5212 * The bootchart tool can now show cgroup information in the
5213 graphs it generates.
5215 * The CFS CPU quota cgroup attribute is now exposed for
5216 services. The new CPUQuota= switch has been added for this
5217 which takes a percentage value. Setting this will have the
5218 result that a service may never get more CPU time than the
5219 specified percentage, even if the machine is otherwise idle.
5221 * systemd-networkd learned IPIP and SIT tunnel support.
5223 * LSB init scripts exposing a dependency on $network will now
5224 get a dependency on network-online.target rather than simply
5225 network.target. This should bring LSB handling closer to
5226 what it was on SysV systems.
5228 * A new fsck.repair= kernel option has been added to control
5229 how fsck shall deal with unclean file systems at boot.
5231 * The (.ini) configuration file parser will now silently
5232 ignore sections whose name begins with "X-". This may be
5233 used to maintain application-specific extension sections in unit
5236 * machined gained a new API to query the IP addresses of
5237 registered containers. "machinectl status" has been updated
5238 to show these addresses in its output.
5240 * A new call sd_uid_get_display() has been added to the
5241 sd-login APIs for querying the "primary" session of a
5242 user. The "primary" session of the user is elected from the
5243 user's sessions and generally a graphical session is
5244 preferred over a text one.
5246 * A minimal systemd-resolved daemon has been added. It
5247 currently simply acts as a companion to systemd-networkd and
5248 manages resolv.conf based on per-interface DNS
5249 configuration, possibly supplied via DHCP. In the long run
5250 we hope to extend this into a local DNSSEC enabled DNS and
5253 * The systemd-networkd-wait-online tool is now enabled by
5254 default. It will delay network-online.target until a network
5255 connection has been configured. The tool primarily integrates
5256 with networkd, but will also make a best effort to make sense
5257 of network configuration performed in some other way.
5259 * Two new service options StartupCPUShares= and
5260 StartupBlockIOWeight= have been added that work similarly to
5261 CPUShares= and BlockIOWeight= however only apply during
5262 system startup. This is useful to prioritize certain services
5263 differently during bootup than during normal runtime.
5265 * hostnamed has been changed to prefer the statically
5266 configured hostname in /etc/hostname (unless set to
5267 'localhost' or empty) over any dynamic one supplied by
5268 dhcp. With this change, the rules for picking the hostname
5269 match more closely the rules of other configuration settings
5270 where the local administrator's configuration in /etc always
5271 overrides any other settings.
5273 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
5274 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
5275 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
5276 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
5277 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
5278 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
5279 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
5280 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
5281 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
5282 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
5283 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
5284 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
5285 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
5286 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
5287 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
5288 Lindskog, WaLyong Cho, Will Woods, Zbigniew
5291 — Beijing, 2014-05-28
5295 * When restoring the screen brightness at boot, stay away from
5296 the darkest setting or from the lowest 5% of the available
5297 range, depending on which is the larger value of both. This
5298 should effectively protect the user from rebooting into a
5299 black screen, should the brightness have been set to minimum
5302 * sd-login gained a new sd_machine_get_class() call to
5303 determine the class ("vm" or "container") of a machine
5304 registered with machined.
5306 * sd-login gained new calls
5307 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
5308 to query the identity of the peer of a local AF_UNIX
5309 connection. They operate similarly to their sd_pid_get_xyz()
5312 * PID 1 will now maintain a system-wide system state engine
5313 with the states "starting", "running", "degraded",
5314 "maintenance", "stopping". These states are bound to system
5315 startup, normal runtime, runtime with at least one failed
5316 service, rescue/emergency mode and system shutdown. This
5317 state is shown in the "systemctl status" output when no unit
5318 name is passed. It is useful to determine system state, in
5319 particularly when doing so for many systems or containers at
5322 * A new command "list-machines" has been added to "systemctl"
5323 that lists all local OS containers and shows their system
5324 state (see above), if systemd runs inside of them.
5326 * systemctl gained a new "-r" switch to recursively enumerate
5327 units on all local containers, when used with the
5328 "list-unit" command (which is the default one that is
5329 executed when no parameters are specified).
5331 * The GPT automatic partition discovery logic will now honour
5332 two GPT partition flags: one may be set on a partition to
5333 cause it to be mounted read-only, and the other may be set
5334 on a partition to ignore it during automatic discovery.
5336 * Two new GPT type UUIDs have been added for automatic root
5337 partition discovery, for 32-bit and 64-bit ARM. This is not
5338 particularly useful for discovering the root directory on
5339 these architectures during bare-metal boots (since UEFI is
5340 not common there), but still very useful to allow booting of
5341 ARM disk images in nspawn with the -i option.
5343 * MAC addresses of interfaces created with nspawn's
5344 --network-interface= switch will now be generated from the
5345 machine name, and thus be stable between multiple invocations
5348 * logind will now automatically remove all IPC objects owned
5349 by a user if she or he fully logs out. This makes sure that
5350 users who are logged out cannot continue to consume IPC
5351 resources. This covers SysV memory, semaphores and message
5352 queues as well as POSIX shared memory and message
5353 queues. Traditionally, SysV and POSIX IPC had no life-cycle
5354 limits. With this functionality, that is corrected. This may
5355 be turned off by using the RemoveIPC= switch of logind.conf.
5357 * The systemd-machine-id-setup and tmpfiles tools gained a
5358 --root= switch to operate on a specific root directory,
5361 * journald can now forward logged messages to the TTYs of all
5362 logged in users ("wall"). This is the default for all
5363 emergency messages now.
5365 * A new tool systemd-journal-remote has been added to stream
5366 journal log messages across the network.
5368 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
5369 controller trees are mounted into it. Note that the
5370 directories mounted beneath it are not read-only. This is a
5371 security measure and is particularly useful because glibc
5372 actually includes a search logic to pick any tmpfs it can
5373 find to implement shm_open() if /dev/shm is not available
5374 (which it might very well be in namespaced setups).
5376 * machinectl gained a new "poweroff" command to cleanly power
5377 down a local OS container.
5379 * The PrivateDevices= unit file setting will now also drop the
5380 CAP_MKNOD capability from the capability bound set, and
5381 imply DevicePolicy=closed.
5383 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
5384 comprehensively on all long-running systemd services where
5385 this is appropriate.
5387 * systemd-udevd will now run in a disassociated mount
5388 namespace. To mount directories from udev rules, make sure to
5389 pull in mount units via SYSTEMD_WANTS properties.
5391 * The kdbus support gained support for uploading policy into
5392 the kernel. sd-bus gained support for creating "monitoring"
5393 connections that can eavesdrop into all bus communication
5394 for debugging purposes.
5396 * Timestamps may now be specified in seconds since the UNIX
5397 epoch Jan 1st, 1970 by specifying "@" followed by the value
5400 * Native tcpwrap support in systemd has been removed. tcpwrap
5401 is old code, not really maintained anymore and has serious
5402 shortcomings, and better options such as firewalls
5403 exist. For setups that require tcpwrap usage, please
5404 consider invoking your socket-activated service via tcpd,
5405 like on traditional inetd.
5407 * A new system.conf configuration option
5408 DefaultTimerAccuracySec= has been added that controls the
5409 default AccuracySec= setting of .timer units.
5411 * Timer units gained a new WakeSystem= switch. If enabled,
5412 timers configured this way will cause the system to resume
5413 from system suspend (if the system supports that, which most
5416 * Timer units gained a new Persistent= switch. If enabled,
5417 timers configured this way will save to disk when they have
5418 been last triggered. This information is then used on next
5419 reboot to possible execute overdue timer events, that
5420 could not take place because the system was powered off.
5421 This enables simple anacron-like behaviour for timer units.
5423 * systemctl's "list-timers" will now also list the time a
5424 timer unit was last triggered in addition to the next time
5425 it will be triggered.
5427 * systemd-networkd will now assign predictable IPv4LL
5428 addresses to its local interfaces.
5430 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
5431 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
5432 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
5433 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
5434 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
5435 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
5436 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
5437 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
5440 — Berlin, 2014-03-25
5444 * A new unit file setting RestrictAddressFamilies= has been
5445 added to restrict which socket address families unit
5446 processes gain access to. This takes address family names
5447 like "AF_INET" or "AF_UNIX", and is useful to minimize the
5448 attack surface of services via exotic protocol stacks. This
5449 is built on seccomp system call filters.
5451 * Two new unit file settings RuntimeDirectory= and
5452 RuntimeDirectoryMode= have been added that may be used to
5453 manage a per-daemon runtime directories below /run. This is
5454 an alternative for setting up directory permissions with
5455 tmpfiles snippets, and has the advantage that the runtime
5456 directory's lifetime is bound to the daemon runtime and that
5457 the daemon starts up with an empty directory each time. This
5458 is particularly useful when writing services that drop
5459 privileges using the User= or Group= setting.
5461 * The DeviceAllow= unit setting now supports globbing for
5462 matching against device group names.
5464 * The systemd configuration file system.conf gained new
5465 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
5466 DefaultMemoryAccounting= to globally turn on/off accounting
5467 for specific resources (cgroups) for all units. These
5468 settings may still be overridden individually in each unit
5471 * systemd-gpt-auto-generator is now able to discover /srv and
5472 root partitions in addition to /home and swap partitions. It
5473 also supports LUKS-encrypted partitions now. With this in
5474 place, automatic discovery of partitions to mount following
5475 the Discoverable Partitions Specification
5476 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
5477 is now a lot more complete. This allows booting without
5478 /etc/fstab and without root= on the kernel command line on
5479 systems prepared appropriately.
5481 * systemd-nspawn gained a new --image= switch which allows
5482 booting up disk images and Linux installations on any block
5483 device that follow the Discoverable Partitions Specification
5484 (see above). This means that installations made with
5485 appropriately updated installers may now be started and
5486 deployed using container managers, completely
5487 unmodified. (We hope that libvirt-lxc will add support for
5488 this feature soon, too.)
5490 * systemd-nspawn gained a new --network-macvlan= setting to
5491 set up a private macvlan interface for the
5492 container. Similarly, systemd-networkd gained a new
5493 Kind=macvlan setting in .netdev files.
5495 * systemd-networkd now supports configuring local addresses
5498 * A new tool systemd-network-wait-online has been added to
5499 synchronously wait for network connectivity using
5502 * The sd-bus.h bus API gained a new sd_bus_track object for
5503 tracking the life-cycle of bus peers. Note that sd-bus.h is
5504 still not a public API though (unless you specify
5505 --enable-kdbus on the configure command line, which however
5506 voids your warranty and you get no API stability guarantee).
5508 * The $XDG_RUNTIME_DIR runtime directories for each user are
5509 now individual tmpfs instances, which has the benefit of
5510 introducing separate pools for each user, with individual
5511 size limits, and thus making sure that unprivileged clients
5512 can no longer negatively impact the system or other users by
5513 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
5514 RuntimeDirectorySize= has been introduced that allows
5515 controlling the default size limit for all users. It
5516 defaults to 10% of the available physical memory. This is no
5517 replacement for quotas on tmpfs though (which the kernel
5518 still does not support), as /dev/shm and /tmp are still
5519 shared resources used by both the system and unprivileged
5522 * logind will now automatically turn off automatic suspending
5523 on laptop lid close when more than one display is
5524 connected. This was previously expected to be implemented
5525 individually in desktop environments (such as GNOME),
5526 however has been added to logind now, in order to fix a
5527 boot-time race where a desktop environment might not have
5528 been started yet and thus not been able to take an inhibitor
5529 lock at the time where logind already suspends the system
5530 due to a closed lid.
5532 * logind will now wait at least 30s after each system
5533 suspend/resume cycle, and 3min after system boot before
5534 suspending the system due to a closed laptop lid. This
5535 should give USB docking stations and similar enough time to
5536 be probed and configured after system resume and boot in
5537 order to then act as suspend blocker.
5539 * systemd-run gained a new --property= setting which allows
5540 initialization of resource control properties (and others)
5541 for the created scope or service unit. Example: "systemd-run
5542 --property=BlockIOWeight=10 updatedb" may be used to run
5543 updatedb at a low block IO scheduling weight.
5545 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
5546 now also work in --scope mode.
5548 * When systemd is compiled with kdbus support, basic support
5549 for enforced policies is now in place. (Note that enabling
5550 kdbus still voids your warranty and no API compatibility
5553 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
5554 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5555 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
5556 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
5557 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
5558 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
5559 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
5560 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
5561 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
5562 Zbigniew Jędrzejewski-Szmek
5564 — Berlin, 2014-03-12
5568 * systemd will now relabel /dev after loading the SMACK policy
5569 according to SMACK rules.
5571 * A new unit file option AppArmorProfile= has been added to
5572 set the AppArmor profile for the processes of a unit.
5574 * A new condition check ConditionArchitecture= has been added
5575 to conditionalize units based on the system architecture, as
5576 reported by uname()'s "machine" field.
5578 * systemd-networkd now supports matching on the system
5579 virtualization, architecture, kernel command line, host name
5582 * logind is now a lot more aggressive when suspending the
5583 machine due to a closed laptop lid. Instead of acting only
5584 on the lid close action, it will continuously watch the lid
5585 status and act on it. This is useful for laptops where the
5586 power button is on the outside of the chassis so that it can
5587 be reached without opening the lid (such as the Lenovo
5588 Yoga). On those machines, logind will now immediately
5589 re-suspend the machine if the power button has been
5590 accidentally pressed while the laptop was suspended and in a
5591 backpack or similar.
5593 * logind will now watch SW_DOCK switches and inhibit reaction
5594 to the lid switch if it is pressed. This means that logind
5595 will not suspend the machine anymore if the lid is closed
5596 and the system is docked, if the laptop supports SW_DOCK
5597 notifications via the input layer. Note that ACPI docking
5598 stations do not generate this currently. Also note that this
5599 logic is usually not fully sufficient and Desktop
5600 Environments should take a lid switch inhibitor lock when an
5601 external display is connected, as systemd will not watch
5604 * nspawn will now make use of the devices cgroup controller by
5605 default, and only permit creation of and access to the usual
5606 API device nodes like /dev/null or /dev/random, as well as
5607 access to (but not creation of) the pty devices.
