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.
46 * sd-boot acquired new loader configuration settings to optionally turn
47 off Windows and MacOS boot partition discovery as well as
48 reboot-into-firmware menu items. It is also able to pick a better
49 screen resolution for HiDPI systems, and now provides loader
50 configuration settings to change the resolution explicitly.
52 * The elogind-resolve tool has been renamed to resolvectl (it also
53 remains available under the old name, for compatibility), and its
54 interface is now verb-based, similar in style to the other <xyz>ctl
55 tools, such as systemctl or loginctl.
57 * The resolvectl/elogind-resolve tool also provides 'resolveconf'
58 compatiblity. It may be symlinked under the 'resolveconf' name, in
59 which case it will take arguments and input compatible with the
60 Debian and FreeBSD resolvconf tool.
62 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
63 where the system initially suspends, and after a time-out resumes and
66 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
67 set the client will only send a DUID as client identifier.
69 * The nss-elogind glibc NSS module will now enumerate dynamic users and
70 groups in effect. Previously, it could resolve UIDs/GIDs to user
71 names/groups and vice versa, but did not support enumeration.
73 * journald's Compress= configuration setting now optionally accepts a
74 byte threshold value. All journal objects larger than this threshold
75 will be compressed, smaller ones will not. Previously this threshold
76 was not configurable and set to 512.
78 * A new system.conf setting NoNewPrivileges= is now available which may
79 be used to turn off acquisition of new privileges system-wide
80 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
81 for all its children). Note that turning this option on means setuid
82 binaries and file system capabilities lose their special powers.
83 While turning on this option is a big step towards a more secure
84 system, doing so is likely to break numerous pre-existing UNIX tools,
85 in particular su and sudo.
87 * A new service elogind-time-sync-wait.service has been added. If
88 enabled it will delay the time-sync.target unit at boot until time
89 synchronization has been recieved from the network. This
90 functionality is useful on systems lacking a local RTC or where it is
91 acceptable that the boot process shall be delayed by external network
94 * When hibernating, elogind will now inform the kernel of the image
95 write offset, on kernels new enough to support this. This means swap
96 files should work for hibernation now.
98 * When loading unit files, elogind will now look for drop-in unit files
99 extensions in additional places. Previously, for a unit file name
100 "foo-bar-baz.service" it would look for dropin files in
101 "foo-bar-baz.service.d/*.conf". Now, it will also look in
102 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
103 service name truncated after all inner dashes. This scheme allows
104 writing drop-ins easily that apply to a whole set of unit files at
105 once. It's particularly useful for mount and slice units (as their
106 naming is prefix based), but is also useful for service and other
107 units, for packages that install multiple unit files at once,
108 following a strict naming regime of beginning the unit file name with
109 the package's name. Two new specifiers are now supported in unit
110 files to match this: %j and %J are replaced by the part of the unit
111 name following the last dash.
113 * Unit files and other configuration files that support specifier
114 expansion now understand another two new specifiers: %T and %V will
115 resolve to /tmp and /var/tmp respectively, or whatever temporary
116 directory has been set for the calling user.
118 * The ExecStart= lines of unit files are no longer required to
119 reference absolute paths. If non-absolute paths are specified the
120 specified binary name is searched within the service manager's
121 built-in $PATH, which may be queried with 'elogind-path
122 search-binaries-default'. It's generally recommended to continue to
123 use absolute paths for all binaries specified in unit files.
125 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
126 support alternative debuggers, for example lldb. The old name
127 continues to be available however, for compatibility reasons. Use the
128 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
129 to pick an alternative debugger instead of the default gdb.
131 * systemctl and the other tools will now output escape sequences that
132 generate proper clickable hyperlinks in various terminal emulators
133 where useful (for example, in the "systemctl status" output you can
134 now click on the unit file name to quickly open it in the
135 editor/viewer of your choice). Note that not all terminal emulators
136 support this functionality yet, but many do. Unfortunately, the
137 "less" pager doesn't support this yet, hence this functionality is
138 currently automatically turned off when a pager is started (which
139 happens quite often due to auto-paging). We hope to remove this
140 limitation as soon as "less" learns these escape sequences. This new
141 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
142 environment variable. For details on these escape sequences see:
143 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
145 * networkd's .network files now support a new IPv6MTUBytes= option for
146 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
147 option in the [Route] section to configure the MTU to use for
148 specific routes. It also gained support for configuration of the DHCP
149 "UserClass" option through the new UserClass= setting. It gained
150 three new options in the new [CAN] section for configuring CAN
151 networks. The MULTICAST and ALLMULTI interface flags may now be
152 controlled explicitly with the new Multicast= and AllMulticast=
155 * networkd will now automatically make use of the kernel's route
156 expiration feature, if it is available.
158 * udevd's .link files now support setting the number of receive and
159 transmit channels, using the RxChannels=, TxChannels=,
160 OtherChannels=, CombinedChannels= settings.
162 * Support for UDPSegmentationOffload= has been removed, given its
163 limited support in hardware, and waning software support.
165 * networkd's .netdev files now support creating "netdevsim" interfaces.
167 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
168 to query the unit belonging to a specific kernel control group.
170 * elogind-analyze gained a new verb "cat-config", which may be used to
171 dump the contents of any configuration file, with all its matching
172 drop-in files added in, and honouring the usual search and masking
173 logic applied to elogind configuration files. For example use
174 "elogind-analyze cat-config elogind/system.conf" to get the complete
175 system configuration file of elogind how it would be loaded by PID 1
176 itself. Similar to this, various tools such as elogind-tmpfiles or
177 elogind-sysusers, gained a new option "--cat-config", which does the
178 corresponding operation for their own configuration settings. For
179 example, "elogind-tmpfiles --cat-config" will now output the full
180 list of tmpfiles.d/ lines in place.
182 * timedatectl gained two new verbs "timesync-status" (to show the
183 current NTP synchronization state of elogind-timesyncd) and
184 "show-timesync" (to show bus properties of elogind-timesyncd).
186 * elogind-timesyncd gained a bus interface on which it exposes details
189 * elogind-nspawn gained a new --rlimit= switch for setting initial
190 resource limits for the container payload. There's a new switch
191 --hostname= to explicitly override the container's hostname. A new
192 --no-new-privileges= switch may be used to control the
193 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
194 --oom-score-adjust= switch controls the OOM scoring adjustment value
195 for the payload. The new --cpu-affinity= switch controls the CPU
196 affinity of the container payload. The new --resolv-conf= switch
197 allows more detailed control of /etc/resolv.conf handling of the
198 container. Similarly, the new --timezone= switch allows more detailed
199 control of /etc/localtime handling of the container.
201 * elogind-detect-virt gained a new --list switch, which will print a
202 list of all currently known VM and container environments.
204 * Support for "Portable Services" has been added, see
205 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
206 experimental, but this is expected to change soon. Reflecting this
207 experimental state, the "portablectl" binary is not installed into
208 /usr/bin yet. The binary has to be called with the full path
209 /usr/lib/elogind/portablectl instead.
211 * journalctl's and systemctl's -o switch now knows a new log output
212 mode "with-unit". The output it generates is very similar to the
213 regular "short" mode, but displays the unit name instead of the
214 syslog tag for each log line. Also, the date is shown with timezone
215 information. This mode is probably more useful than the classic
216 "short" output mode for most purposes, except where pixel-perfect
217 compatibility with classic /var/log/messages formatting is required.
219 * A new --dump-bus-properties switch has been added to the elogind
220 binary, which may be used to dump all supported D-Bus properties.
222 * sd-bus gained a set of new calls:
223 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
224 enable/disable the "floating" state of a bus slot object,
225 i.e. whether the slot object pins the bus it is allocated for into
226 memory or if the bus slot object gets disconnected when the bus goes
227 away. sd_bus_open_with_description(),
228 sd_bus_open_user_with_description(),
229 sd_bus_open_system_with_description() may be used to allocate bus
230 objects and set their description string already during allocation.
232 * sd-event gained support for watching inotify events from the event
233 loop, in an efficient way, sharing inotify handles between multiple
234 users. For this a new function sd_event_add_inotify() has been added.
236 * sd-event and sd-bus gained support for calling special user-supplied
237 destructor functions for userdata pointers associated with
238 sd_event_source, sd_bus_slot and sd_bus_track objects.
240 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
242 * PID 1 will now automatically reschedule .timer units whenever the
243 local timezone changes. (They previously got rescheduled
244 automatically when the system clock changed.)
246 * New documentation has been added to document cgroups delegation,
247 portable services and the various code quality tools we have set up:
249 https://github.com/systemd/systemd/blob/master/doc/CGROUP_DELEGATION.md
250 https://github.com/systemd/systemd/blob/master/doc/PORTABLE_SERVICES.md
251 https://github.com/systemd/systemd/blob/master/doc/CODE_QUALITY.md
253 * pam_elogind will now look for PAM userdata fields elogind.memory_max,
254 elogind.tasks_max, elogind.cpu_weight, elogind.io_weight set by
255 earlier PAM modules. The data in these fields is used to initialize
256 the session scope's resource properties. Thus external PAM modules
257 may now configure per-session limits, for example sourced from
258 external user databases.
260 * socket units with Accept=yes will now maintain a "refused" counter in
261 addition to the existing "accepted" counter, counting connections
262 refused due to the enforced limits.
264 * The "elogind-path search-binaries-default" command may now be use to
265 query the default, built-in $PATH PID 1 will pass to the services it
268 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
269 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Antique, Arnaud
270 Rebillout, Brian J. Murrell, Bruno Vernay, Chris Lesiak, Christian
271 Brauner, Christian Hesse, Daniel Dao, Daniel Lin, Danylo Korostil,
272 Davide Cavalca, David Tardon, Dimitri John Ledkov, Dmitriy Geels,
273 Douglas Christman, Elia Geretto, emelenas, Evegeny Vereshchagin, Evgeny
274 Vereshchagin, Felipe Sateler, Feng Sun, Filipe Brandenburger, Franck
275 Bui, futpib, Giuseppe Scrivano, Guillem Jover, guixxx, Hans de Goede,
276 Henrique Dante de Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko,
277 Ivan Shapovalov, James Cowgill, Jan Janssen, Jan Synacek, Jared
278 Kazimir, João Paulo Rechi Vita, Joost Heitbrink, juergbi, Jui-Chi Ricky
279 Liang, Kai-Heng Feng, Karol Augustin, Krzysztof Nowicki, Lauri
280 Tirkkonen, Lennart Poettering, Leonard, Long Li, Luca Boccassi, Lucas
281 Werkmeister, Marcel Hoppe, Marc Kleine-Budde, Mario Limonciello, Martin
282 Jansa, Martin Wilck, Mathieu Malaterre, Matteo F. Vescovi, Michael
283 Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal Sekletar,
284 Mike Gilbert, Mikhail Kasimov, Milan, Milan Broz, mourikwa, Muhammet
285 Kara, Nicolas Boichat, Omer Katz, Paride Legovini, Paul Menzel, Paul
286 Milliken, Peter A. Bigot, Peter Hutterer, Peter Jones, Philip Sequeira,
287 Philip Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de
288 Araujo, Rosen Penev, rubensa, Ryan Gonzalez, Salvo 'LtWorf' Tomaselli,
289 Sebastian Reichel, Sergio Lindo Mansilla, Stefan Schweter, Stephen
290 Hemminger, Stuart Hayes, Susant Sahani, Sylvain Plantefève, Thomas
291 H. P. Andersen, Tobias Jungel, Tomasz Torcz, Vito Caputo, Will Dietz,
292 Will Thompson, xginn8, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
298 * The MemoryAccounting= unit property now defaults to on. After
299 discussions with the upstream control group maintainers we learnt
300 that the negative impact of cgroup memory accounting on current
301 kernels is finally relatively minimal, so that it should be safe to
302 enable this by default without affecting system performance. Besides
303 memory accounting only task accounting is turned on by default, all
304 other forms of resource accounting (CPU, IO, IP) remain off for now,
305 because it's not clear yet that their impact is small enough to move
306 from opt-in to opt-out. We recommend downstreams to leave memory
307 accounting on by default if kernel 4.14 or higher is primarily
308 used. On very resource constrained systems or when support for old
309 kernels is a necessity, -Dmemory-accounting-default=false can be used
310 to revert this change.
312 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
313 %udev_rules_update) and the journal catalog (%journal_catalog_update)
314 from the upgrade scriptlets of individual packages now do nothing.
315 Transfiletriggers have been added which will perform those updates
316 once at the end of the transaction.
318 Similar transfiletriggers have been added to execute any sysctl.d
319 and binfmt.d rules. Thus, it should be unnecessary to provide any
320 scriptlets to execute this configuration from package installation
323 * elogind-sysusers gained a mode where the configuration to execute is
324 specified on the command line, but this configuration is not executed
325 directly, but instead it is merged with the configuration on disk,
326 and the result is executed. This is useful for package installation
327 scripts which want to create the user before installing any files on
328 disk (in case some of those files are owned by that user), while
329 still allowing local admin overrides.
331 This functionality is exposed to rpm scriptlets through a new
332 %sysusers_create_package macro. Old %sysusers_create and
333 %sysusers_create_inline macros are deprecated.
335 A transfiletrigger for sysusers.d configuration is now installed,
336 which means that it should be unnecessary to call elogind-sysusers from
337 package installation scripts, unless the package installs any files
338 owned by those newly-created users, in which case
339 %sysusers_create_package should be used.
341 * Analogous change has been done for elogind-tmpfiles: it gained a mode
342 where the command-line configuration is merged with the configuration
343 on disk. This is exposed as the new %tmpfiles_create_package macro,
344 and %tmpfiles_create is deprecated. A transfiletrigger is installed
345 for tmpfiles.d, hence it should be unnecessary to call elogind-tmpfiles
346 from package installation scripts.
348 * sysusers.d configuration for a user may now also specify the group
349 number, in addition to the user number ("u username 123:456"), or
350 without the user number ("u username -:456").
352 * Configution items for elogind-sysusers can now be specified as
353 positional arguments when the new --inline switch is used.
355 * The login shell of users created through sysusers.d may now be
356 specified (previously, it was always /bin/sh for root and
357 /sbin/nologin for other users).
359 * elogind-analyze gained a new --global switch to look at global user
360 configuration. It also gained a unit-paths verb to list the unit load
361 paths that are compiled into elogind (which can be used with
362 --elogind, --user, or --global).
364 * udevadm trigger gained a new --settle/-w option to wait for any
365 triggered events to finish (but just those, and not any other events
366 which are triggered meanwhile).
368 * The action that elogind takes when the lid is closed and the
369 machine is connected to external power can now be configured using
370 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
371 was determined by HandleLidSwitch=, and, for backwards compatibility,
372 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
374 * journalctl will periodically call sd_journal_process() to make it
375 resilient against inotify queue overruns when journal files are
376 rotated very quickly.
378 * Two new functions in libelogind — sd_bus_get_n_queued_read and
379 sd_bus_get_n_queued_write — may be used to check the number of
380 pending bus messages.
382 * elogind gained a new
383 org.freedesktop.elogind1.Manager.AttachProcessesToUnit dbus call
384 which can be used to migrate foreign processes to scope and service
385 units. The primary user for this new API is elogind itself: the
386 elogind --user instance uses this call of the elogind --system
387 instance to migrate processes if it itself gets the request to
388 migrate processes and the kernel refuses this due to access
389 restrictions. Thanks to this "elogind-run --scope --user …" works
390 again in pure cgroups v2 environments when invoked from the user
393 * A new TemporaryFileSystem= setting can be used to mask out part of
394 the real file system tree with tmpfs mounts. This may be combined
395 with BindPaths= and BindReadOnlyPaths= to hide files or directories
396 not relevant to the unit, while still allowing some paths lower in
397 the tree to be accessed.
399 ProtectHome=tmpfs may now be used to hide user home and runtime
400 directories from units, in a way that is mostly equivalent to
401 "TemporaryFileSystem=/home /run/user /root".
403 * Non-service units are now started with KeyringMode=shared by default.
404 This means that mount and swapon and other mount tools have access
405 to keys in the main keyring.
407 * /sys/fs/bpf is now mounted automatically.
409 * QNX virtualization is now detected by elogind-detect-virt and may
410 be used in ConditionVirtualization=.
412 * IPAccounting= may now be enabled also for slice units.
414 * A new -Dsplit-bin= build configuration switch may be used to specify
415 whether bin and sbin directories are merged, or if they should be
416 included separately in $PATH and various listings of executable
417 directories. The build configuration scripts will try to autodetect
418 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
419 system, but distributions are encouraged to configure this
422 * A new -Dok-color= build configuration switch may be used to change
423 the colour of "OK" status messages.
425 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
426 PrivateNetwork=yes was buggy in previous versions of elogind. This
427 means that after the upgrade and daemon-reexec, any such units must
430 * INCOMPATIBILITY: as announced in the NEWS for 237, elogind-tmpfiles
431 will not exclude read-only files owned by root from cleanup.
433 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
434 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
435 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
436 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
437 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
438 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
439 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
440 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
441 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
442 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
443 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
444 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
445 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
446 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
447 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
448 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
454 * Some keyboards come with a zoom see-saw or rocker which until now got
455 mapped to the Linux "zoomin/out" keys in hwdb. However, these
456 keycodes are not recognized by any major desktop. They now produce
457 Up/Down key events so that they can be used for scrolling.
459 * INCOMPATIBILITY: elogind-tmpfiles' "f" lines changed behaviour
460 slightly: previously, if an argument was specified for lines of this
461 type (i.e. the right-most column was set) this string was appended to
462 existing files each time elogind-tmpfiles was run. This behaviour was
463 different from what the documentation said, and not particularly
464 useful, as repeated elogind-tmpfiles invocations would not be
465 idempotent and grow such files without bounds. With this release
466 behaviour has been altered slightly, to match what the documentation
467 says: lines of this type only have an effect if the indicated files
468 don't exist yet, and only then the argument string is written to the
471 * FUTURE INCOMPATIBILITY: In elogind v238 we intend to slightly change
472 elogind-tmpfiles behaviour: previously, read-only files owned by root
473 were always excluded from the file "aging" algorithm (i.e. the
474 automatic clean-up of directories like /tmp based on
475 atime/mtime/ctime). We intend to drop this restriction, and age files
476 by default even when owned by root and read-only. This behaviour was
477 inherited from older tools, but there have been requests to remove
478 it, and it's not obvious why this restriction was made in the first
479 place. Please speak up now, if you are aware of software that reqires
480 this behaviour, otherwise we'll remove the restriction in v238.
482 * A new environment variable $SYSTEMD_OFFLINE is now understood by
483 systemctl. It takes a boolean argument. If on, systemctl assumes it
484 operates on an "offline" OS tree, and will not attempt to talk to the
485 service manager. Previously, this mode was implicitly enabled if a
486 chroot() environment was detected, and this new environment variable
487 now provides explicit control.
489 * .path and .socket units may now be created transiently, too.
490 Previously only service, mount, automount and timer units were
491 supported as transient units. The elogind-run tool has been updated
492 to expose this new functionality, you may hence use it now to bind
493 arbitrary commands to path or socket activation on-the-fly from the
494 command line. Moreover, almost all properties are now exposed for the
495 unit types that already supported transient operation.
497 * The elogind-mount command gained support for a new --owner= parameter
498 which takes a user name, which is then resolved and included in uid=
499 and gid= mount options string of the file system to mount.
501 * A new unit condition ConditionControlGroupController= has been added
502 that checks whether a specific cgroup controller is available.
504 * Unit files, udev's .link files, and elogind-networkd's .netdev and
505 .network files all gained support for a new condition
506 ConditionKernelVersion= for checking against specific kernel
509 * In elogind-networkd, the [IPVLAN] section in .netdev files gained
510 support for configuring device flags in the Flags= setting. In the
511 same files, the [Tunnel] section gained support for configuring
512 AllowLocalRemote=. The [Route] section in .network files gained
513 support for configuring InitialCongestionWindow=,
514 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
515 understands RapidCommit=.
517 * elogind-networkd's DHCPv6 support gained support for Prefix
520 * sd-bus gained support for a new "watch-bind" feature. When this
521 feature is enabled, an sd_bus connection may be set up to connect to
522 an AF_UNIX socket in the file system as soon as it is created. This
523 functionality is useful for writing early-boot services that
524 automatically connect to the system bus as soon as it is started,
525 without ugly time-based polling. elogind-networkd and
526 elogind-resolved have been updated to make use of this
527 functionality. busctl exposes this functionality in a new
528 --watch-bind= command line switch.
530 * sd-bus will now optionally synthesize a local "Connected" signal as
531 soon as a D-Bus connection is set up fully. This message mirrors the
532 already existing "Disconnected" signal which is synthesized when the
533 connection is terminated. This signal is generally useful but
534 particularly handy in combination with the "watch-bind" feature
535 described above. Synthesizing of this message has to be requested
536 explicitly through the new API call sd_bus_set_connected_signal(). In
537 addition a new call sd_bus_is_ready() has been added that checks
538 whether a connection is fully set up (i.e. between the "Connected" and
539 "Disconnected" signals).
541 * sd-bus gained two new calls sd_bus_request_name_async() and
542 sd_bus_release_name_async() for asynchronously registering bus
543 names. Similar, there is now sd_bus_add_match_async() for installing
544 a signal match asynchronously. All of elogind's own services have
545 been updated to make use of these calls. Doing these operations
546 asynchronously has two benefits: it reduces the risk of deadlocks in
547 case of cyclic dependencies between bus services, and it speeds up
548 service initialization since synchronization points for bus
549 round-trips are removed.
551 * sd-bus gained two new calls sd_bus_match_signal() and
552 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
553 and sd_bus_add_match_async() but instead of taking a D-Bus match
554 string take match fields as normal function parameters.
556 * sd-bus gained two new calls sd_bus_set_sender() and
557 sd_bus_message_set_sender() for setting the sender name of outgoing
558 messages (either for all outgoing messages or for just one specific
559 one). These calls are only useful in direct connections as on
560 brokered connections the broker fills in the sender anyway,
561 overwriting whatever the client filled in.
563 * sd-event gained a new pseudo-handle that may be specified on all API
564 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
565 used this refers to the default event loop object of the calling
566 thread. Note however that this does not implicitly allocate one —
567 which has to be done prior by using sd_event_default(). Similarly
568 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
569 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
570 to the default bus of the specified type of the calling thread. Here
571 too this does not implicitly allocate bus connection objects, this
572 has to be done prior with sd_bus_default() and friends.
574 * sd-event gained a new call pair
575 sd_event_source_{get|set}_io_fd_own(). This may be used to request
576 automatic closure of the file descriptor an IO event source watches
577 when the event source is destroyed.
579 * elogind-networkd gained support for natively configuring WireGuard
582 * In previous versions elogind synthesized user records both for the
583 "nobody" (UID 65534) and "root" (UID 0) users in nss-elogind and
584 internally. In order to simplify distribution-wide renames of the
585 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
586 new transitional flag file has been added: if
587 /etc/elogind/dont-synthesize-nobody exists synthesizing of the 65534
588 user and group record within the elogind codebase is disabled.
590 * elogind-notify gained a new --uid= option for selecting the source
591 user/UID to use for notification messages sent to the service
594 * journalctl gained a new --grep= option to list only entries in which
595 the message matches a certain pattern. By default matching is case
596 insensitive if the pattern is lowercase, and case sensitive
597 otherwise. Option --case-sensitive=yes|no can be used to override
598 this an specify case sensitivity or case insensitivity.
600 * There's now a "elogind-analyze service-watchdogs" command for printing
601 the current state of the service runtime watchdog, and optionally
602 enabling or disabling the per-service watchdogs system-wide if given a
603 boolean argument (i.e. the concept you configure in WatchdogSec=), for
604 debugging purposes. There's also a kernel command line option
605 elogind.service_watchdogs= for controlling the same.
607 * Two new "log-level" and "log-target" options for elogind-analyze were
608 addded that merge the now deprecated get-log-level, set-log-level and
609 get-log-target, set-log-target pairs. The deprecated options are still
610 understood for backwards compatibility. The two new options print the
611 current value when no arguments are given, and set them when a
612 level/target is given as an argument.
614 * sysusers.d's "u" lines now optionally accept both a UID and a GID
615 specification, separated by a ":" character, in order to create users
616 where UID and GID do not match.
618 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
619 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
620 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
621 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
622 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
623 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
624 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
625 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
626 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
627 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
628 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
629 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
630 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
631 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
632 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
633 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
634 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
635 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
636 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
637 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
644 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
645 in v235 has been extended to also set the dummy.ko module option
646 numdummies=0, preventing the kernel from automatically creating
647 dummy0. All dummy interfaces must now be explicitly created.
649 * Unknown '%' specifiers in configuration files are now rejected. This
650 applies to units and tmpfiles.d configuration. Any percent characters
651 that are followed by a letter or digit that are not supposed to be
652 interpreted as the beginning of a specifier should be escaped by
653 doubling ("%%"). (So "size=5%" is still accepted, as well as
654 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
655 valid specifiers today.)
657 * systemd-resolved now maintains a new dynamic
658 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
659 recommended to make /etc/resolv.conf a symlink to it. This file
660 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
661 includes dynamically acquired search domains, achieving more correct
662 DNS resolution by software that bypasses local DNS APIs such as NSS.
664 * The "uaccess" udev tag has been dropped from /dev/kvm and
665 /dev/dri/renderD*. These devices now have the 0666 permissions by
666 default (but this may be changed at build-time). /dev/dri/renderD*
667 will now be owned by the "render" group along with /dev/kfd.
669 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
670 systemd-journal-gatewayd.service and
671 systemd-journal-upload.service. This means "nss-systemd" must be
672 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
673 services are resolved properly.
675 * In /etc/fstab two new mount options are now understood:
676 x-systemd.makefs and x-systemd.growfs. The former has the effect that
677 the configured file system is formatted before it is mounted, the
678 latter that the file system is resized to the full block device size
679 after it is mounted (i.e. if the file system is smaller than the
680 partition it resides on, it's grown). This is similar to the fsck
681 logic in /etc/fstab, and pulls in systemd-makefs@.service and
682 systemd-growfs@.service as necessary, similar to
683 systemd-fsck@.service. Resizing is currently only supported on ext4
686 * In systemd-networkd, the IPv6 RA logic now optionally may announce
687 DNS server and domain information.
689 * Support for the LUKS2 on-disk format for encrypted partitions has
690 been added. This requires libcryptsetup2 during compilation and
693 * The systemd --user instance will now signal "readiness" when its
694 basic.target unit has been reached, instead of when the run queue ran
695 empty for the first time.
697 * Tmpfiles.d with user configuration are now also supported.
698 systemd-tmpfiles gained a new --user switch, and snippets placed in
699 ~/.config/user-tmpfiles.d/ and corresponding directories will be
700 executed by systemd-tmpfiles --user running in the new
701 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
702 running in the user session.
704 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
705 %S resolves to the top-level state directory (/var/lib for the system
706 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
707 top-level cache directory (/var/cache for the system instance,
708 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
709 logs directory (/var/log for the system instance,
710 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
711 existing %t specifier, that resolves to the top-level runtime
712 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
715 * journalctl learnt a new parameter --output-fields= for limiting the
716 set of journal fields to output in verbose and JSON output modes.
718 * systemd-timesyncd's configuration file gained a new option
719 RootDistanceMaxSec= for setting the maximum root distance of servers
720 it'll use, as well as the new options PollIntervalMinSec= and
721 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
723 * bootctl gained a new command "list" for listing all available boot
724 menu items on systems that follow the boot loader specification.
726 * systemctl gained a new --dry-run switch that shows what would be done
727 instead of doing it, and is currently supported by the shutdown and
730 * ConditionSecurity= can now detect the TOMOYO security module.
732 * Unit file [Install] sections are now also respected in unit drop-in
733 files. This is intended to be used by drop-ins under /usr/lib/.
735 * systemd-firstboot may now also set the initial keyboard mapping.
737 * Udev "changed" events for devices which are exposed as systemd
738 .device units are now propagated to units specified in
739 ReloadPropagatedFrom= as reload requests.
741 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
742 unit template name (i.e. a name in the form of 'foobar@.service',
743 without the instance component between the '@' and - the '.'), then
744 the escaped sysfs path of the device is automatically used as the
747 * SystemCallFilter= in unit files has been extended so that an "errno"
748 can be specified individually for each system call. Example:
749 SystemCallFilter=~uname:EILSEQ.
751 * The cgroup delegation logic has been substantially updated. Delegate=
752 now optionally takes a list of controllers (instead of a boolean, as
753 before), which lists the controllers to delegate at least.
755 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
757 * A new LogLevelMax= setting configures the maximum log level any
758 process of the service may log at (i.e. anything with a lesser
759 priority than what is specified is automatically dropped). A new
760 LogExtraFields= setting allows configuration of additional journal
761 fields to attach to all log records generated by any of the unit's
764 * New StandardInputData= and StandardInputText= settings along with the
765 new option StandardInput=data may be used to configure textual or
766 binary data that shall be passed to the executed service process via
767 standard input, encoded in-line in the unit file.
769 * StandardInput=, StandardOutput= and StandardError= may now be used to
770 connect stdin/stdout/stderr of executed processes directly with a
771 file or AF_UNIX socket in the file system, using the new "file:" option.
773 * A new unit file option CollectMode= has been added, that allows
774 tweaking the garbage collection logic for units. It may be used to
775 tell systemd to garbage collect units that have failed automatically
776 (normally it only GCs units that exited successfully). systemd-run
777 and systemd-mount expose this new functionality with a new -G option.
779 * "machinectl bind" may now be used to bind mount non-directories
780 (i.e. regularfiles, devices, fifos, sockets).
782 * systemd-analyze gained a new verb "calendar" for validating and
783 testing calendar time specifications to use for OnCalendar= in timer
784 units. Besides validating the expression it will calculate the next
785 time the specified expression would elapse.
787 * In addition to the pre-existing FailureAction= unit file setting
788 there's now SuccessAction=, for configuring a shutdown action to
789 execute when a unit completes successfully. This is useful in
790 particular inside containers that shall terminate after some workload
791 has been completed. Also, both options are now supported for all unit
792 types, not just services.
794 * networkds's IP rule support gained two new options
795 IncomingInterface= and OutgoingInterface= for configuring the incoming
796 and outgoing interfaces of configured rules. systemd-networkd also
797 gained support for "vxcan" network devices.
799 * networkd gained a new setting RequiredForOnline=, taking a
800 boolean. If set, systemd-wait-online will take it into consideration
801 when determining that the system is up, otherwise it will ignore the
802 interface for this purpose.
804 * The sd_notify() protocol gained support for a new operation: with
805 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
806 store again, ahead of POLLHUP or POLLERR when they are removed
809 * A new document doc/UIDS-GIDS.md has been added to the source tree,
810 that documents the UID/GID range and assignment assumptions and
811 requirements of systemd.
813 * The watchdog device PID 1 will ping may now be configured through the
814 WatchdogDevice= configuration file setting, or by setting the
815 systemd.watchdog_service= kernel commandline option.
817 * systemd-resolved's gained support for registering DNS-SD services on
818 the local network using MulticastDNS. Services may either be
819 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
820 the same dir below /run, /usr/lib), or through its D-Bus API.
822 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
823 extend the effective start, runtime, and stop time. The service must
824 continue to send EXTEND_TIMEOUT_USEC within the period specified to
825 prevent the service manager from making the service as timedout.
827 * elogind-resolved's DNSSEC support gained support for RFC 8080
828 (Ed25519 keys and signatures).
830 * The elogind-resolve command line tool gained a new set of options
831 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
832 --set-nta= and --revert to configure per-interface DNS configuration
833 dynamically during runtime. It's useful for pushing DNS information
834 into elogind-resolved from DNS hook scripts that various interface
835 managing software supports (such as pppd).
837 * elogind-nspawn gained a new --network-namespace-path= command line
838 option, which may be used to make a container join an existing
839 network namespace, by specifying a path to a "netns" file.
841 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
842 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
843 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
844 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
845 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
846 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
847 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
848 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
849 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
850 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
851 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
852 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
853 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
854 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
855 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
856 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
857 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
858 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
859 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
860 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
861 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
862 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
863 Jędrzejewski-Szmek, Zeal Jagannatha
869 * INCOMPATIBILITY: systemd-logind.service and other long-running
870 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
871 communication with the outside. This generally improves security of
872 the system, and is in almost all cases a safe and good choice, as
873 these services do not and should not provide any network-facing
874 functionality. However, systemd-logind uses the glibc NSS API to
875 query the user database. This creates problems on systems where NSS
876 is set up to directly consult network services for user database
877 lookups. In particular, this creates incompatibilities with the
878 "nss-nis" module, which attempts to directly contact the NIS/YP
879 network servers it is configured for, and will now consistently
880 fail. In such cases, it is possible to turn off IP sandboxing for
881 systemd-logind.service (set IPAddressDeny= in its [Service] section
882 to the empty string, via a .d/ unit file drop-in). Downstream
883 distributions might want to update their nss-nis packaging to include
884 such a drop-in snippet, accordingly, to hide this incompatibility
885 from the user. Another option is to make use of glibc's nscd service
886 to proxy such network requests through a privilege-separated, minimal
887 local caching daemon, or to switch to more modern technologies such
888 sssd, whose NSS hook-ups generally do not involve direct network
889 access. In general, we think it's definitely time to question the
890 implementation choices of nss-nis, i.e. whether it's a good idea
891 today to embed a network-facing loadable module into all local
892 processes that need to query the user database, including the most
893 trivial and benign ones, such as "ls". For more details about
894 IPAddressDeny= see below.
896 * A new modprobe.d drop-in is now shipped by default that sets the
897 bonding module option max_bonds=0. This overrides the kernel default,
898 to avoid conflicts and ambiguity as to whether or not bond0 should be
899 managed by systemd-networkd or not. This resolves multiple issues
900 with bond0 properties not being applied, when bond0 is configured
901 with systemd-networkd. Distributors may choose to not package this,
902 however in that case users will be prevented from correctly managing
903 bond0 interface using systemd-networkd.