5609 * We will now ship a default .network file for
5610 systemd-networkd that automatically configures DHCP for
5611 network interfaces created by nspawn's --network-veth or
5612 --network-bridge= switches.
5614 * systemd will now understand the usual M, K, G, T suffixes
5615 according to SI conventions (i.e. to the base 1000) when
5616 referring to throughput and hardware metrics. It will stay
5617 with IEC conventions (i.e. to the base 1024) for software
5618 metrics, according to what is customary according to
5619 Wikipedia. We explicitly document which base applies for
5620 each configuration option.
5622 * The DeviceAllow= setting in unit files now supports a syntax
5623 to whitelist an entire group of devices node majors at once,
5624 based on the /proc/devices listing. For example, with the
5625 string "char-pts", it is now possible to whitelist all
5626 current and future pseudo-TTYs at once.
5628 * sd-event learned a new "post" event source. Event sources of
5629 this type are triggered by the dispatching of any event
5630 source of a type that is not "post". This is useful for
5631 implementing clean-up and check event sources that are
5632 triggered by other work being done in the program.
5634 * systemd-networkd is no longer statically enabled, but uses
5635 the usual [Install] sections so that it can be
5636 enabled/disabled using systemctl. It still is enabled by
5639 * When creating a veth interface pair with systemd-nspawn, the
5640 host side will now be prefixed with "vb-" if
5641 --network-bridge= is used, and with "ve-" if --network-veth
5642 is used. This way, it is easy to distinguish these cases on
5643 the host, for example to apply different configuration to
5644 them with systemd-networkd.
5646 * The compatibility libraries for libsystemd-journal.so,
5647 libsystem-id128.so, libsystemd-login.so and
5648 libsystemd-daemon.so do not make use of IFUNC
5649 anymore. Instead, we now build libsystemd.so multiple times
5650 under these alternative names. This means that the footprint
5651 is drastically increased, but given that these are
5652 transitional compatibility libraries, this should not matter
5653 much. This change has been made necessary to support the ARM
5654 platform for these compatibility libraries, as the ARM
5655 toolchain is not really at the same level as the toolchain
5656 for other architectures like x86 and does not support
5657 IFUNC. Please make sure to use --enable-compat-libs only
5658 during a transitional period!
5660 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
5661 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5662 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
5663 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
5664 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
5665 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
5666 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
5667 Zbigniew Jędrzejewski-Szmek
5669 — Berlin, 2014-02-24
5673 * A new component "systemd-networkd" has been added that can
5674 be used to configure local network interfaces statically or
5675 via DHCP. It is capable of bringing up bridges, VLANs, and
5676 bonding. Currently, no hook-ups for interactive network
5677 configuration are provided. Use this for your initrd,
5678 container, embedded, or server setup if you need a simple,
5679 yet powerful, network configuration solution. This
5680 configuration subsystem is quite nifty, as it allows wildcard
5681 hotplug matching in interfaces. For example, with a single
5682 configuration snippet, you can configure that all Ethernet
5683 interfaces showing up are automatically added to a bridge,
5684 or similar. It supports link-sensing and more.
5686 * A new tool "systemd-socket-proxyd" has been added which can
5687 act as a bidirectional proxy for TCP sockets. This is
5688 useful for adding socket activation support to services that
5689 do not actually support socket activation, including virtual
5690 machines and the like.
5692 * Add a new tool to save/restore rfkill state on
5695 * Save/restore state of keyboard backlights in addition to
5696 display backlights on shutdown/boot.
5698 * udev learned a new SECLABEL{} construct to label device
5699 nodes with a specific security label when they appear. For
5700 now, only SECLABEL{selinux} is supported, but the syntax is
5701 prepared for additional security frameworks.
5703 * udev gained a new scheme to configure link-level attributes
5704 from files in /etc/systemd/network/*.link. These files can
5705 match against MAC address, device path, driver name and type,
5706 and will apply attributes like the naming policy, link speed,
5707 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
5708 address assignment policy (randomized, ...).
5710 * The configuration of network interface naming rules for
5711 "permanent interface names" has changed: a new NamePolicy=
5712 setting in the [Link] section of .link files determines the
5713 priority of possible naming schemes (onboard, slot, MAC,
5714 path). The default value of this setting is determined by
5715 /usr/lib/net/links/99-default.link. Old
5716 80-net-name-slot.rules udev configuration file has been
5717 removed, so local configuration overriding this file should
5718 be adapted to override 99-default.link instead.
5720 * When the User= switch is used in a unit file, also
5721 initialize $SHELL= based on the user database entry.
5723 * systemd no longer depends on libdbus. All communication is
5724 now done with sd-bus, systemd's low-level bus library
5727 * kdbus support has been added to PID 1 itself. When kdbus is
5728 enabled, this causes PID 1 to set up the system bus and
5729 enable support for a new ".busname" unit type that
5730 encapsulates bus name activation on kdbus. It works a little
5731 bit like ".socket" units, except for bus names. A new
5732 generator has been added that converts classic dbus1 service
5733 activation files automatically into native systemd .busname
5736 * sd-bus: add a light-weight vtable implementation that allows
5737 defining objects on the bus with a simple static const
5738 vtable array of its methods, signals and properties.
5740 * systemd will not generate or install static dbus
5741 introspection data anymore to /usr/share/dbus-1/interfaces,
5742 as the precise format of these files is unclear, and
5743 nothing makes use of it.
5745 * A proxy daemon is now provided to proxy clients connecting
5746 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
5747 compatibility with classic D-Bus.
5749 * A bus driver implementation has been added that supports the
5750 classic D-Bus bus driver calls on kdbus, also for
5751 compatibility purposes.
5753 * A new API "sd-event.h" has been added that implements a
5754 minimal event loop API built around epoll. It provides a
5755 couple of features that direct epoll usage is lacking:
5756 prioritization of events, scales to large numbers of timer
5757 events, per-event timer slack (accuracy), system-wide
5758 coalescing of timer events, exit handlers, watchdog
5759 supervision support using systemd's sd_notify() API, child
5762 * A new API "sd-rntl.h" has been added that provides an API
5763 around the route netlink interface of the kernel, similar in
5764 style to "sd-bus.h".
5766 * A new API "sd-dhcp-client.h" has been added that provides a
5767 small DHCPv4 client-side implementation. This is used by
5770 * There is a new kernel command line option
5771 "systemd.restore_state=0|1". When set to "0", none of the
5772 systemd tools will restore saved runtime state to hardware
5773 devices. More specifically, the rfkill and backlight states
5776 * The FsckPassNo= compatibility option in mount/service units
5777 has been removed. The fstab generator will now add the
5778 necessary dependencies automatically, and does not require
5779 PID1's support for that anymore.
5781 * journalctl gained a new switch, --list-boots, that lists
5782 recent boots with their times and boot IDs.
5784 * The various tools like systemctl, loginctl, timedatectl,
5785 busctl, systemd-run, ... have gained a new switch "-M" to
5786 connect to a specific, local OS container (as direct
5787 connection, without requiring SSH). This works on any
5788 container that is registered with machined, such as those
5789 created by libvirt-lxc or nspawn.
5791 * systemd-run and systemd-analyze also gained support for "-H"
5792 to connect to remote hosts via SSH. This is particularly
5793 useful for systemd-run because it enables queuing of jobs
5794 onto remote systems.
5796 * machinectl gained a new command "login" to open a getty
5797 login in any local container. This works with any container
5798 that is registered with machined (such as those created by
5799 libvirt-lxc or nspawn), and which runs systemd inside.
5801 * machinectl gained a new "reboot" command that may be used to
5802 trigger a reboot on a specific container that is registered
5803 with machined. This works on any container that runs an init
5804 system of some kind.
5806 * systemctl gained a new "list-timers" command to print a nice
5807 listing of installed timer units with the times they elapse
5810 * Alternative reboot() parameters may now be specified on the
5811 "systemctl reboot" command line and are passed to the
5812 reboot() system call.
5814 * systemctl gained a new --job-mode= switch to configure the
5815 mode to queue a job with. This is a more generic version of
5816 --fail, --irreversible, and --ignore-dependencies, which are
5817 still available but not advertised anymore.
5819 * /etc/systemd/system.conf gained new settings to configure
5820 various default timeouts of units, as well as the default
5821 start limit interval and burst. These may still be overridden
5824 * PID1 will now export on the bus profile data of the security
5825 policy upload process (such as the SELinux policy upload to
5828 * journald: when forwarding logs to the console, include
5829 timestamps (following the setting in
5830 /sys/module/printk/parameters/time).
5832 * OnCalendar= in timer units now understands the special
5833 strings "yearly" and "annually". (Both are equivalent)
5835 * The accuracy of timer units is now configurable with the new
5836 AccuracySec= setting. It defaults to 1min.
5838 * A new dependency type JoinsNamespaceOf= has been added that
5839 allows running two services within the same /tmp and network
5840 namespace, if PrivateNetwork= or PrivateTmp= are used.
5842 * A new command "cat" has been added to systemctl. It outputs
5843 the original unit file of a unit, and concatenates the
5844 contents of additional "drop-in" unit file snippets, so that
5845 the full configuration is shown.
5847 * systemctl now supports globbing on the various "list-xyz"
5848 commands, like "list-units" or "list-sockets", as well as on
5849 those commands which take multiple unit names.
5851 * journalctl's --unit= switch gained support for globbing.
5853 * All systemd daemons now make use of the watchdog logic so
5854 that systemd automatically notices when they hang.
5856 * If the $container_ttys environment variable is set,
5857 getty-generator will automatically spawn a getty for each
5858 listed tty. This is useful for container managers to request
5859 login gettys to be spawned on as many ttys as needed.
5861 * %h, %s, %U specifier support is not available anymore when
5862 used in unit files for PID 1. This is because NSS calls are
5863 not safe from PID 1. They stay available for --user
5864 instances of systemd, and as special case for the root user.
5866 * loginctl gained a new "--no-legend" switch to turn off output
5869 * The "sd-login.h" API gained three new calls:
5870 sd_session_is_remote(), sd_session_get_remote_user(),
5871 sd_session_get_remote_host() to query information about
5874 * The udev hardware database now also carries vendor/product
5875 information of SDIO devices.
5877 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
5878 determine whether watchdog notifications are requested by
5881 * Socket-activated per-connection services now include a
5882 short description of the connection parameters in the
5885 * tmpfiles gained a new "--boot" option. When this is not used,
5886 only lines where the command character is not suffixed with
5887 "!" are executed. When this option is specified, those
5888 options are executed too. This partitions tmpfiles
5889 directives into those that can be safely executed at any
5890 time, and those which should be run only at boot (for
5891 example, a line that creates /run/nologin).
5893 * A new API "sd-resolve.h" has been added which provides a simple
5894 asynchronous wrapper around glibc NSS host name resolution
5895 calls, such as getaddrinfo(). In contrast to glibc's
5896 getaddrinfo_a(), it does not use signals. In contrast to most
5897 other asynchronous name resolution libraries, this one does
5898 not reimplement DNS, but reuses NSS, so that alternate
5899 host name resolution systems continue to work, such as mDNS,
5900 LDAP, etc. This API is based on libasyncns, but it has been
5901 cleaned up for inclusion in systemd.
5903 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
5904 "sd-daemon.h" are no longer found in individual libraries
5905 libsystemd-journal.so, libsystemd-login.so,
5906 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
5907 merged them into a single library, libsystemd.so, which
5908 provides all symbols. The reason for this is cyclic
5909 dependencies, as these libraries tend to use each other's
5910 symbols. So far, we have managed to workaround that by linking
5911 a copy of a good part of our code into each of these
5912 libraries again and again, which, however, makes certain
5913 things hard to do, like sharing static variables. Also, it
5914 substantially increases footprint. With this change, there
5915 is only one library for the basic APIs systemd
5916 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
5917 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
5918 library as well, however are subject to the --enable-kdbus
5919 switch (see below). Note that "sd-dhcp-client.h" is not part
5920 of this library (this is because it only consumes, never
5921 provides, services of/to other APIs). To make the transition
5922 easy from the separate libraries to the unified one, we
5923 provide the --enable-compat-libs compile-time switch which
5924 will generate stub libraries that are compatible with the
5925 old ones but redirect all calls to the new one.
5927 * All of the kdbus logic and the new APIs "sd-bus.h",
5928 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
5929 and "sd-utf8.h" are compile-time optional via the
5930 "--enable-kdbus" switch, and they are not compiled in by
5931 default. To make use of kdbus, you have to explicitly enable
5932 the switch. Note however, that neither the kernel nor the
5933 userspace API for all of this is considered stable yet. We
5934 want to maintain the freedom to still change the APIs for
5935 now. By specifying this build-time switch, you acknowledge
5936 that you are aware of the instability of the current
5939 * Also, note that while kdbus is pretty much complete,
5940 it lacks one thing: proper policy support. This means you
5941 can build a fully working system with all features; however,
5942 it will be highly insecure. Policy support will be added in
5943 one of the next releases, at the same time that we will
5944 declare the APIs stable.
5946 * When the kernel command line argument "kdbus" is specified,
5947 systemd will automatically load the kdbus.ko kernel module. At
5948 this stage of development, it is only useful for testing kdbus
5949 and should not be used in production. Note: if "--enable-kdbus"
5950 is specified, and the kdbus.ko kernel module is available, and
5951 "kdbus" is added to the kernel command line, the entire system
5952 runs with kdbus instead of dbus-daemon, with the above mentioned
5953 problem of missing the system policy enforcement. Also a future
5954 version of kdbus.ko or a newer systemd will not be compatible with
5955 each other, and will unlikely be able to boot the machine if only
5956 one of them is updated.
5958 * systemctl gained a new "import-environment" command which
5959 uploads the caller's environment (or parts thereof) into the
5960 service manager so that it is inherited by services started
5961 by the manager. This is useful to upload variables like
5962 $DISPLAY into the user service manager.
5964 * A new PrivateDevices= switch has been added to service units
5965 which allows running a service with a namespaced /dev
5966 directory that does not contain any device nodes for
5967 physical devices. More specifically, it only includes devices
5968 such as /dev/null, /dev/urandom, and /dev/zero which are API
5971 * logind has been extended to support behaviour like VT
5972 switching on seats that do not support a VT. This makes
5973 multi-session available on seats that are not the first seat
5974 (seat0), and on systems where kernel support for VTs has
5975 been disabled at compile-time.