905 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
906 which print the logging level and target of the system manager. They
907 complement the existing "set-log-level" and "set-log-target" verbs
908 used to change those values.
910 * journald.conf gained a new boolean setting ReadKMsg= which defaults
911 to on. If turned off kernel log messages will not be read by
912 systemd-journald or included in the logs. It also gained a new
913 setting LineMax= for configuring the maximum line length in
914 STDOUT/STDERR log streams. The new default for this value is 48K, up
915 from the previous hardcoded 2048.
917 * A new unit setting RuntimeDirectoryPreserve= has been added, which
918 allows more detailed control of what to do with a runtime directory
919 configured with RuntimeDirectory= (i.e. a directory below /run or
920 $XDG_RUNTIME_DIR) after a unit is stopped.
922 * The RuntimeDirectory= setting for units gained support for creating
923 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
924 one top-level directory.
926 * Units gained new options StateDirectory=, CacheDirectory=,
927 LogsDirectory= and ConfigurationDirectory= which are closely related
928 to RuntimeDirectory= but manage per-service directories below
929 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
930 possible to write unit files which when activated automatically gain
931 properly owned service specific directories in these locations, thus
932 making unit files self-contained and increasing compatibility with
933 stateless systems and factory reset where /etc or /var are
934 unpopulated at boot. Matching these new settings there's also
935 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
936 ConfigurationDirectoryMode= for configuring the access mode of these
937 directories. These settings are particularly useful in combination
938 with DynamicUser=yes as they provide secure, properly-owned,
939 writable, and stateful locations for storage, excluded from the
940 sandbox that such services live in otherwise.
942 * Automake support has been removed from this release. systemd is now
945 * systemd-journald will now aggressively cache client metadata during
946 runtime, speeding up log write performance under pressure. This comes
947 at a small price though: as much of the metadata is read
948 asynchronously from /proc/ (and isn't implicitly attached to log
949 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
950 metadata stored alongside a log entry might be slightly
951 out-of-date. Previously it could only be slightly newer than the log
952 message. The time window is small however, and given that the kernel
953 is unlikely to be improved anytime soon in this regard, this appears
956 * nss-myhostname/systemd-resolved will now by default synthesize an
957 A/AAAA resource record for the "_gateway" hostname, pointing to the
958 current default IP gateway. Previously it did that for the "gateway"
959 name, hampering adoption, as some distributions wanted to leave that
960 host name open for local use. The old behaviour may still be
961 requested at build time.
963 * systemd-networkd's [Address] section in .network files gained a new
964 Scope= setting for configuring the IP address scope. The [Network]
965 section gained a new boolean setting ConfigureWithoutCarrier= that
966 tells systemd-networkd to ignore link sensing when configuring the
967 device. The [DHCP] section gained a new Anonymize= boolean option for
968 turning on a number of options suggested in RFC 7844. A new
969 [RoutingPolicyRule] section has been added for configuring the IP
970 routing policy. The [Route] section has gained support for a new
971 Type= setting which permits configuring
972 blackhole/unreachable/prohibit routes.
974 * The [VRF] section in .netdev files gained a new Table= setting for
975 configuring the routing table to use. The [Tunnel] section gained a
976 new Independent= boolean field for configuring tunnels independent of
977 an underlying network interface. The [Bridge] section gained a new
978 GroupForwardMask= option for configuration of propagation of link
979 local frames between bridge ports.
981 * The WakeOnLan= setting in .link files gained support for a number of
982 new modes. A new TCP6SegmentationOffload= setting has been added for
983 configuring TCP/IPv6 hardware segmentation offload.
985 * The IPv6 RA sender implementation may now optionally send out RDNSS
986 and RDNSSL records to supply DNS configuration to peers.
988 * systemd-nspawn gained support for a new --system-call-filter= command
989 line option for adding and removing entries in the default system
990 call filter it applies. Moreover systemd-nspawn has been changed to
991 implement a system call whitelist instead of a blacklist.
993 * systemd-run gained support for a new --pipe command line option. If
994 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
995 are directly passed on to the activated transient service
996 executable. This allows invoking arbitrary processes as systemd
997 services (for example to take benefit of dependency management,
998 accounting management, resource management or log management that is
999 done automatically for services) — while still allowing them to be
1000 integrated in a classic UNIX shell pipeline.
1002 * When a service sends RELOAD=1 via sd_notify() and reload propagation
1003 using ReloadPropagationTo= is configured, a reload is now propagated
1004 to configured units. (Previously this was only done on explicitly
1005 requested reloads, using "systemctl reload" or an equivalent
1008 * For each service unit a restart counter is now kept: it is increased
1009 each time the service is restarted due to Restart=, and may be
1010 queried using "systemctl show -p NRestarts …".
1012 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
1013 @signal and @timer have been added, for usage with SystemCallFilter=
1014 in unit files and the new --system-call-filter= command line option
1015 of systemd-nspawn (see above).
1017 * ExecStart= lines in unit files gained two new modifiers: when a
1018 command line is prefixed with "!" the command will be executed as
1019 configured, except for the credentials applied by
1020 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
1021 "+", but does still apply namespacing options unlike "+". There's
1022 also "!!" now, which is mostly identical, but becomes a NOP on
1023 systems that support ambient capabilities. This is useful to write
1024 unit files that work with ambient capabilities where possible but
1025 automatically fall back to traditional privilege dropping mechanisms
1026 on systems where this is not supported.
1028 * ListenNetlink= settings in socket units now support RDMA netlink
1031 * A new unit file setting LockPersonality= has been added which permits
1032 locking down the chosen execution domain ("personality") of a service
1035 * A new special target "getty-pre.target" has been added, which is
1036 ordered before all text logins, and may be used to order services
1037 before textual logins acquire access to the console.
1039 * systemd will now attempt to load the virtio-rng.ko kernel module very
1040 early on if a VM environment supporting this is detected. This should
1041 improve entropy during early boot in virtualized environments.
1043 * A _netdev option is now supported in /etc/crypttab that operates in a
1044 similar way as the same option in /etc/fstab: it permits configuring
1045 encrypted devices that need to be ordered after the network is up.
1046 Following this logic, two new special targets
1047 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
1048 added that are to cryptsetup.target what remote-fs.target and
1049 remote-fs-pre.target are to local-fs.target.
1051 * Service units gained a new UnsetEnvironment= setting which permits
1052 unsetting specific environment variables for services that are
1053 normally passed to it (for example in order to mask out locale
1054 settings for specific services that can't deal with it).
1056 * Units acquired a new boolean option IPAccounting=. When turned on, IP
1057 traffic accounting (packet count as well as byte count) is done for
1058 the service, and shown as part of "systemctl status" or "systemd-run
1061 * Service units acquired two new options IPAddressAllow= and
1062 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
1063 for configuring a simple IP access control list for all sockets of
1064 the unit. These options are available also on .slice and .socket
1065 units, permitting flexible access list configuration for individual
1066 services as well as groups of services (as defined by a slice unit),
1067 including system-wide. Note that IP ACLs configured this way are
1068 enforced on every single IPv4 and IPv6 socket created by any process
1069 of the service unit, and apply to ingress as well as egress traffic.
1071 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
1072 structured log message is generated each time the unit is stopped,
1073 containing information about the consumed resources of this
1076 * A new setting KeyringMode= has been added to unit files, which may be
1077 used to control how the kernel keyring is set up for executed
1080 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
1081 "systemctl kexec" and "systemctl exit" are now always asynchronous in
1082 behaviour (that is: these commands return immediately after the
1083 operation was enqueued instead of waiting for the operation to
1084 complete). Previously, "systemctl poweroff" and "systemctl reboot"
1085 were asynchronous on systems using systemd-logind (i.e. almost
1086 always, and like they were on sysvinit), and the other three commands
1087 were unconditionally synchronous. With this release this is cleaned
1088 up, and callers will see the same asynchronous behaviour on all
1089 systems for all five operations.
1091 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
1094 * .timer units now accept calendar specifications in other timezones
1095 than UTC or the local timezone.
1097 * The tmpfiles snippet var.conf has been changed to create
1098 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
1099 the "utmp" group already, and it appears to be generally understood
1100 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
1101 databases. Previously this was implemented correctly for all these
1102 databases excepts btmp, which has been opened up like this now
1103 too. Note that while the other databases are world-readable
1104 (i.e. 0644), btmp is not and remains more restrictive.
1106 * The systemd-resolve tool gained a new --reset-server-features
1107 switch. When invoked like this systemd-resolved will forget
1108 everything it learnt about the features supported by the configured
1109 upstream DNS servers, and restarts the feature probing logic on the
1110 next resolver look-up for them at the highest feature level
1113 * The status dump systemd-resolved sends to the logs upon receiving
1114 SIGUSR1 now also includes information about all DNS servers it is
1115 configured to use, and the features levels it probed for them.
1117 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
1118 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
1119 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
1120 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
1121 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
1122 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
1123 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
1124 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
1125 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
1126 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
1127 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
1128 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
1129 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
1130 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
1131 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
1132 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
1133 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
1134 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
1135 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
1136 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1138 — Berlin, 2017-10-06
1142 * Meson is now supported as build system in addition to Automake. It is
1143 our plan to remove Automake in one of our next releases, so that
1144 Meson becomes our exclusive build system. Hence, please start using
1145 the Meson build system in your downstream packaging. There's plenty
1146 of documentation around how to use Meson, the extremely brief
1149 ./autogen.sh && ./configure && make && sudo make install
1153 meson build && ninja -C build && sudo ninja -C build install
1155 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
1156 which permits configuring a timeout on the time a job is
1157 running. This is particularly useful for setting timeouts on jobs for
1160 * Unit files gained two new options ConditionUser= and ConditionGroup=
1161 for conditionalizing units based on the identity of the user/group
1162 running a systemd user instance.
1164 * systemd-networkd now understands a new FlowLabel= setting in the
1165 [VXLAN] section of .network files, as well as a Priority= in
1166 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
1167 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
1168 gained support for configuration of GENEVE links, and IPv6 address
1169 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
1171 * .link files now understand a new Port= setting.
1173 * systemd-networkd's DHCP support gained support for DHCP option 119
1174 (domain search list).
1176 * systemd-networkd gained support for serving IPv6 address ranges using
1177 the Router Advertisment protocol. The new .network configuration
1178 section [IPv6Prefix] may be used to configure the ranges to
1179 serve. This is implemented based on a new, minimal, native server
1180 implementation of RA.
1182 * journalctl's --output= switch gained support for a new parameter
1183 "short-iso-precise" for a mode where timestamps are shown as precise
1186 * systemd-udevd's "net_id" builtin may now generate stable network
1187 interface names from IBM PowerVM VIO devices as well as ACPI platform
1190 * MulticastDNS support in systemd-resolved may now be explicitly
1191 enabled/disabled using the new MulticastDNS= configuration file
1194 * systemd-resolved may now optionally use libidn2 instead of the libidn
1195 for processing internationalized domain names. Support for libidn2
1196 should be considered experimental and should not be enabled by
1199 * "machinectl pull-tar" and related call may now do verification of
1200 downloaded images using SUSE-style .sha256 checksum files in addition
1201 to the already existing support for validating using Ubuntu-style
1204 * sd-bus gained support for a new sd_bus_message_appendv() call which
1205 is va_list equivalent of sd_bus_message_append().
1207 * sd-boot gained support for validating images using SHIM/MOK.
1209 * The SMACK code learnt support for "onlycap".
1211 * systemd-mount --umount is now much smarter in figuring out how to
1212 properly unmount a device given its mount or device path.
1214 * The code to call libnss_dns as a fallback from libnss_resolve when
1215 the communication with systemd-resolved fails was removed. This
1216 fallback was redundant and interfered with the [!UNAVAIL=return]
1217 suffix. See nss-resolve(8) for the recommended configuration.
1219 * systemd-logind may now be restarted without losing state. It stores
1220 the file descriptors for devices it manages in the system manager
1221 using the FDSTORE= mechanism. Please note that further changes in
1222 other components may be required to make use of this (for example
1223 Xorg has code to listen for stops of systemd-logind and terminate
1224 itself when logind is stopped or restarted, in order to avoid using
1225 stale file descriptors for graphical devices, which is now
1226 counterproductive and must be reverted in order for restarts of
1227 systemd-logind to be safe. See
1228 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
1230 * All kernel install plugins are called with the environment variable
1231 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
1232 /etc/machine-id. If the file is missing or empty, the variable is
1233 empty and BOOT_DIR_ABS is the path of a temporary directory which is
1234 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
1235 is empty, all plugins should not put anything in BOOT_DIR_ABS.
1237 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
1238 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
1239 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
1240 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
1241 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
1242 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
1243 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
1244 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
1245 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
1246 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
1247 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
1248 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
1249 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
1250 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
1251 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
1252 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
1253 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
1254 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
1255 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
1256 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
1257 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
1258 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
1259 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
1260 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
1261 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
1262 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
1263 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
1264 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
1265 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
1268 — Berlin, 2017-07-12
1272 * The "hybrid" control group mode has been modified to improve
1273 compatibility with "legacy" cgroups-v1 setups. Specifically, the
1274 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
1275 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
1276 cgroups-v1 hierarchy), the only externally visible change being that
1277 the cgroups-v2 hierarchy is also mounted, to
1278 /sys/fs/cgroup/unified. This should provide a large degree of
1279 compatibility with "legacy" cgroups-v1, while taking benefit of the
1280 better management capabilities of cgroups-v2.
1282 * The default control group setup mode may be selected both a boot-time
1283 via a set of kernel command line parameters (specifically:
1284 systemd.unified_cgroup_hierarchy= and
1285 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
1286 default selected on the configure command line
1287 (--with-default-hierarchy=). The upstream default is "hybrid"
1288 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
1289 this will change in a future systemd version to be "unified" (pure
1290 cgroups-v2 mode). The third option for the compile time option is
1291 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
1292 distributions to default to "hybrid" mode for release distributions,
1293 starting with v233. We recommend "unified" for development
1294 distributions (specifically: distributions such as Fedora's rawhide)
1295 as that's where things are headed in the long run. Use "legacy" for
1296 greatest stability and compatibility only.
1298 * Note one current limitation of "unified" and "hybrid" control group
1299 setup modes: the kernel currently does not permit the systemd --user
1300 instance (i.e. unprivileged code) to migrate processes between two
1301 disconnected cgroup subtrees, even if both are managed and owned by
1302 the user. This effectively means "systemd-run --user --scope" doesn't
1303 work when invoked from outside of any "systemd --user" service or
1304 scope. Specifically, it is not supported from session scopes. We are
1305 working on fixing this in a future systemd version. (See #3388 for
1306 further details about this.)
1308 * DBus policy files are now installed into /usr rather than /etc. Make
1309 sure your system has dbus >= 1.9.18 running before upgrading to this
1310 version, or override the install path with --with-dbuspolicydir= .
1312 * All python scripts shipped with systemd (specifically: the various
1313 tests written in Python) now require Python 3.
1315 * systemd unit tests can now run standalone (without the source or
1316 build directories), and can be installed into /usr/lib/systemd/tests/
1317 with 'make install-tests'.
1319 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
1320 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
1323 * Support for the %c, %r, %R specifiers in unit files has been
1324 removed. Specifiers are not supposed to be dependent on configuration
1325 in the unit file itself (so that they resolve the same regardless
1326 where used in the unit files), but these specifiers were influenced
1327 by the Slice= option.
1329 * The shell invoked by debug-shell.service now defaults to /bin/sh in
1330 all cases. If distributions want to use a different shell for this
1331 purpose (for example Fedora's /sbin/sushell) they need to specify
1332 this explicitly at configure time using --with-debug-shell=.
1334 * The confirmation spawn prompt has been reworked to offer the
1337 (c)ontinue, proceed without asking anymore
1338 (D)ump, show the state of the unit
1339 (f)ail, don't execute the command and pretend it failed
1341 (i)nfo, show a short summary of the unit
1342 (j)obs, show jobs that are in progress
1343 (s)kip, don't execute the command and pretend it succeeded
1344 (y)es, execute the command
1346 The 'n' choice for the confirmation spawn prompt has been removed,
1347 because its meaning was confusing.
1349 The prompt may now also be redirected to an alternative console by
1350 specifying the console as parameter to systemd.confirm_spawn=.
1352 * Services of Type=notify require a READY=1 notification to be sent
1353 during startup. If no such message is sent, the service now fails,
1354 even if the main process exited with a successful exit code.
1356 * Services that fail to start up correctly now always have their
1357 ExecStopPost= commands executed. Previously, they'd enter "failed"
1358 state directly, without executing these commands.
1360 * The option MulticastDNS= of network configuration files has acquired
1361 an actual implementation. With MulticastDNS=yes a host can resolve
1362 names of remote hosts and reply to mDNS A and AAAA requests.
1364 * When units are about to be started an additional check is now done to
1365 ensure that all dependencies of type BindsTo= (when used in
1366 combination with After=) have been started.
1368 * systemd-analyze gained a new verb "syscall-filter" which shows which
1369 system call groups are defined for the SystemCallFilter= unit file
1370 setting, and which system calls they contain.
1372 * A new system call filter group "@filesystem" has been added,
1373 consisting of various file system related system calls. Group
1374 "@reboot" has been added, covering reboot, kexec and shutdown related
1375 calls. Finally, group "@swap" has been added covering swap
1376 configuration related calls.
1378 * A new unit file option RestrictNamespaces= has been added that may be
1379 used to restrict access to the various process namespace types the
1380 Linux kernel provides. Specifically, it may be used to take away the
1381 right for a service unit to create additional file system, network,
1382 user, and other namespaces. This sandboxing option is particularly
1383 relevant due to the high amount of recently discovered namespacing
1384 related vulnerabilities in the kernel.
1386 * systemd-udev's .link files gained support for a new AutoNegotiation=
1387 setting for configuring Ethernet auto-negotiation.
1389 * systemd-networkd's .network files gained support for a new
1390 ListenPort= setting in the [DHCP] section to explicitly configure the
1391 UDP client port the DHCP client shall listen on.
1393 * .network files gained a new Unmanaged= boolean setting for explicitly
1394 excluding one or more interfaces from management by systemd-networkd.
1396 * The systemd-networkd ProxyARP= option has been renamed to
1397 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
1398 renamed to ReduceARPProxy=. The old names continue to be available
1401 * systemd-networkd gained support for configuring IPv6 Proxy NDP
1402 addresses via the new IPv6ProxyNDPAddress= .network file setting.
1404 * systemd-networkd's bonding device support gained support for two new
1405 configuration options ActiveSlave= and PrimarySlave=.
1407 * The various options in the [Match] section of .network files gained
1408 support for negative matching.
1410 * New systemd-specific mount options are now understood in /etc/fstab:
1412 x-systemd.mount-timeout= may be used to configure the maximum
1413 permitted runtime of the mount command.
1415 x-systemd.device-bound may be set to bind a mount point to its
1416 backing device unit, in order to automatically remove a mount point
1417 if its backing device is unplugged. This option may also be
1418 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
1419 on the block device, which is now automatically set for all CDROM
1420 drives, so that mounted CDs are automatically unmounted when they are
1421 removed from the drive.
1423 x-systemd.after= and x-systemd.before= may be used to explicitly
1424 order a mount after or before another unit or mount point.
1426 * Enqueued start jobs for device units are now automatically garbage
1427 collected if there are no jobs waiting for them anymore.
1429 * systemctl list-jobs gained two new switches: with --after, for every
1430 queued job the jobs it's waiting for are shown; with --before the
1431 jobs which it's blocking are shown.
1433 * systemd-nspawn gained support for ephemeral boots from disk images
1434 (or in other words: --ephemeral and --image= may now be
1435 combined). Moreover, ephemeral boots are now supported for normal
1436 directories, even if the backing file system is not btrfs. Of course,
1437 if the file system does not support file system snapshots or
1438 reflinks, the initial copy operation will be relatively expensive, but
1439 this should still be suitable for many use cases.
1441 * Calendar time specifications in .timer units now support
1442 specifications relative to the end of a month by using "~" instead of
1443 "-" as separator between month and day. For example, "*-02~03" means
1444 "the third last day in February". In addition a new syntax for
1445 repeated events has been added using the "/" character. For example,
1446 "9..17/2:00" means "every two hours from 9am to 5pm".
1448 * systemd-socket-proxyd gained a new parameter --connections-max= for
1449 configuring the maximum number of concurrent connections.
1451 * sd-id128 gained a new API for generating unique IDs for the host in a
1452 way that does not leak the machine ID. Specifically,
1453 sd_id128_get_machine_app_specific() derives an ID based on the
1454 machine ID a in well-defined, non-reversible, stable way. This is
1455 useful whenever an identifier for the host is needed but where the
1456 identifier shall not be useful to identify the system beyond the
1457 scope of the application itself. (Internally this uses HMAC-SHA256 as
1458 keyed hash function using the machine ID as input.)
1460 * NotifyAccess= gained a new supported value "exec". When set
1461 notifications are accepted from all processes systemd itself invoked,
1462 including all control processes.
1464 * .nspawn files gained support for defining overlay mounts using the
1465 Overlay= and OverlayReadOnly= options. Previously this functionality
1466 was only available on the systemd-nspawn command line.
1468 * systemd-nspawn's --bind= and --overlay= options gained support for
1469 bind/overlay mounts whose source lies within the container tree by
1470 prefixing the source path with "+".
1472 * systemd-nspawn's --bind= and --overlay= options gained support for
1473 automatically allocating a temporary source directory in /var/tmp
1474 that is removed when the container dies. Specifically, if the source
1475 directory is specified as empty string this mechanism is selected. An
1476 example usage is --overlay=+/var::/var, which creates an overlay
1477 mount based on the original /var contained in the image, overlayed
1478 with a temporary directory in the host's /var/tmp. This way changes
1479 to /var are automatically flushed when the container shuts down.
1481 * systemd-nspawn --image= option does now permit raw file system block
1482 devices (in addition to images containing partition tables, as
1485 * The disk image dissection logic in systemd-nspawn gained support for
1486 automatically setting up LUKS encrypted as well as Verity protected
1487 partitions. When a container is booted from an encrypted image the
1488 passphrase is queried at start-up time. When a container with Verity
1489 data is started, the root hash is search in a ".roothash" file
1490 accompanying the disk image (alternatively, pass the root hash via
1491 the new --root-hash= command line option).
1493 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
1494 be used to dissect disk images the same way as systemd-nspawn does
1495 it, following the Bootable Partition Specification. It may even be
1496 used to mount disk images with complex partition setups (including
1497 LUKS and Verity partitions) to a local host directory, in order to
1498 inspect them. This tool is not considered public API (yet), and is
1499 thus not installed into /usr/bin. Please do not rely on its
1500 existence, since it might go away or be changed in later systemd
1503 * A new generator "systemd-verity-generator" has been added, similar in
1504 style to "systemd-cryptsetup-generator", permitting automatic setup of
1505 Verity root partitions when systemd boots up. In order to make use of
1506 this your partition setup should follow the Discoverable Partitions
1507 Specification, and the GPT partition ID of the root file system
1508 partition should be identical to the upper 128bit of the Verity root
1509 hash. The GPT partition ID of the Verity partition protecting it
1510 should be the lower 128bit of the Verity root hash. If the partition
1511 image follows this model it is sufficient to specify a single
1512 "roothash=" kernel command line argument to both configure which root
1513 image and verity partition to use as well as the root hash for
1514 it. Note that systemd-nspawn's Verity support follows the same
1515 semantics, meaning that disk images with proper Verity data in place
1516 may be booted in containers with systemd-nspawn as well as on
1517 physical systems via the verity generator. Also note that the "mkosi"
1518 tool available at https://github.com/systemd/mkosi has been updated
1519 to generate Verity protected disk images following this scheme. In
1520 fact, it has been updated to generate disk images that optionally
1521 implement a complete UEFI SecureBoot trust chain, involving a signed
1522 kernel and initrd image that incorporates such a root hash as well as
1523 a Verity-enabled root partition.
1525 * The hardware database (hwdb) udev supports has been updated to carry
1526 accelerometer quirks.
1528 * All system services are now run with a fresh kernel keyring set up
1529 for them. The invocation ID is stored by default in it, thus
1530 providing a safe, non-overridable way to determine the invocation
1533 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
1534 options for bind mounting arbitrary paths in a service-specific
1535 way. When these options are used, arbitrary host or service files and
1536 directories may be mounted to arbitrary locations in the service's
1539 * Documentation has been added that lists all of systemd's low-level
1540 environment variables:
1542 https://github.com/systemd/systemd/blob/master/doc/ENVIRONMENT.md
1544 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
1545 whether a specific socket file descriptor matches a specified socket
1548 * systemd-firstboot has been updated to check for the
1549 systemd.firstboot= kernel command line option. It accepts a boolean
1550 and when set to false the first boot questions are skipped.
1552 * systemd-fstab-generator has been updated to check for the
1553 systemd.volatile= kernel command line option, which either takes an
1554 optional boolean parameter or the special value "state". If used the
1555 system may be booted in a "volatile" boot mode. Specifically,
1556 "systemd.volatile" is used, the root directory will be mounted as
1557 tmpfs, and only /usr is mounted from the actual root file system. If
1558 "systemd.volatile=state" is used, the root directory will be mounted
1559 as usual, but /var is mounted as tmpfs. This concept provides similar
1560 functionality as systemd-nspawn's --volatile= option, but provides it
1561 on physical boots. Use this option for implementing stateless
1562 systems, or testing systems with all state and/or configuration reset
1563 to the defaults. (Note though that many distributions are not
1564 prepared to boot up without a populated /etc or /var, though.)
1566 * systemd-gpt-auto-generator gained support for LUKS encrypted root
1567 partitions. Previously it only supported LUKS encrypted partitions
1568 for all other uses, except for the root partition itself.
1570 * Socket units gained support for listening on AF_VSOCK sockets for
1571 communication in virtualized QEMU environments.
1573 * The "configure" script gained a new option --with-fallback-hostname=
1574 for specifying the fallback hostname to use if none is configured in
1575 /etc/hostname. For example, by specifying
1576 --with-fallback-hostname=fedora it is possible to default to a
1577 hostname of "fedora" on pristine installations.
1579 * systemd-cgls gained support for a new --unit= switch for listing only
1580 the control groups of a specific unit. Similar --user-unit= has been
1581 added for listing only the control groups of a specific user unit.
1583 * systemd-mount gained a new --umount switch for unmounting a mount or
1584 automount point (and all mount/automount points below it).
1586 * systemd will now refuse full configuration reloads (via systemctl
1587 daemon-reload and related calls) unless at least 16MiB of free space
1588 are available in /run. This is a safety precaution in order to ensure
1589 that generators can safely operate after the reload completed.
1591 * A new unit file option RootImage= has been added, which has a similar
1592 effect as RootDirectory= but mounts the service's root directory from
1593 a disk image instead of plain directory. This logic reuses the same
1594 image dissection and mount logic that systemd-nspawn already uses,
1595 and hence supports any disk images systemd-nspawn supports, including
1596 those following the Discoverable Partition Specification, as well as
1597 Verity enabled images. This option enables systemd to run system
1598 services directly off disk images acting as resource bundles,
1599 possibly even including full integrity data.
1601 * A new MountAPIVFS= unit file option has been added, taking a boolean
1602 argument. If enabled /proc, /sys and /dev (collectively called the
1603 "API VFS") will be mounted for the service. This is only relevant if
1604 RootDirectory= or RootImage= is used for the service, as these mounts
1605 are of course in place in the host mount namespace anyway.
1607 * systemd-nspawn gained support for a new --pivot-root= switch. If
1608 specified the root directory within the container image is pivoted to
1609 the specified mount point, while the original root disk is moved to a
1610 different place. This option enables booting of ostree images
1611 directly with systemd-nspawn.
1613 * The systemd build scripts will no longer complain if the NTP server
1614 addresses are not changed from the defaults. Google now supports
1615 these NTP servers officially. We still recommend downstreams to
1616 properly register an NTP pool with the NTP pool project though.
1618 * coredumpctl gained a new "--reverse" option for printing the list
1619 of coredumps in reverse order.
1621 * coredumpctl will now show additional information about truncated and
1622 inaccessible coredumps, as well as coredumps that are still being
1623 processed. It also gained a new --quiet switch for suppressing
1624 additional informational message in its output.
1626 * coredumpctl gained support for only showing coredumps newer and/or
1627 older than specific timestamps, using the new --since= and --until=
1628 options, reminiscent of journalctl's options by the same name.
1630 * The systemd-coredump logic has been improved so that it may be reused
1631 to collect backtraces in non-compiled languages, for example in
1632 scripting languages such as Python.
1634 * machinectl will now show the UID shift of local containers, if user
1635 namespacing is enabled for them.
1637 * systemd will now optionally run "environment generator" binaries at
1638 configuration load time. They may be used to add environment
1639 variables to the environment block passed to services invoked. One
1640 user environment generator is shipped by default that sets up
1641 environment variables based on files dropped into /etc/environment.d
1642 and ~/.config/environment.d/.
1644 * systemd-resolved now includes the new, recently published 2017 DNSSEC
1647 * hostnamed has been updated to report a new chassis type of
1648 "convertible" to cover "foldable" laptops that can both act as a
1649 tablet and as a laptop, such as various Lenovo Yoga devices.
1651 Contributions from: Adrián López, Alexander Galanin, Alexander
1652 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
1653 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
1654 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
1655 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
1656 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
1657 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
1658 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
1659 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
1660 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
1661 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
1662 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
1663 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
1664 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
1665 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
1666 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
1667 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
1668 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
1669 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
1670 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
1671 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
1672 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
1673 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
1674 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
1675 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
1676 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
1677 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
1678 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
1681 — Berlin, 2017-03-01
1685 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
1686 RestrictAddressFamilies= enabled. These sandboxing options should
1687 generally be compatible with the various external udev call-out
1688 binaries we are aware of, however there may be exceptions, in
1689 particular when exotic languages for these call-outs are used. In
1690 this case, consider turning off these settings locally.
1692 * The new RemoveIPC= option can be used to remove IPC objects owned by
1693 the user or group of a service when that service exits.
1695 * The new ProtectKernelModules= option can be used to disable explicit
1696 load and unload operations of kernel modules by a service. In
1697 addition access to /usr/lib/modules is removed if this option is set.
1699 * ProtectSystem= option gained a new value "strict", which causes the
1700 whole file system tree with the exception of /dev, /proc, and /sys,
1701 to be remounted read-only for a service.
1703 * The new ProtectKernelTunables= option can be used to disable
1704 modification of configuration files in /sys and /proc by a service.
1705 Various directories and files are remounted read-only, so access is
1706 restricted even if the file permissions would allow it.
1708 * The new ProtectControlGroups= option can be used to disable write
1709 access by a service to /sys/fs/cgroup.
1711 * Various systemd services have been hardened with
1712 ProtectKernelTunables=yes, ProtectControlGroups=yes,
1713 RestrictAddressFamilies=.
1715 * Support for dynamically creating users for the lifetime of a service
1716 has been added. If DynamicUser=yes is specified, user and group IDs
1717 will be allocated from the range 61184..65519 for the lifetime of the
1718 service. They can be resolved using the new nss-systemd.so NSS
1719 module. The module must be enabled in /etc/nsswitch.conf. Services
1720 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
1721 any resources allocated by the service will be cleaned up when the
1722 service exits. They also have ProtectHome=read-only and
1723 ProtectSystem=strict enabled, so they are not able to make any
1724 permanent modifications to the system.
1726 * The nss-systemd module also always resolves root and nobody, making
1727 it possible to have no /etc/passwd or /etc/group files in minimal
1728 container or chroot environments.
1730 * Services may be started with their own user namespace using the new
1731 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
1732 under which the service is running are mapped. All other users are
1735 * Support for the cgroup namespace has been added to systemd-nspawn. If
1736 supported by kernel, the container system started by systemd-nspawn
1737 will have its own view of the cgroup hierarchy. This new behaviour
1738 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
1740 * The new MemorySwapMax= option can be used to limit the maximum swap
1741 usage under the unified cgroup hierarchy.
1743 * Support for the CPU controller in the unified cgroup hierarchy has
1744 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
1745 options. This controller requires out-of-tree patches for the kernel
1746 and the support is provisional.
1748 * Mount and automount units may now be created transiently
1749 (i.e. dynamically at runtime via the bus API, instead of requiring
1750 unit files in the file system).
1752 * systemd-mount is a new tool which may mount file systems – much like
1753 mount(8), optionally pulling in additional dependencies through
1754 transient .mount and .automount units. For example, this tool
1755 automatically runs fsck on a backing block device before mounting,
1756 and allows the automount logic to be used dynamically from the
1757 command line for establishing mount points. This tool is particularly
1758 useful when dealing with removable media, as it will ensure fsck is
1759 run – if necessary – before the first access and that the file system
1760 is quickly unmounted after each access by utilizing the automount
1761 logic. This maximizes the chance that the file system on the
1762 removable media stays in a clean state, and if it isn't in a clean
1763 state is fixed automatically.
1765 * LazyUnmount=yes option for mount units has been added to expose the
1766 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
1769 * /efi will be used as the mount point of the EFI boot partition, if
1770 the directory is present, and the mount point was not configured
1771 through other means (e.g. fstab). If /efi directory does not exist,
1772 /boot will be used as before. This makes it easier to automatically
1773 mount the EFI partition on systems where /boot is used for something
1776 * When operating on GPT disk images for containers, systemd-nspawn will
1777 now mount the ESP to /boot or /efi according to the same rules as PID
1778 1 running on a host. This allows tools like "bootctl" to operate
1779 correctly within such containers, in order to make container images
1780 bootable on physical systems.