5977 * If a process holds a delay lock for system sleep or shutdown
5978 and fails to release it in time, we will now log its
5979 identity. This makes it easier to identify processes that
5980 cause slow suspends or power-offs.
5982 * When parsing /etc/crypttab, support for a new key-slot=
5983 option as supported by Debian is added. It allows indicating
5984 which LUKS slot to use on disk, speeding up key loading.
5986 * The sd_journal_sendv() API call has been checked and
5987 officially declared to be async-signal-safe so that it may
5988 be invoked from signal handlers for logging purposes.
5990 * Boot-time status output is now enabled automatically after a
5991 short timeout if boot does not progress, in order to give
5992 the user an indication what she or he is waiting for.
5994 * The boot-time output has been improved to show how much time
5995 remains until jobs expire.
5997 * The KillMode= switch in service units gained a new possible
5998 value "mixed". If set, and the unit is shut down, then the
5999 initial SIGTERM signal is sent only to the main daemon
6000 process, while the following SIGKILL signal is sent to
6001 all remaining processes of the service.
6003 * When a scope unit is registered, a new property "Controller"
6004 may be set. If set to a valid bus name, systemd will send a
6005 RequestStop() signal to this name when it would like to shut
6006 down the scope. This may be used to hook manager logic into
6007 the shutdown logic of scope units. Also, scope units may now
6008 be put in a special "abandoned" state, in which case the
6009 manager process which created them takes no further
6010 responsibilities for it.
6012 * When reading unit files, systemd will now verify
6013 the access mode of these files, and warn about certain
6014 suspicious combinations. This has been added to make it
6015 easier to track down packaging bugs where unit files are
6016 marked executable or world-writable.
6018 * systemd-nspawn gained a new "--setenv=" switch to set
6019 container-wide environment variables. The similar option in
6020 systemd-activate was renamed from "--environment=" to
6021 "--setenv=" for consistency.
6023 * systemd-nspawn has been updated to create a new kdbus domain
6024 for each container that is invoked, thus allowing each
6025 container to have its own set of system and user buses,
6026 independent of the host.
6028 * systemd-nspawn gained a new --drop-capability= switch to run
6029 the container with less capabilities than the default. Both
6030 --drop-capability= and --capability= now take the special
6031 string "all" for dropping or keeping all capabilities.
6033 * systemd-nspawn gained new switches for executing containers
6034 with specific SELinux labels set.
6036 * systemd-nspawn gained a new --quiet switch to not generate
6037 any additional output but the container's own console
6040 * systemd-nspawn gained a new --share-system switch to run a
6041 container without PID namespacing enabled.
6043 * systemd-nspawn gained a new --register= switch to control
6044 whether the container is registered with systemd-machined or
6045 not. This is useful for containers that do not run full
6046 OS images, but only specific apps.
6048 * systemd-nspawn gained a new --keep-unit which may be used
6049 when invoked as the only program from a service unit, and
6050 results in registration of the unit service itself in
6051 systemd-machined, instead of a newly opened scope unit.
6053 * systemd-nspawn gained a new --network-interface= switch for
6054 moving arbitrary interfaces to the container. The new
6055 --network-veth switch creates a virtual Ethernet connection
6056 between host and container. The new --network-bridge=
6057 switch then allows assigning the host side of this virtual
6058 Ethernet connection to a bridge device.
6060 * systemd-nspawn gained a new --personality= switch for
6061 setting the kernel personality for the container. This is
6062 useful when running a 32-bit container on a 64-bit host. A
6063 similar option Personality= is now also available for service
6066 * logind will now also track a "Desktop" identifier for each
6067 session which encodes the desktop environment of it. This is
6068 useful for desktop environments that want to identify
6069 multiple running sessions of itself easily.
6071 * A new SELinuxContext= setting for service units has been
6072 added that allows setting a specific SELinux execution
6073 context for a service.
6075 * Most systemd client tools will now honour $SYSTEMD_LESS for
6076 settings of the "less" pager. By default, these tools will
6077 override $LESS to allow certain operations to work, such as
6078 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
6079 influence this logic.
6081 * systemd's "seccomp" hook-up has been changed to make use of
6082 the libseccomp library instead of using its own
6083 implementation. This has benefits for portability among
6086 * For usage together with SystemCallFilter=, a new
6087 SystemCallErrorNumber= setting has been introduced that
6088 allows configuration of a system error number to be returned
6089 on filtered system calls, instead of immediately killing the
6090 process. Also, SystemCallArchitectures= has been added to
6091 limit access to system calls of a particular architecture
6092 (in order to turn off support for unused secondary
6093 architectures). There is also a global
6094 SystemCallArchitectures= setting in system.conf now to turn
6095 off support for non-native system calls system-wide.
6097 * systemd requires a kernel with a working name_to_handle_at(),
6098 please see the kernel config requirements in the README file.
6100 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
6101 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
6102 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
6103 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
6104 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
6105 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
6106 Elia Pinto, Florian Weimer, George McCollister, Goffredo
6107 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
6108 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
6109 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
6110 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
6111 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
6112 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
6113 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
6114 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
6115 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
6116 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
6117 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
6118 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
6119 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
6120 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
6121 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
6122 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
6123 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
6125 — Berlin, 2014-02-20
6129 * logind has gained support for facilitating privileged input
6130 and drm device access for unprivileged clients. This work is
6131 useful to allow Wayland display servers (and similar
6132 programs, such as kmscon) to run under the user's ID and
6133 access input and drm devices which are normally
6134 protected. When this is used (and the kernel is new enough)
6135 logind will "mute" IO on the file descriptors passed to
6136 Wayland as long as it is in the background and "unmute" it
6137 if it returns into the foreground. This allows secure
6138 session switching without allowing background sessions to
6139 eavesdrop on input and display data. This also introduces
6140 session switching support if VT support is turned off in the
6141 kernel, and on seats that are not seat0.
6143 * A new kernel command line option luks.options= is understood
6144 now which allows specifying LUKS options for usage for LUKS
6145 encrypted partitions specified with luks.uuid=.
6147 * tmpfiles.d(5) snippets may now use specifier expansion in
6148 path names. More specifically %m, %b, %H, %v, are now
6149 replaced by the local machine id, boot id, hostname, and
6150 kernel version number.
6152 * A new tmpfiles.d(5) command "m" has been introduced which
6153 may be used to change the owner/group/access mode of a file
6154 or directory if it exists, but do nothing if it does not.
6156 * This release removes high-level support for the
6157 MemorySoftLimit= cgroup setting. The underlying kernel
6158 cgroup attribute memory.soft_limit= is currently badly
6159 designed and likely to be removed from the kernel API in its
6160 current form, hence we should not expose it for now.
6162 * The memory.use_hierarchy cgroup attribute is now enabled for
6163 all cgroups systemd creates in the memory cgroup
6164 hierarchy. This option is likely to be come the built-in
6165 default in the kernel anyway, and the non-hierarchical mode
6166 never made much sense in the intrinsically hierarchical
6169 * A new field _SYSTEMD_SLICE= is logged along with all journal
6170 messages containing the slice a message was generated
6171 from. This is useful to allow easy per-customer filtering of
6172 logs among other things.
6174 * systemd-journald will no longer adjust the group of journal
6175 files it creates to the "systemd-journal" group. Instead we
6176 rely on the journal directory to be owned by the
6177 "systemd-journal" group, and its setgid bit set, so that the
6178 kernel file system layer will automatically enforce that
6179 journal files inherit this group assignment. The reason for
6180 this change is that we cannot allow NSS look-ups from
6181 journald which would be necessary to resolve
6182 "systemd-journal" to a numeric GID, because this might
6183 create deadlocks if NSS involves synchronous queries to
6184 other daemons (such as nscd, or sssd) which in turn are
6185 logging clients of journald and might block on it, which
6186 would then dead lock. A tmpfiles.d(5) snippet included in
6187 systemd will make sure the setgid bit and group are
6188 properly set on the journal directory if it exists on every
6189 boot. However, we recommend adjusting it manually after
6190 upgrades too (or from RPM scriptlets), so that the change is
6191 not delayed until next reboot.
6193 * Backlight and random seed files in /var/lib/ have moved into
6194 the /var/lib/systemd/ directory, in order to centralize all
6195 systemd generated files in one directory.
6197 * Boot time performance measurements (as displayed by
6198 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
6199 performance information if that's available to determine how
6200 much time BIOS and boot loader initialization required. With
6201 a sufficiently new BIOS you hence no longer need to boot
6202 with Gummiboot to get access to such information.
6204 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
6205 Cristian Rodríguez, Dave Reisner, David Herrmann, David
6206 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
6207 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
6208 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
6209 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
6210 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6212 — Berlin, 2013-10-02
6216 * The Restart= option for services now understands a new
6217 on-watchdog setting, which will restart the service
6218 automatically if the service stops sending out watchdog keep
6219 alive messages (as configured with WatchdogSec=).
6221 * The getty generator (which is responsible for bringing up a
6222 getty on configured serial consoles) will no longer only
6223 start a getty on the primary kernel console but on all
6224 others, too. This makes the order in which console= is
6225 specified on the kernel command line less important.
6227 * libsystemd-logind gained a new sd_session_get_vt() call to
6228 retrieve the VT number of a session.
6230 * If the option "tries=0" is set for an entry of /etc/crypttab
6231 its passphrase is queried indefinitely instead of any
6232 maximum number of tries.
6234 * If a service with a configure PID file terminates its PID
6235 file will now be removed automatically if it still exists
6236 afterwards. This should put an end to stale PID files.
6238 * systemd-run will now also take relative binary path names
6239 for execution and no longer insists on absolute paths.
6241 * InaccessibleDirectories= and ReadOnlyDirectories= now take
6242 paths that are optionally prefixed with "-" to indicate that
6243 it should not be considered a failure if they do not exist.
6245 * journalctl -o (and similar commands) now understands a new
6246 output mode "short-precise", it is similar to "short" but
6247 shows timestamps with usec accuracy.
6249 * The option "discard" (as known from Debian) is now
6250 synonymous to "allow-discards" in /etc/crypttab. In fact,
6251 "discard" is preferred now (since it is easier to remember
6254 * Some licensing clean-ups were made, so that more code is now
6255 LGPL-2.1 licensed than before.
6257 * A minimal tool to save/restore the display backlight
6258 brightness across reboots has been added. It will store the
6259 backlight setting as late as possible at shutdown, and
6260 restore it as early as possible during reboot.
6262 * A logic to automatically discover and enable home and swap
6263 partitions on GPT disks has been added. With this in place
6264 /etc/fstab becomes optional for many setups as systemd can
6265 discover certain partitions located on the root disk
6266 automatically. Home partitions are recognized under their
6267 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
6268 partitions are recognized under their GPT type ID
6269 0657fd6da4ab43c484e50933c84b4f4f.
6271 * systemd will no longer pass any environment from the kernel
6272 or initrd to system services. If you want to set an
6273 environment for all services, do so via the kernel command
6274 line systemd.setenv= assignment.
6276 * The systemd-sysctl tool no longer natively reads the file
6277 /etc/sysctl.conf. If desired, the file should be symlinked
6278 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
6279 legacy support by a symlink rather than built-in code, it
6280 also makes the otherwise hidden order of application of the
6281 different files visible. (Note that this partly reverts to a
6282 pre-198 application order of sysctl knobs!)
6284 * The "systemctl set-log-level" and "systemctl dump" commands
6285 have been moved to systemd-analyze.
6287 * systemd-run learned the new --remain-after-exit switch,
6288 which causes the scope unit not to be cleaned up
6289 automatically after the process terminated.
6291 * tmpfiles learned a new --exclude-prefix= switch to exclude
6292 certain paths from operation.
6294 * journald will now automatically flush all messages to disk
6295 as soon as a message at the log level CRIT, ALERT or EMERG
6298 Contributions from: Andrew Cook, Brandon Philips, Christian
6299 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
6300 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
6301 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
6302 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
6303 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
6304 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
6305 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
6306 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
6307 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
6308 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
6309 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
6310 William Giokas, Zbigniew Jędrzejewski-Szmek
6312 — Berlin, 2013-09-13
6316 * The documentation has been updated to cover the various new
6317 concepts introduced with 205.
6319 * Unit files now understand the new %v specifier which
6320 resolves to the kernel version string as returned by "uname
6323 * systemctl now supports filtering the unit list output by
6324 load state, active state and sub state, using the new
6327 * "systemctl status" will now show the results of the
6328 condition checks (like ConditionPathExists= and similar) of
6329 the last start attempts of the unit. They are also logged to
6332 * "journalctl -b" may now be used to look for boot output of a
6333 specific boot. Try "journalctl -b -1" for the previous boot,
6334 but the syntax is substantially more powerful.
6336 * "journalctl --show-cursor" has been added which prints the
6337 cursor string the last shown log line. This may then be used
6338 with the new "journalctl --after-cursor=" switch to continue
6339 browsing logs from that point on.
6341 * "journalctl --force" may now be used to force regeneration
6344 * Creation of "dead" device nodes has been moved from udev
6345 into kmod and tmpfiles. Previously, udev would read the kmod
6346 databases to pre-generate dead device nodes based on meta
6347 information contained in kernel modules, so that these would
6348 be auto-loaded on access rather then at boot. As this
6349 does not really have much to do with the exposing actual
6350 kernel devices to userspace this has always been slightly
6351 alien in the udev codebase. Following the new scheme kmod
6352 will now generate a runtime snippet for tmpfiles from the
6353 module meta information and it now is tmpfiles' job to the
6354 create the nodes. This also allows overriding access and
6355 other parameters for the nodes using the usual tmpfiles
6356 facilities. As side effect this allows us to remove the
6357 CAP_SYS_MKNOD capability bit from udevd entirely.
6359 * logind's device ACLs may now be applied to these "dead"
6360 devices nodes too, thus finally allowing managed access to
6361 devices such as /dev/snd/sequencer without loading the
6362 backing module right-away.
6364 * A new RPM macro has been added that may be used to apply
6365 tmpfiles configuration during package installation.
6367 * systemd-detect-virt and ConditionVirtualization= now can
6368 detect User-Mode-Linux machines (UML).