1782 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
1784 * Two new user session targets have been added to support running
1785 graphical sessions under the systemd --user instance:
1786 graphical-session.target and graphical-session-pre.target. See
1787 systemd.special(7) for a description of how those targets should be
1790 * The vconsole initialization code has been significantly reworked to
1791 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
1792 support unicode keymaps. Font and keymap configuration will now be
1793 copied to all allocated virtual consoles.
1795 * FreeBSD's bhyve virtualization is now detected.
1797 * Information recorded in the journal for core dumps now includes the
1798 contents of /proc/mountinfo and the command line of the process at
1799 the top of the process hierarchy (which is usually the init process
1802 * systemd-journal-gatewayd learned the --directory= option to serve
1803 files from the specified location.
1805 * journalctl --root=… can be used to peruse the journal in the
1806 /var/log/ directories inside of a container tree. This is similar to
1807 the existing --machine= option, but does not require the container to
1810 * The hardware database has been extended to support
1811 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
1814 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
1815 specify the click rate for mice which include a horizontal wheel with
1816 a click rate that is different than the one for the vertical wheel.
1818 * systemd-run gained a new --wait option that makes service execution
1819 synchronous. (Specifically, the command will not return until the
1820 specified service binary exited.)
1822 * systemctl gained a new --wait option that causes the start command to
1823 wait until the units being started have terminated again.
1825 * A new journal output mode "short-full" has been added which displays
1826 timestamps with abbreviated English day names and adds a timezone
1827 suffix. Those timestamps include more information than the default
1828 "short" output mode, and can be passed directly to journalctl's
1829 --since= and --until= options.
1831 * /etc/resolv.conf will be bind-mounted into containers started by
1832 systemd-nspawn, if possible, so any changes to resolv.conf contents
1833 are automatically propagated to the container.
1835 * The number of instances for socket-activated services originating
1836 from a single IP address can be limited with
1837 MaxConnectionsPerSource=, extending the existing setting of
1840 * systemd-networkd gained support for vcan ("Virtual CAN") interface
1843 * .netdev and .network configuration can now be extended through
1846 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
1847 Segmentation Offload, Generic Receive Offload, Large Receive Offload
1848 can be enabled and disabled using the new UDPSegmentationOffload=,
1849 TCPSegmentationOffload=, GenericSegmentationOffload=,
1850 GenericReceiveOffload=, LargeReceiveOffload= options in the
1851 [Link] section of .link files.
1853 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
1854 Port VLAN ID can be configured for bridge devices using the new STP=,
1855 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
1856 section of .netdev files.
1858 * The route table to which routes received over DHCP or RA should be
1859 added can be configured with the new RouteTable= option in the [DHCP]
1860 and [IPv6AcceptRA] sections of .network files.
1862 * The Address Resolution Protocol can be disabled on links managed by
1863 systemd-networkd using the ARP=no setting in the [Link] section of
1866 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
1867 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
1868 encode information about the result and exit codes of the current
1869 service runtime cycle.
1871 * systemd-sysctl will now configure kernel parameters in the order
1872 they occur in the configuration files. This matches what sysctl
1873 has been traditionally doing.
1875 * kernel-install "plugins" that are executed to perform various
1876 tasks after a new kernel is added and before an old one is removed
1877 can now return a special value to terminate the procedure and
1878 prevent any later plugins from running.
1880 * Journald's SplitMode=login setting has been deprecated. It has been
1881 removed from documentation, and its use is discouraged. In a future
1882 release it will be completely removed, and made equivalent to current
1883 default of SplitMode=uid.
1885 * Storage=both option setting in /etc/systemd/coredump.conf has been
1886 removed. With fast LZ4 compression storing the core dump twice is not
1889 * The --share-system systemd-nspawn option has been replaced with an
1890 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
1891 this functionality is discouraged. In addition the variables
1892 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
1893 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
1894 individual namespaces.
1896 * "machinectl list" now shows the IP address of running containers in
1897 the output, as well as OS release information.
1899 * "loginctl list" now shows the TTY of each session in the output.
1901 * sd-bus gained new API calls sd_bus_track_set_recursive(),
1902 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
1903 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
1904 tracking objects in a "recursive" mode, where a single client can be
1905 counted multiple times, if it takes multiple references.
1907 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
1908 sd_bus_get_exit_on_disconnect(). They may be used to to make a
1909 process using sd-bus automatically exit if the bus connection is
1912 * Bus clients of the service manager may now "pin" loaded units into
1913 memory, by taking an explicit reference on them. This is useful to
1914 ensure the client can retrieve runtime data about the service even
1915 after the service completed execution. Taking such a reference is
1916 available only for privileged clients and should be helpful to watch
1917 running services in a race-free manner, and in particular collect
1918 information about exit statuses and results.
1920 * The nss-resolve module has been changed to strictly return UNAVAIL
1921 when communication via D-Bus with resolved failed, and NOTFOUND when
1922 a lookup completed but was negative. This means it is now possible to
1923 neatly configure fallbacks using nsswitch.conf result checking
1924 expressions. Taking benefit of this, the new recommended
1925 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
1927 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
1929 * A new setting CtrlAltDelBurstAction= has been added to
1930 /etc/systemd/system.conf which may be used to configure the precise
1931 behaviour if the user on the console presses Ctrl-Alt-Del more often
1932 than 7 times in 2s. Previously this would unconditionally result in
1933 an expedited, immediate reboot. With this new setting the precise
1934 operation may be configured in more detail, and also turned off
1937 * In .netdev files two new settings RemoteChecksumTx= and
1938 RemoteChecksumRx= are now understood that permit configuring the
1939 remote checksumming logic for VXLAN networks.
1941 * The service manager learnt a new "invocation ID" concept for invoked
1942 services. Each runtime cycle of a service will get a new invocation
1943 ID (a 128bit random UUID) assigned that identifies the current
1944 run of the service uniquely and globally. A new invocation ID
1945 is generated each time a service starts up. The journal will store
1946 the invocation ID of a service along with any logged messages, thus
1947 making the invocation ID useful for matching the online runtime of a
1948 service with the offline log data it generated in a safe way without
1949 relying on synchronized timestamps. In many ways this new service
1950 invocation ID concept is similar to the kernel's boot ID concept that
1951 uniquely and globally identifies the runtime of each boot. The
1952 invocation ID of a service is passed to the service itself via an
1953 environment variable ($INVOCATION_ID). A new bus call
1954 GetUnitByInvocationID() has been added that is similar to GetUnit()
1955 but instead of retrieving the bus path for a unit by its name
1956 retrieves it by its invocation ID. The returned path is valid only as
1957 long as the passed invocation ID is current.
1959 * systemd-resolved gained a new "DNSStubListener" setting in
1960 resolved.conf. It either takes a boolean value or the special values
1961 "udp" and "tcp", and configures whether to enable the stub DNS
1962 listener on 127.0.0.53:53.
1964 * IP addresses configured via networkd may now carry additional
1965 configuration settings supported by the kernel. New options include:
1966 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
1967 PrefixRoute=, AutoJoin=.
1969 * The PAM configuration fragment file for "user@.service" shipped with
1970 systemd (i.e. the --user instance of systemd) has been stripped to
1971 the minimum necessary to make the system boot. Previously, it
1972 contained Fedora-specific stanzas that did not apply to other
1973 distributions. It is expected that downstream distributions add
1974 additional configuration lines, matching their needs to this file,
1975 using it only as rough template of what systemd itself needs. Note
1976 that this reduced fragment does not even include an invocation of
1977 pam_limits which most distributions probably want to add, even though
1978 systemd itself does not need it. (There's also the new build time
1979 option --with-pamconfdir=no to disable installation of the PAM
1982 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
1983 capability is now also dropped from its set (in addition to
1984 CAP_SYS_MKNOD as before).
1986 * In service unit files it is now possible to connect a specific named
1987 file descriptor with stdin/stdout/stdout of an executed service. The
1988 name may be specified in matching .socket units using the
1989 FileDescriptorName= setting.
1991 * A number of journal settings may now be configured on the kernel
1992 command line. Specifically, the following options are now understood:
1993 systemd.journald.max_level_console=,
1994 systemd.journald.max_level_store=,
1995 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
1996 systemd.journald.max_level_wall=.
1998 * "systemctl is-enabled --full" will now show by which symlinks a unit
1999 file is enabled in the unit dependency tree.
2001 * Support for VeraCrypt encrypted partitions has been added to the
2002 "cryptsetup" logic and /etc/crypttab.
2004 * systemd-detect-virt gained support for a new --private-users switch
2005 that checks whether the invoking processes are running inside a user
2006 namespace. Similar, a new special value "private-users" for the
2007 existing ConditionVirtualization= setting has been added, permitting
2008 skipping of specific units in user namespace environments.
2010 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
2011 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
2012 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
2013 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
2014 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
2015 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
2016 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
2017 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
2018 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
2019 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
2020 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
2021 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
2022 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
2023 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
2024 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
2025 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
2026 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
2027 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
2028 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
2029 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
2030 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
2031 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
2032 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
2033 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
2034 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
2035 Jędrzejewski-Szmek, Zeal Jagannatha
2037 — Santa Fe, 2016-11-03
2041 * In service units the various ExecXYZ= settings have been extended
2042 with an additional special character as first argument of the
2043 assigned value: if the character '+' is used the specified command
2044 line it will be run with full privileges, regardless of User=,
2045 Group=, CapabilityBoundingSet= and similar options. The effect is
2046 similar to the existing PermissionsStartOnly= option, but allows
2047 configuration of this concept for each executed command line
2050 * Services may now alter the service watchdog timeout at runtime by
2051 sending a WATCHDOG_USEC= message via sd_notify().
2053 * MemoryLimit= and related unit settings now optionally take percentage
2054 specifications. The percentage is taken relative to the amount of
2055 physical memory in the system (or in case of containers, the assigned
2056 amount of memory). This allows scaling service resources neatly with
2057 the amount of RAM available on the system. Similarly, systemd-logind's
2058 RuntimeDirectorySize= option now also optionally takes percentage
2061 * In similar fashion TasksMax= takes percentage values now, too. The
2062 value is taken relative to the configured maximum number of processes
2063 on the system. The per-service task maximum has been changed to 15%
2064 using this functionality. (Effectively this is an increase of 512 →
2065 4915 for service units, given the kernel's default pid_max setting.)
2067 * Calendar time specifications in .timer units now understand a ".."
2068 syntax for time ranges. Example: "4..7:10" may now be used for
2069 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
2072 * The InaccessableDirectories=, ReadOnlyDirectories= and
2073 ReadWriteDirectories= unit file settings have been renamed to
2074 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
2075 applied to all kinds of file nodes, and not just directories, with
2076 the exception of symlinks. Specifically these settings may now be
2077 used on block and character device nodes, UNIX sockets and FIFOS as
2078 well as regular files. The old names of these settings remain
2079 available for compatibility.
2081 * systemd will now log about all service processes it kills forcibly
2082 (using SIGKILL) because they remained after the clean shutdown phase
2083 of the service completed. This should help identifying services that
2084 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
2085 systemd-logind's configuration a similar log message is generated for
2086 processes killed at the end of each session due to this setting.
2088 * systemd will now set the $JOURNAL_STREAM environment variable for all
2089 services whose stdout/stderr are connected to the Journal (which
2090 effectively means by default: all services). The variable contains
2091 the device and inode number of the file descriptor used for
2092 stdout/stderr. This may be used by invoked programs to detect whether
2093 their stdout/stderr is connected to the Journal, in which case they
2094 can switch over to direct Journal communication, thus being able to
2095 pass extended, structured metadata along with their log messages. As
2096 one example, this is now used by glib's logging primitives.
2098 * When using systemd's default tmp.mount unit for /tmp, the mount point
2099 will now be established with the "nosuid" and "nodev" options. This
2100 avoids privilege escalation attacks that put traps and exploits into
2101 /tmp. However, this might cause problems if you e. g. put container
2102 images or overlays into /tmp; if you need this, override tmp.mount's
2103 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
2106 * systemd now supports the "memory" cgroup controller also on
2109 * The systemd-cgtop tool now optionally takes a control group path as
2110 command line argument. If specified, the control group list shown is
2111 limited to subgroups of that group.
2113 * The SystemCallFilter= unit file setting gained support for
2114 pre-defined, named system call filter sets. For example
2115 SystemCallFilter=@clock is now an effective way to make all clock
2116 changing-related system calls unavailable to a service. A number of
2117 similar pre-defined groups are defined. Writing system call filters
2118 for system services is simplified substantially with this new
2119 concept. Accordingly, all of systemd's own, long-running services now
2120 enable system call filtering based on this, by default.
2122 * A new service setting MemoryDenyWriteExecute= has been added, taking
2123 a boolean value. If turned on, a service may no longer create memory
2124 mappings that are writable and executable at the same time. This
2125 enhances security for services where this is enabled as it becomes
2126 harder to dynamically write and then execute memory in exploited
2127 service processes. This option has been enabled for all of systemd's
2128 own long-running services.
2130 * A new RestrictRealtime= service setting has been added, taking a
2131 boolean argument. If set the service's processes may no longer
2132 acquire realtime scheduling. This improves security as realtime
2133 scheduling may otherwise be used to easily freeze the system.
2135 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
2136 value. This may be used for requesting that the system manager inside
2137 of the container reports start-up completion to nspawn which then
2138 propagates this notification further to the service manager
2139 supervising nspawn itself. A related option NotifyReady= in .nspawn
2140 files has been added too. This functionality allows ordering of the
2141 start-up of multiple containers using the usual systemd ordering
2144 * machinectl gained a new command "stop" that is an alias for
2147 * systemd-resolved gained support for contacting DNS servers on
2148 link-local IPv6 addresses.
2150 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
2151 its caches. A method call for requesting the same operation has been
2152 added to the bus API too, and is made available via "systemd-resolve
2155 * systemd-resolve gained a new --status switch. If passed a brief
2156 summary of the used DNS configuration with per-interface information
2159 * resolved.conf gained a new Cache= boolean option, defaulting to
2160 on. If turned off local DNS caching is disabled. This comes with a
2161 performance penalty in particular when DNSSEC is enabled. Note that
2162 resolved disables its internal caching implicitly anyway, when the
2163 configured DNS server is on a host-local IP address such as ::1 or
2164 127.0.0.1, thus automatically avoiding double local caching.
2166 * systemd-resolved now listens on the local IP address 127.0.0.53:53
2167 for DNS requests. This improves compatibility with local programs
2168 that do not use the libc NSS or systemd-resolved's bus APIs for name
2169 resolution. This minimal DNS service is only available to local
2170 programs and does not implement the full DNS protocol, but enough to
2171 cover local DNS clients. A new, static resolv.conf file, listing just
2172 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
2173 now recommended to make /etc/resolv.conf a symlink to this file in
2174 order to route all DNS lookups to systemd-resolved, regardless if
2175 done via NSS, the bus API or raw DNS packets. Note that this local
2176 DNS service is not as fully featured as the libc NSS or
2177 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
2178 used to deliver link-local address information (as this implies
2179 sending a local interface index along), LLMNR/mDNS support via this
2180 interface is severely restricted. It is thus strongly recommended for
2181 all applications to use the libc NSS API or native systemd-resolved
2184 * systemd-networkd's bridge support learned a new setting
2185 VLANFiltering= for controlling VLAN filtering. Moreover a new section
2186 in .network files has been added for configuring VLAN bridging in
2187 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
2189 * systemd-networkd's IPv6 Router Advertisement code now makes use of
2190 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
2191 now be acquired without relying on DHCPv6. Two new options
2192 UseDomains= and UseDNS= have been added to configure this behaviour.
2194 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
2195 renamed IPv6AcceptRA=, without altering its behaviour. The old
2196 setting name remains available for compatibility reasons.
2198 * The systemd-networkd VTI/VTI6 tunneling support gained new options
2199 Key=, InputKey= and OutputKey=.
2201 * systemd-networkd gained support for VRF ("Virtual Routing Function")
2202 interface configuration.
2204 * "systemctl edit" may now be used to create new unit files by
2205 specifying the --force switch.
2207 * sd-event gained a new function sd_event_get_iteration() for
2208 requesting the current iteration counter of the event loop. It starts
2209 at zero and is increased by one with each event loop iteration.
2211 * A new rpm macro %systemd_ordering is provided by the macros.systemd
2212 file. It can be used in lieu of %systemd_requires in packages which
2213 don't use any systemd functionality and are intended to be installed
2214 in minimal containers without systemd present. This macro provides
2215 ordering dependencies to ensure that if the package is installed in
2216 the same rpm transaction as systemd, systemd will be installed before
2217 the scriptlets for the package are executed, allowing unit presets
2220 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
2221 been added to simplify packaging of generators.
2223 * The os-release file gained VERSION_CODENAME field for the
2224 distribution nickname (e.g. VERSION_CODENAME=woody).
2226 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
2227 can be set to disable parsing of metadata and the creation
2228 of persistent symlinks for that device.
2230 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
2231 to make them available to logged-in users has been reverted.
2233 * Much of the common code of the various systemd components is now
2234 built into an internal shared library libsystemd-shared-231.so
2235 (incorporating the systemd version number in the name, to be updated
2236 with future releases) that the components link to. This should
2237 decrease systemd footprint both in memory during runtime and on
2238 disk. Note that the shared library is not for public use, and is
2239 neither API nor ABI stable, but is likely to change with every new
2240 released update. Packagers need to make sure that binaries
2241 linking to libsystemd-shared.so are updated in step with the
2244 * Configuration for "mkosi" is now part of the systemd
2245 repository. mkosi is a tool to easily build legacy-free OS images,
2246 and is available on github: https://github.com/systemd/mkosi. If
2247 "mkosi" is invoked in the build tree a new raw OS image is generated
2248 incorporating the systemd sources currently being worked on and a
2249 clean, fresh distribution installation. The generated OS image may be
2250 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
2251 UEFI PC. This functionality is particularly useful to easily test
2252 local changes made to systemd in a pristine, defined environment. See
2253 doc/HACKING for details.
2255 * configure learned the --with-support-url= option to specify the
2256 distribution's bugtracker.
2258 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
2259 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
2260 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
2261 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
2262 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
2263 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
2264 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
2265 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
2266 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
2267 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
2268 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
2269 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
2270 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
2271 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
2272 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
2273 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
2274 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
2275 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
2276 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
2278 — Berlin, 2016-07-25
2282 * DNSSEC is now turned on by default in systemd-resolved (in
2283 "allow-downgrade" mode), but may be turned off during compile time by
2284 passing "--with-default-dnssec=no" to "configure" (and of course,
2285 during runtime with DNSSEC= in resolved.conf). We recommend
2286 downstreams to leave this on at least during development cycles and
2287 report any issues with the DNSSEC logic upstream. We are very
2288 interested in collecting feedback about the DNSSEC validator and its
2289 limitations in the wild. Note however, that DNSSEC support is
2290 probably nothing downstreams should turn on in stable distros just
2291 yet, as it might create incompatibilities with a few DNS servers and
2292 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
2293 automatically whenever we detect such incompatible setups, but there
2294 might be systems we do not cover yet. Hence: please help us testing
2295 the DNSSEC code, leave this on where you can, report back, but then
2296 again don't consider turning this on in your stable, LTS or
2297 production release just yet. (Note that you have to enable
2298 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
2299 and its DNSSEC mode for host name resolution from local
2302 * systemd-resolve conveniently resolves DANE records with the --tlsa
2303 option and OPENPGPKEY records with the --openpgp option. It also
2304 supports dumping raw DNS record data via the new --raw= switch.
2306 * systemd-logind will now by default terminate user processes that are
2307 part of the user session scope unit (session-XX.scope) when the user
2308 logs out. This behavior is controlled by the KillUserProcesses=
2309 setting in logind.conf, and the previous default of "no" is now
2310 changed to "yes". This means that user sessions will be properly
2311 cleaned up after, but additional steps are necessary to allow
2312 intentionally long-running processes to survive logout.
2314 While the user is logged in at least once, user@.service is running,
2315 and any service that should survive the end of any individual login
2316 session can be started at a user service or scope using systemd-run.
2317 systemd-run(1) man page has been extended with an example which shows
2318 how to run screen in a scope unit underneath user@.service. The same
2319 command works for tmux.
2321 After the user logs out of all sessions, user@.service will be
2322 terminated too, by default, unless the user has "lingering" enabled.
2323 To effectively allow users to run long-term tasks even if they are
2324 logged out, lingering must be enabled for them. See loginctl(1) for
2325 details. The default polkit policy was modified to allow users to
2326 set lingering for themselves without authentication.
2328 Previous defaults can be restored at compile time by the
2329 --without-kill-user-processes option to "configure".
2331 * systemd-logind gained new configuration settings SessionsMax= and
2332 InhibitorsMax=, both with a default of 8192. It will not register new
2333 user sessions or inhibitors above this limit.
2335 * systemd-logind will now reload configuration on SIGHUP.
2337 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
2338 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
2339 enable. Also, support for the "io" cgroup controller in the unified
2340 hierarchy has been added, so that the "memory", "pids" and "io" are
2341 now the controllers that are supported on the unified hierarchy.
2343 WARNING: it is not possible to use previous systemd versions with
2344 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
2345 is necessary to also update systemd in the initramfs if using the
2346 unified hierarchy. An updated SELinux policy is also required.
2348 * LLDP support has been extended, and both passive (receive-only) and
2349 active (sender) modes are supported. Passive mode ("routers-only") is
2350 enabled by default in systemd-networkd. Active LLDP mode is enabled
2351 by default for containers on the internal network. The "networkctl
2352 lldp" command may be used to list information gathered. "networkctl
2353 status" will also show basic LLDP information on connected peers now.
2355 * The IAID and DUID unique identifier sent in DHCP requests may now be
2356 configured for the system and each .network file managed by
2357 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
2359 * systemd-networkd gained support for configuring proxy ARP support for
2360 each interface, via the ProxyArp= setting in .network files. It also
2361 gained support for configuring the multicast querier feature of
2362 bridge devices, via the new MulticastQuerier= setting in .netdev
2363 files. Similarly, snooping on the IGMP traffic can be controlled
2364 via the new setting MulticastSnooping=.
2366 A new setting PreferredLifetime= has been added for addresses
2367 configured in .network file to configure the lifetime intended for an
2370 The systemd-networkd DHCP server gained the option EmitRouter=, which
2371 defaults to yes, to configure whether the DHCP Option 3 (Router)
2374 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
2375 systemd-socket-activate and installed into /usr/bin. It is now fully
2378 * systemd-journald now uses separate threads to flush changes to disk
2379 when closing journal files, thus reducing impact of slow disk I/O on
2380 logging performance.
2382 * The sd-journal API gained two new calls
2383 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
2384 can be used to open journal files using file descriptors instead of
2385 file or directory paths. sd_journal_open_container() has been
2386 deprecated, sd_journal_open_directory_fd() should be used instead
2387 with the flag SD_JOURNAL_OS_ROOT.
2389 * journalctl learned a new output mode "-o short-unix" that outputs log
2390 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
2391 UTC). It also gained support for a new --no-hostname setting to
2392 suppress the hostname column in the family of "short" output modes.
2394 * systemd-ask-password now optionally skips printing of the password to
2395 stdout with --no-output which can be useful in scripts.
2397 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
2398 (devices tagged with ID_MAKER_TOOL) are now tagged with
2399 "uaccess" and are available to logged in users.
2401 * The DeviceAllow= unit setting now supports specifiers (with "%").
2403 * "systemctl show" gained a new --value switch, which allows print a
2404 only the contents of a specific unit property, without also printing
2405 the property's name. Similar support was added to "show*" verbs
2406 of loginctl and machinectl that output "key=value" lists.
2408 * A new unit type "generated" was added for files dynamically generated
2409 by generator tools. Similarly, a new unit type "transient" is used
2410 for unit files created using the runtime API. "systemctl enable" will
2411 refuse to operate on such files.
2413 * A new command "systemctl revert" has been added that may be used to
2414 revert to the vendor version of a unit file, in case local changes
2415 have been made by adding drop-ins or overriding the unit file.
2417 * "machinectl clean" gained a new verb to automatically remove all or
2418 just hidden container images.
2420 * systemd-tmpfiles gained support for a new line type "e" for emptying
2421 directories, if they exist, without creating them if they don't.
2423 * systemd-nspawn gained support for automatically patching the UID/GIDs
2424 of the owners and the ACLs of all files and directories in a
2425 container tree to match the UID/GID user namespacing range selected
2426 for the container invocation. This mode is enabled via the new
2427 --private-users-chown switch. It also gained support for
2428 automatically choosing a free, previously unused UID/GID range when
2429 starting a container, via the new --private-users=pick setting (which
2430 implies --private-users-chown). Together, these options for the first
2431 time make user namespacing for nspawn containers fully automatic and
2432 thus deployable. The systemd-nspawn@.service template unit file has
2433 been changed to use this functionality by default.
2435 * systemd-nspawn gained a new --network-zone= switch, that allows
2436 creating ad-hoc virtual Ethernet links between multiple containers,
2437 that only exist as long as at least one container referencing them is
2438 running. This allows easy connecting of multiple containers with a
2439 common link that implements an Ethernet broadcast domain. Each of
2440 these network "zones" may be named relatively freely by the user, and
2441 may be referenced by any number of containers, but each container may
2442 only reference one of these "zones". On the lower level, this is
2443 implemented by an automatically managed bridge network interface for
2444 each zone, that is created when the first container referencing its
2445 zone is created and removed when the last one referencing its zone
2448 * The default start timeout may now be configured on the kernel command
2449 line via systemd.default_timeout_start_sec=. It was already
2450 configurable via the DefaultTimeoutStartSec= option in
2451 /etc/systemd/system.conf.
2453 * Socket units gained a new TriggerLimitIntervalSec= and
2454 TriggerLimitBurst= setting to configure a limit on the activation
2455 rate of the socket unit.
2457 * The LimitNICE= setting now optionally takes normal UNIX nice values
2458 in addition to the raw integer limit value. If the specified
2459 parameter is prefixed with "+" or "-" and is in the range -20..19 the
2460 value is understood as UNIX nice value. If not prefixed like this it
2461 is understood as raw RLIMIT_NICE limit.
2463 * Note that the effect of the PrivateDevices= unit file setting changed
2464 slightly with this release: the per-device /dev file system will be
2465 mounted read-only from this version on, and will have "noexec"
2466 set. This (minor) change of behavior might cause some (exceptional)
2467 legacy software to break, when PrivateDevices=yes is set for its
2468 service. Please leave PrivateDevices= off if you run into problems
2471 * systemd-bootchart has been split out to a separate repository:
2472 https://github.com/systemd/systemd-bootchart
2474 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
2475 merged into the kernel in its current form.
2477 * The compatibility libraries libsystemd-daemon.so,
2478 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
2479 which have been deprecated since systemd-209 have been removed along
2480 with the corresponding pkg-config files. All symbols provided by
2481 those libraries are provided by libsystemd.so.
2483 * The Capabilities= unit file setting has been removed (it is ignored
2484 for backwards compatibility). AmbientCapabilities= and
2485 CapabilityBoundingSet= should be used instead.
2487 * A new special target has been added, initrd-root-device.target,
2488 which creates a synchronization point for dependencies of the root
2489 device in early userspace. Initramfs builders must ensure that this
2490 target is now included in early userspace.
2492 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
2493 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
2494 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
2495 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
2496 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
2497 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
2498 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
2499 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
2500 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
2501 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
2502 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
2503 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
2504 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
2505 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
2506 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
2507 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
2508 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
2509 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
2510 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
2511 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
2512 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
2513 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
2514 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
2515 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
2516 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
2519 — Fairfax, 2016-05-21
2523 * The systemd-resolved DNS resolver service has gained a substantial
2524 set of new features, most prominently it may now act as a DNSSEC
2525 validating stub resolver. DNSSEC mode is currently turned off by
2526 default, but is expected to be turned on by default in one of the
2527 next releases. For now, we invite everybody to test the DNSSEC logic
2528 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
2529 service also gained a full set of D-Bus interfaces, including calls
2530 to configure DNS and DNSSEC settings per link (for use by external
2531 network management software). systemd-resolved and systemd-networkd
2532 now distinguish between "search" and "routing" domains. The former
2533 are used to qualify single-label names, the latter are used purely
2534 for routing lookups within certain domains to specific links.
2535 resolved now also synthesizes RRs for all entries from /etc/hosts.
2537 * The systemd-resolve tool (which is a client utility for
2538 systemd-resolved) has been improved considerably and is now fully
2539 supported and documented. Hence it has moved from /usr/lib/systemd to
2542 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
2545 * The coredump collection logic has been reworked: when a coredump is
2546 collected it is now written to disk, compressed and processed
2547 (including stacktrace extraction) from a new instantiated service
2548 systemd-coredump@.service, instead of directly from the
2549 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
2550 processing large coredumps can take up a substantial amount of
2551 resources and time, and this previously happened entirely outside of
2552 systemd's service supervision. With the new logic the core_pattern
2553 hook only does minimal metadata collection before passing off control
2554 to the new instantiated service, which is configured with a time
2555 limit, a nice level and other settings to minimize negative impact on
2556 the rest of the system. Also note that the new logic will honour the
2557 RLIMIT_CORE setting of the crashed process, which now allows users
2558 and processes to turn off coredumping for their processes by setting
2561 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
2562 and all forked processes by default. Previously, PID 1 would leave
2563 the setting at "0" for all processes, as set by the kernel. Note that
2564 the resource limit traditionally has no effect on the generated
2565 coredumps on the system if the /proc/sys/kernel/core_pattern hook
2566 logic is used. Since the limit is now honoured (see above) its
2567 default has been changed so that the coredumping logic is enabled by
2568 default for all processes, while allowing specific opt-out.
2570 * When the stacktrace is extracted from processes of system users, this
2571 is now done as "systemd-coredump" user, in order to sandbox this
2572 potentially security sensitive parsing operation. (Note that when
2573 processing coredumps of normal users this is done under the user ID
2574 of process that crashed, as before.) Packagers should take notice
2575 that it is now necessary to create the "systemd-coredump" system user
2576 and group at package installation time.
2578 * The systemd-activate socket activation testing tool gained support
2579 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
2580 and --seqpacket switches. It also has been extended to support both
2581 new-style and inetd-style file descriptor passing. Use the new
2582 --inetd switch to request inetd-style file descriptor passing.
2584 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
2585 variable, which takes a boolean value. If set to false, ANSI color
2586 output is disabled in the tools even when run on a terminal that
2589 * The VXLAN support in networkd now supports two new settings
2590 DestinationPort= and PortRange=.
2592 * A new systemd.machine_id= kernel command line switch has been added,
2593 that may be used to set the machine ID in /etc/machine-id if it is
2594 not initialized yet. This command line option has no effect if the
2595 file is already initialized.
2597 * systemd-nspawn gained a new --as-pid2 switch that invokes any
2598 specified command line as PID 2 rather than PID 1 in the
2599 container. In this mode PID 1 is a minimal stub init process that
2600 implements the special POSIX and Linux semantics of PID 1 regarding
2601 signal and child process management. Note that this stub init process
2602 is implemented in nspawn itself and requires no support from the
2603 container image. This new logic is useful to support running
2604 arbitrary commands in the container, as normal processes are
2605 generally not prepared to run as PID 1.
2607 * systemd-nspawn gained a new --chdir= switch for setting the current
2608 working directory for the process started in the container.
2610 * "journalctl /dev/sda" will now output all kernel log messages for
2611 specified device from the current boot, in addition to all devices
2612 that are parents of it. This should make log output about devices
2613 pretty useful, as long as kernel drivers attach enough metadata to
2614 the log messages. (The usual SATA drivers do.)
2616 * The sd-journal API gained two new calls
2617 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
2618 that report whether log data from /run or /var has been found.
2620 * journalctl gained a new switch "--fields" that prints all journal
2621 record field names currently in use in the journal. This is backed
2622 by two new sd-journal API calls sd_journal_enumerate_fields() and
2623 sd_journal_restart_fields().
2625 * Most configurable timeouts in systemd now expect an argument of
2626 "infinity" to turn them off, instead of "0" as before. The semantics
2627 from now on is that a timeout of "0" means "now", and "infinity"
2628 means "never". To maintain backwards compatibility, "0" continues to
2629 turn off previously existing timeout settings.
2631 * "systemctl reload-or-try-restart" has been renamed to "systemctl
2632 try-reload-or-restart" to clarify what it actually does: the "try"
2633 logic applies to both reloading and restarting, not just restarting.
2634 The old name continues to be accepted for compatibility.
2636 * On boot-up, when PID 1 detects that the system clock is behind the
2637 release date of the systemd version in use, the clock is now set
2638 to the latter. Previously, this was already done in timesyncd, in order
2639 to avoid running with clocks set to the various clock epochs such as
2640 1902, 1938 or 1970. With this change the logic is now done in PID 1
2641 in addition to timesyncd during early boot-up, so that it is enforced
2642 before the first process is spawned by systemd. Note that the logic
2643 in timesyncd remains, as it is more comprehensive and ensures
2644 clock monotonicity by maintaining a persistent timestamp file in
2645 /var. Since /var is generally not available in earliest boot or the
2646 initrd, this part of the logic remains in timesyncd, and is not done
2649 * Support for tweaking details in net_cls.class_id through the
2650 NetClass= configuration directive has been removed, as the kernel
2651 people have decided to deprecate that controller in cgroup v2.
2652 Userspace tools such as nftables are moving over to setting rules
2653 that are specific to the full cgroup path of a task, which obsoletes
2654 these controllers anyway. The NetClass= directive is kept around for
2655 legacy compatibility reasons. For a more in-depth description of the
2656 kernel change, please refer to the respective upstream commit:
2658 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
2660 * A new service setting RuntimeMaxSec= has been added that may be used
2661 to specify a maximum runtime for a service. If the timeout is hit, the
2662 service is terminated and put into a failure state.
2664 * A new service setting AmbientCapabilities= has been added. It allows
2665 configuration of additional Linux process capabilities that are
2666 passed to the activated processes. This is only available on very
2669 * The process resource limit settings in service units may now be used
2670 to configure hard and soft limits individually.
2672 * The various libsystemd APIs such as sd-bus or sd-event now publicly
2673 expose support for gcc's __attribute__((cleanup())) C extension.
2674 Specifically, for many object destructor functions alternative
2675 versions have been added that have names suffixed with "p" and take a
2676 pointer to a pointer to the object to destroy, instead of just a
2677 pointer to the object itself. This is useful because these destructor
2678 functions may be used directly as parameters to the cleanup
2679 construct. Internally, systemd has been a heavy user of this GCC
2680 extension for a long time, and with this change similar support is
2681 now available to consumers of the library outside of systemd. Note
2682 that by using this extension in your sources compatibility with old
2683 and strictly ANSI compatible C compilers is lost. However, all gcc or
2684 LLVM versions of recent years support this extension.
2686 * Timer units gained support for a new setting RandomizedDelaySec= that
2687 allows configuring some additional randomized delay to the configured
2688 time. This is useful to spread out timer events to avoid load peaks in
2689 clusters or larger setups.
2691 * Calendar time specifications now support sub-second accuracy.
2693 * Socket units now support listening on SCTP and UDP-lite protocol
2696 * The sd-event API now comes with a full set of man pages.
2698 * Older versions of systemd contained experimental support for
2699 compressing journal files and coredumps with the LZ4 compressor that
2700 was not compatible with the lz4 binary (due to API limitations of the
2701 lz4 library). This support has been removed; only support for files
2702 compatible with the lz4 binary remains. This LZ4 logic is now
2703 officially supported and no longer considered experimental.
2705 * The dkr image import logic has been removed again from importd. dkr's
2706 micro-services focus doesn't fit into the machine image focus of
2707 importd, and quickly got out of date with the upstream dkr API.
2709 * Creation of the /run/lock/lockdev/ directory was dropped from
2710 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
2711 been available for many years. If you still need this, you need to
2712 create your own tmpfiles.d config file with:
2714 d /run/lock/lockdev 0775 root lock -
2716 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
2717 and RebootArgument= have been moved from the [Service] section of
2718 unit files to [Unit], and they are now supported on all unit types,
2719 not just service units. Of course, elogind will continue to
2720 understand these settings also at the old location, in order to
2721 maintain compatibility.
2723 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
2724 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
2725 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
2726 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
2727 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
2728 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
2729 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
2730 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
2731 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
2732 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
2733 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
2734 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
2735 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
2736 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
2737 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
2738 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
2739 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
2740 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
2741 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2743 — Berlin, 2016-02-11
2747 * A number of properties previously only settable in unit
2748 files are now also available as properties to set when
2749 creating transient units programmatically via the bus, as it
2750 is exposed with systemd-run's --property=
2751 setting. Specifically, these are: SyslogIdentifier=,
2752 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
2753 EnvironmentFile=, ReadWriteDirectories=,
2754 ReadOnlyDirectories=, InaccessibleDirectories=,
2755 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
2757 * When creating transient services via the bus API it is now
2758 possible to pass in a set of file descriptors to use as
2759 STDIN/STDOUT/STDERR for the invoked process.
2761 * Slice units may now be created transiently via the bus APIs,
2762 similar to the way service and scope units may already be
2763 created transiently.
2765 * Wherever systemd expects a calendar timestamp specification
2766 (like in journalctl's --since= and --until= switches) UTC
2767 timestamps are now supported. Timestamps suffixed with "UTC"
2768 are now considered to be in Universal Time Coordinated
2769 instead of the local timezone. Also, timestamps may now
2770 optionally be specified with sub-second accuracy. Both of
2771 these additions also apply to recurring calendar event
2772 specification, such as OnCalendar= in timer units.
2774 * journalctl gained a new "--sync" switch that asks the
2775 journal daemon to write all so far unwritten log messages to
2776 disk and sync the files, before returning.
2778 * systemd-tmpfiles learned two new line types "q" and "Q" that
2779 operate like "v", but also set up a basic btrfs quota
2780 hierarchy when used on a btrfs file system with quota
2783 * tmpfiles' "v", "q" and "Q" will now create a plain directory
2784 instead of a subvolume (even on a btrfs file system) if the
2785 root directory is a plain directory, and not a
2786 subvolume. This should simplify things with certain chroot()
2787 environments which are not aware of the concept of btrfs
2790 * systemd-detect-virt gained a new --chroot switch to detect
2791 whether execution takes place in a chroot() environment.
2793 * CPUAffinity= now takes CPU index ranges in addition to
2796 * The various memory-related resource limit settings (such as
2797 LimitAS=) now understand the usual K, M, G, ... suffixes to
2798 the base of 1024 (IEC). Similar, the time-related resource
2799 limit settings understand the usual min, h, day, ...
2802 * There's a new system.conf setting DefaultTasksMax= to
2803 control the default TasksMax= setting for services and
2804 scopes running on the system. (TasksMax= is the primary
2805 setting that exposes the "pids" cgroup controller on systemd
2806 and was introduced in the previous systemd release.) The
2807 setting now defaults to 512, which means services that are
2808 not explicitly configured otherwise will only be able to
2809 create 512 processes or threads at maximum, from this
2810 version on. Note that this means that thread- or
2811 process-heavy services might need to be reconfigured to set
2812 TasksMax= to a higher value. It is sufficient to set
2813 TasksMax= in these specific unit files to a higher value, or
2814 even "infinity". Similar, there's now a logind.conf setting
2815 UserTasksMax= that defaults to 4096 and limits the total
2816 number of processes or tasks each user may own
2817 concurrently. nspawn containers also have the TasksMax=
2818 value set by default now, to 8192. Note that all of this
2819 only has an effect if the "pids" cgroup controller is
2820 enabled in the kernel. The general benefit of these changes
2821 should be a more robust and safer system, that provides a
2822 certain amount of per-service fork() bomb protection.
2824 * systemd-nspawn gained the new --network-veth-extra= switch
2825 to define additional and arbitrarily-named virtual Ethernet
2826 links between the host and the container.
2828 * A new service execution setting PassEnvironment= has been
2829 added that allows importing select environment variables
2830 from PID1's environment block into the environment block of
2833 * Timer units gained support for a new RemainAfterElapse=
2834 setting which takes a boolean argument. It defaults to on,
2835 exposing behaviour unchanged to previous releases. If set to
2836 off, timer units are unloaded after they elapsed if they
2837 cannot elapse again. This is particularly useful for
2838 transient timer units, which shall not stay around longer
2839 than until they first elapse.
2841 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
2842 default now (the kernel default is 16). This is beneficial
2843 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
2844 allows substantially larger numbers of queued
2845 datagrams. This should increase the capability of systemd to
2846 parallelize boot-up, as logging and sd_notify() are unlikely
2847 to stall execution anymore. If you need to change the value
2848 from the new defaults, use the usual sysctl.d/ snippets.
2850 * The compression framing format used by the journal or
2851 coredump processing has changed to be in line with what the
2852 official LZ4 tools generate. LZ4 compression support in
2853 systemd was considered unsupported previously, as the format
2854 was not compatible with the normal tools. With this release
2855 this has changed now, and it is hence safe for downstream
2856 distributions to turn it on. While not compressing as well
2857 as the XZ, LZ4 is substantially faster, which makes
2858 it a good default choice for the compression logic in the
2859 journal and in coredump handling.
2861 * Any reference to /etc/mtab has been dropped from
2862 systemd. The file has been obsolete since a while, but
2863 systemd refused to work on systems where it was incorrectly
2864 set up (it should be a symlink or non-existent). Please make
2865 sure to update to util-linux 2.27.1 or newer in conjunction
2866 with this systemd release, which also drops any reference to
2867 /etc/mtab. If you maintain a distribution make sure that no
2868 software you package still references it, as this is a
2869 likely source of bugs. There's also a glibc bug pending,
2870 asking for removal of any reference to this obsolete file:
2872 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
2874 Note that only util-linux versions built with
2875 --enable-libmount-force-mountinfo are supported.
2877 * Support for the ".snapshot" unit type has been removed. This
2878 feature turned out to be little useful and little used, and
2879 has now been removed from the core and from systemctl.
2881 * The dependency types RequiresOverridable= and
2882 RequisiteOverridable= have been removed from systemd. They
2883 have been used only very sparingly to our knowledge and
2884 other options that provide a similar effect (such as
2885 systemctl --mode=ignore-dependencies) are much more useful
2886 and commonly used. Moreover, they were only half-way
2887 implemented as the option to control behaviour regarding
2888 these dependencies was never added to systemctl. By removing
2889 these dependency types the execution engine becomes a bit
2890 simpler. Unit files that use these dependencies should be
2891 changed to use the non-Overridable dependency types
2892 instead. In fact, when parsing unit files with these
2893 options, that's what systemd will automatically convert them
2894 too, but it will also warn, asking users to fix the unit
2895 files accordingly. Removal of these dependency types should
2896 only affect a negligible number of unit files in the wild.
2898 * Behaviour of networkd's IPForward= option changed
2899 (again). It will no longer maintain a per-interface setting,
2900 but propagate one way from interfaces where this is enabled
2901 to the global kernel setting. The global setting will be
2902 enabled when requested by a network that is set up, but
2903 never be disabled again. This change was made to make sure
2904 IPv4 and IPv6 behaviour regarding packet forwarding is
2905 similar (as the Linux IPv6 stack does not support
2906 per-interface control of this setting) and to minimize
2909 * In unit files the behaviour of %u, %U, %h, %s has
2910 changed. These specifiers will now unconditionally resolve
2911 to the various user database fields of the user that the
2912 systemd instance is running as, instead of the user
2913 configured in the specific unit via User=. Note that this
2914 effectively doesn't change much, as resolving of these
2915 specifiers was already turned off in the --system instance
2916 of systemd, as we cannot do NSS lookups from PID 1. In the
2917 --user instance of systemd these specifiers where correctly
2918 resolved, but hardly made any sense, since the user instance
2919 lacks privileges to do user switches anyway, and User= is
2920 hence useless. Moreover, even in the --user instance of
2921 systemd behaviour was awkward as it would only take settings
2922 from User= assignment placed before the specifier into
2923 account. In order to unify and simplify the logic around
2924 this the specifiers will now always resolve to the
2925 credentials of the user invoking the manager (which in case
2926 of PID 1 is the root user).
2928 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
2929 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
2930 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
2931 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
2932 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
2933 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
2934 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
2935 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
2936 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
2937 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
2938 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
2939 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
2940 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
2941 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
2944 — Berlin, 2015-11-18
2948 * systemd now depends on util-linux v2.27. More specifically,
2949 the newly added mount monitor feature in libmount now
2950 replaces systemd's former own implementation.
2952 * libmount mandates /etc/mtab not to be regular file, and
2953 systemd now enforces this condition at early boot.
2954 /etc/mtab has been deprecated and warned about for a very
2955 long time, so systems running systemd should already have
2956 stopped having this file around as anything else than a
2957 symlink to /proc/self/mounts.
2959 * Support for the "pids" cgroup controller has been added. It
2960 allows accounting the number of tasks in a cgroup and
2961 enforcing limits on it. This adds two new setting
2962 TasksAccounting= and TasksMax= to each unit, as well as a
2963 global option DefaultTasksAccounting=.
2965 * Support for the "net_cls" cgroup controller has been added.
2966 It allows assigning a net class ID to each task in the
2967 cgroup, which can then be used in firewall rules and traffic
2968 shaping configurations. Note that the kernel netfilter net
2969 class code does not currently work reliably for ingress
2970 packets on unestablished sockets.
2972 This adds a new config directive called NetClass= to CGroup
2973 enabled units. Allowed values are positive numbers for fixed
2974 assignments and "auto" for picking a free value
2977 * 'systemctl is-system-running' now returns 'offline' if the
2978 system is not booted with systemd. This command can now be
2979 used as a substitute for 'systemd-notify --booted'.
2981 * Watchdog timeouts have been increased to 3 minutes for all
2982 in-tree service files. Apparently, disk IO issues are more
2983 frequent than we hoped, and user reported >1 minute waiting
2986 * 'machine-id-commit' functionality has been merged into
2987 'machine-id-setup --commit'. The separate binary has been
2990 * The WorkingDirectory= directive in unit files may now be set
2991 to the special value '~'. In this case, the working
2992 directory is set to the home directory of the user
2993 configured in User=.
2995 * "machinectl shell" will now open the shell in the home
2996 directory of the selected user by default.
2998 * The CrashChVT= configuration file setting is renamed to
2999 CrashChangeVT=, following our usual logic of not
3000 abbreviating unnecessarily. The old directive is still
3001 supported for compat reasons. Also, this directive now takes
3002 an integer value between 1 and 63, or a boolean value. The
3003 formerly supported '-1' value for disabling stays around for
3006 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
3007 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
3008 RootDirectory= properties can now be set for transient
3011 * The systemd-analyze tool gained a new "set-log-target" verb
3012 to change the logging target the system manager logs to
3013 dynamically during runtime. This is similar to how
3014 "systemd-analyze set-log-level" already changes the log
3017 * In nspawn /sys is now mounted as tmpfs, with only a selected
3018 set of subdirectories mounted in from the real sysfs. This
3019 enhances security slightly, and is useful for ensuring user
3020 namespaces work correctly.
3022 * Support for USB FunctionFS activation has been added. This
3023 allows implementation of USB gadget services that are
3024 activated as soon as they are requested, so that they don't
3025 have to run continuously, similar to classic socket
3028 * The "systemctl exit" command now optionally takes an
3029 additional parameter that sets the exit code to return from
3030 the systemd manager when exiting. This is only relevant when
3031 running the systemd user instance, or when running the
3032 system instance in a container.
3034 * sd-bus gained the new API calls sd_bus_path_encode_many()
3035 and sd_bus_path_decode_many() that allow easy encoding and
3036 decoding of multiple identifier strings inside a D-Bus
3037 object path. Another new call sd_bus_default_flush_close()
3038 has been added to flush and close per-thread default
3041 * systemd-cgtop gained support for a -M/--machine= switch to
3042 show the control groups within a certain container only.
3044 * "systemctl kill" gained support for an optional --fail
3045 switch. If specified the requested operation will fail of no
3046 processes have been killed, because the unit had no
3047 processes attached, or similar.
3049 * A new systemd.crash_reboot=1 kernel command line option has
3050 been added that triggers a reboot after crashing. This can
3051 also be set through CrashReboot= in systemd.conf.
3053 * The RuntimeDirectory= setting now understands unit
3054 specifiers like %i or %f.
3056 * A new (still internal) library API sd-ipv4acd has been added,
3057 that implements address conflict detection for IPv4. It's
3058 based on code from sd-ipv4ll, and will be useful for
3059 detecting DHCP address conflicts.
3061 * File descriptors passed during socket activation may now be
3062 named. A new API sd_listen_fds_with_names() is added to
3063 access the names. The default names may be overridden,
3064 either in the .socket file using the FileDescriptorName=
3065 parameter, or by passing FDNAME= when storing the file
3066 descriptors using sd_notify().
3068 * systemd-networkd gained support for:
3070 - Setting the IPv6 Router Advertisement settings via
3071 IPv6AcceptRouterAdvertisements= in .network files.
3073 - Configuring the HelloTimeSec=, MaxAgeSec= and
3074 ForwardDelaySec= bridge parameters in .netdev files.
3076 - Configuring PreferredSource= for static routes in
3079 * The "ask-password" framework used to query for LUKS harddisk
3080 passwords or SSL passwords during boot gained support for
3081 caching passwords in the kernel keyring, if it is
3082 available. This makes sure that the user only has to type in
3083 a passphrase once if there are multiple objects to unlock
3084 with the same one. Previously, such password caching was
3085 available only when Plymouth was used; this moves the
3086 caching logic into the systemd codebase itself. The
3087 "systemd-ask-password" utility gained a new --keyname=
3088 switch to control which kernel keyring key to use for
3089 caching a password in. This functionality is also useful for
3090 enabling display managers such as gdm to automatically
3091 unlock the user's GNOME keyring if its passphrase, the
3092 user's password and the harddisk password are the same, if
3093 gdm-autologin is used.
3095 * When downloading tar or raw images using "machinectl
3096 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
3097 file is now also downloaded, if it is available and stored
3098 next to the image file.
3100 * Units of type ".socket" gained a new boolean setting
3101 Writable= which is only useful in conjunction with
3102 ListenSpecial=. If true, enables opening the specified
3103 special file in O_RDWR mode rather than O_RDONLY mode.
3105 * systemd-rfkill has been reworked to become a singleton
3106 service that is activated through /dev/rfkill on each rfkill
3107 state change and saves the settings to disk. This way,
3108 systemd-rfkill is now compatible with devices that exist
3109 only intermittendly, and even restores state if the previous
3110 system shutdown was abrupt rather than clean.
3112 * The journal daemon gained support for vacuuming old journal
3113 files controlled by the number of files that shall remain,
3114 in addition to the already existing control by size and by
3115 date. This is useful as journal interleaving performance
3116 degrades with too many separate journal files, and allows
3117 putting an effective limit on them. The new setting defaults
3118 to 100, but this may be changed by setting SystemMaxFiles=
3119 and RuntimeMaxFiles= in journald.conf. Also, the
3120 "journalctl" tool gained the new --vacuum-files= switch to
3121 manually vacuum journal files to leave only the specified
3122 number of files in place.
3124 * udev will now create /dev/disk/by-path links for ATA devices
3125 on kernels where that is supported.
3127 * Galician, Serbian, Turkish and Korean translations were added.
3129 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
3130 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
3131 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
3132 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
3133 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
3134 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
3135 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
3136 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
3137 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
3138 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
3139 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
3140 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
3141 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
3142 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
3143 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
3144 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
3145 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
3146 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
3148 — Berlin, 2015-10-07
3152 * The DHCP implementation of systemd-networkd gained a set of
3155 - The DHCP server now supports emitting DNS and NTP
3156 information. It may be enabled and configured via
3157 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
3158 and NTP information is enabled, but no servers are
3159 configured, the corresponding uplink information (if there
3160 is any) is propagated.
3162 - Server and client now support transmission and reception
3163 of timezone information. It can be configured via the
3164 newly introduced network options UseTimezone=,
3165 EmitTimezone=, and Timezone=. Transmission of timezone
3166 information is enabled between host and containers by
3167 default now: the container will change its local timezone
3168 to what the host has set.
3170 - Lease timeouts can now be configured via
3171 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
3173 - The DHCP server improved on the stability of
3174 leases. Clients are more likely to get the same lease
3175 information back, even if the server loses state.
3177 - The DHCP server supports two new configuration options to
3178 control the lease address pool metrics, PoolOffset= and
3181 * The encapsulation limit of tunnels in systemd-networkd may
3182 now be configured via 'EncapsulationLimit='. It allows
3183 modifying the maximum additional levels of encapsulation
3184 that are permitted to be prepended to a packet.
3186 * systemd now supports the concept of user buses replacing
3187 session buses, if used with dbus-1.10 (and enabled via dbus
3188 --enable-user-session). It previously only supported this on
3189 kdbus-enabled systems, and this release expands this to
3190 'dbus-daemon' systems.
3192 * systemd-networkd now supports predictable interface names
3195 * systemd now optionally supports the new Linux kernel
3196 "unified" control group hierarchy. If enabled via the kernel
3197 command-line option 'systemd.unified_cgroup_hierarchy=1',
3198 systemd will try to mount the unified cgroup hierarchy
3199 directly on /sys/fs/cgroup. If not enabled, or not
3200 available, systemd will fall back to the legacy cgroup
3201 hierarchy setup, as before. Host system and containers can
3202 mix and match legacy and unified hierarchies as they
3203 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
3204 environment variable to individually select the hierarchy to
3205 use for executed containers. By default, nspawn will use the
3206 unified hierarchy for the containers if the host uses the
3207 unified hierarchy, and the legacy hierarchy otherwise.
3208 Please note that at this point the unified hierarchy is an
3209 experimental kernel feature and is likely to change in one
3210 of the next kernel releases. Therefore, it should not be
3211 enabled by default in downstream distributions yet. The
3212 minimum required kernel version for the unified hierarchy to
3213 work is 4.2. Note that when the unified hierarchy is used
3214 for the first time delegated access to controllers is
3215 safe. Because of this systemd-nspawn containers will get
3216 access to controllers now, as will systemd user
3217 sessions. This means containers and user sessions may now
3218 manage their own resources, partitioning up what the system
3221 * A new special scope unit "init.scope" has been introduced
3222 that encapsulates PID 1 of the system. It may be used to
3223 determine resource usage and enforce resource limits on PID
3224 1 itself. PID 1 hence moved out of the root of the control
3227 * The cgtop tool gained support for filtering out kernel
3228 threads when counting tasks in a control group. Also, the
3229 count of processes is now recursively summed up by
3230 default. Two options -k and --recursive= have been added to
3231 revert to old behaviour. The tool has also been updated to
3232 work correctly in containers now.
3234 * systemd-nspawn's --bind= and --bind-ro= options have been
3235 extended to allow creation of non-recursive bind mounts.
3237 * libsystemd gained two new calls sd_pid_get_cgroup() and
3238 sd_peer_get_cgroup() which return the control group path of
3239 a process or peer of a connected AF_UNIX socket. This
3240 function call is particularly useful when implementing
3241 delegated subtrees support in the control group hierarchy.
3243 * The "sd-event" event loop API of libsystemd now supports
3244 correct dequeuing of real-time signals, without losing
3247 * When systemd requests a PolicyKit decision when managing
3248 units it will now add additional fields to the request,
3249 including unit name and desired operation. This enables more
3250 powerful PolicyKit policies, that make decisions depending
3251 on these parameters.
3253 * nspawn learnt support for .nspawn settings files, that may
3254 accompany the image files or directories of containers, and
3255 may contain additional settings for the container. This is
3256 an alternative to configuring container parameters via the
3257 nspawn command line.
3259 Contributions from: Cristian Rodríguez, Daniel Mack, David
3260 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
3261 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
3262 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
3263 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
3264 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
3265 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
3266 Andersen, Tom Gundersen, Torstein Husebø
3268 — Berlin, 2015-09-08
3272 * machinectl gained a new verb 'shell' which opens a fresh
3273 shell on the target container or the host. It is similar to
3274 the existing 'login' command of machinectl, but spawns the
3275 shell directly without prompting for username or
3276 password. The pseudo machine '.host' now refers to the local
3277 host and is used by default. Hence, 'machinectl shell' can
3278 be used as replacement for 'su -' which spawns a session as
3279 a fresh systemd unit in a way that is fully isolated from
3280 the originating session.
3282 * systemd-networkd learned to cope with private-zone DHCP
3283 options and allows other programs to query the values.
3285 * SELinux access control when enabling/disabling units is no
3286 longer enforced with this release. The previous
3287 implementation was incorrect, and a new corrected
3288 implementation is not yet available. As unit file operations
3289 are still protected via PolicyKit and D-Bus policy this is
3290 not a security problem. Yet, distributions which care about
3291 optimal SELinux support should probably not stabilize on
3294 * sd-bus gained support for matches of type "arg0has=", that
3295 test for membership of strings in string arrays sent in bus
3298 * systemd-resolved now dumps the contents of its DNS and LLMNR
3299 caches to the logs on reception of the SIGUSR1 signal. This
3300 is useful to debug DNS behaviour.
3302 * The coredumpctl tool gained a new --directory= option to
3303 operate on journal files in a specific directory.
3305 * "systemctl reboot" and related commands gained a new
3306 "--message=" option which may be used to set a free-text
3307 wall message when shutting down or rebooting the
3308 system. This message is also logged, which is useful for
3309 figuring out the reason for a reboot or shutdown a
3312 * The "systemd-resolve-host" tool's -i switch now takes
3313 network interface numbers as alternative to interface names.
3315 * A new unit file setting for services has been introduced:
3316 UtmpMode= allows configuration of how precisely systemd
3317 handles utmp and wtmp entries for the service if this is
3318 enabled. This allows writing services that appear similar to
3319 user sessions in the output of the "w", "who", "last" and
3322 * systemd-resolved will now locally synthesize DNS resource
3323 records for the "localhost" and "gateway" domains as well as
3324 the local hostname. This should ensure that clients querying
3325 RRs via resolved will get similar results as those going via
3326 NSS, if nss-myhostname is enabled.
3328 Contributions from: Alastair Hughes, Alex Crawford, Daniel
3329 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
3330 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
3331 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
3332 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
3333 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
3334 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
3335 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
3336 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
3337 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
3338 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
3339 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3341 — Berlin, 2015-08-27
3345 * The systemd-efi-boot-generator functionality was merged into
3346 systemd-gpt-auto-generator.
3348 * systemd-networkd now supports Group Policy for vxlan
3349 devices. It can be enabled via the new boolean configuration
3350 option called 'GroupPolicyExtension='.
3352 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
3353 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
3354 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
3356 — Berlin, 2015-07-31
3360 * The python-systemd code has been removed from the systemd repository.
3361 A new repository has been created which accommodates the code from
3362 now on, and we kindly ask distributions to create a separate package
3363 for this: https://github.com/systemd/python-systemd
3365 * The systemd daemon will now reload its main configuration
3366 (/etc/systemd/system.conf) on daemon-reload.
3368 * sd-dhcp now exposes vendor specific extensions via
3369 sd_dhcp_lease_get_vendor_specific().
3371 * systemd-networkd gained a number of new configuration options.
3373 - A new boolean configuration option for TAP devices called
3374 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
3375 device, thus allowing to send and receive GSO packets.
3377 - A new tunnel configuration option called 'CopyDSCP='.
3378 If enabled, the DSCP field of ip6 tunnels is copied into the
3379 decapsulated packet.
3381 - A set of boolean bridge configuration options were added.
3382 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
3383 and 'UnicastFlood=' are now parsed by networkd and applied to the
3384 respective bridge link device via the respective IFLA_BRPORT_*
3387 - A new string configuration option to override the hostname sent
3388 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
3389 is true, networkd will use the configured hostname instead of the
3390 system hostname when sending DHCP requests.
3392 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
3393 networkd will configure the IPv6 flow-label of the tunnel device
3394 according to RFC2460.
3396 - The 'macvtap' virtual network devices are now supported, similar to
3397 the already supported 'macvlan' devices.
3399 * systemd-resolved now implements RFC5452 to improve resilience against
3400 cache poisoning. Additionally, source port randomization is enabled
3401 by default to further protect against DNS spoofing attacks.
3403 * nss-mymachines now supports translating UIDs and GIDs of running
3404 containers with user-namespaces enabled. If a container 'foo'
3405 translates a host uid 'UID' to the container uid 'TUID', then
3406 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
3407 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
3408 mapped as 'vg-foo-TGID'.
3410 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
3411 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
3412 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
3413 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
3414 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
3415 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
3416 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
3417 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
3418 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
3419 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
3421 — Berlin, 2015-07-29
3425 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
3426 There are no known issues with current sysfs, and udev does not need
3427 or should be used to work around such bugs.
3429 * udev does no longer enable USB HID power management. Several reports
3430 indicate, that some devices cannot handle that setting.
3432 * The udev accelerometer helper was removed. The functionality
3433 is now fully included in iio-sensor-proxy. But this means,
3434 older iio-sensor-proxy versions will no longer provide
3435 accelerometer/orientation data with this systemd version.
3436 Please upgrade iio-sensor-proxy to version 1.0.
3438 * networkd gained a new configuration option IPv6PrivacyExtensions=
3439 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
3440 for Stateless Address") on selected networks.
3442 * For the sake of fewer build-time dependencies and less code in the
3443 main repository, the python bindings are about to be removed in the
3444 next release. A new repository has been created which accommodates
3445 the code from now on, and we kindly ask distributions to create a
3446 separate package for this. The removal will take place in v223.
3448 https://github.com/systemd/python-systemd
3450 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
3451 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
3452 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
3453 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
3454 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
3455 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
3456 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
3457 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
3458 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3459 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3461 — Berlin, 2015-07-07
3465 * The sd-bus.h and sd-event.h APIs have now been declared
3466 stable and have been added to the official interface of
3467 libsystemd.so. sd-bus implements an alternative D-Bus client
3468 library, that is relatively easy to use, very efficient and
3469 supports both classic D-Bus as well as kdbus as transport
3470 backend. sd-event is a generic event loop abstraction that
3471 is built around Linux epoll, but adds features such as event
3472 prioritization or efficient timer handling. Both APIs are good
3473 choices for C programs looking for a bus and/or event loop
3474 implementation that is minimal and does not have to be
3475 portable to other kernels.
3477 * kdbus support is no longer compile-time optional. It is now
3478 always built-in. However, it can still be disabled at
3479 runtime using the kdbus=0 kernel command line setting, and
3480 that setting may be changed to default to off, by specifying
3481 --disable-kdbus at build-time. Note though that the kernel
3482 command line setting has no effect if the kdbus.ko kernel
3483 module is not installed, in which case kdbus is (obviously)
3484 also disabled. We encourage all downstream distributions to
3485 begin testing kdbus by adding it to the kernel images in the
3486 development distributions, and leaving kdbus support in
3489 * The minimal required util-linux version has been bumped to
3492 * Support for chkconfig (--enable-chkconfig) was removed in
3493 favor of calling an abstraction tool
3494 /lib/systemd/systemd-sysv-install. This needs to be
3495 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
3496 in README for details.
3498 * If there's a systemd unit and a SysV init script for the
3499 same service name, and the user executes "systemctl enable"
3500 for it (or a related call), then this will now enable both
3501 (or execute the related operation on both), not just the
3504 * The libudev API documentation has been converted from gtkdoc
3507 * gudev has been removed from the systemd tree, it is now an
3510 * The systemd-cgtop tool learnt a new --raw switch to generate
3511 "raw" (machine parsable) output.
3513 * networkd's IPForwarding= .network file setting learnt the
3514 new setting "kernel", which ensures that networkd does not
3515 change the IP forwarding sysctl from the default kernel
3518 * The systemd-logind bus API now exposes a new boolean
3519 property "Docked" that reports whether logind considers the
3520 system "docked", i.e. connected to a docking station or not.
3522 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
3523 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
3524 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
3525 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
3526 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
3527 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
3528 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
3529 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
3530 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
3531 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
3532 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
3533 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
3534 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
3535 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3536 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
3537 Fink, Zbigniew Jędrzejewski-Szmek
3539 — Berlin, 2015-06-19
3543 * The gudev library has been extracted into a separate repository
3544 available at: https://git.gnome.org/browse/libgudev/
3545 It is now managed as part of the Gnome project. Distributions
3546 are recommended to pass --disable-gudev to systemd and use
3547 gudev from the Gnome project instead. gudev is still included
3548 in systemd, for now. It will be removed soon, though. Please
3549 also see the announcement-thread on systemd-devel:
3550 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
3552 * systemd now exposes a CPUUsageNSec= property for each
3553 service unit on the bus, that contains the overall consumed
3554 CPU time of a service (the sum of what each process of the
3555 service consumed). This value is only available if
3556 CPUAccounting= is turned on for a service, and is then shown
3557 in the "systemctl status" output.
3559 * Support for configuring alternative mappings of the old SysV
3560 runlevels to systemd targets has been removed. They are now
3561 hardcoded in a way that runlevels 2, 3, 4 all map to
3562 multi-user.target and 5 to graphical.target (which
3563 previously was already the default behaviour).
3565 * The auto-mounter logic gained support for mount point
3566 expiry, using a new TimeoutIdleSec= setting in .automount
3567 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
3569 * The EFI System Partition (ESP) as mounted to /boot by
3570 systemd-efi-boot-generator will now be unmounted
3571 automatically after 2 minutes of not being used. This should
3572 minimize the risk of ESP corruptions.
3574 * New /etc/fstab options x-systemd.requires= and
3575 x-systemd.requires-mounts-for= are now supported to express
3576 additional dependencies for mounts. This is useful for
3577 journalling file systems that support external journal
3578 devices or overlay file systems that require underlying file
3579 systems to be mounted.
3581 * systemd does not support direct live-upgrades (via systemctl
3582 daemon-reexec) from versions older than v44 anymore. As no
3583 distribution we are aware of shipped such old versions in a
3584 stable release this should not be problematic.
3586 * When systemd forks off a new per-connection service instance
3587 it will now set the $REMOTE_ADDR environment variable to the
3588 remote IP address, and $REMOTE_PORT environment variable to
3589 the remote IP port. This behaviour is similar to the
3590 corresponding environment variables defined by CGI.
3592 * systemd-networkd gained support for uplink failure
3593 detection. The BindCarrier= option allows binding interface
3594 configuration dynamically to the link sense of other
3595 interfaces. This is useful to achieve behaviour like in
3598 * systemd-networkd gained support for configuring the DHCP
3599 client identifier to use when requesting leases.
3601 * systemd-networkd now has a per-network UseNTP= option to
3602 configure whether NTP server information acquired via DHCP
3603 is passed on to services like systemd-timesyncd.
3605 * systemd-networkd gained support for vti6 tunnels.
3607 * Note that systemd-networkd manages the sysctl variable
3608 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
3609 it is configured for since v219. The variable controls IP
3610 forwarding, and is a per-interface alternative to the global
3611 /proc/sys/net/ipv[46]/ip_forward. This setting is
3612 configurable in the IPForward= option, which defaults to
3613 "no". This means if networkd is used for an interface it is
3614 no longer sufficient to set the global sysctl option to turn
3615 on IP forwarding! Instead, the .network file option
3616 IPForward= needs to be turned on! Note that the
3617 implementation of this behaviour was broken in v219 and has
3620 * Many bonding and vxlan options are now configurable in
3623 * systemd-nspawn gained a new --property= setting to set unit
3624 properties for the container scope. This is useful for
3625 setting resource parameters (e.g. "CPUShares=500") on
3626 containers started from the command line.
3628 * systemd-nspawn gained a new --private-users= switch to make
3629 use of user namespacing available on recent Linux kernels.