6370 * journald will now implicitly log the effective capabilities
6371 set of processes in the message metadata.
6373 * systemd-cryptsetup has gained support for TrueCrypt volumes.
6375 * The initrd interface has been simplified (more specifically,
6376 support for passing performance data via environment
6377 variables and fsck results via files in /run has been
6378 removed). These features were non-essential, and are
6379 nowadays available in a much nicer way by having systemd in
6380 the initrd serialize its state and have the hosts systemd
6381 deserialize it again.
6383 * The udev "keymap" data files and tools to apply keyboard
6384 specific mappings of scan to key codes, and force-release
6385 scan code lists have been entirely replaced by a udev
6386 "keyboard" builtin and a hwdb data file.
6388 * systemd will now honour the kernel's "quiet" command line
6389 argument also during late shutdown, resulting in a
6390 completely silent shutdown when used.
6392 * There's now an option to control the SO_REUSEPORT socket
6393 option in .socket units.
6395 * Instance units will now automatically get a per-template
6396 subslice of system.slice unless something else is explicitly
6397 configured. For example, instances of sshd@.service will now
6398 implicitly be placed in system-sshd.slice rather than
6399 system.slice as before.
6401 * Test coverage support may now be enabled at build time.
6403 Contributions from: Dave Reisner, Frederic Crozat, Harald
6404 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
6405 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
6406 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
6407 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
6408 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
6409 Giokas, Zbigniew Jędrzejewski-Szmek
6411 — Berlin, 2013-07-23
6415 * Two new unit types have been introduced:
6417 Scope units are very similar to service units, however, are
6418 created out of pre-existing processes — instead of PID 1
6419 forking off the processes. By using scope units it is
6420 possible for system services and applications to group their
6421 own child processes (worker processes) in a powerful way
6422 which then maybe used to organize them, or kill them
6423 together, or apply resource limits on them.
6425 Slice units may be used to partition system resources in an
6426 hierarchical fashion and then assign other units to them. By
6427 default there are now three slices: system.slice (for all
6428 system services), user.slice (for all user sessions),
6429 machine.slice (for VMs and containers).
6431 Slices and scopes have been introduced primarily in
6432 context of the work to move cgroup handling to a
6433 single-writer scheme, where only PID 1
6434 creates/removes/manages cgroups.
6436 * There's a new concept of "transient" units. In contrast to
6437 normal units these units are created via an API at runtime,
6438 not from configuration from disk. More specifically this
6439 means it is now possible to run arbitrary programs as
6440 independent services, with all execution parameters passed
6441 in via bus APIs rather than read from disk. Transient units
6442 make systemd substantially more dynamic then it ever was,
6443 and useful as a general batch manager.
6445 * logind has been updated to make use of scope and slice units
6446 for managing user sessions. As a user logs in he will get
6447 his own private slice unit, to which all sessions are added
6448 as scope units. We also added support for automatically
6449 adding an instance of user@.service for the user into the
6450 slice. Effectively logind will no longer create cgroup
6451 hierarchies on its own now, it will defer entirely to PID 1
6452 for this by means of scope, service and slice units. Since
6453 user sessions this way become entities managed by PID 1
6454 the output of "systemctl" is now a lot more comprehensive.
6456 * A new mini-daemon "systemd-machined" has been added which
6457 may be used by virtualization managers to register local
6458 VMs/containers. nspawn has been updated accordingly, and
6459 libvirt will be updated shortly. machined will collect a bit
6460 of meta information about the VMs/containers, and assign
6461 them their own scope unit (see above). The collected
6462 meta-data is then made available via the "machinectl" tool,
6463 and exposed in "ps" and similar tools. machined/machinectl
6464 is compile-time optional.
6466 * As discussed earlier, the low-level cgroup configuration
6467 options ControlGroup=, ControlGroupModify=,
6468 ControlGroupPersistent=, ControlGroupAttribute= have been
6469 removed. Please use high-level attribute settings instead as
6470 well as slice units.
6472 * A new bus call SetUnitProperties() has been added to alter
6473 various runtime parameters of a unit. This is primarily
6474 useful to alter cgroup parameters dynamically in a nice way,
6475 but will be extended later on to make more properties
6476 modifiable at runtime. systemctl gained a new set-properties
6477 command that wraps this call.
6479 * A new tool "systemd-run" has been added which can be used to
6480 run arbitrary command lines as transient services or scopes,
6481 while configuring a number of settings via the command
6482 line. This tool is currently very basic, however already
6483 very useful. We plan to extend this tool to even allow
6484 queuing of execution jobs with time triggers from the
6485 command line, similar in fashion to "at".
6487 * nspawn will now inform the user explicitly that kernels with
6488 audit enabled break containers, and suggest the user to turn
6491 * Support for detecting the IMA and AppArmor security
6492 frameworks with ConditionSecurity= has been added.
6494 * journalctl gained a new "-k" switch for showing only kernel
6495 messages, mimicking dmesg output; in addition to "--user"
6496 and "--system" switches for showing only user's own logs
6499 * systemd-delta can now show information about drop-in
6500 snippets extending unit files.
6502 * libsystemd-bus has been substantially updated but is still
6503 not available as public API.
6505 * systemd will now look for the "debug" argument on the kernel
6506 command line and enable debug logging, similar to what
6507 "systemd.log_level=debug" already did before.
6509 * "systemctl set-default", "systemctl get-default" has been
6510 added to configure the default.target symlink, which
6511 controls what to boot into by default.
6513 * "systemctl set-log-level" has been added as a convenient
6514 way to raise and lower systemd logging threshold.
6516 * "systemd-analyze plot" will now show the time the various
6517 generators needed for execution, as well as information
6518 about the unit file loading.
6520 * libsystemd-journal gained a new sd_journal_open_files() call
6521 for opening specific journal files. journactl also gained a
6522 new switch to expose this new functionality. Previously we
6523 only supported opening all files from a directory, or all
6524 files from the system, as opening individual files only is
6525 racy due to journal file rotation.
6527 * systemd gained the new DefaultEnvironment= setting in
6528 /etc/systemd/system.conf to set environment variables for
6531 * If a privileged process logs a journal message with the
6532 OBJECT_PID= field set, then journald will automatically
6533 augment this with additional OBJECT_UID=, OBJECT_GID=,
6534 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
6535 system services want to log events about specific client
6536 processes. journactl/systemctl has been updated to make use
6537 of this information if all log messages regarding a specific
6540 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
6541 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
6542 Reisner, David Coppa, David King, David Strauss, Eelco
6543 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
6544 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
6545 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
6546 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
6547 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
6548 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
6549 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
6550 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
6551 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
6552 Łukasz Stelmach, 장동준
6556 * The Python bindings gained some minimal support for the APIs
6557 exposed by libsystemd-logind.
6559 * ConditionSecurity= gained support for detecting SMACK. Since
6560 this condition already supports SELinux and AppArmor we only
6561 miss IMA for this. Patches welcome!
6563 Contributions from: Karol Lewandowski, Lennart Poettering,
6564 Zbigniew Jędrzejewski-Szmek
6568 * systemd-nspawn will now create /etc/resolv.conf if
6569 necessary, before bind-mounting the host's file onto it.
6571 * systemd-nspawn will now store meta information about a
6572 container on the container's cgroup as extended attribute
6573 fields, including the root directory.
6575 * The cgroup hierarchy has been reworked in many ways. All
6576 objects any of the components systemd creates in the cgroup
6577 tree are now suffixed. More specifically, user sessions are
6578 now placed in cgroups suffixed with ".session", users in
6579 cgroups suffixed with ".user", and nspawn containers in
6580 cgroups suffixed with ".nspawn". Furthermore, all cgroup
6581 names are now escaped in a simple scheme to avoid collision
6582 of userspace object names with kernel filenames. This work
6583 is preparation for making these objects relocatable in the
6584 cgroup tree, in order to allow easy resource partitioning of
6585 these objects without causing naming conflicts.
6587 * systemctl list-dependencies gained the new switches
6588 --plain, --reverse, --after and --before.
6590 * systemd-inhibit now shows the process name of processes that
6591 have taken an inhibitor lock.
6593 * nss-myhostname will now also resolve "localhost"
6594 implicitly. This makes /etc/hosts an optional file and
6595 nicely handles that on IPv6 ::1 maps to both "localhost" and
6598 * libsystemd-logind.so gained a new call
6599 sd_get_machine_names() to enumerate running containers and
6600 VMs (currently only supported by very new libvirt and
6601 nspawn). sd_login_monitor can now be used to watch
6602 VMs/containers coming and going.
6604 * .include is not allowed recursively anymore, and only in
6605 unit files. Usually it is better to use drop-in snippets in
6606 .d/*.conf anyway, as introduced with systemd 198.
6608 * systemd-analyze gained a new "critical-chain" command that
6609 determines the slowest chain of units run during system
6610 boot-up. It is very useful for tracking down where
6611 optimizing boot time is the most beneficial.
6613 * systemd will no longer allow manipulating service paths in
6614 the name=systemd:/system cgroup tree using ControlGroup= in
6615 units. (But is still fine with it in all other dirs.)
6617 * There's a new systemd-nspawn@.service service file that may
6618 be used to easily run nspawn containers as system
6619 services. With the container's root directory in
6620 /var/lib/container/foobar it is now sufficient to run
6621 "systemctl start systemd-nspawn@foobar.service" to boot it.
6623 * systemd-cgls gained a new parameter "--machine" to list only
6624 the processes within a certain container.
6626 * ConditionSecurity= now can check for "apparmor". We still
6627 are lacking checks for SMACK and IMA for this condition
6628 check though. Patches welcome!
6630 * A new configuration file /etc/systemd/sleep.conf has been
6631 added that may be used to configure which kernel operation
6632 systemd is supposed to execute when "suspend", "hibernate"
6633 or "hybrid-sleep" is requested. This makes the new kernel
6634 "freeze" state accessible to the user.
6636 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
6637 the passed argument if applicable.
6639 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
6640 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
6641 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
6642 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
6643 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
6644 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
6645 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
6650 * The output of 'systemctl list-jobs' got some polishing. The
6651 '--type=' argument may now be passed more than once. A new
6652 command 'systemctl list-sockets' has been added which shows
6653 a list of kernel sockets systemd is listening on with the
6654 socket units they belong to, plus the units these socket
6657 * The experimental libsystemd-bus library got substantial
6658 updates to work in conjunction with the (also experimental)
6659 kdbus kernel project. It works well enough to exchange
6660 messages with some sophistication. Note that kdbus is not
6661 ready yet, and the library is mostly an elaborate test case
6662 for now, and not installable.
6664 * systemd gained a new unit 'systemd-static-nodes.service'
6665 that generates static device nodes earlier during boot, and
6666 can run in conjunction with udev.
6668 * libsystemd-login gained a new call sd_pid_get_user_unit()
6669 to retrieve the user systemd unit a process is running
6670 in. This is useful for systems where systemd is used as
6673 * systemd-nspawn now places all containers in the new /machine
6674 top-level cgroup directory in the name=systemd
6675 hierarchy. libvirt will soon do the same, so that we get a
6676 uniform separation of /system, /user and /machine for system
6677 services, user processes and containers/virtual
6678 machines. This new cgroup hierarchy is also useful to stick
6679 stable names to specific container instances, which can be
6680 recognized later this way (this name may be controlled
6681 via systemd-nspawn's new -M switch). libsystemd-login also
6682 gained a new call sd_pid_get_machine_name() to retrieve the
6683 name of the container/VM a specific process belongs to.
6685 * bootchart can now store its data in the journal.
6687 * libsystemd-journal gained a new call
6688 sd_journal_add_conjunction() for AND expressions to the
6689 matching logic. This can be used to express more complex
6690 logical expressions.
6692 * journactl can now take multiple --unit= and --user-unit=
6695 * The cryptsetup logic now understands the "luks.key=" kernel
6696 command line switch for specifying a file to read the
6697 decryption key from. Also, if a configured key file is not
6698 found the tool will now automatically fall back to prompting
6701 * Python systemd.journal module was updated to wrap recently
6702 added functions from libsystemd-journal. The interface was
6703 changed to bring the low level interface in s.j._Reader
6704 closer to the C API, and the high level interface in
6705 s.j.Reader was updated to wrap and convert all data about
6708 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
6709 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
6710 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
6711 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
6712 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
6713 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6717 * journalctl --update-catalog now understands a new --root=
6718 option to operate on catalogs found in a different root
6721 * During shutdown after systemd has terminated all running
6722 services a final killing loop kills all remaining left-over
6723 processes. We will now print the name of these processes
6724 when we send SIGKILL to them, since this usually indicates a
6727 * If /etc/crypttab refers to password files stored on
6728 configured mount points automatic dependencies will now be
6729 generated to ensure the specific mount is established first
6730 before the key file is attempted to be read.
6732 * 'systemctl status' will now show information about the
6733 network sockets a socket unit is listening on.
6735 * 'systemctl status' will also shown information about any
6736 drop-in configuration file for units. (Drop-In configuration
6737 files in this context are files such as
6738 /etc/systemd/systemd/foobar.service.d/*.conf)
6740 * systemd-cgtop now optionally shows summed up CPU times of
6741 cgroups. Press '%' while running cgtop to switch between
6742 percentage and absolute mode. This is useful to determine
6743 which cgroups use up the most CPU time over the entire
6744 runtime of the system. systemd-cgtop has also been updated
6745 to be 'pipeable' for processing with further shell tools.
6747 * 'hostnamectl set-hostname' will now allow setting of FQDN
6750 * The formatting and parsing of time span values has been
6751 changed. The parser now understands fractional expressions
6752 such as "5.5h". The formatter will now output fractional
6753 expressions for all time spans under 1min, i.e. "5.123456s"
6754 rather than "5s 123ms 456us". For time spans under 1s
6755 millisecond values are shown, for those under 1ms
6756 microsecond values are shown. This should greatly improve
6757 all time-related output of systemd.
6759 * libsystemd-login and libsystemd-journal gained new
6760 functions for querying the poll() events mask and poll()
6761 timeout value for integration into arbitrary event
6764 * localectl gained the ability to list available X11 keymaps
6765 (models, layouts, variants, options).