3631 * systemd-nspawn may now be called as part of a shell pipeline
3632 in which case the pipes used for stdin and stdout are passed
3633 directly to the process invoked in the container, without
3634 indirection via a pseudo tty.
3636 * systemd-nspawn gained a new switch to control the UNIX
3637 signal to use when killing the init process of the container
3640 * systemd-nspawn gained a new --overlay= switch for mounting
3641 overlay file systems into the container using the new kernel
3644 * When a container image is imported via systemd-importd and
3645 the host file system is not btrfs, a loopback block device
3646 file is created in /var/lib/machines.raw with a btrfs file
3647 system inside. It is then mounted to /var/lib/machines to
3648 enable btrfs features for container management. The loopback
3649 file and btrfs file system is grown as needed when container
3650 images are imported via systemd-importd.
3652 * systemd-machined/systemd-importd gained support for btrfs
3653 quota, to enforce container disk space limits on disk. This
3654 is exposed in "machinectl set-limit".
3656 * systemd-importd now can import containers from local .tar,
3657 .raw and .qcow2 images, and export them to .tar and .raw. It
3658 can also import dkr v2 images now from the network (on top
3661 * systemd-importd gained support for verifying downloaded
3662 images with gpg2 (previously only gpg1 was supported).
3664 * systemd-machined, systemd-logind, systemd: most bus calls
3665 are now accessible to unprivileged processes via
3666 PolicyKit. Also, systemd-logind will now allow users to kill
3667 their own sessions without further privileges or
3670 * systemd-shutdownd has been removed. This service was
3671 previously responsible for implementing scheduled shutdowns
3672 as exposed in /usr/bin/shutdown's time parameter. This
3673 functionality has now been moved into systemd-logind and is
3674 accessible via a bus interface.
3676 * "systemctl reboot" gained a new switch --firmware-setup that
3677 can be used to reboot into the EFI firmware setup, if that
3678 is available. systemd-logind now exposes an API on the bus
3679 to trigger such reboots, in case graphical desktop UIs want
3680 to cover this functionality.
3682 * "systemctl enable", "systemctl disable" and "systemctl mask"
3683 now support a new "--now" switch. If specified the units
3684 that are enabled will also be started, and the ones
3685 disabled/masked also stopped.
3687 * The Gummiboot EFI boot loader tool has been merged into
3688 systemd, and renamed to "systemd-boot". The bootctl tool has been
3689 updated to support systemd-boot.
3691 * An EFI kernel stub has been added that may be used to create
3692 kernel EFI binaries that contain not only the actual kernel,
3693 but also an initrd, boot splash, command line and OS release
3694 information. This combined binary can then be signed as a
3695 single image, so that the firmware can verify it all in one
3696 step. systemd-boot has special support for EFI binaries created
3697 like this and can extract OS release information from them
3698 and show them in the boot menu. This functionality is useful
3699 to implement cryptographically verified boot schemes.
3701 * Optional support has been added to systemd-fsck to pass
3702 fsck's progress report to an AF_UNIX socket in the file
3705 * udev will no longer create device symlinks for all block
3706 devices by default. A blacklist for excluding special block
3707 devices from this logic has been turned into a whitelist
3708 that requires picking block devices explicitly that require
3711 * A new (currently still internal) API sd-device.h has been
3712 added to libsystemd. This modernized API is supposed to
3713 replace libudev eventually. In fact, already much of libudev
3714 is now just a wrapper around sd-device.h.
3716 * A new hwdb database for storing metadata about pointing
3717 stick devices has been added.
3719 * systemd-tmpfiles gained support for setting file attributes
3720 similar to the "chattr" tool with new 'h' and 'H' lines.
3722 * systemd-journald will no longer unconditionally set the
3723 btrfs NOCOW flag on new journal files. This is instead done
3724 with tmpfiles snippet using the new 'h' line type. This
3725 allows easy disabling of this logic, by masking the
3726 journal-nocow.conf tmpfiles file.
3728 * systemd-journald will now translate audit message types to
3729 human readable identifiers when writing them to the
3730 journal. This should improve readability of audit messages.
3732 * The LUKS logic gained support for the offset= and skip=
3733 options in /etc/crypttab, as previously implemented by
3736 * /usr/lib/os-release gained a new optional field VARIANT= for
3737 distributions that support multiple variants (such as a
3738 desktop edition, a server edition, ...)
3740 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
3741 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
3742 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
3743 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
3744 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
3745 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
3746 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
3747 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
3748 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
3749 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
3750 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
3751 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
3752 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
3753 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
3754 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
3755 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
3756 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
3757 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
3758 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
3759 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
3760 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
3761 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
3762 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
3763 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
3764 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
3765 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
3766 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
3768 — Berlin, 2015-05-22
3772 * Introduce a new API "sd-hwdb.h" for querying the hardware
3773 metadata database. With this minimal interface one can query
3774 and enumerate the udev hwdb, decoupled from the old libudev
3775 library. libudev's interface for this is now only a wrapper
3776 around sd-hwdb. A new tool systemd-hwdb has been added to
3777 interface with and update the database.
3779 * When any of systemd's tools copies files (for example due to
3780 tmpfiles' C lines) a btrfs reflink will attempted first,
3781 before bytewise copying is done.
3783 * systemd-nspawn gained a new --ephemeral switch. When
3784 specified a btrfs snapshot is taken of the container's root
3785 directory, and immediately removed when the container
3786 terminates again. Thus, a container can be started whose
3787 changes never alter the container's root directory, and are
3788 lost on container termination. This switch can also be used
3789 for starting a container off the root file system of the
3790 host without affecting the host OS. This switch is only
3791 available on btrfs file systems.
3793 * systemd-nspawn gained a new --template= switch. It takes the
3794 path to a container tree to use as template for the tree
3795 specified via --directory=, should that directory be
3796 missing. This allows instantiating containers dynamically,
3797 on first run. This switch is only available on btrfs file
3800 * When a .mount unit refers to a mount point on which multiple
3801 mounts are stacked, and the .mount unit is stopped all of
3802 the stacked mount points will now be unmounted until no
3803 mount point remains.
3805 * systemd now has an explicit notion of supported and
3806 unsupported unit types. Jobs enqueued for unsupported unit
3807 types will now fail with an "unsupported" error code. More
3808 specifically .swap, .automount and .device units are not
3809 supported in containers, .busname units are not supported on
3810 non-kdbus systems. .swap and .automount are also not
3811 supported if their respective kernel compile time options
3814 * machinectl gained support for two new "copy-from" and
3815 "copy-to" commands for copying files from a running
3816 container to the host or vice versa.
3818 * machinectl gained support for a new "bind" command to bind
3819 mount host directories into local containers. This is
3820 currently only supported for nspawn containers.
3822 * networkd gained support for configuring bridge forwarding
3823 database entries (fdb) from .network files.
3825 * A new tiny daemon "systemd-importd" has been added that can
3826 download container images in tar, raw, qcow2 or dkr formats,
3827 and make them available locally in /var/lib/machines, so
3828 that they can run as nspawn containers. The daemon can GPG
3829 verify the downloads (not supported for dkr, since it has no
3830 provisions for verifying downloads). It will transparently
3831 decompress bz2, xz, gzip compressed downloads if necessary,
3832 and restore sparse files on disk. The daemon uses privilege
3833 separation to ensure the actual download logic runs with
3834 fewer privileges than the daemon itself. machinectl has
3835 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
3836 make the functionality of importd available to the
3837 user. With this in place the Fedora and Ubuntu "Cloud"
3838 images can be downloaded and booted as containers unmodified
3839 (the Fedora images lack the appropriate GPG signature files
3840 currently, so they cannot be verified, but this will change
3841 soon, hopefully). Note that downloading images is currently
3842 only fully supported on btrfs.
3844 * machinectl is now able to list container images found in
3845 /var/lib/machines, along with some metadata about sizes of
3846 disk and similar. If the directory is located on btrfs and
3847 quota is enabled, this includes quota display. A new command
3848 "image-status" has been added that shows additional
3849 information about images.
3851 * machinectl is now able to clone container images
3852 efficiently, if the underlying file system (btrfs) supports
3853 it, with the new "machinectl clone" command. It also
3854 gained commands for renaming and removing images, as well as
3855 marking them read-only or read-write (supported also on
3856 legacy file systems).
3858 * networkd gained support for collecting LLDP network
3859 announcements, from hardware that supports this. This is
3860 shown in networkctl output.
3862 * systemd-run gained support for a new -t (--pty) switch for
3863 invoking a binary on a pty whose input and output is
3864 connected to the invoking terminal. This allows executing
3865 processes as system services while interactively
3866 communicating with them via the terminal. Most interestingly
3867 this is supported across container boundaries. Invoking
3868 "systemd-run -t /bin/bash" is an alternative to running a
3869 full login session, the difference being that the former
3870 will not register a session, nor go through the PAM session
3873 * tmpfiles gained support for a new "v" line type for creating
3874 btrfs subvolumes. If the underlying file system is a legacy
3875 file system, this automatically degrades to creating a
3876 normal directory. Among others /var/lib/machines is now
3877 created like this at boot, should it be missing.
3879 * The directory /var/lib/containers/ has been deprecated and
3880 been replaced by /var/lib/machines. The term "machines" has
3881 been used in the systemd context as generic term for both
3882 VMs and containers, and hence appears more appropriate for
3883 this, as the directory can also contain raw images bootable
3886 * systemd-nspawn when invoked with -M but without --directory=
3887 or --image= is now capable of searching for the container
3888 root directory, subvolume or disk image automatically, in
3889 /var/lib/machines. systemd-nspawn@.service has been updated
3890 to make use of this, thus allowing it to be used for raw
3893 * A new machines.target unit has been introduced that is
3894 supposed to group all containers/VMs invoked as services on
3895 the system. systemd-nspawn@.service has been updated to
3896 integrate with that.
3898 * machinectl gained a new "start" command, for invoking a
3899 container as a service. "machinectl start foo" is mostly
3900 equivalent to "systemctl start systemd-nspawn@foo.service",
3901 but handles escaping in a nicer way.
3903 * systemd-nspawn will now mount most of the cgroupfs tree
3904 read-only into each container, with the exception of the
3905 container's own subtree in the name=systemd hierarchy.
3907 * journald now sets the special FS_NOCOW file flag for its
3908 journal files. This should improve performance on btrfs, by
3909 avoiding heavy fragmentation when journald's write-pattern
3910 is used on COW file systems. It degrades btrfs' data
3911 integrity guarantees for the files to the same levels as for
3912 ext3/ext4 however. This should be OK though as journald does
3913 its own data integrity checks and all its objects are
3914 checksummed on disk. Also, journald should handle btrfs disk
3915 full events a lot more gracefully now, by processing SIGBUS
3916 errors, and not relying on fallocate() anymore.
3918 * When journald detects that journal files it is writing to
3919 have been deleted it will immediately start new journal
3922 * systemd now provides a way to store file descriptors
3923 per-service in PID 1. This is useful for daemons to ensure
3924 that fds they require are not lost during a daemon
3925 restart. The fds are passed to the daemon on the next
3926 invocation in the same way socket activation fds are
3927 passed. This is now used by journald to ensure that the
3928 various sockets connected to all the system's stdout/stderr
3929 are not lost when journald is restarted. File descriptors
3930 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
3931 an extension to sd_notify(). Note that a limit is enforced
3932 on the number of fds a service can store in PID 1, and it
3933 defaults to 0, so that no fds may be stored, unless this is
3934 explicitly turned on.
3936 * The default TERM variable to use for units connected to a
3937 terminal, when no other value is explicitly is set is now
3938 vt220 rather than vt102. This should be fairly safe still,
3939 but allows PgUp/PgDn work.
3941 * The /etc/crypttab option header= as known from Debian is now
3944 * "loginctl user-status" and "loginctl session-status" will
3945 now show the last 10 lines of log messages of the
3946 user/session following the status output. Similar,
3947 "machinectl status" will show the last 10 log lines
3948 associated with a virtual machine or container
3949 service. (Note that this is usually not the log messages
3950 done in the VM/container itself, but simply what the
3951 container manager logs. For nspawn this includes all console
3954 * "loginctl session-status" without further argument will now
3955 show the status of the session of the caller. Similar,
3956 "lock-session", "unlock-session", "activate",
3957 "enable-linger", "disable-linger" may now be called without
3958 session/user parameter in which case they apply to the
3959 caller's session/user.
3961 * An X11 session scriptlet is now shipped that uploads
3962 $DISPLAY and $XAUTHORITY into the environment of the systemd
3963 --user daemon if a session begins. This should improve
3964 compatibility with X11 enabled applications run as systemd
3967 * Generators are now subject to masking via /etc and /run, the
3968 same way as unit files.
3970 * networkd .network files gained support for configuring
3971 per-link IPv4/IPv6 packet forwarding as well as IPv4
3972 masquerading. This is by default turned on for veth links to
3973 containers, as registered by systemd-nspawn. This means that
3974 nspawn containers run with --network-veth will now get
3975 automatic routed access to the host's networks without any
3976 further configuration or setup, as long as networkd runs on
3979 * systemd-nspawn gained the --port= (-p) switch to expose TCP
3980 or UDP posts of a container on the host. With this in place
3981 it is possible to run containers with private veth links
3982 (--network-veth), and have their functionality exposed on
3983 the host as if their services were running directly on the
3986 * systemd-nspawn's --network-veth switch now gained a short
3987 version "-n", since with the changes above it is now truly
3988 useful out-of-the-box. The systemd-nspawn@.service has been
3989 updated to make use of it too by default.
3991 * systemd-nspawn will now maintain a per-image R/W lock, to
3992 ensure that the same image is not started more than once
3993 writable. (It's OK to run an image multiple times
3994 simultaneously in read-only mode.)
3996 * systemd-nspawn's --image= option is now capable of
3997 dissecting and booting MBR and GPT disk images that contain
3998 only a single active Linux partition. Previously it
3999 supported only GPT disk images with proper GPT type
4000 IDs. This allows running cloud images from major
4001 distributions directly with systemd-nspawn, without
4004 * In addition to collecting mouse dpi data in the udev
4005 hardware database, there's now support for collecting angle
4006 information for mouse scroll wheels. The database is
4007 supposed to guarantee similar scrolling behavior on mice
4008 that it knows about. There's also support for collecting
4009 information about Touchpad types.
4011 * udev's input_id built-in will now also collect touch screen
4012 dimension data and attach it to probed devices.
4014 * /etc/os-release gained support for a Distribution Privacy
4017 * networkd gained support for creating "ipvlan", "gretap",
4018 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
4020 * systemd-tmpfiles gained support for "a" lines for setting
4023 * systemd-nspawn will now mount /tmp in the container to
4024 tmpfs, automatically.
4026 * systemd now exposes the memory.usage_in_bytes cgroup
4027 attribute and shows it for each service in the "systemctl
4028 status" output, if available.
4030 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
4031 immediate reboot is triggered. This useful if shutdown is
4032 hung and is unable to complete, to expedite the
4033 operation. Note that this kind of reboot will still unmount
4034 all file systems, and hence should not result in fsck being
4037 * A .device unit for an optical block device will now be
4038 considered active only when a medium is in the drive. Also,
4039 mount units are now bound to their backing devices thus
4040 triggering automatic unmounting when devices become
4041 unavailable. With this in place systemd will now
4042 automatically unmount left-over mounts when a CD-ROM is
4043 ejected or an USB stick is yanked from the system.
4045 * networkd-wait-online now has support for waiting for
4046 specific interfaces only (with globbing), and for giving up
4047 after a configurable timeout.
4049 * networkd now exits when idle. It will be automatically
4050 restarted as soon as interfaces show up, are removed or
4051 change state. networkd will stay around as long as there is
4052 at least one DHCP state machine or similar around, that keep
4055 * networkd may now configure IPv6 link-local addressing in
4056 addition to IPv4 link-local addressing.
4058 * The IPv6 "token" for use in SLAAC may now be configured for
4059 each .network interface in networkd.
4061 * Routes configured with networkd may now be assigned a scope
4064 * networkd's [Match] sections now support globbing and lists
4065 of multiple space-separated matches per item.
4067 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
4068 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
4069 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
4070 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
4071 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
4072 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
4073 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
4074 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
4075 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
4076 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
4077 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
4078 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
4079 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
4080 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
4081 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
4082 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
4083 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
4084 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
4085 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
4086 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
4087 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
4088 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
4089 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
4090 Hoffmann, Zbigniew Jędrzejewski-Szmek
4092 — Berlin, 2015-02-16
4096 * When querying unit file enablement status (for example via
4097 "systemctl is-enabled"), a new state "indirect" is now known
4098 which indicates that a unit might not be enabled itself, but
4099 another unit listed in its Also= setting might be.
4101 * Similar to the various existing ConditionXYZ= settings for
4102 units, there are now matching AssertXYZ= settings. While
4103 failing conditions cause a unit to be skipped, but its job
4104 to succeed, failing assertions declared like this will cause
4105 a unit start operation and its job to fail.
4107 * hostnamed now knows a new chassis type "embedded".
4109 * systemctl gained a new "edit" command. When used on a unit
4110 file, this allows extending unit files with .d/ drop-in
4111 configuration snippets or editing the full file (after
4112 copying it from /usr/lib to /etc). This will invoke the
4113 user's editor (as configured with $EDITOR), and reload the
4114 modified configuration after editing.
4116 * "systemctl status" now shows the suggested enablement state
4117 for a unit, as declared in the (usually vendor-supplied)
4118 system preset files.
4120 * nss-myhostname will now resolve the single-label host name
4121 "gateway" to the locally configured default IP routing
4122 gateways, ordered by their metrics. This assigns a stable
4123 name to the used gateways, regardless which ones are
4124 currently configured. Note that the name will only be
4125 resolved after all other name sources (if nss-myhostname is
4126 configured properly) and should hence not negatively impact
4127 systems that use the single-label host name "gateway" in
4130 * systemd-inhibit now allows filtering by mode when listing
4133 * Scope and service units gained a new "Delegate" boolean
4134 property, which, when set, allows processes running inside the
4135 unit to further partition resources. This is primarily
4136 useful for systemd user instances as well as container
4139 * journald will now pick up audit messages directly from
4140 the kernel, and log them like any other log message. The
4141 audit fields are split up and fully indexed. This means that
4142 journalctl in many ways is now a (nicer!) alternative to
4143 ausearch, the traditional audit client. Note that this
4144 implements only a minimal audit client. If you want the
4145 special audit modes like reboot-on-log-overflow, please use
4146 the traditional auditd instead, which can be used in
4147 parallel to journald.
4149 * The ConditionSecurity= unit file option now understands the
4150 special string "audit" to check whether auditing is
4153 * journalctl gained two new commands --vacuum-size= and
4154 --vacuum-time= to delete old journal files until the
4155 remaining ones take up no more than the specified size on disk,
4156 or are not older than the specified time.
4158 * A new, native PPPoE library has been added to sd-network,
4159 systemd's library of light-weight networking protocols. This
4160 library will be used in a future version of networkd to
4161 enable PPPoE communication without an external pppd daemon.
4163 * The busctl tool now understands a new "capture" verb that
4164 works similar to "monitor", but writes a packet capture
4165 trace to STDOUT that can be redirected to a file which is
4166 compatible with libcap's capture file format. This can then
4167 be loaded in Wireshark and similar tools to inspect bus
4170 * The busctl tool now understands a new "tree" verb that shows
4171 the object trees of a specific service on the bus, or of all
4174 * The busctl tool now understands a new "introspect" verb that
4175 shows all interfaces and members of objects on the bus,
4176 including their signature and values. This is particularly
4177 useful to get more information about bus objects shown by
4178 the new "busctl tree" command.
4180 * The busctl tool now understands new verbs "call",
4181 "set-property" and "get-property" for invoking bus method
4182 calls, setting and getting bus object properties in a
4185 * busctl gained a new --augment-creds= argument that controls
4186 whether the tool shall augment credential information it
4187 gets from the bus with data from /proc, in a possibly
4190 * nspawn's --link-journal= switch gained two new values
4191 "try-guest" and "try-host" that work like "guest" and
4192 "host", but do not fail if the host has no persistent
4193 journalling enabled. -j is now equivalent to
4194 --link-journal=try-guest.
4196 * macvlan network devices created by nspawn will now have
4197 stable MAC addresses.
4199 * A new SmackProcessLabel= unit setting has been added, which
4200 controls the SMACK security label processes forked off by
4201 the respective unit shall use.
4203 * If compiled with --enable-xkbcommon, systemd-localed will
4204 verify x11 keymap settings by compiling the given keymap. It
4205 will spew out warnings if the compilation fails. This
4206 requires libxkbcommon to be installed.
4208 * When a coredump is collected, a larger number of metadata
4209 fields is now collected and included in the journal records
4210 created for it. More specifically, control group membership,
4211 environment variables, memory maps, working directory,
4212 chroot directory, /proc/$PID/status, and a list of open file
4213 descriptors is now stored in the log entry.
4215 * The udev hwdb now contains DPI information for mice. For
4218 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
4220 * All systemd programs that read standalone configuration
4221 files in /etc now also support a corresponding series of
4222 .conf.d configuration directories in /etc/, /run/,
4223 /usr/local/lib/, /usr/lib/, and (if configured with
4224 --enable-split-usr) /lib/. In particular, the following
4225 configuration files now have corresponding configuration
4226 directories: system.conf user.conf, logind.conf,
4227 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
4228 resolved.conf, timesyncd.conf, journal-remote.conf, and
4229 journal-upload.conf. Note that distributions should use the
4230 configuration directories in /usr/lib/; the directories in
4231 /etc/ are reserved for the system administrator.
4233 * systemd-rfkill will no longer take the rfkill device name
4234 into account when storing rfkill state on disk, as the name
4235 might be dynamically assigned and not stable. Instead, the
4236 ID_PATH udev variable combined with the rfkill type (wlan,
4237 bluetooth, ...) is used.
4239 * A new service systemd-machine-id-commit.service has been
4240 added. When used on systems where /etc is read-only during
4241 boot, and /etc/machine-id is not initialized (but an empty
4242 file), this service will copy the temporary machine ID
4243 created as replacement into /etc after the system is fully
4244 booted up. This is useful for systems that are freshly
4245 installed with a non-initialized machine ID, but should get
4246 a fixed machine ID for subsequent boots.
4248 * networkd's .netdev files now provide a large set of
4249 configuration parameters for VXLAN devices. Similarly, the
4250 bridge port cost parameter is now configurable in .network
4251 files. There's also new support for configuring IP source
4252 routing. networkd .link files gained support for a new
4253 OriginalName= match that is useful to match against the
4254 original interface name the kernel assigned. .network files
4255 may include MTU= and MACAddress= fields for altering the MTU
4256 and MAC address while being connected to a specific network
4259 * The LUKS logic gained supported for configuring
4260 UUID-specific key files. There's also new support for naming
4261 LUKS device from the kernel command line, using the new
4262 luks.name= argument.
4264 * Timer units may now be transiently created via the bus API
4265 (this was previously already available for scope and service
4266 units). In addition it is now possible to create multiple
4267 transient units at the same time with a single bus call. The
4268 "systemd-run" tool has been updated to make use of this for
4269 running commands on a specified time, in at(1)-style.
4271 * tmpfiles gained support for "t" lines, for assigning
4272 extended attributes to files. Among other uses this may be
4273 used to assign SMACK labels to files.
4275 Contributions from: Alin Rauta, Alison Chaiken, Andrej
4276 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
4277 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
4278 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
4279 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
4280 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
4281 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
4282 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
4283 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
4284 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
4285 Michael Biebl, Michael Chapman, Michael Marineau, Michal
4286 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
4287 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
4288 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
4289 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
4290 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4291 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
4292 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
4294 — Berlin, 2014-12-10
4298 * journalctl gained the new options -t/--identifier= to match
4299 on the syslog identifier (aka "tag"), as well as --utc to
4300 show log timestamps in the UTC timezone. journalctl now also
4301 accepts -n/--lines=all to disable line capping in a pager.
4303 * journalctl gained a new switch, --flush, that synchronously
4304 flushes logs from /run/log/journal to /var/log/journal if
4305 persistent storage is enabled. systemd-journal-flush.service
4306 now waits until the operation is complete.
4308 * Services can notify the manager before they start a reload
4309 (by sending RELOADING=1) or shutdown (by sending
4310 STOPPING=1). This allows the manager to track and show the
4311 internal state of daemons and closes a race condition when
4312 the process is still running but has closed its D-Bus
4315 * Services with Type=oneshot do not have to have any ExecStart
4318 * User units are now loaded also from
4319 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
4320 /run/systemd/user directory that was already previously
4321 supported, but is under the control of the user.
4323 * Job timeouts (i.e. time-outs on the time a job that is
4324 queued stays in the run queue) can now optionally result in
4325 immediate reboot or power-off actions (JobTimeoutAction= and
4326 JobTimeoutRebootArgument=). This is useful on ".target"
4327 units, to limit the maximum time a target remains
4328 undispatched in the run queue, and to trigger an emergency
4329 operation in such a case. This is now used by default to
4330 turn off the system if boot-up (as defined by everything in
4331 basic.target) hangs and does not complete for at least
4332 15min. Also, if power-off or reboot hang for at least 30min
4333 an immediate power-off/reboot operation is triggered. This
4334 functionality is particularly useful to increase reliability
4335 on embedded devices, but also on laptops which might
4336 accidentally get powered on when carried in a backpack and
4337 whose boot stays stuck in a hard disk encryption passphrase
4340 * systemd-logind can be configured to also handle lid switch
4341 events even when the machine is docked or multiple displays
4342 are attached (HandleLidSwitchDocked= option).
4344 * A helper binary and a service have been added which can be
4345 used to resume from hibernation in the initramfs. A
4346 generator will parse the resume= option on the kernel
4347 command line to trigger resume.
4349 * A user console daemon systemd-consoled has been
4350 added. Currently, it is a preview, and will so far open a
4351 single terminal on each session of the user marked as
4352 Desktop=systemd-console.
4354 * Route metrics can be specified for DHCP routes added by
4357 * The SELinux context of socket-activated services can be set
4358 from the information provided by the networking stack
4359 (SELinuxContextFromNet= option).
4361 * Userspace firmware loading support has been removed and
4362 the minimum supported kernel version is thus bumped to 3.7.
4364 * Timeout for udev workers has been increased from 1 to 3
4365 minutes, but a warning will be printed after 1 minute to
4366 help diagnose kernel modules that take a long time to load.
4368 * Udev rules can now remove tags on devices with TAG-="foobar".
4370 * systemd's readahead implementation has been removed. In many
4371 circumstances it didn't give expected benefits even for
4372 rotational disk drives and was becoming less relevant in the
4373 age of SSDs. As none of the developers has been using
4374 rotating media anymore, and nobody stepped up to actively
4375 maintain this component of systemd it has now been removed.
4377 * Swap units can use Options= to specify discard options.
4378 Discard options specified for swaps in /etc/fstab are now
4381 * Docker containers are now detected as a separate type of
4384 * The Password Agent protocol gained support for queries where
4385 the user input is shown, useful e.g. for user names.
4386 systemd-ask-password gained a new --echo option to turn that
4389 * The default sysctl.d/ snippets will now set:
4391 net.core.default_qdisc = fq_codel
4393 This selects Fair Queuing Controlled Delay as the default
4394 queuing discipline for network interfaces. fq_codel helps
4395 fight the network bufferbloat problem. It is believed to be
4396 a good default with no tuning required for most workloads.
4397 Downstream distributions may override this choice. On 10Gbit
4398 servers that do not do forwarding, "fq" may perform better.
4399 Systems without a good clocksource should use "pfifo_fast".
4401 * If kdbus is enabled during build a new option BusPolicy= is
4402 available for service units, that allows locking all service
4403 processes into a stricter bus policy, in order to limit
4404 access to various bus services, or even hide most of them
4405 from the service's view entirely.
4407 * networkctl will now show the .network and .link file
4408 networkd has applied to a specific interface.
4410 * sd-login gained a new API call sd_session_get_desktop() to
4411 query which desktop environment has been selected for a
4414 * UNIX utmp support is now compile-time optional to support
4415 legacy-free systems.
4417 * systemctl gained two new commands "add-wants" and
4418 "add-requires" for pulling in units from specific targets
4421 * If the word "rescue" is specified on the kernel command line
4422 the system will now boot into rescue mode (aka
4423 rescue.target), which was previously available only by
4424 specifying "1" or "systemd.unit=rescue.target" on the kernel
4425 command line. This new kernel command line option nicely
4426 mirrors the already existing "emergency" kernel command line
4429 * New kernel command line options mount.usr=, mount.usrflags=,
4430 mount.usrfstype= have been added that match root=, rootflags=,
4431 rootfstype= but allow mounting a specific file system to
4434 * The $NOTIFY_SOCKET is now also passed to control processes of
4435 services, not only the main process.
4437 * This version reenables support for fsck's -l switch. This
4438 means at least version v2.25 of util-linux is required for
4439 operation, otherwise dead-locks on device nodes may
4440 occur. Again: you need to update util-linux to at least
4441 v2.25 when updating systemd to v217.
4443 * The "multi-seat-x" tool has been removed from systemd, as
4444 its functionality has been integrated into X servers 1.16,
4445 and the tool is hence redundant. It is recommended to update
4446 display managers invoking this tool to simply invoke X
4447 directly from now on, again.
4449 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
4450 message flag has been added for all of systemd's PolicyKit
4451 authenticated method calls has been added. In particular
4452 this now allows optional interactive authorization via
4453 PolicyKit for many of PID1's privileged operations such as
4454 unit file enabling and disabling.
4456 * "udevadm hwdb --update" learnt a new switch "--usr" for
4457 placing the rebuilt hardware database in /usr instead of
4458 /etc. When used only hardware database entries stored in
4459 /usr will be used, and any user database entries in /etc are
4460 ignored. This functionality is useful for vendors to ship a
4461 pre-built database on systems where local configuration is
4462 unnecessary or unlikely.
4464 * Calendar time specifications in .timer units now also
4465 understand the strings "semi-annually", "quarterly" and
4466 "minutely" as shortcuts (in addition to the preexisting
4467 "anually", "hourly", ...).
4469 * systemd-tmpfiles will now correctly create files in /dev
4470 at boot which are marked for creation only at boot. It is
4471 recommended to always create static device nodes with 'c!'
4472 and 'b!', so that they are created only at boot and not
4473 overwritten at runtime.
4475 * When the watchdog logic is used for a service (WatchdogSec=)
4476 and the watchdog timeout is hit the service will now be
4477 terminated with SIGABRT (instead of just SIGTERM), in order
4478 to make sure a proper coredump and backtrace is
4479 generated. This ensures that hanging services will result in
4480 similar coredump/backtrace behaviour as services that hit a
4483 Contributions from: Andreas Henriksson, Andrei Borzenkov,
4484 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
4485 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
4486 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
4487 Herrmann, David Sommerseth, David Strauss, Emil Renner
4488 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
4489 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
4490 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
4491 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
4492 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
4493 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
4494 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
4495 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
4496 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
4497 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
4498 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
4499 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
4500 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
4501 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
4502 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
4503 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
4506 — Berlin, 2014-10-28
4510 * timedated no longer reads NTP implementation unit names from
4511 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
4512 implementations should add a
4514 Conflicts=systemd-timesyncd.service
4516 to their unit files to take over and replace systemd's NTP
4517 default functionality.
4519 * systemd-sysusers gained a new line type "r" for configuring
4520 which UID/GID ranges to allocate system users/groups
4521 from. Lines of type "u" may now add an additional column
4522 that specifies the home directory for the system user to be
4523 created. Also, systemd-sysusers may now optionally read user
4524 information from STDIN instead of a file. This is useful for
4525 invoking it from RPM preinst scriptlets that need to create
4526 users before the first RPM file is installed since these
4527 files might need to be owned by them. A new
4528 %sysusers_create_inline RPM macro has been introduced to do
4529 just that. systemd-sysusers now updates the shadow files as
4530 well as the user/group databases, which should enhance
4531 compatibility with certain tools like grpck.
4533 * A number of bus APIs of PID 1 now optionally consult
4534 PolicyKit to permit access for otherwise unprivileged
4535 clients under certain conditions. Note that this currently
4536 doesn't support interactive authentication yet, but this is
4537 expected to be added eventually, too.
4539 * /etc/machine-info now has new fields for configuring the
4540 deployment environment of the machine, as well as the
4541 location of the machine. hostnamectl has been updated with
4542 new command to update these fields.
4544 * systemd-timesyncd has been updated to automatically acquire
4545 NTP server information from systemd-networkd, which might
4546 have been discovered via DHCP.
4548 * systemd-resolved now includes a caching DNS stub resolver
4549 and a complete LLMNR name resolution implementation. A new
4550 NSS module "nss-resolve" has been added which can be used
4551 instead of glibc's own "nss-dns" to resolve hostnames via
4552 systemd-resolved. Hostnames, addresses and arbitrary RRs may
4553 be resolved via systemd-resolved D-Bus APIs. In contrast to
4554 the glibc internal resolver systemd-resolved is aware of
4555 multi-homed system, and keeps DNS server and caches separate
4556 and per-interface. Queries are sent simultaneously on all
4557 interfaces that have DNS servers configured, in order to
4558 properly handle VPNs and local LANs which might resolve
4559 separate sets of domain names. systemd-resolved may acquire
4560 DNS server information from systemd-networkd automatically,
4561 which in turn might have discovered them via DHCP. A tool
4562 "systemd-resolve-host" has been added that may be used to
4563 query the DNS logic in resolved. systemd-resolved implements
4564 IDNA and automatically uses IDNA or UTF-8 encoding depending
4565 on whether classic DNS or LLMNR is used as transport. In the
4566 next releases we intend to add a DNSSEC and mDNS/DNS-SD
4567 implementation to systemd-resolved.
4569 * A new NSS module nss-mymachines has been added, that
4570 automatically resolves the names of all local registered
4571 containers to their respective IP addresses.
4573 * A new client tool "networkctl" for systemd-networkd has been
4574 added. It currently is entirely passive and will query
4575 networking configuration from udev, rtnetlink and networkd,
4576 and present it to the user in a very friendly
4577 way. Eventually, we hope to extend it to become a full
4578 control utility for networkd.
4580 * .socket units gained a new DeferAcceptSec= setting that
4581 controls the kernels' TCP_DEFER_ACCEPT sockopt for
4582 TCP. Similarly, support for controlling TCP keep-alive
4583 settings has been added (KeepAliveTimeSec=,
4584 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
4585 turning off Nagle's algorithm on TCP has been added
4588 * logind learned a new session type "web", for use in projects
4589 like Cockpit which register web clients as PAM sessions.
4591 * timer units with at least one OnCalendar= setting will now
4592 be started only after time-sync.target has been
4593 reached. This way they will not elapse before the system
4594 clock has been corrected by a local NTP client or
4595 similar. This is particular useful on RTC-less embedded
4596 machines, that come up with an invalid system clock.
4598 * systemd-nspawn's --network-veth= switch should now result in
4599 stable MAC addresses for both the outer and the inner side
4602 * systemd-nspawn gained a new --volatile= switch for running
4603 container instances with /etc or /var unpopulated.
4605 * The kdbus client code has been updated to use the new Linux
4606 3.17 memfd subsystem instead of the old kdbus-specific one.
4608 * systemd-networkd's DHCP client and server now support
4609 FORCERENEW. There are also new configuration options to
4610 configure the vendor client identifier and broadcast mode
4613 * systemd will no longer inform the kernel about the current
4614 timezone, as this is necessarily incorrect and racy as the
4615 kernel has no understanding of DST and similar
4616 concepts. This hence means FAT timestamps will be always
4617 considered UTC, similar to what Android is already
4618 doing. Also, when the RTC is configured to the local time
4619 (rather than UTC) systemd will never synchronize back to it,
4620 as this might confuse Windows at a later boot.
4622 * systemd-analyze gained a new command "verify" for offline
4623 validation of unit files.
4625 * systemd-networkd gained support for a couple of additional
4626 settings for bonding networking setups. Also, the metric for
4627 statically configured routes may now be configured. For
4628 network interfaces where this is appropriate the peer IP
4629 address may now be configured.
4631 * systemd-networkd's DHCP client will no longer request
4632 broadcasting by default, as this tripped up some networks.
4633 For hardware where broadcast is required the feature should
4634 be switched back on using RequestBroadcast=yes.
4636 * systemd-networkd will now set up IPv4LL addresses (when
4637 enabled) even if DHCP is configured successfully.
4639 * udev will now default to respect network device names given
4640 by the kernel when the kernel indicates that these are
4641 predictable. This behavior can be tweaked by changing
4642 NamePolicy= in the relevant .link file.
4644 * A new library systemd-terminal has been added that
4645 implements full TTY stream parsing and rendering. This
4646 library is supposed to be used later on for implementing a
4647 full userspace VT subsystem, replacing the current kernel
4650 * A new tool systemd-journal-upload has been added to push
4651 journal data to a remote system running
4652 systemd-journal-remote.
4654 * journald will no longer forward all local data to another
4655 running syslog daemon. This change has been made because
4656 rsyslog (which appears to be the most commonly used syslog
4657 implementation these days) no longer makes use of this, and
4658 instead pulls the data out of the journal on its own. Since
4659 forwarding the messages to a non-existent syslog server is
4660 more expensive than we assumed we have now turned this
4661 off. If you run a syslog server that is not a recent rsyslog
4662 version, you have to turn this option on again
4663 (ForwardToSyslog= in journald.conf).
4665 * journald now optionally supports the LZ4 compressor for
4666 larger journal fields. This compressor should perform much
4667 better than XZ which was the previous default.
4669 * machinectl now shows the IP addresses of local containers,
4670 if it knows them, plus the interface name of the container.
4672 * A new tool "systemd-escape" has been added that makes it
4673 easy to escape strings to build unit names and similar.
4675 * sd_notify() messages may now include a new ERRNO= field
4676 which is parsed and collected by systemd and shown among the
4677 "systemctl status" output for a service.
4679 * A new component "systemd-firstboot" has been added that
4680 queries the most basic systemd information (timezone,
4681 hostname, root password) interactively on first
4682 boot. Alternatively it may also be used to provision these
4683 things offline on OS images installed into directories.
4685 * The default sysctl.d/ snippets will now set
4687 net.ipv4.conf.default.promote_secondaries=1
4689 This has the benefit of no flushing secondary IP addresses
4690 when primary addresses are removed.
4692 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
4693 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
4694 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
4695 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
4696 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
4697 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
4698 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
4699 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
4700 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
4701 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
4702 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
4703 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
4704 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
4705 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
4706 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
4708 — Berlin, 2014-08-19
4712 * A new tool systemd-sysusers has been added. This tool
4713 creates system users and groups in /etc/passwd and
4714 /etc/group, based on static declarative system user/group
4715 definitions in /usr/lib/sysusers.d/. This is useful to
4716 enable factory resets and volatile systems that boot up with
4717 an empty /etc directory, and thus need system users and
4718 groups created during early boot. systemd now also ships
4719 with two default sysusers.d/ files for the most basic
4720 users and groups systemd and the core operating system
4723 * A new tmpfiles snippet has been added that rebuilds the
4724 essential files in /etc on boot, should they be missing.
4726 * A directive for ensuring automatic clean-up of
4727 /var/cache/man/ has been removed from the default
4728 configuration. This line should now be shipped by the man
4729 implementation. The necessary change has been made to the
4730 man-db implementation. Note that you need to update your man
4731 implementation to one that ships this line, otherwise no
4732 automatic clean-up of /var/cache/man will take place.
4734 * A new condition ConditionNeedsUpdate= has been added that
4735 may conditionalize services to only run when /etc or /var
4736 are "older" than the vendor operating system resources in
4737 /usr. This is useful for reconstructing or updating /etc
4738 after an offline update of /usr or a factory reset, on the
4739 next reboot. Services that want to run once after such an
4740 update or reset should use this condition and order
4741 themselves before the new systemd-update-done.service, which
4742 will mark the two directories as fully updated. A number of
4743 service files have been added making use of this, to rebuild
4744 the udev hardware database, the journald message catalog and
4745 dynamic loader cache (ldconfig). The systemd-sysusers tool
4746 described above also makes use of this now. With this in
4747 place it is now possible to start up a minimal operating
4748 system with /etc empty cleanly. For more information on the
4749 concepts involved see this recent blog story:
4751 http://0pointer.de/blog/projects/stateless.html
4753 * A new system group "input" has been introduced, and all
4754 input device nodes get this group assigned. This is useful
4755 for system-level software to get access to input devices. It
4756 complements what is already done for "audio" and "video".
4758 * systemd-networkd learnt minimal DHCPv4 server support in
4759 addition to the existing DHCPv4 client support. It also
4760 learnt DHCPv6 client and IPv6 Router Solicitation client
4761 support. The DHCPv4 client gained support for static routes
4762 passed in from the server. Note that the [DHCPv4] section
4763 known in older systemd-networkd versions has been renamed to
4764 [DHCP] and is now also used by the DHCPv6 client. Existing
4765 .network files using settings of this section should be
4766 updated, though compatibility is maintained. Optionally, the
4767 client hostname may now be sent to the DHCP server.
4769 * networkd gained support for vxlan virtual networks as well
4770 as tun/tap and dummy devices.
4772 * networkd gained support for automatic allocation of address
4773 ranges for interfaces from a system-wide pool of
4774 addresses. This is useful for dynamically managing a large
4775 number of interfaces with a single network configuration
4776 file. In particular this is useful to easily assign
4777 appropriate IP addresses to the veth links of a large number
4778 of nspawn instances.
4780 * RPM macros for processing sysusers, sysctl and binfmt
4781 drop-in snippets at package installation time have been
4784 * The /etc/os-release file should now be placed in
4785 /usr/lib/os-release. The old location is automatically
4786 created as symlink. /usr/lib is the more appropriate
4787 location of this file, since it shall actually describe the
4788 vendor operating system shipped in /usr, and not the
4789 configuration stored in /etc.
4791 * .mount units gained a new boolean SloppyOptions= setting
4792 that maps to mount(8)'s -s option which enables permissive
4793 parsing of unknown mount options.
4795 * tmpfiles learnt a new "L+" directive which creates a symlink
4796 but (unlike "L") deletes a pre-existing file first, should
4797 it already exist and not already be the correct
4798 symlink. Similarly, "b+", "c+" and "p+" directives have been
4799 added as well, which create block and character devices, as
4800 well as fifos in the filesystem, possibly removing any
4801 pre-existing files of different types.
4803 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
4804 'argument' field (which so far specified the source to
4805 symlink/copy the files from) is now optional. If omitted the
4806 same file os copied from /usr/share/factory/ suffixed by the
4807 full destination path. This is useful for populating /etc
4808 with essential files, by copying them from vendor defaults
4809 shipped in /usr/share/factory/etc.
4811 * A new command "systemctl preset-all" has been added that
4812 applies the service preset settings to all installed unit
4813 files. A new switch --preset-mode= has been added that
4814 controls whether only enable or only disable operations
4817 * A new command "systemctl is-system-running" has been added
4818 that allows checking the overall state of the system, for
4819 example whether it is fully up and running.
4821 * When the system boots up with an empty /etc, the equivalent
4822 to "systemctl preset-all" is executed during early boot, to
4823 make sure all default services are enabled after a factory
4826 * systemd now contains a minimal preset file that enables the
4827 most basic services systemd ships by default.
4829 * Unit files' [Install] section gained a new DefaultInstance=
4830 field for defining the default instance to create if a
4831 template unit is enabled with no instance specified.
4833 * A new passive target cryptsetup-pre.target has been added
4834 that may be used by services that need to make they run and
4835 finish before the first LUKS cryptographic device is set up.
4837 * The /dev/loop-control and /dev/btrfs-control device nodes
4838 are now owned by the "disk" group by default, opening up
4839 access to this group.
4841 * systemd-coredump will now automatically generate a
4842 stack trace of all core dumps taking place on the system,
4843 based on elfutils' libdw library. This stack trace is logged
4846 * systemd-coredump may now optionally store coredumps directly
4847 on disk (in /var/lib/systemd/coredump, possibly compressed),
4848 instead of storing them unconditionally in the journal. This
4849 mode is the new default. A new configuration file
4850 /etc/systemd/coredump.conf has been added to configure this
4851 and other parameters of systemd-coredump.
4853 * coredumpctl gained a new "info" verb to show details about a
4854 specific coredump. A new switch "-1" has also been added
4855 that makes sure to only show information about the most
4856 recent entry instead of all entries. Also, as the tool is
4857 generally useful now the "systemd-" prefix of the binary
4858 name has been removed. Distributions that want to maintain
4859 compatibility with the old name should add a symlink from
4860 the old name to the new name.
4862 * journald's SplitMode= now defaults to "uid". This makes sure
4863 that unprivileged users can access their own coredumps with
4864 coredumpctl without restrictions.
4866 * New kernel command line options "systemd.wants=" (for
4867 pulling an additional unit during boot), "systemd.mask="
4868 (for masking a specific unit for the boot), and
4869 "systemd.debug-shell" (for enabling the debug shell on tty9)
4870 have been added. This is implemented in the new generator
4871 "systemd-debug-generator".
4873 * systemd-nspawn will now by default filter a couple of
4874 syscalls for containers, among them those required for
4875 kernel module loading, direct x86 IO port access, swap
4876 management, and kexec. Most importantly though
4877 open_by_handle_at() is now prohibited for containers,
4878 closing a hole similar to a recently discussed vulnerability
4879 in docker regarding access to files on file hierarchies the
4880 container should normally not have access to. Note that, for
4881 nspawn, we generally make no security claims anyway (and
4882 this is explicitly documented in the man page), so this is
4883 just a fix for one of the most obvious problems.
4885 * A new man page file-hierarchy(7) has been added that
4886 contains a minimized, modernized version of the file system
4887 layout systemd expects, similar in style to the FHS
4888 specification or hier(5). A new tool systemd-path(1) has
4889 been added to query many of these paths for the local
4892 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
4893 longer done. Since the directory now has a per-user size
4894 limit, and is cleaned on logout this appears unnecessary,
4895 in particular since this now brings the lifecycle of this
4896 directory closer in line with how IPC objects are handled.
4898 * systemd.pc now exports a number of additional directories,
4899 including $libdir (which is useful to identify the library
4900 path for the primary architecture of the system), and a
4901 couple of drop-in directories.
4903 * udev's predictable network interface names now use the dev_port
4904 sysfs attribute, introduced in linux 3.15 instead of dev_id to
4905 distinguish between ports of the same PCI function. dev_id should
4906 only be used for ports using the same HW address, hence the need
4909 * machined has been updated to export the OS version of a
4910 container (read from /etc/os-release and
4911 /usr/lib/os-release) on the bus. This is now shown in
4912 "machinectl status" for a machine.
4914 * A new service setting RestartForceExitStatus= has been
4915 added. If configured to a set of exit signals or process
4916 return values, the service will be restarted when the main
4917 daemon process exits with any of them, regardless of the
4920 * systemctl's -H switch for connecting to remote systemd
4921 machines has been extended so that it may be used to
4922 directly connect to a specific container on the
4923 host. "systemctl -H root@foobar:waldi" will now connect as
4924 user "root" to host "foobar", and then proceed directly to
4925 the container named "waldi". Note that currently you have to
4926 authenticate as user "root" for this to work, as entering
4927 containers is a privileged operation.
4929 Contributions from: Andreas Henriksson, Benjamin Steinwender,
4930 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
4931 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
4932 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
4933 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
4934 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
4935 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
4936 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
4937 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
4938 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
4939 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
4940 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
4942 — Berlin, 2014-07-03
4946 * As an experimental feature, udev now tries to lock the
4947 disk device node (flock(LOCK_SH|LOCK_NB)) while it
4948 executes events for the disk or any of its partitions.
4949 Applications like partitioning programs can lock the
4950 disk device node (flock(LOCK_EX)) and claim temporary
4951 device ownership that way; udev will entirely skip all event
4952 handling for this disk and its partitions. If the disk
4953 was opened for writing, the close will trigger a partition
4954 table rescan in udev's "watch" facility, and if needed
4955 synthesize "change" events for the disk and all its partitions.
4956 This is now unconditionally enabled, and if it turns out to
4957 cause major problems, we might turn it on only for specific
4958 devices, or might need to disable it entirely. Device Mapper
4959 devices are excluded from this logic.
4961 * We temporarily dropped the "-l" switch for fsck invocations,
4962 since they collide with the flock() logic above. util-linux
4963 upstream has been changed already to avoid this conflict,
4964 and we will readd "-l" as soon as util-linux with this
4965 change has been released.
4967 * The dependency on libattr has been removed. Since a long
4968 time, the extended attribute calls have moved to glibc, and
4969 libattr is thus unnecessary.
4971 * Virtualization detection works without privileges now. This
4972 means the systemd-detect-virt binary no longer requires
4973 CAP_SYS_PTRACE file capabilities, and our daemons can run
4974 with fewer privileges.
4976 * systemd-networkd now runs under its own "systemd-network"
4977 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
4978 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
4979 loses the ability to write to files owned by root this way.
4981 * Similarly, systemd-resolved now runs under its own
4982 "systemd-resolve" user with no capabilities remaining.
4984 * Similarly, systemd-bus-proxyd now runs under its own
4985 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
4987 * systemd-networkd gained support for setting up "veth"
4988 virtual Ethernet devices for container connectivity, as well
4989 as GRE and VTI tunnels.
4991 * systemd-networkd will no longer automatically attempt to
4992 manually load kernel modules necessary for certain tunnel
4993 transports. Instead, it is assumed the kernel loads them
4994 automatically when required. This only works correctly on
4995 very new kernels. On older kernels, please consider adding
4996 the kernel modules to /etc/modules-load.d/ as a work-around.
4998 * The resolv.conf file systemd-resolved generates has been
4999 moved to /run/systemd/resolve/. If you have a symlink from
5000 /etc/resolv.conf, it might be necessary to correct it.
5002 * Two new service settings, ProtectHome= and ProtectSystem=,
5003 have been added. When enabled, they will make the user data
5004 (such as /home) inaccessible or read-only and the system
5005 (such as /usr) read-only, for specific services. This allows
5006 very light-weight per-service sandboxing to avoid
5007 modifications of user data or system files from
5008 services. These two new switches have been enabled for all
5009 of systemd's long-running services, where appropriate.
5011 * Socket units gained new SocketUser= and SocketGroup=
5012 settings to set the owner user and group of AF_UNIX sockets
5013 and FIFOs in the file system.
5015 * Socket units gained a new RemoveOnStop= setting. If enabled,
5016 all FIFOS and sockets in the file system will be removed
5017 when the specific socket unit is stopped.
5019 * Socket units gained a new Symlinks= setting. It takes a list
5020 of symlinks to create to file system sockets or FIFOs
5021 created by the specific Unix sockets. This is useful to
5022 manage symlinks to socket nodes with the same life-cycle as
5025 * The /dev/log socket and /dev/initctl FIFO have been moved to
5026 /run, and have been replaced by symlinks. This allows
5027 connecting to these facilities even if PrivateDevices=yes is
5028 used for a service (which makes /dev/log itself unavailable,
5029 but /run is left). This also has the benefit of ensuring
5030 that /dev only contains device nodes, directories and
5031 symlinks, and nothing else.
5033 * sd-daemon gained two new calls sd_pid_notify() and
5034 sd_pid_notifyf(). They are similar to sd_notify() and
5035 sd_notifyf(), but allow overriding of the source PID of
5036 notification messages if permissions permit this. This is
5037 useful to send notify messages on behalf of a different
5038 process (for example, the parent process). The
5039 systemd-notify tool has been updated to make use of this
5040 when sending messages (so that notification messages now
5041 originate from the shell script invoking systemd-notify and
5042 not the systemd-notify process itself. This should minimize
5043 a race where systemd fails to associate notification
5044 messages to services when the originating process already
5047 * A new "on-abnormal" setting for Restart= has been added. If
5048 set, it will result in automatic restarts on all "abnormal"
5049 reasons for a process to exit, which includes unclean
5050 signals, core dumps, timeouts and watchdog timeouts, but
5051 does not include clean and unclean exit codes or clean
5052 signals. Restart=on-abnormal is an alternative for
5053 Restart=on-failure for services that shall be able to
5054 terminate and avoid restarts on certain errors, by
5055 indicating so with an unclean exit code. Restart=on-failure
5056 or Restart=on-abnormal is now the recommended setting for
5057 all long-running services.
5059 * If the InaccessibleDirectories= service setting points to a
5060 mount point (or if there are any submounts contained within
5061 it), it is now attempted to completely unmount it, to make
5062 the file systems truly unavailable for the respective
5065 * The ReadOnlyDirectories= service setting and
5066 systemd-nspawn's --read-only parameter are now recursively
5067 applied to all submounts, too.
5069 * Mount units may now be created transiently via the bus APIs.
5071 * The support for SysV and LSB init scripts has been removed
5072 from the systemd daemon itself. Instead, it is now
5073 implemented as a generator that creates native systemd units
5074 from these scripts when needed. This enables us to remove a
5075 substantial amount of legacy code from PID 1, following the
5076 fact that many distributions only ship a very small number
5077 of LSB/SysV init scripts nowadays.
5079 * Privileged Xen (dom0) domains are not considered
5080 virtualization anymore by the virtualization detection
5081 logic. After all, they generally have unrestricted access to
5082 the hardware and usually are used to manage the unprivileged
5085 * systemd-tmpfiles gained a new "C" line type, for copying
5086 files or entire directories.
5088 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
5089 lines. So far, they have been non-globbing versions of the
5090 latter, and have thus been redundant. In future, it is
5091 recommended to only use "z". "m" has hence been removed
5092 from the documentation, even though it stays supported.
5094 * A tmpfiles snippet to recreate the most basic structure in
5095 /var has been added. This is enough to create the /var/run →
5096 /run symlink and create a couple of structural
5097 directories. This allows systems to boot up with an empty or
5098 volatile /var. Of course, while with this change, the core OS
5099 now is capable with dealing with a volatile /var, not all
5100 user services are ready for it. However, we hope that sooner
5101 or later, many service daemons will be changed upstream so
5102 that they are able to automatically create their necessary
5103 directories in /var at boot, should they be missing. This is
5104 the first step to allow state-less systems that only require
5105 the vendor image for /usr to boot.
5107 * systemd-nspawn has gained a new --tmpfs= switch to mount an
5108 empty tmpfs instance to a specific directory. This is
5109 particularly useful for making use of the automatic
5110 reconstruction of /var (see above), by passing --tmpfs=/var.
5112 * Access modes specified in tmpfiles snippets may now be
5113 prefixed with "~", which indicates that they shall be masked
5114 by whether the existing file or directory is currently
5115 writable, readable or executable at all. Also, if specified,
5116 the sgid/suid/sticky bits will be masked for all
5119 * A new passive target unit "network-pre.target" has been
5120 added which is useful for services that shall run before any
5121 network is configured, for example firewall scripts.
5123 * The "floppy" group that previously owned the /dev/fd*
5124 devices is no longer used. The "disk" group is now used
5125 instead. Distributions should probably deprecate usage of
5128 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
5129 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
5130 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
5131 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
5132 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
5133 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
5136 — Berlin, 2014-06-11
5140 * A new "systemd-timesyncd" daemon has been added for
5141 synchronizing the system clock across the network. It
5142 implements an SNTP client. In contrast to NTP
5143 implementations such as chrony or the NTP reference server,
5144 this only implements a client side, and does not bother with
5145 the full NTP complexity, focusing only on querying time from
5146 one remote server and synchronizing the local clock to
5147 it. Unless you intend to serve NTP to networked clients or
5148 want to connect to local hardware clocks, this simple NTP
5149 client should be more than appropriate for most
5150 installations. The daemon runs with minimal privileges, and
5151 has been hooked up with networkd to only operate when
5152 network connectivity is available. The daemon saves the
5153 current clock to disk every time a new NTP sync has been
5154 acquired, and uses this to possibly correct the system clock
5155 early at bootup, in order to accommodate for systems that
5156 lack an RTC such as the Raspberry Pi and embedded devices,
5157 and to make sure that time monotonically progresses on these
5158 systems, even if it is not always correct. To make use of
5159 this daemon, a new system user and group "systemd-timesync"
5160 needs to be created on installation of systemd.
5162 * The queue "seqnum" interface of libudev has been disabled, as
5163 it was generally incompatible with device namespacing as
5164 sequence numbers of devices go "missing" if the devices are
5165 part of a different namespace.
5167 * "systemctl list-timers" and "systemctl list-sockets" gained
5168 a --recursive switch for showing units of these types also
5169 for all local containers, similar in style to the already
5170 supported --recursive switch for "systemctl list-units".
5172 * A new RebootArgument= setting has been added for service
5173 units, which may be used to specify a kernel reboot argument
5174 to use when triggering reboots with StartLimitAction=.
5176 * A new FailureAction= setting has been added for service
5177 units which may be used to specify an operation to trigger
5178 when a service fails. This works similarly to
5179 StartLimitAction=, but unlike it, controls what is done
5180 immediately rather than only after several attempts to
5181 restart the service in question.
5183 * hostnamed got updated to also expose the kernel name,
5184 release, and version on the bus. This is useful for
5185 executing commands like hostnamectl with the -H switch.
5186 systemd-analyze makes use of this to properly display
5187 details when running non-locally.
5189 * The bootchart tool can now show cgroup information in the
5190 graphs it generates.
5192 * The CFS CPU quota cgroup attribute is now exposed for
5193 services. The new CPUQuota= switch has been added for this
5194 which takes a percentage value. Setting this will have the
5195 result that a service may never get more CPU time than the
5196 specified percentage, even if the machine is otherwise idle.
5198 * systemd-networkd learned IPIP and SIT tunnel support.
5200 * LSB init scripts exposing a dependency on $network will now
5201 get a dependency on network-online.target rather than simply
5202 network.target. This should bring LSB handling closer to
5203 what it was on SysV systems.
5205 * A new fsck.repair= kernel option has been added to control
5206 how fsck shall deal with unclean file systems at boot.
5208 * The (.ini) configuration file parser will now silently
5209 ignore sections whose name begins with "X-". This may be
5210 used to maintain application-specific extension sections in unit
5213 * machined gained a new API to query the IP addresses of
5214 registered containers. "machinectl status" has been updated
5215 to show these addresses in its output.
5217 * A new call sd_uid_get_display() has been added to the
5218 sd-login APIs for querying the "primary" session of a
5219 user. The "primary" session of the user is elected from the
5220 user's sessions and generally a graphical session is
5221 preferred over a text one.
5223 * A minimal systemd-resolved daemon has been added. It
5224 currently simply acts as a companion to systemd-networkd and
5225 manages resolv.conf based on per-interface DNS
5226 configuration, possibly supplied via DHCP. In the long run
5227 we hope to extend this into a local DNSSEC enabled DNS and
5230 * The systemd-networkd-wait-online tool is now enabled by
5231 default. It will delay network-online.target until a network
5232 connection has been configured. The tool primarily integrates
5233 with networkd, but will also make a best effort to make sense
5234 of network configuration performed in some other way.
5236 * Two new service options StartupCPUShares= and
5237 StartupBlockIOWeight= have been added that work similarly to
5238 CPUShares= and BlockIOWeight= however only apply during
5239 system startup. This is useful to prioritize certain services
5240 differently during bootup than during normal runtime.
5242 * hostnamed has been changed to prefer the statically
5243 configured hostname in /etc/hostname (unless set to
5244 'localhost' or empty) over any dynamic one supplied by
5245 dhcp. With this change, the rules for picking the hostname
5246 match more closely the rules of other configuration settings
5247 where the local administrator's configuration in /etc always
5248 overrides any other settings.
5250 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
5251 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
5252 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
5253 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
5254 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
5255 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
5256 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
5257 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
5258 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
5259 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
5260 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
5261 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
5262 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
5263 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
5264 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
5265 Lindskog, WaLyong Cho, Will Woods, Zbigniew
5268 — Beijing, 2014-05-28
5272 * When restoring the screen brightness at boot, stay away from
5273 the darkest setting or from the lowest 5% of the available
5274 range, depending on which is the larger value of both. This
5275 should effectively protect the user from rebooting into a
5276 black screen, should the brightness have been set to minimum
5279 * sd-login gained a new sd_machine_get_class() call to
5280 determine the class ("vm" or "container") of a machine
5281 registered with machined.
5283 * sd-login gained new calls
5284 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
5285 to query the identity of the peer of a local AF_UNIX
5286 connection. They operate similarly to their sd_pid_get_xyz()
5289 * PID 1 will now maintain a system-wide system state engine
5290 with the states "starting", "running", "degraded",
5291 "maintenance", "stopping". These states are bound to system
5292 startup, normal runtime, runtime with at least one failed
5293 service, rescue/emergency mode and system shutdown. This
5294 state is shown in the "systemctl status" output when no unit
5295 name is passed. It is useful to determine system state, in
5296 particularly when doing so for many systems or containers at
5299 * A new command "list-machines" has been added to "systemctl"
5300 that lists all local OS containers and shows their system
5301 state (see above), if systemd runs inside of them.
5303 * systemctl gained a new "-r" switch to recursively enumerate
5304 units on all local containers, when used with the
5305 "list-unit" command (which is the default one that is
5306 executed when no parameters are specified).
5308 * The GPT automatic partition discovery logic will now honour
5309 two GPT partition flags: one may be set on a partition to
5310 cause it to be mounted read-only, and the other may be set
5311 on a partition to ignore it during automatic discovery.
5313 * Two new GPT type UUIDs have been added for automatic root
5314 partition discovery, for 32-bit and 64-bit ARM. This is not
5315 particularly useful for discovering the root directory on
5316 these architectures during bare-metal boots (since UEFI is
5317 not common there), but still very useful to allow booting of
5318 ARM disk images in nspawn with the -i option.
5320 * MAC addresses of interfaces created with nspawn's
5321 --network-interface= switch will now be generated from the
5322 machine name, and thus be stable between multiple invocations
5325 * logind will now automatically remove all IPC objects owned
5326 by a user if she or he fully logs out. This makes sure that
5327 users who are logged out cannot continue to consume IPC
5328 resources. This covers SysV memory, semaphores and message
5329 queues as well as POSIX shared memory and message
5330 queues. Traditionally, SysV and POSIX IPC had no life-cycle
5331 limits. With this functionality, that is corrected. This may
5332 be turned off by using the RemoveIPC= switch of logind.conf.
5334 * The systemd-machine-id-setup and tmpfiles tools gained a
5335 --root= switch to operate on a specific root directory,
5338 * journald can now forward logged messages to the TTYs of all
5339 logged in users ("wall"). This is the default for all
5340 emergency messages now.
5342 * A new tool systemd-journal-remote has been added to stream
5343 journal log messages across the network.
5345 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
5346 controller trees are mounted into it. Note that the
5347 directories mounted beneath it are not read-only. This is a
5348 security measure and is particularly useful because glibc
5349 actually includes a search logic to pick any tmpfs it can
5350 find to implement shm_open() if /dev/shm is not available
5351 (which it might very well be in namespaced setups).
5353 * machinectl gained a new "poweroff" command to cleanly power
5354 down a local OS container.
5356 * The PrivateDevices= unit file setting will now also drop the
5357 CAP_MKNOD capability from the capability bound set, and
5358 imply DevicePolicy=closed.
5360 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
5361 comprehensively on all long-running systemd services where
5362 this is appropriate.
5364 * systemd-udevd will now run in a disassociated mount
5365 namespace. To mount directories from udev rules, make sure to
5366 pull in mount units via SYSTEMD_WANTS properties.
5368 * The kdbus support gained support for uploading policy into
5369 the kernel. sd-bus gained support for creating "monitoring"
5370 connections that can eavesdrop into all bus communication
5371 for debugging purposes.
5373 * Timestamps may now be specified in seconds since the UNIX
5374 epoch Jan 1st, 1970 by specifying "@" followed by the value
5377 * Native tcpwrap support in systemd has been removed. tcpwrap
5378 is old code, not really maintained anymore and has serious
5379 shortcomings, and better options such as firewalls
5380 exist. For setups that require tcpwrap usage, please
5381 consider invoking your socket-activated service via tcpd,
5382 like on traditional inetd.
5384 * A new system.conf configuration option
5385 DefaultTimerAccuracySec= has been added that controls the
5386 default AccuracySec= setting of .timer units.
5388 * Timer units gained a new WakeSystem= switch. If enabled,
5389 timers configured this way will cause the system to resume
5390 from system suspend (if the system supports that, which most
5393 * Timer units gained a new Persistent= switch. If enabled,
5394 timers configured this way will save to disk when they have
5395 been last triggered. This information is then used on next
5396 reboot to possible execute overdue timer events, that
5397 could not take place because the system was powered off.
5398 This enables simple anacron-like behaviour for timer units.
5400 * systemctl's "list-timers" will now also list the time a
5401 timer unit was last triggered in addition to the next time
5402 it will be triggered.
5404 * systemd-networkd will now assign predictable IPv4LL
5405 addresses to its local interfaces.
5407 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
5408 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
5409 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
5410 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
5411 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
5412 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
5413 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
5414 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
5417 — Berlin, 2014-03-25
5421 * A new unit file setting RestrictAddressFamilies= has been
5422 added to restrict which socket address families unit
5423 processes gain access to. This takes address family names
5424 like "AF_INET" or "AF_UNIX", and is useful to minimize the
5425 attack surface of services via exotic protocol stacks. This
5426 is built on seccomp system call filters.
5428 * Two new unit file settings RuntimeDirectory= and
5429 RuntimeDirectoryMode= have been added that may be used to
5430 manage a per-daemon runtime directories below /run. This is
5431 an alternative for setting up directory permissions with
5432 tmpfiles snippets, and has the advantage that the runtime
5433 directory's lifetime is bound to the daemon runtime and that
5434 the daemon starts up with an empty directory each time. This
5435 is particularly useful when writing services that drop
5436 privileges using the User= or Group= setting.
5438 * The DeviceAllow= unit setting now supports globbing for
5439 matching against device group names.
5441 * The systemd configuration file system.conf gained new
5442 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
5443 DefaultMemoryAccounting= to globally turn on/off accounting
5444 for specific resources (cgroups) for all units. These
5445 settings may still be overridden individually in each unit
5448 * systemd-gpt-auto-generator is now able to discover /srv and
5449 root partitions in addition to /home and swap partitions. It
5450 also supports LUKS-encrypted partitions now. With this in
5451 place, automatic discovery of partitions to mount following
5452 the Discoverable Partitions Specification
5453 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
5454 is now a lot more complete. This allows booting without
5455 /etc/fstab and without root= on the kernel command line on
5456 systems prepared appropriately.
5458 * systemd-nspawn gained a new --image= switch which allows
5459 booting up disk images and Linux installations on any block
5460 device that follow the Discoverable Partitions Specification
5461 (see above). This means that installations made with
5462 appropriately updated installers may now be started and
5463 deployed using container managers, completely
5464 unmodified. (We hope that libvirt-lxc will add support for
5465 this feature soon, too.)
5467 * systemd-nspawn gained a new --network-macvlan= setting to
5468 set up a private macvlan interface for the
5469 container. Similarly, systemd-networkd gained a new
5470 Kind=macvlan setting in .netdev files.
5472 * systemd-networkd now supports configuring local addresses
5475 * A new tool systemd-network-wait-online has been added to
5476 synchronously wait for network connectivity using
5479 * The sd-bus.h bus API gained a new sd_bus_track object for
5480 tracking the life-cycle of bus peers. Note that sd-bus.h is
5481 still not a public API though (unless you specify
5482 --enable-kdbus on the configure command line, which however
5483 voids your warranty and you get no API stability guarantee).