6767 * 'systemd-analyze dot' gained the ability to filter for
6768 specific units via shell-style globs, to create smaller,
6769 more useful graphs. I.e. it is now possible to create simple
6770 graphs of all the dependencies between only target units, or
6771 of all units that Avahi has dependencies with.
6773 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
6774 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
6775 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
6776 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
6777 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
6778 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
6779 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
6783 * The boot-time readahead implementation for rotating media
6784 will now read the read-ahead data in multiple passes which
6785 consist of all read requests made in equidistant time
6786 intervals. This means instead of strictly reading read-ahead
6787 data in its physical order on disk we now try to find a
6788 middle ground between physical and access time order.
6790 * /etc/os-release files gained a new BUILD_ID= field for usage
6791 on operating systems that provide continuous builds of OS
6794 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
6795 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
6796 William Douglas, Zbigniew Jędrzejewski-Szmek
6800 * systemd-python gained an API exposing libsystemd-daemon.
6802 * The SMACK setup logic gained support for uploading CIPSO
6805 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
6806 ReadOnlyDirectories= and InaccessibleDirectories= has
6807 changed. The private /tmp and /var/tmp directories are now
6808 shared by all processes of a service (which means
6809 ExecStartPre= may now leave data in /tmp that ExecStart= of
6810 the same service can still access). When a service is
6811 stopped its temporary directories are immediately deleted
6812 (normal clean-up with tmpfiles is still done in addition to
6815 * By default, systemd will now set a couple of sysctl
6816 variables in the kernel: the safe sysrq options are turned
6817 on, IP route verification is turned on, and source routing
6818 disabled. The recently added hardlink and softlink
6819 protection of the kernel is turned on. These settings should
6820 be reasonably safe, and good defaults for all new systems.
6822 * The predictable network naming logic may now be turned off
6823 with a new kernel command line switch: net.ifnames=0.
6825 * A new libsystemd-bus module has been added that implements a
6826 pretty complete D-Bus client library. For details see:
6828 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
6830 * journald will now explicitly flush the journal files to disk
6831 at the latest 5min after each write. The file will then also
6832 be marked offline until the next write. This should increase
6833 reliability in case of a crash. The synchronization delay
6834 can be configured via SyncIntervalSec= in journald.conf.
6836 * There's a new remote-fs-setup.target unit that can be used
6837 to pull in specific services when at least one remote file
6838 system is to be mounted.
6840 * There are new targets timers.target and paths.target as
6841 canonical targets to pull user timer and path units in
6842 from. This complements sockets.target with a similar
6843 purpose for socket units.
6845 * libudev gained a new call udev_device_set_attribute_value()
6846 to set sysfs attributes of a device.
6848 * The udev daemon now sets the default number of worker
6849 processes executed in parallel based on the number of available
6850 CPUs instead of the amount of available RAM. This is supposed
6851 to provide a more reliable default and limit a too aggressive
6852 parallelism for setups with 1000s of devices connected.
6854 Contributions from: Auke Kok, Colin Walters, Cristian
6855 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
6856 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
6857 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
6858 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
6859 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
6860 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
6861 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
6862 Zbigniew Jędrzejewski-Szmek
6866 * Configuration of unit files may now be extended via drop-in
6867 files without having to edit/override the unit files
6868 themselves. More specifically, if the administrator wants to
6869 change one value for a service file foobar.service he can
6870 now do so by dropping in a configuration snippet into
6871 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
6872 will load all these snippets and apply them on top of the
6873 main unit configuration file, possibly extending or
6874 overriding its settings. Using these drop-in snippets is
6875 generally nicer than the two earlier options for changing
6876 unit files locally: copying the files from
6877 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
6878 them there; or creating a new file in /etc/systemd/system/
6879 that incorporates the original one via ".include". Drop-in
6880 snippets into these .d/ directories can be placed in any
6881 directory systemd looks for units in, and the usual
6882 overriding semantics between /usr/lib, /etc and /run apply
6885 * Most unit file settings which take lists of items can now be
6886 reset by assigning the empty string to them. For example,
6887 normally, settings such as Environment=FOO=BAR append a new
6888 environment variable assignment to the environment block,
6889 each time they are used. By assigning Environment= the empty
6890 string the environment block can be reset to empty. This is
6891 particularly useful with the .d/*.conf drop-in snippets
6892 mentioned above, since this adds the ability to reset list
6893 settings from vendor unit files via these drop-ins.
6895 * systemctl gained a new "list-dependencies" command for
6896 listing the dependencies of a unit recursively.
6898 * Inhibitors are now honored and listed by "systemctl
6899 suspend", "systemctl poweroff" (and similar) too, not only
6900 GNOME. These commands will also list active sessions by
6903 * Resource limits (as exposed by the various control group
6904 controllers) can now be controlled dynamically at runtime
6905 for all units. More specifically, you can now use a command
6906 like "systemctl set-cgroup-attr foobar.service cpu.shares
6907 2000" to alter the CPU shares a specific service gets. These
6908 settings are stored persistently on disk, and thus allow the
6909 administrator to easily adjust the resource usage of
6910 services with a few simple commands. This dynamic resource
6911 management logic is also available to other programs via the
6912 bus. Almost any kernel cgroup attribute and controller is
6915 * systemd-vconsole-setup will now copy all font settings to
6916 all allocated VTs, where it previously applied them only to
6919 * libsystemd-login gained the new sd_session_get_tty() API
6922 * This release drops support for a few legacy or
6923 distribution-specific LSB facility names when parsing init
6924 scripts: $x-display-manager, $mail-transfer-agent,
6925 $mail-transport-agent, $mail-transfer-agent, $smtp,
6926 $null. Also, the mail-transfer-agent.target unit backing
6927 this has been removed. Distributions which want to retain
6928 compatibility with this should carry the burden for
6929 supporting this themselves and patch support for these back
6930 in, if they really need to. Also, the facilities $syslog and
6931 $local_fs are now ignored, since systemd does not support
6932 early-boot LSB init scripts anymore, and these facilities
6933 are implied anyway for normal services. syslog.target has
6936 * There are new bus calls on PID1's Manager object for
6937 cancelling jobs, and removing snapshot units. Previously,
6938 both calls were only available on the Job and Snapshot
6941 * systemd-journal-gatewayd gained SSL support.
6943 * The various "environment" files, such as /etc/locale.conf
6944 now support continuation lines with a backslash ("\") as
6945 last character in the line, similarly in style (but different)
6946 to how this is supported in shells.
6948 * For normal user processes the _SYSTEMD_USER_UNIT= field is
6949 now implicitly appended to every log entry logged. systemctl
6950 has been updated to filter by this field when operating on a
6951 user systemd instance.
6953 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
6954 CAP_AUDIT_CONTROL capabilities to the capabilities set for
6955 the container. This makes it easier to boot unmodified
6956 Fedora systems in a container, which however still requires
6957 audit=0 to be passed on the kernel command line. Auditing in
6958 kernel and userspace is unfortunately still too broken in
6959 context of containers, hence we recommend compiling it out
6960 of the kernel or using audit=0. Hopefully this will be fixed
6961 one day for good in the kernel.
6963 * nspawn gained the new --bind= and --bind-ro= parameters to
6964 bind mount specific directories from the host into the
6967 * nspawn will now mount its own devpts file system instance
6968 into the container, in order not to leak pty devices from
6969 the host into the container.
6971 * systemd will now read the firmware boot time performance
6972 information from the EFI variables, if the used boot loader
6973 supports this, and takes it into account for boot performance
6974 analysis via "systemd-analyze". This is currently supported
6975 only in conjunction with Gummiboot, but could be supported
6976 by other boot loaders too. For details see:
6978 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
6980 * A new generator has been added that automatically mounts the
6981 EFI System Partition (ESP) to /boot, if that directory
6982 exists, is empty, and no other file system has been
6983 configured to be mounted there.
6985 * logind will now send out PrepareForSleep(false) out
6986 unconditionally, after coming back from suspend. This may be
6987 used by applications as asynchronous notification for
6988 system resume events.
6990 * "systemctl unlock-sessions" has been added, that allows
6991 unlocking the screens of all user sessions at once, similar
6992 to how "systemctl lock-sessions" already locked all users
6993 sessions. This is backed by a new D-Bus call UnlockSessions().
6995 * "loginctl seat-status" will now show the master device of a
6996 seat. (i.e. the device of a seat that needs to be around for
6997 the seat to be considered available, usually the graphics
7000 * tmpfiles gained a new "X" line type, that allows
7001 configuration of files and directories (with wildcards) that
7002 shall be excluded from automatic cleanup ("aging").
7004 * udev default rules set the device node permissions now only
7005 at "add" events, and do not change them any longer with a
7006 later "change" event.
7008 * The log messages for lid events and power/sleep keypresses
7009 now carry a message ID.
7011 * We now have a substantially larger unit test suite, but this
7012 continues to be work in progress.
7014 * udevadm hwdb gained a new --root= parameter to change the
7015 root directory to operate relative to.
7017 * logind will now issue a background sync() request to the kernel
7018 early at shutdown, so that dirty buffers are flushed to disk early
7019 instead of at the last moment, in order to optimize shutdown
7022 * A new bootctl tool has been added that is an interface for
7023 certain boot loader operations. This is currently a preview
7024 and is likely to be extended into a small mechanism daemon
7025 like timedated, localed, hostnamed, and can be used by
7026 graphical UIs to enumerate available boot options, and
7027 request boot into firmware operations.
7029 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
7030 the rest of the package. It also has been updated to work
7031 correctly in initrds.
7033 * Policykit previously has been runtime optional, and is now
7034 also compile time optional via a configure switch.
7036 * systemd-analyze has been reimplemented in C. Also "systemctl
7037 dot" has moved into systemd-analyze.
7039 * "systemctl status" with no further parameters will now print
7040 the status of all active or failed units.
7042 * Operations such as "systemctl start" can now be executed
7043 with a new mode "--irreversible" which may be used to queue
7044 operations that cannot accidentally be reversed by a later
7045 job queuing. This is by default used to make shutdown
7046 requests more robust.
7048 * The Python API of systemd now gained a new module for
7049 reading journal files.
7051 * A new tool kernel-install has been added that can install
7052 kernel images according to the Boot Loader Specification:
7054 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
7056 * Boot time console output has been improved to provide
7057 animated boot time output for hanging jobs.
7059 * A new tool systemd-activate has been added which can be used
7060 to test socket activation with, directly from the command
7061 line. This should make it much easier to test and debug
7062 socket activation in daemons.
7064 * journalctl gained a new "--reverse" (or -r) option to show
7065 journal output in reverse order (i.e. newest line first).
7067 * journalctl gained a new "--pager-end" (or -e) option to jump
7068 to immediately jump to the end of the journal in the
7069 pager. This is only supported in conjunction with "less".
7071 * journalctl gained a new "--user-unit=" option, that works
7072 similarly to "--unit=" but filters for user units rather than
7075 * A number of unit files to ease adoption of systemd in
7076 initrds has been added. This moves some minimal logic from
7077 the various initrd implementations into systemd proper.
7079 * The journal files are now owned by a new group
7080 "systemd-journal", which exists specifically to allow access
7081 to the journal, and nothing else. Previously, we used the
7082 "adm" group for that, which however possibly covers more
7083 than just journal/log file access. This new group is now
7084 already used by systemd-journal-gatewayd to ensure this
7085 daemon gets access to the journal files and as little else
7086 as possible. Note that "make install" will also set FS ACLs
7087 up for /var/log/journal to give "adm" and "wheel" read
7088 access to it, in addition to "systemd-journal" which owns
7089 the journal files. We recommend that packaging scripts also
7090 add read access to "adm" + "wheel" to /var/log/journal, and
7091 all existing/future journal files. To normal users and
7092 administrators little changes, however packagers need to
7093 ensure to create the "systemd-journal" system group at
7094 package installation time.
7096 * The systemd-journal-gatewayd now runs as unprivileged user
7097 systemd-journal-gateway:systemd-journal-gateway. Packaging
7098 scripts need to create these system user/group at
7101 * timedated now exposes a new boolean property CanNTP that
7102 indicates whether a local NTP service is available or not.
7104 * systemd-detect-virt will now also detect xen PVs
7106 * The pstore file system is now mounted by default, if it is
7109 * In addition to the SELinux and IMA policies we will now also
7110 load SMACK policies at early boot.
7112 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
7113 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
7114 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
7115 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
7116 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
7117 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
7118 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
7119 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
7120 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
7121 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
7122 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
7123 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
7124 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
7125 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
7129 * Timer units now support calendar time events in addition to
7130 monotonic time events. That means you can now trigger a unit
7131 based on a calendar time specification such as "Thu,Fri
7132 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
7133 or fifth day of any month of the year 2013, given that it is
7134 a thursday or friday. This brings timer event support
7135 considerably closer to cron's capabilities. For details on
7136 the supported calendar time specification language see
7139 * udev now supports a number of different naming policies for
7140 network interfaces for predictable names, and a combination
7141 of these policies is now the default. Please see this wiki
7142 document for details:
7144 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
7146 * Auke Kok's bootchart implementation has been added to the
7147 systemd tree. It is an optional component that can graph the
7148 boot in quite some detail. It is one of the best bootchart
7149 implementations around and minimal in its code and
7152 * nss-myhostname has been integrated into the systemd source
7153 tree. nss-myhostname guarantees that the local hostname
7154 always stays resolvable via NSS. It has been a weak
7155 requirement of systemd-hostnamed since a long time, and
7156 since its code is actually trivial we decided to just
7157 include it in systemd's source tree. It can be turned off
7158 with a configure switch.
7160 * The read-ahead logic is now capable of properly detecting
7161 whether a btrfs file system is on SSD or rotating media, in
7162 order to optimize the read-ahead scheme. Previously, it was
7163 only capable of detecting this on traditional file systems
7166 * In udev, additional device properties are now read from the
7167 IAB in addition to the OUI database. Also, Bluetooth company
7168 identities are attached to the devices as well.
7170 * In service files %U may be used as specifier that is
7171 replaced by the configured user name of the service.
7173 * nspawn may now be invoked without a controlling TTY. This
7174 makes it suitable for invocation as its own service. This
7175 may be used to set up a simple containerized server system
7176 using only core OS tools.