5485 * The $XDG_RUNTIME_DIR runtime directories for each user are
5486 now individual tmpfs instances, which has the benefit of
5487 introducing separate pools for each user, with individual
5488 size limits, and thus making sure that unprivileged clients
5489 can no longer negatively impact the system or other users by
5490 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
5491 RuntimeDirectorySize= has been introduced that allows
5492 controlling the default size limit for all users. It
5493 defaults to 10% of the available physical memory. This is no
5494 replacement for quotas on tmpfs though (which the kernel
5495 still does not support), as /dev/shm and /tmp are still
5496 shared resources used by both the system and unprivileged
5499 * logind will now automatically turn off automatic suspending
5500 on laptop lid close when more than one display is
5501 connected. This was previously expected to be implemented
5502 individually in desktop environments (such as GNOME),
5503 however has been added to logind now, in order to fix a
5504 boot-time race where a desktop environment might not have
5505 been started yet and thus not been able to take an inhibitor
5506 lock at the time where logind already suspends the system
5507 due to a closed lid.
5509 * logind will now wait at least 30s after each system
5510 suspend/resume cycle, and 3min after system boot before
5511 suspending the system due to a closed laptop lid. This
5512 should give USB docking stations and similar enough time to
5513 be probed and configured after system resume and boot in
5514 order to then act as suspend blocker.
5516 * systemd-run gained a new --property= setting which allows
5517 initialization of resource control properties (and others)
5518 for the created scope or service unit. Example: "systemd-run
5519 --property=BlockIOWeight=10 updatedb" may be used to run
5520 updatedb at a low block IO scheduling weight.
5522 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
5523 now also work in --scope mode.
5525 * When systemd is compiled with kdbus support, basic support
5526 for enforced policies is now in place. (Note that enabling
5527 kdbus still voids your warranty and no API compatibility
5530 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
5531 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5532 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
5533 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
5534 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
5535 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
5536 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
5537 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
5538 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
5539 Zbigniew Jędrzejewski-Szmek
5541 — Berlin, 2014-03-12
5545 * systemd will now relabel /dev after loading the SMACK policy
5546 according to SMACK rules.
5548 * A new unit file option AppArmorProfile= has been added to
5549 set the AppArmor profile for the processes of a unit.
5551 * A new condition check ConditionArchitecture= has been added
5552 to conditionalize units based on the system architecture, as
5553 reported by uname()'s "machine" field.
5555 * systemd-networkd now supports matching on the system
5556 virtualization, architecture, kernel command line, host name
5559 * logind is now a lot more aggressive when suspending the
5560 machine due to a closed laptop lid. Instead of acting only
5561 on the lid close action, it will continuously watch the lid
5562 status and act on it. This is useful for laptops where the
5563 power button is on the outside of the chassis so that it can
5564 be reached without opening the lid (such as the Lenovo
5565 Yoga). On those machines, logind will now immediately
5566 re-suspend the machine if the power button has been
5567 accidentally pressed while the laptop was suspended and in a
5568 backpack or similar.
5570 * logind will now watch SW_DOCK switches and inhibit reaction
5571 to the lid switch if it is pressed. This means that logind
5572 will not suspend the machine anymore if the lid is closed
5573 and the system is docked, if the laptop supports SW_DOCK
5574 notifications via the input layer. Note that ACPI docking
5575 stations do not generate this currently. Also note that this
5576 logic is usually not fully sufficient and Desktop
5577 Environments should take a lid switch inhibitor lock when an
5578 external display is connected, as systemd will not watch
5581 * nspawn will now make use of the devices cgroup controller by
5582 default, and only permit creation of and access to the usual
5583 API device nodes like /dev/null or /dev/random, as well as
5584 access to (but not creation of) the pty devices.
5586 * We will now ship a default .network file for
5587 systemd-networkd that automatically configures DHCP for
5588 network interfaces created by nspawn's --network-veth or
5589 --network-bridge= switches.
5591 * systemd will now understand the usual M, K, G, T suffixes
5592 according to SI conventions (i.e. to the base 1000) when
5593 referring to throughput and hardware metrics. It will stay
5594 with IEC conventions (i.e. to the base 1024) for software
5595 metrics, according to what is customary according to
5596 Wikipedia. We explicitly document which base applies for
5597 each configuration option.
5599 * The DeviceAllow= setting in unit files now supports a syntax
5600 to whitelist an entire group of devices node majors at once,
5601 based on the /proc/devices listing. For example, with the
5602 string "char-pts", it is now possible to whitelist all
5603 current and future pseudo-TTYs at once.
5605 * sd-event learned a new "post" event source. Event sources of
5606 this type are triggered by the dispatching of any event
5607 source of a type that is not "post". This is useful for
5608 implementing clean-up and check event sources that are
5609 triggered by other work being done in the program.
5611 * systemd-networkd is no longer statically enabled, but uses
5612 the usual [Install] sections so that it can be
5613 enabled/disabled using systemctl. It still is enabled by
5616 * When creating a veth interface pair with systemd-nspawn, the
5617 host side will now be prefixed with "vb-" if
5618 --network-bridge= is used, and with "ve-" if --network-veth
5619 is used. This way, it is easy to distinguish these cases on
5620 the host, for example to apply different configuration to
5621 them with systemd-networkd.
5623 * The compatibility libraries for libsystemd-journal.so,
5624 libsystem-id128.so, libsystemd-login.so and
5625 libsystemd-daemon.so do not make use of IFUNC
5626 anymore. Instead, we now build libsystemd.so multiple times
5627 under these alternative names. This means that the footprint
5628 is drastically increased, but given that these are
5629 transitional compatibility libraries, this should not matter
5630 much. This change has been made necessary to support the ARM
5631 platform for these compatibility libraries, as the ARM
5632 toolchain is not really at the same level as the toolchain
5633 for other architectures like x86 and does not support
5634 IFUNC. Please make sure to use --enable-compat-libs only
5635 during a transitional period!
5637 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
5638 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5639 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
5640 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
5641 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
5642 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
5643 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
5644 Zbigniew Jędrzejewski-Szmek
5646 — Berlin, 2014-02-24
5650 * A new component "systemd-networkd" has been added that can
5651 be used to configure local network interfaces statically or
5652 via DHCP. It is capable of bringing up bridges, VLANs, and
5653 bonding. Currently, no hook-ups for interactive network
5654 configuration are provided. Use this for your initrd,
5655 container, embedded, or server setup if you need a simple,
5656 yet powerful, network configuration solution. This
5657 configuration subsystem is quite nifty, as it allows wildcard
5658 hotplug matching in interfaces. For example, with a single
5659 configuration snippet, you can configure that all Ethernet
5660 interfaces showing up are automatically added to a bridge,
5661 or similar. It supports link-sensing and more.
5663 * A new tool "systemd-socket-proxyd" has been added which can
5664 act as a bidirectional proxy for TCP sockets. This is
5665 useful for adding socket activation support to services that
5666 do not actually support socket activation, including virtual
5667 machines and the like.
5669 * Add a new tool to save/restore rfkill state on
5672 * Save/restore state of keyboard backlights in addition to
5673 display backlights on shutdown/boot.
5675 * udev learned a new SECLABEL{} construct to label device
5676 nodes with a specific security label when they appear. For
5677 now, only SECLABEL{selinux} is supported, but the syntax is
5678 prepared for additional security frameworks.
5680 * udev gained a new scheme to configure link-level attributes
5681 from files in /etc/systemd/network/*.link. These files can
5682 match against MAC address, device path, driver name and type,
5683 and will apply attributes like the naming policy, link speed,
5684 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
5685 address assignment policy (randomized, ...).
5687 * The configuration of network interface naming rules for
5688 "permanent interface names" has changed: a new NamePolicy=
5689 setting in the [Link] section of .link files determines the
5690 priority of possible naming schemes (onboard, slot, MAC,
5691 path). The default value of this setting is determined by
5692 /usr/lib/net/links/99-default.link. Old
5693 80-net-name-slot.rules udev configuration file has been
5694 removed, so local configuration overriding this file should
5695 be adapted to override 99-default.link instead.
5697 * When the User= switch is used in a unit file, also
5698 initialize $SHELL= based on the user database entry.
5700 * systemd no longer depends on libdbus. All communication is
5701 now done with sd-bus, systemd's low-level bus library
5704 * kdbus support has been added to PID 1 itself. When kdbus is
5705 enabled, this causes PID 1 to set up the system bus and
5706 enable support for a new ".busname" unit type that
5707 encapsulates bus name activation on kdbus. It works a little
5708 bit like ".socket" units, except for bus names. A new
5709 generator has been added that converts classic dbus1 service
5710 activation files automatically into native systemd .busname
5713 * sd-bus: add a light-weight vtable implementation that allows
5714 defining objects on the bus with a simple static const
5715 vtable array of its methods, signals and properties.
5717 * systemd will not generate or install static dbus
5718 introspection data anymore to /usr/share/dbus-1/interfaces,
5719 as the precise format of these files is unclear, and
5720 nothing makes use of it.
5722 * A proxy daemon is now provided to proxy clients connecting
5723 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
5724 compatibility with classic D-Bus.
5726 * A bus driver implementation has been added that supports the
5727 classic D-Bus bus driver calls on kdbus, also for
5728 compatibility purposes.
5730 * A new API "sd-event.h" has been added that implements a
5731 minimal event loop API built around epoll. It provides a
5732 couple of features that direct epoll usage is lacking:
5733 prioritization of events, scales to large numbers of timer
5734 events, per-event timer slack (accuracy), system-wide
5735 coalescing of timer events, exit handlers, watchdog
5736 supervision support using systemd's sd_notify() API, child
5739 * A new API "sd-rntl.h" has been added that provides an API
5740 around the route netlink interface of the kernel, similar in
5741 style to "sd-bus.h".
5743 * A new API "sd-dhcp-client.h" has been added that provides a
5744 small DHCPv4 client-side implementation. This is used by
5747 * There is a new kernel command line option
5748 "systemd.restore_state=0|1". When set to "0", none of the
5749 systemd tools will restore saved runtime state to hardware
5750 devices. More specifically, the rfkill and backlight states
5753 * The FsckPassNo= compatibility option in mount/service units
5754 has been removed. The fstab generator will now add the
5755 necessary dependencies automatically, and does not require
5756 PID1's support for that anymore.
5758 * journalctl gained a new switch, --list-boots, that lists
5759 recent boots with their times and boot IDs.
5761 * The various tools like systemctl, loginctl, timedatectl,
5762 busctl, systemd-run, ... have gained a new switch "-M" to
5763 connect to a specific, local OS container (as direct
5764 connection, without requiring SSH). This works on any
5765 container that is registered with machined, such as those
5766 created by libvirt-lxc or nspawn.
5768 * systemd-run and systemd-analyze also gained support for "-H"
5769 to connect to remote hosts via SSH. This is particularly
5770 useful for systemd-run because it enables queuing of jobs
5771 onto remote systems.
5773 * machinectl gained a new command "login" to open a getty
5774 login in any local container. This works with any container
5775 that is registered with machined (such as those created by
5776 libvirt-lxc or nspawn), and which runs systemd inside.
5778 * machinectl gained a new "reboot" command that may be used to
5779 trigger a reboot on a specific container that is registered
5780 with machined. This works on any container that runs an init
5781 system of some kind.
5783 * systemctl gained a new "list-timers" command to print a nice
5784 listing of installed timer units with the times they elapse
5787 * Alternative reboot() parameters may now be specified on the
5788 "systemctl reboot" command line and are passed to the
5789 reboot() system call.
5791 * systemctl gained a new --job-mode= switch to configure the
5792 mode to queue a job with. This is a more generic version of
5793 --fail, --irreversible, and --ignore-dependencies, which are
5794 still available but not advertised anymore.
5796 * /etc/systemd/system.conf gained new settings to configure
5797 various default timeouts of units, as well as the default
5798 start limit interval and burst. These may still be overridden
5801 * PID1 will now export on the bus profile data of the security
5802 policy upload process (such as the SELinux policy upload to
5805 * journald: when forwarding logs to the console, include
5806 timestamps (following the setting in
5807 /sys/module/printk/parameters/time).
5809 * OnCalendar= in timer units now understands the special
5810 strings "yearly" and "annually". (Both are equivalent)
5812 * The accuracy of timer units is now configurable with the new
5813 AccuracySec= setting. It defaults to 1min.
5815 * A new dependency type JoinsNamespaceOf= has been added that
5816 allows running two services within the same /tmp and network
5817 namespace, if PrivateNetwork= or PrivateTmp= are used.
5819 * A new command "cat" has been added to systemctl. It outputs
5820 the original unit file of a unit, and concatenates the
5821 contents of additional "drop-in" unit file snippets, so that
5822 the full configuration is shown.
5824 * systemctl now supports globbing on the various "list-xyz"
5825 commands, like "list-units" or "list-sockets", as well as on
5826 those commands which take multiple unit names.
5828 * journalctl's --unit= switch gained support for globbing.
5830 * All systemd daemons now make use of the watchdog logic so
5831 that systemd automatically notices when they hang.
5833 * If the $container_ttys environment variable is set,
5834 getty-generator will automatically spawn a getty for each
5835 listed tty. This is useful for container managers to request
5836 login gettys to be spawned on as many ttys as needed.
5838 * %h, %s, %U specifier support is not available anymore when
5839 used in unit files for PID 1. This is because NSS calls are
5840 not safe from PID 1. They stay available for --user
5841 instances of systemd, and as special case for the root user.
5843 * loginctl gained a new "--no-legend" switch to turn off output
5846 * The "sd-login.h" API gained three new calls:
5847 sd_session_is_remote(), sd_session_get_remote_user(),
5848 sd_session_get_remote_host() to query information about
5851 * The udev hardware database now also carries vendor/product
5852 information of SDIO devices.
5854 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
5855 determine whether watchdog notifications are requested by
5858 * Socket-activated per-connection services now include a
5859 short description of the connection parameters in the
5862 * tmpfiles gained a new "--boot" option. When this is not used,
5863 only lines where the command character is not suffixed with
5864 "!" are executed. When this option is specified, those
5865 options are executed too. This partitions tmpfiles
5866 directives into those that can be safely executed at any
5867 time, and those which should be run only at boot (for
5868 example, a line that creates /run/nologin).
5870 * A new API "sd-resolve.h" has been added which provides a simple
5871 asynchronous wrapper around glibc NSS host name resolution
5872 calls, such as getaddrinfo(). In contrast to glibc's
5873 getaddrinfo_a(), it does not use signals. In contrast to most
5874 other asynchronous name resolution libraries, this one does
5875 not reimplement DNS, but reuses NSS, so that alternate
5876 host name resolution systems continue to work, such as mDNS,
5877 LDAP, etc. This API is based on libasyncns, but it has been
5878 cleaned up for inclusion in systemd.
5880 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
5881 "sd-daemon.h" are no longer found in individual libraries
5882 libsystemd-journal.so, libsystemd-login.so,
5883 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
5884 merged them into a single library, libsystemd.so, which
5885 provides all symbols. The reason for this is cyclic
5886 dependencies, as these libraries tend to use each other's
5887 symbols. So far, we have managed to workaround that by linking
5888 a copy of a good part of our code into each of these
5889 libraries again and again, which, however, makes certain
5890 things hard to do, like sharing static variables. Also, it
5891 substantially increases footprint. With this change, there
5892 is only one library for the basic APIs systemd
5893 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
5894 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
5895 library as well, however are subject to the --enable-kdbus
5896 switch (see below). Note that "sd-dhcp-client.h" is not part
5897 of this library (this is because it only consumes, never
5898 provides, services of/to other APIs). To make the transition
5899 easy from the separate libraries to the unified one, we
5900 provide the --enable-compat-libs compile-time switch which
5901 will generate stub libraries that are compatible with the
5902 old ones but redirect all calls to the new one.
5904 * All of the kdbus logic and the new APIs "sd-bus.h",
5905 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
5906 and "sd-utf8.h" are compile-time optional via the
5907 "--enable-kdbus" switch, and they are not compiled in by
5908 default. To make use of kdbus, you have to explicitly enable
5909 the switch. Note however, that neither the kernel nor the
5910 userspace API for all of this is considered stable yet. We
5911 want to maintain the freedom to still change the APIs for
5912 now. By specifying this build-time switch, you acknowledge
5913 that you are aware of the instability of the current
5916 * Also, note that while kdbus is pretty much complete,
5917 it lacks one thing: proper policy support. This means you
5918 can build a fully working system with all features; however,
5919 it will be highly insecure. Policy support will be added in
5920 one of the next releases, at the same time that we will
5921 declare the APIs stable.
5923 * When the kernel command line argument "kdbus" is specified,
5924 systemd will automatically load the kdbus.ko kernel module. At
5925 this stage of development, it is only useful for testing kdbus
5926 and should not be used in production. Note: if "--enable-kdbus"
5927 is specified, and the kdbus.ko kernel module is available, and
5928 "kdbus" is added to the kernel command line, the entire system
5929 runs with kdbus instead of dbus-daemon, with the above mentioned
5930 problem of missing the system policy enforcement. Also a future
5931 version of kdbus.ko or a newer systemd will not be compatible with
5932 each other, and will unlikely be able to boot the machine if only
5933 one of them is updated.
5935 * systemctl gained a new "import-environment" command which
5936 uploads the caller's environment (or parts thereof) into the
5937 service manager so that it is inherited by services started
5938 by the manager. This is useful to upload variables like
5939 $DISPLAY into the user service manager.
5941 * A new PrivateDevices= switch has been added to service units
5942 which allows running a service with a namespaced /dev
5943 directory that does not contain any device nodes for
5944 physical devices. More specifically, it only includes devices
5945 such as /dev/null, /dev/urandom, and /dev/zero which are API
5948 * logind has been extended to support behaviour like VT
5949 switching on seats that do not support a VT. This makes
5950 multi-session available on seats that are not the first seat
5951 (seat0), and on systems where kernel support for VTs has
5952 been disabled at compile-time.
5954 * If a process holds a delay lock for system sleep or shutdown
5955 and fails to release it in time, we will now log its
5956 identity. This makes it easier to identify processes that
5957 cause slow suspends or power-offs.
5959 * When parsing /etc/crypttab, support for a new key-slot=
5960 option as supported by Debian is added. It allows indicating
5961 which LUKS slot to use on disk, speeding up key loading.
5963 * The sd_journal_sendv() API call has been checked and
5964 officially declared to be async-signal-safe so that it may
5965 be invoked from signal handlers for logging purposes.
5967 * Boot-time status output is now enabled automatically after a
5968 short timeout if boot does not progress, in order to give
5969 the user an indication what she or he is waiting for.
5971 * The boot-time output has been improved to show how much time
5972 remains until jobs expire.
5974 * The KillMode= switch in service units gained a new possible
5975 value "mixed". If set, and the unit is shut down, then the
5976 initial SIGTERM signal is sent only to the main daemon
5977 process, while the following SIGKILL signal is sent to
5978 all remaining processes of the service.
5980 * When a scope unit is registered, a new property "Controller"
5981 may be set. If set to a valid bus name, systemd will send a
5982 RequestStop() signal to this name when it would like to shut
5983 down the scope. This may be used to hook manager logic into
5984 the shutdown logic of scope units. Also, scope units may now
5985 be put in a special "abandoned" state, in which case the
5986 manager process which created them takes no further
5987 responsibilities for it.
5989 * When reading unit files, systemd will now verify
5990 the access mode of these files, and warn about certain
5991 suspicious combinations. This has been added to make it
5992 easier to track down packaging bugs where unit files are
5993 marked executable or world-writable.
5995 * systemd-nspawn gained a new "--setenv=" switch to set
5996 container-wide environment variables. The similar option in
5997 systemd-activate was renamed from "--environment=" to
5998 "--setenv=" for consistency.
6000 * systemd-nspawn has been updated to create a new kdbus domain
6001 for each container that is invoked, thus allowing each
6002 container to have its own set of system and user buses,
6003 independent of the host.
6005 * systemd-nspawn gained a new --drop-capability= switch to run
6006 the container with less capabilities than the default. Both
6007 --drop-capability= and --capability= now take the special
6008 string "all" for dropping or keeping all capabilities.
6010 * systemd-nspawn gained new switches for executing containers
6011 with specific SELinux labels set.
6013 * systemd-nspawn gained a new --quiet switch to not generate
6014 any additional output but the container's own console
6017 * systemd-nspawn gained a new --share-system switch to run a
6018 container without PID namespacing enabled.
6020 * systemd-nspawn gained a new --register= switch to control
6021 whether the container is registered with systemd-machined or
6022 not. This is useful for containers that do not run full
6023 OS images, but only specific apps.
6025 * systemd-nspawn gained a new --keep-unit which may be used
6026 when invoked as the only program from a service unit, and
6027 results in registration of the unit service itself in
6028 systemd-machined, instead of a newly opened scope unit.
6030 * systemd-nspawn gained a new --network-interface= switch for
6031 moving arbitrary interfaces to the container. The new
6032 --network-veth switch creates a virtual Ethernet connection
6033 between host and container. The new --network-bridge=
6034 switch then allows assigning the host side of this virtual
6035 Ethernet connection to a bridge device.
6037 * systemd-nspawn gained a new --personality= switch for
6038 setting the kernel personality for the container. This is
6039 useful when running a 32-bit container on a 64-bit host. A
6040 similar option Personality= is now also available for service
6043 * logind will now also track a "Desktop" identifier for each
6044 session which encodes the desktop environment of it. This is
6045 useful for desktop environments that want to identify
6046 multiple running sessions of itself easily.
6048 * A new SELinuxContext= setting for service units has been
6049 added that allows setting a specific SELinux execution
6050 context for a service.
6052 * Most systemd client tools will now honour $SYSTEMD_LESS for
6053 settings of the "less" pager. By default, these tools will
6054 override $LESS to allow certain operations to work, such as
6055 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
6056 influence this logic.
6058 * systemd's "seccomp" hook-up has been changed to make use of
6059 the libseccomp library instead of using its own
6060 implementation. This has benefits for portability among
6063 * For usage together with SystemCallFilter=, a new
6064 SystemCallErrorNumber= setting has been introduced that
6065 allows configuration of a system error number to be returned
6066 on filtered system calls, instead of immediately killing the
6067 process. Also, SystemCallArchitectures= has been added to
6068 limit access to system calls of a particular architecture
6069 (in order to turn off support for unused secondary
6070 architectures). There is also a global
6071 SystemCallArchitectures= setting in system.conf now to turn
6072 off support for non-native system calls system-wide.
6074 * systemd requires a kernel with a working name_to_handle_at(),
6075 please see the kernel config requirements in the README file.
6077 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
6078 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
6079 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
6080 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
6081 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
6082 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
6083 Elia Pinto, Florian Weimer, George McCollister, Goffredo
6084 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
6085 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
6086 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
6087 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
6088 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
6089 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
6090 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
6091 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
6092 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
6093 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
6094 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
6095 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
6096 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
6097 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
6098 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
6099 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
6100 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
6102 — Berlin, 2014-02-20
6106 * logind has gained support for facilitating privileged input
6107 and drm device access for unprivileged clients. This work is
6108 useful to allow Wayland display servers (and similar
6109 programs, such as kmscon) to run under the user's ID and
6110 access input and drm devices which are normally
6111 protected. When this is used (and the kernel is new enough)
6112 logind will "mute" IO on the file descriptors passed to
6113 Wayland as long as it is in the background and "unmute" it
6114 if it returns into the foreground. This allows secure
6115 session switching without allowing background sessions to
6116 eavesdrop on input and display data. This also introduces
6117 session switching support if VT support is turned off in the
6118 kernel, and on seats that are not seat0.
6120 * A new kernel command line option luks.options= is understood
6121 now which allows specifying LUKS options for usage for LUKS
6122 encrypted partitions specified with luks.uuid=.
6124 * tmpfiles.d(5) snippets may now use specifier expansion in
6125 path names. More specifically %m, %b, %H, %v, are now
6126 replaced by the local machine id, boot id, hostname, and
6127 kernel version number.
6129 * A new tmpfiles.d(5) command "m" has been introduced which
6130 may be used to change the owner/group/access mode of a file
6131 or directory if it exists, but do nothing if it does not.
6133 * This release removes high-level support for the
6134 MemorySoftLimit= cgroup setting. The underlying kernel
6135 cgroup attribute memory.soft_limit= is currently badly
6136 designed and likely to be removed from the kernel API in its
6137 current form, hence we should not expose it for now.
6139 * The memory.use_hierarchy cgroup attribute is now enabled for
6140 all cgroups systemd creates in the memory cgroup
6141 hierarchy. This option is likely to be come the built-in
6142 default in the kernel anyway, and the non-hierarchical mode
6143 never made much sense in the intrinsically hierarchical
6146 * A new field _SYSTEMD_SLICE= is logged along with all journal
6147 messages containing the slice a message was generated
6148 from. This is useful to allow easy per-customer filtering of
6149 logs among other things.
6151 * systemd-journald will no longer adjust the group of journal
6152 files it creates to the "systemd-journal" group. Instead we
6153 rely on the journal directory to be owned by the
6154 "systemd-journal" group, and its setgid bit set, so that the
6155 kernel file system layer will automatically enforce that
6156 journal files inherit this group assignment. The reason for
6157 this change is that we cannot allow NSS look-ups from
6158 journald which would be necessary to resolve
6159 "systemd-journal" to a numeric GID, because this might
6160 create deadlocks if NSS involves synchronous queries to
6161 other daemons (such as nscd, or sssd) which in turn are
6162 logging clients of journald and might block on it, which
6163 would then dead lock. A tmpfiles.d(5) snippet included in
6164 systemd will make sure the setgid bit and group are
6165 properly set on the journal directory if it exists on every
6166 boot. However, we recommend adjusting it manually after
6167 upgrades too (or from RPM scriptlets), so that the change is
6168 not delayed until next reboot.
6170 * Backlight and random seed files in /var/lib/ have moved into
6171 the /var/lib/systemd/ directory, in order to centralize all
6172 systemd generated files in one directory.
6174 * Boot time performance measurements (as displayed by
6175 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
6176 performance information if that's available to determine how
6177 much time BIOS and boot loader initialization required. With
6178 a sufficiently new BIOS you hence no longer need to boot
6179 with Gummiboot to get access to such information.
6181 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
6182 Cristian Rodríguez, Dave Reisner, David Herrmann, David
6183 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
6184 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
6185 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
6186 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
6187 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6189 — Berlin, 2013-10-02
6193 * The Restart= option for services now understands a new
6194 on-watchdog setting, which will restart the service
6195 automatically if the service stops sending out watchdog keep
6196 alive messages (as configured with WatchdogSec=).
6198 * The getty generator (which is responsible for bringing up a
6199 getty on configured serial consoles) will no longer only
6200 start a getty on the primary kernel console but on all
6201 others, too. This makes the order in which console= is
6202 specified on the kernel command line less important.
6204 * libsystemd-logind gained a new sd_session_get_vt() call to
6205 retrieve the VT number of a session.
6207 * If the option "tries=0" is set for an entry of /etc/crypttab
6208 its passphrase is queried indefinitely instead of any
6209 maximum number of tries.
6211 * If a service with a configure PID file terminates its PID
6212 file will now be removed automatically if it still exists
6213 afterwards. This should put an end to stale PID files.
6215 * systemd-run will now also take relative binary path names
6216 for execution and no longer insists on absolute paths.
6218 * InaccessibleDirectories= and ReadOnlyDirectories= now take
6219 paths that are optionally prefixed with "-" to indicate that
6220 it should not be considered a failure if they do not exist.
6222 * journalctl -o (and similar commands) now understands a new
6223 output mode "short-precise", it is similar to "short" but
6224 shows timestamps with usec accuracy.
6226 * The option "discard" (as known from Debian) is now
6227 synonymous to "allow-discards" in /etc/crypttab. In fact,
6228 "discard" is preferred now (since it is easier to remember
6231 * Some licensing clean-ups were made, so that more code is now
6232 LGPL-2.1 licensed than before.
6234 * A minimal tool to save/restore the display backlight
6235 brightness across reboots has been added. It will store the
6236 backlight setting as late as possible at shutdown, and
6237 restore it as early as possible during reboot.
6239 * A logic to automatically discover and enable home and swap
6240 partitions on GPT disks has been added. With this in place
6241 /etc/fstab becomes optional for many setups as systemd can
6242 discover certain partitions located on the root disk
6243 automatically. Home partitions are recognized under their
6244 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
6245 partitions are recognized under their GPT type ID
6246 0657fd6da4ab43c484e50933c84b4f4f.
6248 * systemd will no longer pass any environment from the kernel
6249 or initrd to system services. If you want to set an
6250 environment for all services, do so via the kernel command
6251 line systemd.setenv= assignment.
6253 * The systemd-sysctl tool no longer natively reads the file
6254 /etc/sysctl.conf. If desired, the file should be symlinked
6255 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
6256 legacy support by a symlink rather than built-in code, it
6257 also makes the otherwise hidden order of application of the
6258 different files visible. (Note that this partly reverts to a
6259 pre-198 application order of sysctl knobs!)
6261 * The "systemctl set-log-level" and "systemctl dump" commands
6262 have been moved to systemd-analyze.
6264 * systemd-run learned the new --remain-after-exit switch,
6265 which causes the scope unit not to be cleaned up
6266 automatically after the process terminated.
6268 * tmpfiles learned a new --exclude-prefix= switch to exclude
6269 certain paths from operation.
6271 * journald will now automatically flush all messages to disk
6272 as soon as a message at the log level CRIT, ALERT or EMERG
6275 Contributions from: Andrew Cook, Brandon Philips, Christian
6276 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
6277 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
6278 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
6279 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
6280 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
6281 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
6282 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
6283 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
6284 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
6285 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
6286 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
6287 William Giokas, Zbigniew Jędrzejewski-Szmek
6289 — Berlin, 2013-09-13
6293 * The documentation has been updated to cover the various new
6294 concepts introduced with 205.
6296 * Unit files now understand the new %v specifier which
6297 resolves to the kernel version string as returned by "uname
6300 * systemctl now supports filtering the unit list output by
6301 load state, active state and sub state, using the new
6304 * "systemctl status" will now show the results of the
6305 condition checks (like ConditionPathExists= and similar) of
6306 the last start attempts of the unit. They are also logged to
6309 * "journalctl -b" may now be used to look for boot output of a
6310 specific boot. Try "journalctl -b -1" for the previous boot,
6311 but the syntax is substantially more powerful.
6313 * "journalctl --show-cursor" has been added which prints the
6314 cursor string the last shown log line. This may then be used
6315 with the new "journalctl --after-cursor=" switch to continue
6316 browsing logs from that point on.
6318 * "journalctl --force" may now be used to force regeneration
6321 * Creation of "dead" device nodes has been moved from udev
6322 into kmod and tmpfiles. Previously, udev would read the kmod
6323 databases to pre-generate dead device nodes based on meta
6324 information contained in kernel modules, so that these would
6325 be auto-loaded on access rather then at boot. As this
6326 does not really have much to do with the exposing actual
6327 kernel devices to userspace this has always been slightly
6328 alien in the udev codebase. Following the new scheme kmod
6329 will now generate a runtime snippet for tmpfiles from the
6330 module meta information and it now is tmpfiles' job to the
6331 create the nodes. This also allows overriding access and
6332 other parameters for the nodes using the usual tmpfiles
6333 facilities. As side effect this allows us to remove the
6334 CAP_SYS_MKNOD capability bit from udevd entirely.
6336 * logind's device ACLs may now be applied to these "dead"
6337 devices nodes too, thus finally allowing managed access to
6338 devices such as /dev/snd/sequencer without loading the
6339 backing module right-away.
6341 * A new RPM macro has been added that may be used to apply
6342 tmpfiles configuration during package installation.
6344 * systemd-detect-virt and ConditionVirtualization= now can
6345 detect User-Mode-Linux machines (UML).
6347 * journald will now implicitly log the effective capabilities
6348 set of processes in the message metadata.
6350 * systemd-cryptsetup has gained support for TrueCrypt volumes.
6352 * The initrd interface has been simplified (more specifically,
6353 support for passing performance data via environment
6354 variables and fsck results via files in /run has been
6355 removed). These features were non-essential, and are
6356 nowadays available in a much nicer way by having systemd in
6357 the initrd serialize its state and have the hosts systemd
6358 deserialize it again.
6360 * The udev "keymap" data files and tools to apply keyboard
6361 specific mappings of scan to key codes, and force-release
6362 scan code lists have been entirely replaced by a udev
6363 "keyboard" builtin and a hwdb data file.
6365 * systemd will now honour the kernel's "quiet" command line
6366 argument also during late shutdown, resulting in a
6367 completely silent shutdown when used.
6369 * There's now an option to control the SO_REUSEPORT socket
6370 option in .socket units.
6372 * Instance units will now automatically get a per-template
6373 subslice of system.slice unless something else is explicitly
6374 configured. For example, instances of sshd@.service will now
6375 implicitly be placed in system-sshd.slice rather than
6376 system.slice as before.
6378 * Test coverage support may now be enabled at build time.
6380 Contributions from: Dave Reisner, Frederic Crozat, Harald
6381 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
6382 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
6383 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
6384 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
6385 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
6386 Giokas, Zbigniew Jędrzejewski-Szmek
6388 — Berlin, 2013-07-23
6392 * Two new unit types have been introduced:
6394 Scope units are very similar to service units, however, are
6395 created out of pre-existing processes — instead of PID 1
6396 forking off the processes. By using scope units it is
6397 possible for system services and applications to group their
6398 own child processes (worker processes) in a powerful way
6399 which then maybe used to organize them, or kill them
6400 together, or apply resource limits on them.
6402 Slice units may be used to partition system resources in an
6403 hierarchical fashion and then assign other units to them. By
6404 default there are now three slices: system.slice (for all
6405 system services), user.slice (for all user sessions),
6406 machine.slice (for VMs and containers).
6408 Slices and scopes have been introduced primarily in
6409 context of the work to move cgroup handling to a
6410 single-writer scheme, where only PID 1
6411 creates/removes/manages cgroups.
6413 * There's a new concept of "transient" units. In contrast to
6414 normal units these units are created via an API at runtime,
6415 not from configuration from disk. More specifically this
6416 means it is now possible to run arbitrary programs as
6417 independent services, with all execution parameters passed
6418 in via bus APIs rather than read from disk. Transient units
6419 make systemd substantially more dynamic then it ever was,
6420 and useful as a general batch manager.
6422 * logind has been updated to make use of scope and slice units
6423 for managing user sessions. As a user logs in he will get
6424 his own private slice unit, to which all sessions are added
6425 as scope units. We also added support for automatically
6426 adding an instance of user@.service for the user into the
6427 slice. Effectively logind will no longer create cgroup
6428 hierarchies on its own now, it will defer entirely to PID 1
6429 for this by means of scope, service and slice units. Since
6430 user sessions this way become entities managed by PID 1
6431 the output of "systemctl" is now a lot more comprehensive.
6433 * A new mini-daemon "systemd-machined" has been added which
6434 may be used by virtualization managers to register local
6435 VMs/containers. nspawn has been updated accordingly, and
6436 libvirt will be updated shortly. machined will collect a bit
6437 of meta information about the VMs/containers, and assign
6438 them their own scope unit (see above). The collected
6439 meta-data is then made available via the "machinectl" tool,
6440 and exposed in "ps" and similar tools. machined/machinectl
6441 is compile-time optional.
6443 * As discussed earlier, the low-level cgroup configuration
6444 options ControlGroup=, ControlGroupModify=,
6445 ControlGroupPersistent=, ControlGroupAttribute= have been
6446 removed. Please use high-level attribute settings instead as
6447 well as slice units.
6449 * A new bus call SetUnitProperties() has been added to alter
6450 various runtime parameters of a unit. This is primarily
6451 useful to alter cgroup parameters dynamically in a nice way,
6452 but will be extended later on to make more properties
6453 modifiable at runtime. systemctl gained a new set-properties
6454 command that wraps this call.
6456 * A new tool "systemd-run" has been added which can be used to
6457 run arbitrary command lines as transient services or scopes,
6458 while configuring a number of settings via the command
6459 line. This tool is currently very basic, however already
6460 very useful. We plan to extend this tool to even allow
6461 queuing of execution jobs with time triggers from the
6462 command line, similar in fashion to "at".
6464 * nspawn will now inform the user explicitly that kernels with
6465 audit enabled break containers, and suggest the user to turn
6468 * Support for detecting the IMA and AppArmor security
6469 frameworks with ConditionSecurity= has been added.
6471 * journalctl gained a new "-k" switch for showing only kernel
6472 messages, mimicking dmesg output; in addition to "--user"
6473 and "--system" switches for showing only user's own logs
6476 * systemd-delta can now show information about drop-in
6477 snippets extending unit files.
6479 * libsystemd-bus has been substantially updated but is still
6480 not available as public API.
6482 * systemd will now look for the "debug" argument on the kernel
6483 command line and enable debug logging, similar to what
6484 "systemd.log_level=debug" already did before.
6486 * "systemctl set-default", "systemctl get-default" has been
6487 added to configure the default.target symlink, which
6488 controls what to boot into by default.
6490 * "systemctl set-log-level" has been added as a convenient
6491 way to raise and lower systemd logging threshold.
6493 * "systemd-analyze plot" will now show the time the various
6494 generators needed for execution, as well as information
6495 about the unit file loading.
6497 * libsystemd-journal gained a new sd_journal_open_files() call
6498 for opening specific journal files. journactl also gained a
6499 new switch to expose this new functionality. Previously we
6500 only supported opening all files from a directory, or all
6501 files from the system, as opening individual files only is
6502 racy due to journal file rotation.
6504 * systemd gained the new DefaultEnvironment= setting in
6505 /etc/systemd/system.conf to set environment variables for
6508 * If a privileged process logs a journal message with the
6509 OBJECT_PID= field set, then journald will automatically
6510 augment this with additional OBJECT_UID=, OBJECT_GID=,
6511 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
6512 system services want to log events about specific client
6513 processes. journactl/systemctl has been updated to make use
6514 of this information if all log messages regarding a specific
6517 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
6518 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
6519 Reisner, David Coppa, David King, David Strauss, Eelco
6520 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
6521 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
6522 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
6523 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
6524 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
6525 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
6526 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
6527 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
6528 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
6529 Łukasz Stelmach, 장동준
6533 * The Python bindings gained some minimal support for the APIs
6534 exposed by libsystemd-logind.
6536 * ConditionSecurity= gained support for detecting SMACK. Since
6537 this condition already supports SELinux and AppArmor we only
6538 miss IMA for this. Patches welcome!
6540 Contributions from: Karol Lewandowski, Lennart Poettering,
6541 Zbigniew Jędrzejewski-Szmek
6545 * systemd-nspawn will now create /etc/resolv.conf if
6546 necessary, before bind-mounting the host's file onto it.
6548 * systemd-nspawn will now store meta information about a
6549 container on the container's cgroup as extended attribute
6550 fields, including the root directory.
6552 * The cgroup hierarchy has been reworked in many ways. All
6553 objects any of the components systemd creates in the cgroup
6554 tree are now suffixed. More specifically, user sessions are
6555 now placed in cgroups suffixed with ".session", users in
6556 cgroups suffixed with ".user", and nspawn containers in
6557 cgroups suffixed with ".nspawn". Furthermore, all cgroup
6558 names are now escaped in a simple scheme to avoid collision
6559 of userspace object names with kernel filenames. This work
6560 is preparation for making these objects relocatable in the
6561 cgroup tree, in order to allow easy resource partitioning of
6562 these objects without causing naming conflicts.
6564 * systemctl list-dependencies gained the new switches
6565 --plain, --reverse, --after and --before.
6567 * systemd-inhibit now shows the process name of processes that
6568 have taken an inhibitor lock.
6570 * nss-myhostname will now also resolve "localhost"
6571 implicitly. This makes /etc/hosts an optional file and
6572 nicely handles that on IPv6 ::1 maps to both "localhost" and
6575 * libsystemd-logind.so gained a new call
6576 sd_get_machine_names() to enumerate running containers and
6577 VMs (currently only supported by very new libvirt and
6578 nspawn). sd_login_monitor can now be used to watch
6579 VMs/containers coming and going.
6581 * .include is not allowed recursively anymore, and only in
6582 unit files. Usually it is better to use drop-in snippets in
6583 .d/*.conf anyway, as introduced with systemd 198.
6585 * systemd-analyze gained a new "critical-chain" command that
6586 determines the slowest chain of units run during system
6587 boot-up. It is very useful for tracking down where
6588 optimizing boot time is the most beneficial.
6590 * systemd will no longer allow manipulating service paths in
6591 the name=systemd:/system cgroup tree using ControlGroup= in
6592 units. (But is still fine with it in all other dirs.)
6594 * There's a new systemd-nspawn@.service service file that may
6595 be used to easily run nspawn containers as system
6596 services. With the container's root directory in
6597 /var/lib/container/foobar it is now sufficient to run
6598 "systemctl start systemd-nspawn@foobar.service" to boot it.
6600 * systemd-cgls gained a new parameter "--machine" to list only
6601 the processes within a certain container.
6603 * ConditionSecurity= now can check for "apparmor". We still
6604 are lacking checks for SMACK and IMA for this condition
6605 check though. Patches welcome!
6607 * A new configuration file /etc/systemd/sleep.conf has been
6608 added that may be used to configure which kernel operation
6609 systemd is supposed to execute when "suspend", "hibernate"
6610 or "hybrid-sleep" is requested. This makes the new kernel
6611 "freeze" state accessible to the user.
6613 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
6614 the passed argument if applicable.
6616 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
6617 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
6618 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
6619 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
6620 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
6621 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
6622 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
6627 * The output of 'systemctl list-jobs' got some polishing. The
6628 '--type=' argument may now be passed more than once. A new
6629 command 'systemctl list-sockets' has been added which shows
6630 a list of kernel sockets systemd is listening on with the
6631 socket units they belong to, plus the units these socket
6634 * The experimental libsystemd-bus library got substantial
6635 updates to work in conjunction with the (also experimental)
6636 kdbus kernel project. It works well enough to exchange
6637 messages with some sophistication. Note that kdbus is not
6638 ready yet, and the library is mostly an elaborate test case
6639 for now, and not installable.
6641 * systemd gained a new unit 'systemd-static-nodes.service'
6642 that generates static device nodes earlier during boot, and
6643 can run in conjunction with udev.
6645 * libsystemd-login gained a new call sd_pid_get_user_unit()
6646 to retrieve the user systemd unit a process is running
6647 in. This is useful for systems where systemd is used as
6650 * systemd-nspawn now places all containers in the new /machine
6651 top-level cgroup directory in the name=systemd
6652 hierarchy. libvirt will soon do the same, so that we get a
6653 uniform separation of /system, /user and /machine for system
6654 services, user processes and containers/virtual
6655 machines. This new cgroup hierarchy is also useful to stick
6656 stable names to specific container instances, which can be
6657 recognized later this way (this name may be controlled
6658 via systemd-nspawn's new -M switch). libsystemd-login also
6659 gained a new call sd_pid_get_machine_name() to retrieve the
6660 name of the container/VM a specific process belongs to.
6662 * bootchart can now store its data in the journal.
6664 * libsystemd-journal gained a new call
6665 sd_journal_add_conjunction() for AND expressions to the
6666 matching logic. This can be used to express more complex
6667 logical expressions.
6669 * journactl can now take multiple --unit= and --user-unit=
6672 * The cryptsetup logic now understands the "luks.key=" kernel
6673 command line switch for specifying a file to read the
6674 decryption key from. Also, if a configured key file is not
6675 found the tool will now automatically fall back to prompting
6678 * Python systemd.journal module was updated to wrap recently
6679 added functions from libsystemd-journal. The interface was
6680 changed to bring the low level interface in s.j._Reader
6681 closer to the C API, and the high level interface in
6682 s.j.Reader was updated to wrap and convert all data about
6685 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
6686 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
6687 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
6688 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
6689 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
6690 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6694 * journalctl --update-catalog now understands a new --root=
6695 option to operate on catalogs found in a different root
6698 * During shutdown after systemd has terminated all running
6699 services a final killing loop kills all remaining left-over
6700 processes. We will now print the name of these processes
6701 when we send SIGKILL to them, since this usually indicates a
6704 * If /etc/crypttab refers to password files stored on
6705 configured mount points automatic dependencies will now be
6706 generated to ensure the specific mount is established first
6707 before the key file is attempted to be read.
6709 * 'systemctl status' will now show information about the
6710 network sockets a socket unit is listening on.
6712 * 'systemctl status' will also shown information about any
6713 drop-in configuration file for units. (Drop-In configuration
6714 files in this context are files such as
6715 /etc/systemd/systemd/foobar.service.d/*.conf)
6717 * systemd-cgtop now optionally shows summed up CPU times of
6718 cgroups. Press '%' while running cgtop to switch between
6719 percentage and absolute mode. This is useful to determine
6720 which cgroups use up the most CPU time over the entire
6721 runtime of the system. systemd-cgtop has also been updated
6722 to be 'pipeable' for processing with further shell tools.
6724 * 'hostnamectl set-hostname' will now allow setting of FQDN
6727 * The formatting and parsing of time span values has been
6728 changed. The parser now understands fractional expressions
6729 such as "5.5h". The formatter will now output fractional
6730 expressions for all time spans under 1min, i.e. "5.123456s"
6731 rather than "5s 123ms 456us". For time spans under 1s
6732 millisecond values are shown, for those under 1ms
6733 microsecond values are shown. This should greatly improve
6734 all time-related output of systemd.
6736 * libsystemd-login and libsystemd-journal gained new
6737 functions for querying the poll() events mask and poll()
6738 timeout value for integration into arbitrary event
6741 * localectl gained the ability to list available X11 keymaps
6742 (models, layouts, variants, options).
6744 * 'systemd-analyze dot' gained the ability to filter for
6745 specific units via shell-style globs, to create smaller,
6746 more useful graphs. I.e. it is now possible to create simple
6747 graphs of all the dependencies between only target units, or
6748 of all units that Avahi has dependencies with.
6750 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
6751 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
6752 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
6753 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
6754 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
6755 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
6756 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
6760 * The boot-time readahead implementation for rotating media
6761 will now read the read-ahead data in multiple passes which
6762 consist of all read requests made in equidistant time
6763 intervals. This means instead of strictly reading read-ahead
6764 data in its physical order on disk we now try to find a
6765 middle ground between physical and access time order.
6767 * /etc/os-release files gained a new BUILD_ID= field for usage
6768 on operating systems that provide continuous builds of OS
6771 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
6772 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
6773 William Douglas, Zbigniew Jędrzejewski-Szmek
6777 * systemd-python gained an API exposing libsystemd-daemon.
6779 * The SMACK setup logic gained support for uploading CIPSO
6782 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
6783 ReadOnlyDirectories= and InaccessibleDirectories= has
6784 changed. The private /tmp and /var/tmp directories are now
6785 shared by all processes of a service (which means
6786 ExecStartPre= may now leave data in /tmp that ExecStart= of
6787 the same service can still access). When a service is
6788 stopped its temporary directories are immediately deleted
6789 (normal clean-up with tmpfiles is still done in addition to
6792 * By default, systemd will now set a couple of sysctl
6793 variables in the kernel: the safe sysrq options are turned
6794 on, IP route verification is turned on, and source routing
6795 disabled. The recently added hardlink and softlink
6796 protection of the kernel is turned on. These settings should
6797 be reasonably safe, and good defaults for all new systems.
6799 * The predictable network naming logic may now be turned off
6800 with a new kernel command line switch: net.ifnames=0.
6802 * A new libsystemd-bus module has been added that implements a
6803 pretty complete D-Bus client library. For details see:
6805 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
6807 * journald will now explicitly flush the journal files to disk
6808 at the latest 5min after each write. The file will then also
6809 be marked offline until the next write. This should increase
6810 reliability in case of a crash. The synchronization delay
6811 can be configured via SyncIntervalSec= in journald.conf.
6813 * There's a new remote-fs-setup.target unit that can be used
6814 to pull in specific services when at least one remote file
6815 system is to be mounted.
6817 * There are new targets timers.target and paths.target as
6818 canonical targets to pull user timer and path units in
6819 from. This complements sockets.target with a similar
6820 purpose for socket units.
6822 * libudev gained a new call udev_device_set_attribute_value()
6823 to set sysfs attributes of a device.
6825 * The udev daemon now sets the default number of worker
6826 processes executed in parallel based on the number of available
6827 CPUs instead of the amount of available RAM. This is supposed
6828 to provide a more reliable default and limit a too aggressive
6829 parallelism for setups with 1000s of devices connected.
6831 Contributions from: Auke Kok, Colin Walters, Cristian
6832 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
6833 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
6834 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
6835 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
6836 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
6837 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
6838 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
6839 Zbigniew Jędrzejewski-Szmek
6843 * Configuration of unit files may now be extended via drop-in
6844 files without having to edit/override the unit files
6845 themselves. More specifically, if the administrator wants to
6846 change one value for a service file foobar.service he can
6847 now do so by dropping in a configuration snippet into
6848 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
6849 will load all these snippets and apply them on top of the
6850 main unit configuration file, possibly extending or
6851 overriding its settings. Using these drop-in snippets is
6852 generally nicer than the two earlier options for changing
6853 unit files locally: copying the files from
6854 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
6855 them there; or creating a new file in /etc/systemd/system/
6856 that incorporates the original one via ".include". Drop-in
6857 snippets into these .d/ directories can be placed in any
6858 directory systemd looks for units in, and the usual
6859 overriding semantics between /usr/lib, /etc and /run apply
6862 * Most unit file settings which take lists of items can now be
6863 reset by assigning the empty string to them. For example,
6864 normally, settings such as Environment=FOO=BAR append a new
6865 environment variable assignment to the environment block,
6866 each time they are used. By assigning Environment= the empty
6867 string the environment block can be reset to empty. This is
6868 particularly useful with the .d/*.conf drop-in snippets
6869 mentioned above, since this adds the ability to reset list
6870 settings from vendor unit files via these drop-ins.
6872 * systemctl gained a new "list-dependencies" command for
6873 listing the dependencies of a unit recursively.
6875 * Inhibitors are now honored and listed by "systemctl
6876 suspend", "systemctl poweroff" (and similar) too, not only
6877 GNOME. These commands will also list active sessions by
6880 * Resource limits (as exposed by the various control group
6881 controllers) can now be controlled dynamically at runtime
6882 for all units. More specifically, you can now use a command
6883 like "systemctl set-cgroup-attr foobar.service cpu.shares
6884 2000" to alter the CPU shares a specific service gets. These
6885 settings are stored persistently on disk, and thus allow the
6886 administrator to easily adjust the resource usage of
6887 services with a few simple commands. This dynamic resource
6888 management logic is also available to other programs via the
6889 bus. Almost any kernel cgroup attribute and controller is
6892 * systemd-vconsole-setup will now copy all font settings to
6893 all allocated VTs, where it previously applied them only to
6896 * libsystemd-login gained the new sd_session_get_tty() API
6899 * This release drops support for a few legacy or
6900 distribution-specific LSB facility names when parsing init
6901 scripts: $x-display-manager, $mail-transfer-agent,
6902 $mail-transport-agent, $mail-transfer-agent, $smtp,
6903 $null. Also, the mail-transfer-agent.target unit backing
6904 this has been removed. Distributions which want to retain
6905 compatibility with this should carry the burden for
6906 supporting this themselves and patch support for these back
6907 in, if they really need to. Also, the facilities $syslog and
6908 $local_fs are now ignored, since systemd does not support
6909 early-boot LSB init scripts anymore, and these facilities
6910 are implied anyway for normal services. syslog.target has
6913 * There are new bus calls on PID1's Manager object for
6914 cancelling jobs, and removing snapshot units. Previously,
6915 both calls were only available on the Job and Snapshot
6918 * systemd-journal-gatewayd gained SSL support.
6920 * The various "environment" files, such as /etc/locale.conf
6921 now support continuation lines with a backslash ("\") as
6922 last character in the line, similarly in style (but different)
6923 to how this is supported in shells.
6925 * For normal user processes the _SYSTEMD_USER_UNIT= field is
6926 now implicitly appended to every log entry logged. systemctl
6927 has been updated to filter by this field when operating on a
6928 user systemd instance.
6930 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
6931 CAP_AUDIT_CONTROL capabilities to the capabilities set for
6932 the container. This makes it easier to boot unmodified
6933 Fedora systems in a container, which however still requires
6934 audit=0 to be passed on the kernel command line. Auditing in
6935 kernel and userspace is unfortunately still too broken in
6936 context of containers, hence we recommend compiling it out
6937 of the kernel or using audit=0. Hopefully this will be fixed
6938 one day for good in the kernel.
6940 * nspawn gained the new --bind= and --bind-ro= parameters to
6941 bind mount specific directories from the host into the
6944 * nspawn will now mount its own devpts file system instance
6945 into the container, in order not to leak pty devices from
6946 the host into the container.
6948 * systemd will now read the firmware boot time performance
6949 information from the EFI variables, if the used boot loader
6950 supports this, and takes it into account for boot performance
6951 analysis via "systemd-analyze". This is currently supported
6952 only in conjunction with Gummiboot, but could be supported
6953 by other boot loaders too. For details see:
6955 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
6957 * A new generator has been added that automatically mounts the
6958 EFI System Partition (ESP) to /boot, if that directory
6959 exists, is empty, and no other file system has been
6960 configured to be mounted there.
6962 * logind will now send out PrepareForSleep(false) out
6963 unconditionally, after coming back from suspend. This may be
6964 used by applications as asynchronous notification for
6965 system resume events.
6967 * "systemctl unlock-sessions" has been added, that allows
6968 unlocking the screens of all user sessions at once, similar
6969 to how "systemctl lock-sessions" already locked all users
6970 sessions. This is backed by a new D-Bus call UnlockSessions().
6972 * "loginctl seat-status" will now show the master device of a
6973 seat. (i.e. the device of a seat that needs to be around for
6974 the seat to be considered available, usually the graphics
6977 * tmpfiles gained a new "X" line type, that allows
6978 configuration of files and directories (with wildcards) that
6979 shall be excluded from automatic cleanup ("aging").
6981 * udev default rules set the device node permissions now only
6982 at "add" events, and do not change them any longer with a
6983 later "change" event.
6985 * The log messages for lid events and power/sleep keypresses
6986 now carry a message ID.
6988 * We now have a substantially larger unit test suite, but this
6989 continues to be work in progress.
6991 * udevadm hwdb gained a new --root= parameter to change the
6992 root directory to operate relative to.
6994 * logind will now issue a background sync() request to the kernel
6995 early at shutdown, so that dirty buffers are flushed to disk early
6996 instead of at the last moment, in order to optimize shutdown
6999 * A new bootctl tool has been added that is an interface for
7000 certain boot loader operations. This is currently a preview
7001 and is likely to be extended into a small mechanism daemon
7002 like timedated, localed, hostnamed, and can be used by
7003 graphical UIs to enumerate available boot options, and
7004 request boot into firmware operations.
7006 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
7007 the rest of the package. It also has been updated to work
7008 correctly in initrds.
7010 * Policykit previously has been runtime optional, and is now
7011 also compile time optional via a configure switch.
7013 * systemd-analyze has been reimplemented in C. Also "systemctl
7014 dot" has moved into systemd-analyze.
7016 * "systemctl status" with no further parameters will now print
7017 the status of all active or failed units.
7019 * Operations such as "systemctl start" can now be executed
7020 with a new mode "--irreversible" which may be used to queue
7021 operations that cannot accidentally be reversed by a later
7022 job queuing. This is by default used to make shutdown
7023 requests more robust.
7025 * The Python API of systemd now gained a new module for
7026 reading journal files.
7028 * A new tool kernel-install has been added that can install
7029 kernel images according to the Boot Loader Specification:
7031 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
7033 * Boot time console output has been improved to provide
7034 animated boot time output for hanging jobs.
7036 * A new tool systemd-activate has been added which can be used
7037 to test socket activation with, directly from the command
7038 line. This should make it much easier to test and debug
7039 socket activation in daemons.
7041 * journalctl gained a new "--reverse" (or -r) option to show
7042 journal output in reverse order (i.e. newest line first).
7044 * journalctl gained a new "--pager-end" (or -e) option to jump
7045 to immediately jump to the end of the journal in the
7046 pager. This is only supported in conjunction with "less".
7048 * journalctl gained a new "--user-unit=" option, that works
7049 similarly to "--unit=" but filters for user units rather than
7052 * A number of unit files to ease adoption of systemd in
7053 initrds has been added. This moves some minimal logic from
7054 the various initrd implementations into systemd proper.
7056 * The journal files are now owned by a new group
7057 "systemd-journal", which exists specifically to allow access
7058 to the journal, and nothing else. Previously, we used the
7059 "adm" group for that, which however possibly covers more
7060 than just journal/log file access. This new group is now
7061 already used by systemd-journal-gatewayd to ensure this
7062 daemon gets access to the journal files and as little else
7063 as possible. Note that "make install" will also set FS ACLs
7064 up for /var/log/journal to give "adm" and "wheel" read
7065 access to it, in addition to "systemd-journal" which owns
7066 the journal files. We recommend that packaging scripts also
7067 add read access to "adm" + "wheel" to /var/log/journal, and
7068 all existing/future journal files. To normal users and
7069 administrators little changes, however packagers need to
7070 ensure to create the "systemd-journal" system group at
7071 package installation time.
7073 * The systemd-journal-gatewayd now runs as unprivileged user
7074 systemd-journal-gateway:systemd-journal-gateway. Packaging
7075 scripts need to create these system user/group at
7078 * timedated now exposes a new boolean property CanNTP that
7079 indicates whether a local NTP service is available or not.
7081 * systemd-detect-virt will now also detect xen PVs
7083 * The pstore file system is now mounted by default, if it is
7086 * In addition to the SELinux and IMA policies we will now also
7087 load SMACK policies at early boot.
7089 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
7090 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
7091 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
7092 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
7093 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
7094 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
7095 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
7096 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
7097 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
7098 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
7099 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
7100 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
7101 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
7102 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
7106 * Timer units now support calendar time events in addition to
7107 monotonic time events. That means you can now trigger a unit
7108 based on a calendar time specification such as "Thu,Fri
7109 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
7110 or fifth day of any month of the year 2013, given that it is
7111 a thursday or friday. This brings timer event support
7112 considerably closer to cron's capabilities. For details on
7113 the supported calendar time specification language see
7116 * udev now supports a number of different naming policies for
7117 network interfaces for predictable names, and a combination
7118 of these policies is now the default. Please see this wiki
7119 document for details:
7121 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
7123 * Auke Kok's bootchart implementation has been added to the
7124 systemd tree. It is an optional component that can graph the
7125 boot in quite some detail. It is one of the best bootchart
7126 implementations around and minimal in its code and
7129 * nss-myhostname has been integrated into the systemd source
7130 tree. nss-myhostname guarantees that the local hostname
7131 always stays resolvable via NSS. It has been a weak
7132 requirement of systemd-hostnamed since a long time, and
7133 since its code is actually trivial we decided to just
7134 include it in systemd's source tree. It can be turned off
7135 with a configure switch.
7137 * The read-ahead logic is now capable of properly detecting
7138 whether a btrfs file system is on SSD or rotating media, in
7139 order to optimize the read-ahead scheme. Previously, it was
7140 only capable of detecting this on traditional file systems
7143 * In udev, additional device properties are now read from the
7144 IAB in addition to the OUI database. Also, Bluetooth company
7145 identities are attached to the devices as well.
7147 * In service files %U may be used as specifier that is
7148 replaced by the configured user name of the service.
7150 * nspawn may now be invoked without a controlling TTY. This
7151 makes it suitable for invocation as its own service. This
7152 may be used to set up a simple containerized server system
7153 using only core OS tools.
7155 * systemd and nspawn can now accept socket file descriptors
7156 when they are started for socket activation. This enables
7157 implementation of socket activated nspawn
7158 containers. i.e. think about autospawning an entire OS image
7159 when the first SSH or HTTP connection is received. We expect
7160 that similar functionality will also be added to libvirt-lxc
7163 * journalctl will now suppress ANSI color codes when
7164 presenting log data.
7166 * systemctl will no longer show control group information for
7167 a unit if the control group is empty anyway.
7169 * logind can now automatically suspend/hibernate/shutdown the
7172 * /etc/machine-info and hostnamed now also expose the chassis
7173 type of the system. This can be used to determine whether
7174 the local system is a laptop, desktop, handset or
7175 tablet. This information may either be configured by the
7176 user/vendor or is automatically determined from ACPI and DMI
7177 information if possible.
7179 * A number of PolicyKit actions are now bound together with
7180 "imply" rules. This should simplify creating UIs because
7181 many actions will now authenticate similar ones as well.
7183 * Unit files learnt a new condition ConditionACPower= which
7184 may be used to conditionalize a unit depending on whether an
7185 AC power source is connected or not, of whether the system
7186 is running on battery power.
7188 * systemctl gained a new "is-failed" verb that may be used in
7189 shell scripts and suchlike to check whether a specific unit
7190 is in the "failed" state.
7192 * The EnvironmentFile= setting in unit files now supports file
7193 globbing, and can hence be used to easily read a number of
7194 environment files at once.
7196 * systemd will no longer detect and recognize specific
7197 distributions. All distribution-specific #ifdeffery has been
7198 removed, systemd is now fully generic and
7199 distribution-agnostic. Effectively, not too much is lost as
7200 a lot of the code is still accessible via explicit configure
7201 switches. However, support for some distribution specific
7202 legacy configuration file formats has been dropped. We
7203 recommend distributions to simply adopt the configuration
7204 files everybody else uses now and convert the old
7205 configuration from packaging scripts. Most distributions
7206 already did that. If that's not possible or desirable,
7207 distributions are welcome to forward port the specific
7208 pieces of code locally from the git history.
7210 * When logging a message about a unit systemd will now always
7211 log the unit name in the message meta data.
7213 * localectl will now also discover system locale data that is
7214 not stored in locale archives, but directly unpacked.
7216 * logind will no longer unconditionally use framebuffer
7217 devices as seat masters, i.e. as devices that are required
7218 to be existing before a seat is considered preset. Instead,
7219 it will now look for all devices that are tagged as
7220 "seat-master" in udev. By default, framebuffer devices will
7221 be marked as such, but depending on local systems, other
7222 devices might be marked as well. This may be used to
7223 integrate graphics cards using closed source drivers (such
7224 as NVidia ones) more nicely into logind. Note however, that
7225 we recommend using the open source NVidia drivers instead,
7226 and no udev rules for the closed-source drivers will be
7227 shipped from us upstream.
7229 Contributions from: Adam Williamson, Alessandro Crismani, Auke
7230 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
7231 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
7232 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
7233 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
7234 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
7235 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
7236 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
7237 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
7238 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
7239 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
7240 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
7245 * udev gained support for loading additional device properties
7246 from an indexed database that is keyed by vendor/product IDs
7247 and similar device identifiers. For the beginning this
7248 "hwdb" is populated with data from the well-known PCI and
7249 USB database, but also includes PNP, ACPI and OID data. In
7250 the longer run this indexed database shall grow into
7251 becoming the one central database for non-essential
7252 userspace device metadata. Previously, data from the PCI/USB
7253 database was only attached to select devices, since the
7254 lookup was a relatively expensive operation due to O(n) time
7255 complexity (with n being the number of entries in the
7256 database). Since this is now O(1), we decided to add in this
7257 data for all devices where this is available, by
7258 default. Note that the indexed database needs to be rebuilt
7259 when new data files are installed. To achieve this you need
7260 to update your packaging scripts to invoke "udevadm hwdb
7261 --update" after installation of hwdb data files. For
7262 RPM-based distributions we introduced the new
7263 %udev_hwdb_update macro for this purpose.
7265 * The Journal gained support for the "Message Catalog", an
7266 indexed database to link up additional information with
7267 journal entries. For further details please check:
7269 https://www.freedesktop.org/wiki/Software/systemd/catalog
7271 The indexed message catalog database also needs to be
7272 rebuilt after installation of message catalog files. Use
7273 "journalctl --update-catalog" for this. For RPM-based
7274 distributions we introduced the %journal_catalog_update
7275 macro for this purpose.
7277 * The Python Journal bindings gained support for the standard
7278 Python logging framework.
7280 * The Journal API gained new functions for checking whether
7281 the underlying file system of a journal file is capable of
7282 properly reporting file change notifications, or whether
7283 applications that want to reflect journal changes "live"
7284 need to recheck journal files continuously in appropriate
7287 * It is now possible to set the "age" field for tmpfiles
7288 entries to 0, indicating that files matching this entry
7289 shall always be removed when the directories are cleaned up.
7291 * coredumpctl gained a new "gdb" verb which invokes gdb
7292 right-away on the selected coredump.
7294 * There's now support for "hybrid sleep" on kernels that
7295 support this, in addition to "suspend" and "hibernate". Use
7296 "systemctl hybrid-sleep" to make use of this.
7298 * logind's HandleSuspendKey= setting (and related settings)
7299 now gained support for a new "lock" setting to simply
7300 request the screen lock on all local sessions, instead of
7301 actually executing a suspend or hibernation.
7303 * systemd will now mount the EFI variables file system by
7306 * Socket units now gained support for configuration of the
7307 SMACK security label.
7309 * timedatectl will now output the time of the last and next
7310 daylight saving change.
7312 * We dropped support for various legacy and distro-specific
7313 concepts, such as insserv, early-boot SysV services
7314 (i.e. those for non-standard runlevels such as 'b' or 'S')
7315 or ArchLinux /etc/rc.conf support. We recommend the
7316 distributions who still need support this to either continue
7317 to maintain the necessary patches downstream, or find a
7318 different solution. (Talk to us if you have questions!)
7320 * Various systemd components will now bypass PolicyKit checks
7321 for root and otherwise handle properly if PolicyKit is not
7322 found to be around. This should fix most issues for
7323 PolicyKit-less systems. Quite frankly this should have been
7324 this way since day one. It is absolutely our intention to
7325 make systemd work fine on PolicyKit-less systems, and we
7326 consider it a bug if something does not work as it should if
7327 PolicyKit is not around.
7329 * For embedded systems it is now possible to build udev and
7330 systemd without blkid and/or kmod support.
7332 * "systemctl switch-root" is now capable of switching root
7333 more than once. I.e. in addition to transitions from the
7334 initrd to the host OS it is now possible to transition to
7335 further OS images from the host. This is useful to implement
7336 offline updating tools.
7338 * Various other additions have been made to the RPM macros
7339 shipped with systemd. Use %udev_rules_update() after
7340 installing new udev rules files. %_udevhwdbdir,
7341 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
7342 %_sysctldir are now available which resolve to the right
7343 directories for packages to place various data files in.
7345 * journalctl gained the new --full switch (in addition to
7346 --all, to disable ellipsation for long messages.
7348 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
7349 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
7350 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
7351 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
7352 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
7353 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
7354 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
7355 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
7356 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
7360 * journalctl gained new --since= and --until= switches to
7361 filter by time. It also now supports nice filtering for
7362 units via --unit=/-u.
7364 * Type=oneshot services may use ExecReload= and do the
7367 * The journal daemon now supports time-based rotation and
7368 vacuuming, in addition to the usual disk-space based
7371 * The journal will now index the available field values for
7372 each field name. This enables clients to show pretty drop
7373 downs of available match values when filtering. The bash
7374 completion of journalctl has been updated