7178 * systemd and nspawn can now accept socket file descriptors
7179 when they are started for socket activation. This enables
7180 implementation of socket activated nspawn
7181 containers. i.e. think about autospawning an entire OS image
7182 when the first SSH or HTTP connection is received. We expect
7183 that similar functionality will also be added to libvirt-lxc
7186 * journalctl will now suppress ANSI color codes when
7187 presenting log data.
7189 * systemctl will no longer show control group information for
7190 a unit if the control group is empty anyway.
7192 * logind can now automatically suspend/hibernate/shutdown the
7195 * /etc/machine-info and hostnamed now also expose the chassis
7196 type of the system. This can be used to determine whether
7197 the local system is a laptop, desktop, handset or
7198 tablet. This information may either be configured by the
7199 user/vendor or is automatically determined from ACPI and DMI
7200 information if possible.
7202 * A number of PolicyKit actions are now bound together with
7203 "imply" rules. This should simplify creating UIs because
7204 many actions will now authenticate similar ones as well.
7206 * Unit files learnt a new condition ConditionACPower= which
7207 may be used to conditionalize a unit depending on whether an
7208 AC power source is connected or not, of whether the system
7209 is running on battery power.
7211 * systemctl gained a new "is-failed" verb that may be used in
7212 shell scripts and suchlike to check whether a specific unit
7213 is in the "failed" state.
7215 * The EnvironmentFile= setting in unit files now supports file
7216 globbing, and can hence be used to easily read a number of
7217 environment files at once.
7219 * systemd will no longer detect and recognize specific
7220 distributions. All distribution-specific #ifdeffery has been
7221 removed, systemd is now fully generic and
7222 distribution-agnostic. Effectively, not too much is lost as
7223 a lot of the code is still accessible via explicit configure
7224 switches. However, support for some distribution specific
7225 legacy configuration file formats has been dropped. We
7226 recommend distributions to simply adopt the configuration
7227 files everybody else uses now and convert the old
7228 configuration from packaging scripts. Most distributions
7229 already did that. If that's not possible or desirable,
7230 distributions are welcome to forward port the specific
7231 pieces of code locally from the git history.
7233 * When logging a message about a unit systemd will now always
7234 log the unit name in the message meta data.
7236 * localectl will now also discover system locale data that is
7237 not stored in locale archives, but directly unpacked.
7239 * logind will no longer unconditionally use framebuffer
7240 devices as seat masters, i.e. as devices that are required
7241 to be existing before a seat is considered preset. Instead,
7242 it will now look for all devices that are tagged as
7243 "seat-master" in udev. By default, framebuffer devices will
7244 be marked as such, but depending on local systems, other
7245 devices might be marked as well. This may be used to
7246 integrate graphics cards using closed source drivers (such
7247 as NVidia ones) more nicely into logind. Note however, that
7248 we recommend using the open source NVidia drivers instead,
7249 and no udev rules for the closed-source drivers will be
7250 shipped from us upstream.
7252 Contributions from: Adam Williamson, Alessandro Crismani, Auke
7253 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
7254 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
7255 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
7256 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
7257 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
7258 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
7259 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
7260 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
7261 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
7262 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
7263 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
7268 * udev gained support for loading additional device properties
7269 from an indexed database that is keyed by vendor/product IDs
7270 and similar device identifiers. For the beginning this
7271 "hwdb" is populated with data from the well-known PCI and
7272 USB database, but also includes PNP, ACPI and OID data. In
7273 the longer run this indexed database shall grow into
7274 becoming the one central database for non-essential
7275 userspace device metadata. Previously, data from the PCI/USB
7276 database was only attached to select devices, since the
7277 lookup was a relatively expensive operation due to O(n) time
7278 complexity (with n being the number of entries in the
7279 database). Since this is now O(1), we decided to add in this
7280 data for all devices where this is available, by
7281 default. Note that the indexed database needs to be rebuilt
7282 when new data files are installed. To achieve this you need
7283 to update your packaging scripts to invoke "udevadm hwdb
7284 --update" after installation of hwdb data files. For
7285 RPM-based distributions we introduced the new
7286 %udev_hwdb_update macro for this purpose.
7288 * The Journal gained support for the "Message Catalog", an
7289 indexed database to link up additional information with
7290 journal entries. For further details please check:
7292 https://www.freedesktop.org/wiki/Software/systemd/catalog
7294 The indexed message catalog database also needs to be
7295 rebuilt after installation of message catalog files. Use
7296 "journalctl --update-catalog" for this. For RPM-based
7297 distributions we introduced the %journal_catalog_update
7298 macro for this purpose.
7300 * The Python Journal bindings gained support for the standard
7301 Python logging framework.
7303 * The Journal API gained new functions for checking whether
7304 the underlying file system of a journal file is capable of
7305 properly reporting file change notifications, or whether
7306 applications that want to reflect journal changes "live"
7307 need to recheck journal files continuously in appropriate
7310 * It is now possible to set the "age" field for tmpfiles
7311 entries to 0, indicating that files matching this entry
7312 shall always be removed when the directories are cleaned up.
7314 * coredumpctl gained a new "gdb" verb which invokes gdb
7315 right-away on the selected coredump.
7317 * There's now support for "hybrid sleep" on kernels that
7318 support this, in addition to "suspend" and "hibernate". Use
7319 "systemctl hybrid-sleep" to make use of this.
7321 * logind's HandleSuspendKey= setting (and related settings)
7322 now gained support for a new "lock" setting to simply
7323 request the screen lock on all local sessions, instead of
7324 actually executing a suspend or hibernation.
7326 * systemd will now mount the EFI variables file system by
7329 * Socket units now gained support for configuration of the
7330 SMACK security label.
7332 * timedatectl will now output the time of the last and next
7333 daylight saving change.
7335 * We dropped support for various legacy and distro-specific
7336 concepts, such as insserv, early-boot SysV services
7337 (i.e. those for non-standard runlevels such as 'b' or 'S')
7338 or ArchLinux /etc/rc.conf support. We recommend the
7339 distributions who still need support this to either continue
7340 to maintain the necessary patches downstream, or find a
7341 different solution. (Talk to us if you have questions!)
7343 * Various systemd components will now bypass PolicyKit checks
7344 for root and otherwise handle properly if PolicyKit is not
7345 found to be around. This should fix most issues for
7346 PolicyKit-less systems. Quite frankly this should have been
7347 this way since day one. It is absolutely our intention to
7348 make systemd work fine on PolicyKit-less systems, and we
7349 consider it a bug if something does not work as it should if
7350 PolicyKit is not around.
7352 * For embedded systems it is now possible to build udev and
7353 systemd without blkid and/or kmod support.
7355 * "systemctl switch-root" is now capable of switching root
7356 more than once. I.e. in addition to transitions from the
7357 initrd to the host OS it is now possible to transition to
7358 further OS images from the host. This is useful to implement
7359 offline updating tools.
7361 * Various other additions have been made to the RPM macros
7362 shipped with systemd. Use %udev_rules_update() after
7363 installing new udev rules files. %_udevhwdbdir,
7364 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
7365 %_sysctldir are now available which resolve to the right
7366 directories for packages to place various data files in.
7368 * journalctl gained the new --full switch (in addition to
7369 --all, to disable ellipsation for long messages.
7371 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
7372 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
7373 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
7374 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
7375 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
7376 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
7377 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
7378 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
7379 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
7383 * journalctl gained new --since= and --until= switches to
7384 filter by time. It also now supports nice filtering for
7385 units via --unit=/-u.
7387 * Type=oneshot services may use ExecReload= and do the
7390 * The journal daemon now supports time-based rotation and
7391 vacuuming, in addition to the usual disk-space based
7394 * The journal will now index the available field values for
7395 each field name. This enables clients to show pretty drop
7396 downs of available match values when filtering. The bash
7397 completion of journalctl has been updated
7398 accordingly. journalctl gained a new switch -F to list all
7399 values a certain field takes in the journal database.
7401 * More service events are now written as structured messages
7402 to the journal, and made recognizable via message IDs.
7404 * The timedated, localed and hostnamed mini-services which
7405 previously only provided support for changing time, locale
7406 and hostname settings from graphical DEs such as GNOME now
7407 also have a minimal (but very useful) text-based client
7408 utility each. This is probably the nicest way to changing
7409 these settings from the command line now, especially since
7410 it lists available options and is fully integrated with bash
7413 * There's now a new tool "systemd-coredumpctl" to list and
7414 extract coredumps from the journal.
7416 * We now install a README each in /var/log/ and
7417 /etc/rc.d/init.d explaining where the system logs and init
7418 scripts went. This hopefully should help folks who go to
7419 that dirs and look into the otherwise now empty void and
7420 scratch their heads.
7422 * When user-services are invoked (by systemd --user) the
7423 $MANAGERPID env var is set to the PID of systemd.
7425 * SIGRTMIN+24 when sent to a --user instance will now result
7426 in immediate termination of systemd.
7428 * gatewayd received numerous feature additions such as a
7429 "follow" mode, for live syncing and filtering.
7431 * browse.html now allows filtering and showing detailed
7432 information on specific entries. Keyboard navigation and
7433 mouse screen support has been added.
7435 * gatewayd/journalctl now supports HTML5/JSON
7436 Server-Sent-Events as output.
7438 * The SysV init script compatibility logic will now
7439 heuristically determine whether a script supports the
7440 "reload" verb, and only then make this available as
7443 * "systemctl status --follow" has been removed, use "journalctl
7446 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
7447 have been removed since they are hardly useful to be
7450 * And I'd like to take the opportunity to specifically mention
7451 Zbigniew for his great contributions. Zbigniew, you rock!
7453 Contributions from: Andrew Eikum, Christian Hesse, Colin
7454 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
7455 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
7456 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
7457 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
7458 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
7459 Jędrzejewski-Szmek, Сковорода Никита Андреевич
7463 * If /etc/vconsole.conf is non-existent or empty we will no
7464 longer load any console font or key map at boot by
7465 default. Instead the kernel defaults will be left
7466 intact. This is definitely the right thing to do, as no
7467 configuration should mean no configuration, and hard-coding
7468 font names that are different on all archs is probably a bad
7469 idea. Also, the kernel default key map and font should be
7470 good enough for most cases anyway, and mostly identical to
7471 the userspace fonts/key maps we previously overloaded them
7472 with. If distributions want to continue to default to a
7473 non-kernel font or key map they should ship a default
7474 /etc/vconsole.conf with the appropriate contents.
7476 Contributions from: Colin Walters, Daniel J Walsh, Dave
7477 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
7478 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
7482 * journalctl gained a new --cursor= switch to show entries
7483 starting from the specified location in the journal.
7485 * We now enforce a size limit on journal entry fields exported
7486 with "-o json" in journalctl. Fields larger than 4K will be
7487 assigned null. This can be turned off with --all.
7489 * An (optional) journal gateway daemon is now available as
7490 "systemd-journal-gatewayd.service". This service provides
7491 access to the journal via HTTP and JSON. This functionality
7492 will be used to implement live log synchronization in both
7493 pull and push modes, but has various other users too, such
7494 as easy log access for debugging of embedded devices. Right
7495 now it is already useful to retrieve the journal via HTTP:
7497 # systemctl start systemd-journal-gatewayd.service
7498 # wget http://localhost:19531/entries
7500 This will download the journal contents in a
7501 /var/log/messages compatible format. The same as JSON:
7503 # curl -H"Accept: application/json" http://localhost:19531/entries
7505 This service is also accessible via a web browser where a
7506 single static HTML5 app is served that uses the JSON logic
7507 to enable the user to do some basic browsing of the
7508 journal. This will be extended later on. Here's an example
7509 screenshot of this app in its current state:
7511 http://0pointer.de/public/journal-gatewayd
7513 Contributions from: Kay Sievers, Lennart Poettering, Robert
7514 Milasan, Tom Gundersen
7518 * The bash completion logic is now available for journalctl
7521 * We do not mount the "cpuset" controller anymore together with
7522 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
7523 started if no parameters are assigned to it. "cpuset" hence
7524 broke code that assumed it could create "cpu" groups and
7527 * journalctl -f will now subscribe to terminal size changes,
7528 and line break accordingly.
7530 Contributions from: Dave Reisner, Kay Sievers, Lennart
7531 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
7535 * nspawn will now create a symlink /etc/localtime in the
7536 container environment, copying the host's timezone
7537 setting. Previously this has been done via a bind mount, but
7538 since symlinks cannot be bind mounted this has now been
7539 changed to create/update the appropriate symlink.
7541 * journalctl -n's line number argument is now optional, and
7542 will default to 10 if omitted.
7544 * journald will now log the maximum size the journal files may
7545 take up on disk. This is particularly useful if the default
7546 built-in logic of determining this parameter from the file
7547 system size is used. Use "systemctl status
7548 systemd-journald.service" to see this information.
7550 * The multi-seat X wrapper tool has been stripped down. As X
7551 is now capable of enumerating graphics devices via udev in a
7552 seat-aware way the wrapper is not strictly necessary
7553 anymore. A stripped down temporary stop-gap is still shipped
7554 until the upstream display managers have been updated to
7555 fully support the new X logic. Expect this wrapper to be
7556 removed entirely in one of the next releases.
7558 * HandleSleepKey= in logind.conf has been split up into
7559 HandleSuspendKey= and HandleHibernateKey=. The old setting
7560 is not available anymore. X11 and the kernel are
7561 distinguishing between these keys and we should too. This
7562 also means the inhibition lock for these keys has been split
7565 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
7566 Poettering, Lukas Nykryn, Václav Pavlín
7570 * Whenever a unit changes state we will now log this to the
7571 journal and show along the unit's own log output in
7574 * ConditionPathIsMountPoint= can now properly detect bind
7575 mount points too. (Previously, a bind mount of one file
7576 system to another place in the same file system could not be
7577 detected as mount, since they shared struct stat's st_dev
7580 * We will now mount the cgroup controllers cpu, cpuacct,
7581 cpuset and the controllers net_cls, net_prio together by
7584 * nspawn containers will now have a virtualized boot
7585 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
7586 over with a randomized ID at container initialization). This
7587 has the effect of making "journalctl -b" do the right thing
7590 * The JSON output journal serialization has been updated not
7591 to generate "endless" list objects anymore, but rather one
7592 JSON object per line. This is more in line how most JSON
7593 parsers expect JSON objects. The new output mode
7594 "json-pretty" has been added to provide similar output, but
7595 neatly aligned for readability by humans.
7597 * We dropped all explicit sync() invocations in the shutdown
7598 code. The kernel does this implicitly anyway in the kernel
7599 reboot() syscall. halt(8)'s -n option is now a compatibility
7602 * We now support virtualized reboot() in containers, as
7603 supported by newer kernels. We will fall back to exit() if
7604 CAP_SYS_REBOOT is not available to the container. Also,
7605 nspawn makes use of this now and will actually reboot the
7606 container if the containerized OS asks for that.
7608 * journalctl will only show local log output by default
7609 now. Use --merge (-m) to show remote log output, too.
7611 * libsystemd-journal gained the new sd_journal_get_usage()
7612 call to determine the current disk usage of all journal
7613 files. This is exposed in the new "journalctl --disk-usage"
7616 * journald gained a new configuration setting SplitMode= in
7617 journald.conf which may be used to control how user journals
7618 are split off. See journald.conf(5) for details.
7620 * A new condition type ConditionFileNotEmpty= has been added.
7622 * tmpfiles' "w" lines now support file globbing, to write
7623 multiple files at once.
7625 * We added Python bindings for the journal submission
7626 APIs. More Python APIs for a number of selected APIs will
7627 likely follow. Note that we intend to add native bindings
7628 only for the Python language, as we consider it common
7629 enough to deserve bindings shipped within systemd. There are
7630 various projects outside of systemd that provide bindings
7631 for languages such as PHP or Lua.
7633 * Many conditions will now resolve specifiers such as %i. In
7634 addition, PathChanged= and related directives of .path units
7635 now support specifiers as well.
7637 * There's now a new RPM macro definition for the system preset
7640 * journald will now warn if it ca not forward a message to the
7641 syslog daemon because its socket is full.
7643 * timedated will no longer write or process /etc/timezone,
7644 except on Debian. As we do not support late mounted /usr
7645 anymore /etc/localtime always being a symlink is now safe,
7646 and hence the information in /etc/timezone is not necessary
7649 * logind will now always reserve one VT for a text getty (VT6
7650 by default). Previously if more than 6 X sessions where
7651 started they took up all the VTs with auto-spawned gettys,
7652 so that no text gettys were available anymore.
7654 * udev will now automatically inform the btrfs kernel logic
7655 about btrfs RAID components showing up. This should make
7656 simple hotplug based btrfs RAID assembly work.
7658 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
7659 (but not for its children which will stay at the kernel
7660 default). This should allow setups with a lot more listening
7663 * systemd will now always pass the configured timezone to the
7664 kernel at boot. timedated will do the same when the timezone
7667 * logind's inhibition logic has been updated. By default,
7668 logind will now handle the lid switch, the power and sleep
7669 keys all the time, even in graphical sessions. If DEs want
7670 to handle these events on their own they should take the new
7671 handle-power-key, handle-sleep-key and handle-lid-switch
7672 inhibitors during their runtime. A simple way to achieve
7673 that is to invoke the DE wrapped in an invocation of:
7675 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
7677 * Access to unit operations is now checked via SELinux taking
7678 the unit file label and client process label into account.
7680 * systemd will now notify the administrator in the journal
7681 when he over-mounts a non-empty directory.
7683 * There are new specifiers that are resolved in unit files,
7684 for the host name (%H), the machine ID (%m) and the boot ID
7687 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
7688 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
7689 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
7690 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
7691 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
7692 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
7693 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
7697 * Support for reading structured kernel messages from
7698 /dev/kmsg has now been added and is enabled by default.
7700 * Support for reading kernel messages from /proc/kmsg has now
7701 been removed. If you want kernel messages in the journal
7702 make sure to run a recent kernel (>= 3.5) that supports
7703 reading structured messages from /dev/kmsg (see
7704 above). /proc/kmsg is now exclusive property of classic
7705 syslog daemons again.
7707 * The libudev API gained the new
7708 udev_device_new_from_device_id() call.
7710 * The logic for file system namespace (ReadOnlyDirectory=,
7711 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
7712 require pivot_root() anymore. This means fewer temporary
7713 directories are created below /tmp for this feature.
7715 * nspawn containers will now see and receive all submounts
7716 made on the host OS below the root file system of the
7719 * Forward Secure Sealing is now supported for Journal files,
7720 which provide cryptographical sealing of journal files so
7721 that attackers cannot alter log history anymore without this
7722 being detectable. Lennart will soon post a blog story about
7723 this explaining it in more detail.
7725 * There are two new service settings RestartPreventExitStatus=
7726 and SuccessExitStatus= which allow configuration of exit
7727 status (exit code or signal) which will be excepted from the
7728 restart logic, resp. consider successful.
7730 * journalctl gained the new --verify switch that can be used
7731 to check the integrity of the structure of journal files and
7732 (if Forward Secure Sealing is enabled) the contents of
7735 * nspawn containers will now be run with /dev/stdin, /dev/fd/
7736 and similar symlinks pre-created. This makes running shells
7737 as container init process a lot more fun.
7739 * The fstab support can now handle PARTUUID= and PARTLABEL=
7742 * A new ConditionHost= condition has been added to match
7743 against the hostname (with globs) and machine ID. This is
7744 useful for clusters where a single OS image is used to
7745 provision a large number of hosts which shall run slightly
7746 different sets of services.
7748 * Services which hit the restart limit will now be placed in a
7751 Contributions from: Bertram Poettering, Dave Reisner, Huang
7752 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
7753 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
7757 * When running in --user mode systemd will now become a
7758 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
7759 tree a lot more organized.
7761 * A new PartOf= unit dependency type has been introduced that
7762 may be used to group services in a natural way.
7764 * "systemctl enable" may now be used to enable instances of
7767 * journalctl now prints error log levels in red, and
7768 warning/notice log levels in bright white. It also supports
7769 filtering by log level now.
7771 * cgtop gained a new -n switch (similar to top), to configure
7772 the maximum number of iterations to run for. It also gained
7773 -b, to run in batch mode (accepting no input).
7775 * The suffix ".service" may now be omitted on most systemctl
7776 command lines involving service unit names.
7778 * There's a new bus call in logind to lock all sessions, as
7779 well as a loginctl verb for it "lock-sessions".
7781 * libsystemd-logind.so gained a new call sd_journal_perror()
7782 that works similar to libc perror() but logs to the journal
7783 and encodes structured information about the error number.
7785 * /etc/crypttab entries now understand the new keyfile-size=
7788 * shutdown(8) now can send a (configurable) wall message when
7789 a shutdown is cancelled.
7791 * The mount propagation mode for the root file system will now
7792 default to "shared", which is useful to make containers work
7793 nicely out-of-the-box so that they receive new mounts from
7794 the host. This can be undone locally by running "mount
7795 --make-rprivate /" if needed.
7797 * The prefdm.service file has been removed. Distributions
7798 should maintain this unit downstream if they intend to keep
7799 it around. However, we recommend writing normal unit files
7800 for display managers instead.
7802 * Since systemd is a crucial part of the OS we will now
7803 default to a number of compiler switches that improve
7804 security (hardening) such as read-only relocations, stack
7805 protection, and suchlike.
7807 * The TimeoutSec= setting for services is now split into
7808 TimeoutStartSec= and TimeoutStopSec= to allow configuration
7809 of individual time outs for the start and the stop phase of
7812 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
7813 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
7814 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
7815 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
7816 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
7817 Gundersen, Zbigniew Jędrzejewski-Szmek
7821 * The journal and id128 C APIs are now fully documented as man
7824 * Extra safety checks have been added when transitioning from
7825 the initial RAM disk to the main system to avoid accidental
7828 * /etc/crypttab entries now understand the new keyfile-offset=
7831 * systemctl -t can now be used to filter by unit load state.
7833 * The journal C API gained the new sd_journal_wait() call to
7834 make writing synchronous journal clients easier.
7836 * journalctl gained the new -D switch to show journals from a
7839 * journalctl now displays a special marker between log
7840 messages of two different boots.
7842 * The journal is now explicitly flushed to /var via a service
7843 systemd-journal-flush.service, rather than implicitly simply
7844 by seeing /var/log/journal to be writable.
7846 * journalctl (and the journal C APIs) can now match for much
7847 more complex expressions, with alternatives and
7850 * When transitioning from the initial RAM disk to the main
7851 system we will now kill all processes in a killing spree to
7852 ensure no processes stay around by accident.
7854 * Three new specifiers may be used in unit files: %u, %h, %s
7855 resolve to the user name, user home directory resp. user
7856 shell. This is useful for running systemd user instances.
7858 * We now automatically rotate journal files if their data
7859 object hash table gets a fill level > 75%. We also size the
7860 hash table based on the configured maximum file size. This
7861 together should lower hash collisions drastically and thus
7862 speed things up a bit.
7864 * journalctl gained the new "--header" switch to introspect
7865 header data of journal files.
7867 * A new setting SystemCallFilters= has been added to services
7868 which may be used to apply blacklists or whitelists to
7869 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
7871 * nspawn gained a new --link-journal= switch (and quicker: -j)
7872 to link the container journal with the host. This makes it
7873 very easy to centralize log viewing on the host for all
7874 guests while still keeping the journal files separated.
7876 * Many bugfixes and optimizations
7878 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
7879 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
7880 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
7885 * Several tools now understand kernel command line arguments,
7886 which are only read when run in an initial RAM disk. They
7887 usually follow closely their normal counterparts, but are
7890 * There's a new tool to analyze the readahead files that are
7891 automatically generated at boot. Use:
7893 /usr/lib/systemd/systemd-readahead analyze /.readahead
7895 * We now provide an early debug shell on tty9 if this enabled. Use:
7897 systemctl enable debug-shell.service
7899 * All plymouth related units have been moved into the Plymouth
7900 package. Please make sure to upgrade your Plymouth version
7903 * systemd-tmpfiles now supports getting passed the basename of
7904 a configuration file only, in which case it will look for it
7905 in all appropriate directories automatically.
7907 * udevadm info now takes a /dev or /sys path as argument, and
7908 does the right thing. Example:
7910 udevadm info /dev/sda
7911 udevadm info /sys/class/block/sda
7913 * systemctl now prints a warning if a unit is stopped but a
7914 unit that might trigger it continues to run. Example: a
7915 service is stopped but the socket that activates it is left
7918 * "systemctl status" will now mention if the log output was
7919 shortened due to rotation since a service has been started.
7921 * The journal API now exposes functions to determine the
7922 "cutoff" times due to rotation.
7924 * journald now understands SIGUSR1 and SIGUSR2 for triggering
7925 immediately flushing of runtime logs to /var if possible,
7926 resp. for triggering immediate rotation of the journal
7929 * It is now considered an error if a service is attempted to
7930 be stopped that is not loaded.
7932 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
7934 * systemd-analyze now supports Python 3
7936 * tmpfiles now supports cleaning up directories via aging
7937 where the first level dirs are always kept around but
7938 directories beneath it automatically aged. This is enabled
7939 by prefixing the age field with '~'.
7941 * Seat objects now expose CanGraphical, CanTTY properties
7942 which is required to deal with very fast bootups where the
7943 display manager might be running before the graphics drivers
7944 completed initialization.
7946 * Seat objects now expose a State property.
7948 * We now include RPM macros for service enabling/disabling
7949 based on the preset logic. We recommend RPM based
7950 distributions to make use of these macros if possible. This
7951 makes it simpler to reuse RPM spec files across
7954 * We now make sure that the collected systemd unit name is
7955 always valid when services log to the journal via
7958 * There's a new man page kernel-command-line(7) detailing all
7959 command line options we understand.
7961 * The fstab generator may now be disabled at boot by passing
7962 fstab=0 on the kernel command line.
7964 * A new kernel command line option modules-load= is now understood
7965 to load a specific kernel module statically, early at boot.
7967 * Unit names specified on the systemctl command line are now
7968 automatically escaped as needed. Also, if file system or
7969 device paths are specified they are automatically turned
7970 into the appropriate mount or device unit names. Example:
7972 systemctl status /home
7973 systemctl status /dev/sda
7975 * The SysVConsole= configuration option has been removed from
7976 system.conf parsing.
7978 * The SysV search path is no longer exported on the D-Bus
7981 * The Names= option has been removed from unit file parsing.
7983 * There's a new man page bootup(7) detailing the boot process.
7985 * Every unit and every generator we ship with systemd now
7986 comes with full documentation. The self-explanatory boot is
7989 * A couple of services gained "systemd-" prefixes in their
7990 name if they wrap systemd code, rather than only external
7991 code. Among them fsck@.service which is now
7992 systemd-fsck@.service.
7994 * The HaveWatchdog property has been removed from the D-Bus
7997 * systemd.confirm_spawn= on the kernel command line should now
8000 * There's a new man page crypttab(5) which details all options
8001 we actually understand.
8003 * systemd-nspawn gained a new --capability= switch to pass
8004 additional capabilities to the container.
8006 * timedated will now read known NTP implementation unit names
8007 from /usr/lib/systemd/ntp-units.d/*.list,
8008 systemd-timedated-ntp.target has been removed.
8010 * journalctl gained a new switch "-b" that lists log data of
8011 the current boot only.
8013 * The notify socket is in the abstract namespace again, in
8014 order to support daemons which chroot() at start-up.
8016 * There is a new Storage= configuration option for journald
8017 which allows configuration of where log data should go. This
8018 also provides a way to disable journal logging entirely, so
8019 that data collected is only forwarded to the console, the
8020 kernel log buffer or another syslog implementation.
8022 * Many bugfixes and optimizations
8024 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
8025 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
8026 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
8027 Shawn Landden, Tom Gundersen
8031 * "systemctl help <unit>" now shows the man page if one is
8034 * Several new man pages have been added.
8036 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
8037 MaxLevelConsole= can now be specified in
8038 journald.conf. These options allow reducing the amount of
8039 data stored on disk or forwarded by the log level.
8041 * TimerSlackNSec= can now be specified in system.conf for
8042 PID1. This allows system-wide power savings.
8044 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
8045 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
8050 * logind is now capable of (optionally) handling power and
8051 sleep keys as well as the lid switch.
8053 * journalctl now understands the syntax "journalctl
8054 /usr/bin/avahi-daemon" to get all log output of a specific
8057 * CapabilityBoundingSet= in system.conf now also influences
8058 the capability bound set of usermode helpers of the kernel.
8060 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
8061 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
8062 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
8063 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
8067 * Note that we skipped 139 releases here in order to set the
8068 new version to something that is greater than both udev's
8069 and systemd's most recent version number.
8071 * udev: all udev sources are merged into the systemd source tree now.
8072 All future udev development will happen in the systemd tree. It
8073 is still fully supported to use the udev daemon and tools without
8074 systemd running, like in initramfs or other init systems. Building
8075 udev though, will require the *build* of the systemd tree, but
8076 udev can be properly *run* without systemd.
8078 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
8079 should be used to create dead device nodes as workarounds for broken
8082 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
8083 no longer supported. udev_monitor_new_from_netlink() needs to be
8084 used to subscribe to events.
8086 * udev: when udevd is started by systemd, processes which are left
8087 behind by forking them off of udev rules, are unconditionally cleaned
8088 up and killed now after the event handling has finished. Services or
8089 daemons must be started as systemd services. Services can be
8090 pulled-in by udev to get started, but they can no longer be directly
8091 forked by udev rules.
8093 * udev: the daemon binary is called systemd-udevd now and installed
8094 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
8095 to adapt to that, create symlink, or rename the binary after building
8098 * libudev no longer provides these symbols:
8099 udev_monitor_from_socket()
8100 udev_queue_get_failed_list_entry()
8101 udev_get_{dev,sys,run}_path()
8102 The versions number was bumped and symbol versioning introduced.
8104 * systemd-loginctl and systemd-journalctl have been renamed
8105 to loginctl and journalctl to match systemctl.
8107 * The config files: /etc/systemd/systemd-logind.conf and
8108 /etc/systemd/systemd-journald.conf have been renamed to
8109 logind.conf and journald.conf. Package updates should rename
8110 the files to the new names on upgrade.
8112 * For almost all files the license is now LGPL2.1+, changed
8113 from the previous GPL2.0+. Exceptions are some minor stuff
8114 of udev (which will be changed to LGPL2.1 eventually, too),
8115 and the MIT licensed sd-daemon.[ch] library that is suitable
8116 to be used as drop-in files.
8118 * systemd and logind now handle system sleep states, in
8119 particular suspending and hibernating.
8121 * logind now implements a sleep/shutdown/idle inhibiting logic
8122 suitable for a variety of uses. Soonishly Lennart will blog
8123 about this in more detail.
8125 * var-run.mount and var-lock.mount are no longer provided
8126 (which previously bind mounted these directories to their new
8127 places). Distributions which have not converted these
8128 directories to symlinks should consider stealing these files
8129 from git history and add them downstream.
8131 * We introduced the Documentation= field for units and added
8132 this to all our shipped units. This is useful to make it
8133 easier to explore the boot and the purpose of the various
8136 * All smaller setup units (such as
8137 systemd-vconsole-setup.service) now detect properly if they
8138 are run in a container and are skipped when
8139 appropriate. This guarantees an entirely noise-free boot in
8140 Linux container environments such as systemd-nspawn.
8142 * A framework for implementing offline system updates is now
8143 integrated, for details see:
8144 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
8146 * A new service type Type=idle is available now which helps us
8147 avoiding ugly interleaving of getty output and boot status
8150 * There's now a system-wide CapabilityBoundingSet= option to
8151 globally reduce the set of capabilities for the
8152 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
8153 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
8154 even CAP_NET_ADMIN system-wide for secure systems.
8156 * There are now system-wide DefaultLimitXXX= options to
8157 globally change the defaults of the various resource limits
8158 for all units started by PID 1.
8160 * Harald Hoyer's systemd test suite has been integrated into
8161 systemd which allows easy testing of systemd builds in qemu
8162 and nspawn. (This is really awesome! Ask us for details!)
8164 * The fstab parser is now implemented as generator, not inside
8167 * systemctl will now warn you if .mount units generated from
8168 /etc/fstab are out of date due to changes in fstab that
8169 have not been read by systemd yet.
8171 * systemd is now suitable for usage in initrds. Dracut has
8172 already been updated to make use of this. With this in place
8173 initrds get a slight bit faster but primarily are much
8174 easier to introspect and debug since "systemctl status" in
8175 the host system can be used to introspect initrd services,
8176 and the journal from the initrd is kept around too.
8178 * systemd-delta has been added, a tool to explore differences
8179 between user/admin configuration and vendor defaults.
8181 * PrivateTmp= now affects both /tmp and /var/tmp.
8183 * Boot time status messages are now much prettier and feature
8184 proper english language. Booting up systemd has never been
8187 * Read-ahead pack files now include the inode number of all
8188 files to pre-cache. When the inode changes the pre-caching
8189 is not attempted. This should be nicer to deal with updated
8190 packages which might result in changes of read-ahead
8193 * We now temporaritly lower the kernel's read_ahead_kb variable
8194 when collecting read-ahead data to ensure the kernel's
8195 built-in read-ahead does not add noise to our measurements
8196 of necessary blocks to pre-cache.
8198 * There's now RequiresMountsFor= to add automatic dependencies
8199 for all mounts necessary for a specific file system path.
8201 * MountAuto= and SwapAuto= have been removed from
8202 system.conf. Mounting file systems at boot has to take place
8205 * nspawn now learned a new switch --uuid= to set the machine
8206 ID on the command line.
8208 * nspawn now learned the -b switch to automatically search
8211 * vt102 is now the default TERM for serial TTYs, upgraded from
8214 * systemd-logind now works on VT-less systems.
8216 * The build tree has been reorganized. The individual
8217 components now have directories of their own.
8219 * A new condition type ConditionPathIsReadWrite= is now available.
8221 * nspawn learned the new -C switch to create cgroups for the
8222 container in other hierarchies.
8224 * We now have support for hardware watchdogs, configurable in
8227 * The scheduled shutdown logic now has a public API.
8229 * We now mount /tmp as tmpfs by default, but this can be
8230 masked and /etc/fstab can override it.
8232 * Since udisks does not make use of /media anymore we are not
8233 mounting a tmpfs on it anymore.
8235 * journalctl gained a new --local switch to only interleave
8236 locally generated journal files.
8238 * We can now load the IMA policy at boot automatically.
8240 * The GTK tools have been split off into a systemd-ui.
8242 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
8243 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
8244 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
8245 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
8246 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
8247 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
8248 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
8249 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
8250 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
8255 * This is mostly a bugfix release
8257 * Support optional initialization of the machine ID from the
8258 KVM or container configured UUID.
8260 * Support immediate reboots with "systemctl reboot -ff"
8262 * Show /etc/os-release data in systemd-analyze output
8264 * Many bugfixes for the journal, including endianness fixes and
8265 ensuring that disk space enforcement works
8267 * sd-login.h is C++ compatible again
8269 * Extend the /etc/os-release format on request of the Debian
8272 * We now refuse non-UTF8 strings used in various configuration
8273 and unit files. This is done to ensure we do not pass invalid
8274 data over D-Bus or expose it elsewhere.
8276 * Register Mimo USB Screens as suitable for automatic seat
8279 * Read SELinux client context from journal clients in a race
8282 * Reorder configuration file lookup order. /etc now always
8283 overrides /run in order to allow the administrator to always
8284 and unconditionally override vendor-supplied or
8285 automatically generated data.
8287 * The various user visible bits of the journal now have man
8288 pages. We still lack man pages for the journal API calls
8291 * We now ship all man pages in HTML format again in the
8294 Contributions from: Dave Reisner, Dirk Eibach, Frederic
8295 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
8296 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
8301 * This is mostly a bugfix release
8303 * systems lacking /etc/os-release are no longer supported.
8305 * Various functionality updates to libsystemd-login.so
8307 * Track class of PAM logins to distinguish greeters from
8310 Contributions from: Kay Sievers, Lennart Poettering, Michael
8315 * This is an important bugfix release for v41.
8317 * Building man pages is now optional which should be useful
8318 for those building systemd from git but unwilling to install
8321 * Watchdog support for supervising services is now usable. In
8322 a future release support for hardware watchdogs
8323 (i.e. /dev/watchdog) will be added building on this.
8325 * Service start rate limiting is now configurable and can be
8326 turned off per service. When a start rate limit is hit a
8327 reboot can automatically be triggered.
8329 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
8331 Contributions from: Benjamin Franzke, Bill Nottingham,
8332 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
8333 Schmidt, Michał Górny, Piotr Drąg
8337 * The systemd binary is installed /usr/lib/systemd/systemd now;
8338 An existing /sbin/init symlink needs to be adapted with the
8341 * The code that loads kernel modules has been ported to invoke
8342 libkmod directly, instead of modprobe. This means we do not
8343 support systems with module-init-tools anymore.
8345 * Watchdog support is now already useful, but still not
8348 * A new kernel command line option systemd.setenv= is
8349 understood to set system wide environment variables
8350 dynamically at boot.
8352 * We now limit the set of capabilities of systemd-journald.
8354 * We now set SIGPIPE to ignore by default, since it only is
8355 useful in shell pipelines, and has little use in general
8356 code. This can be disabled with IgnoreSIPIPE=no in unit
8359 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
8360 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
8365 * This is mostly a bugfix release
8367 * We now expose the reason why a service failed in the
8368 "Result" D-Bus property.
8370 * Rudimentary service watchdog support (will be completed over
8371 the next few releases.)
8373 * When systemd forks off in order execute some service we will
8374 now immediately changes its argv[0] to reflect which process
8375 it will execute. This is useful to minimize the time window
8376 with a generic argv[0], which makes bootcharts more useful
8378 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
8379 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
8380 Mike Kazantsev, Ray Strode
8384 * This is mostly a test release, but incorporates many
8387 * New systemd-cgtop tool to show control groups by their
8390 * Linking against libacl for ACLs is optional again. If
8391 disabled, support tracking device access for active logins
8392 goes becomes unavailable, and so does access to the user
8393 journals by the respective users.
8395 * If a group "adm" exists, journal files are automatically
8396 owned by them, thus allow members of this group full access
8397 to the system journal as well as all user journals.
8399 * The journal now stores the SELinux context of the logging
8400 client for all entries.
8402 * Add C++ inclusion guards to all public headers
8404 * New output mode "cat" in the journal to print only text
8405 messages, without any meta data like date or time.
8407 * Include tiny X server wrapper as a temporary stop-gap to
8408 teach XOrg udev display enumeration. This is used by display
8409 managers such as gdm, and will go away as soon as XOrg
8410 learned native udev hotplugging for display devices.
8412 * Add new systemd-cat tool for executing arbitrary programs
8413 with STDERR/STDOUT connected to the journal. Can also act as
8414 BSD logger replacement, and does so by default.
8416 * Optionally store all locally generated coredumps in the
8417 journal along with meta data.
8419 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
8420 writing short strings to files (for usage for /sys), and for
8421 creating symlinks, character and block device nodes.
8423 * New unit file option ControlGroupPersistent= to make cgroups
8424 persistent, following the mechanisms outlined in
8425 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
8427 * Support multiple local RTCs in a sane way
8429 * No longer monopolize IO when replaying readahead data on
8430 rotating disks, since we might starve non-file-system IO to
8431 death, since fanotify() will not see accesses done by blkid,
8434 * Do not show kernel threads in systemd-cgls anymore, unless
8435 requested with new -k switch.
8437 Contributions from: Dan Horák, Kay Sievers, Lennart
8438 Poettering, Michal Schmidt
8442 * This is mostly a test release, but incorporates many
8445 * The git repository moved to:
8446 git://anongit.freedesktop.org/systemd/systemd
8447 ssh://git.freedesktop.org/git/systemd/systemd
8449 * First release with the journal
8450 http://0pointer.de/blog/projects/the-journal.html
8452 * The journal replaces both systemd-kmsg-syslogd and
8453 systemd-stdout-bridge.
8455 * New sd_pid_get_unit() API call in libsystemd-logind
8457 * Many systemadm clean-ups
8459 * Introduce remote-fs-pre.target which is ordered before all
8460 remote mounts and may be used to start services before all
8463 * Added Mageia support
8465 * Add bash completion for systemd-loginctl
8467 * Actively monitor PID file creation for daemons which exit in
8468 the parent process before having finished writing the PID
8469 file in the daemon process. Daemons which do this need to be
8470 fixed (i.e. PID file creation must have finished before the
8471 parent exits), but we now react a bit more gracefully to them.
8473 * Add colourful boot output, mimicking the well-known output
8474 of existing distributions.
8476 * New option PassCredentials= for socket units, for
8477 compatibility with a recent kernel ABI breakage.
8479 * /etc/rc.local is now hooked in via a generator binary, and
8480 thus will no longer act as synchronization point during
8483 * systemctl list-unit-files now supports --root=.
8485 * systemd-tmpfiles now understands two new commands: z, Z for
8486 relabelling files according to the SELinux database. This is
8487 useful to apply SELinux labels to specific files in /sys,
8490 * Output of SysV services is now forwarded to both the console
8491 and the journal by default, not only just the console.
8493 * New man pages for all APIs from libsystemd-login.
8495 * The build tree got reorganized and the build system is a
8496 lot more modular allowing embedded setups to specifically
8497 select the components of systemd they are interested in.
8499 * Support for Linux systems lacking the kernel VT subsystem is
8502 * configure's --with-rootdir= got renamed to
8503 --with-rootprefix= to follow the naming used by udev and
8506 * Unless specified otherwise we will now install to /usr instead
8507 of /usr/local by default.
8509 * Processes with '@' in argv[0][0] are now excluded from the
8510 final shut-down killing spree, following the logic explained
8512 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
8514 * All processes remaining in a service cgroup when we enter
8515 the START or START_PRE states are now killed with
8516 SIGKILL. That means it is no longer possible to spawn
8517 background processes from ExecStart= lines (which was never
8518 supported anyway, and bad style).
8520 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
8521 reloading of units together.
8523 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
8524 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
8525 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
8526 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
8527 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek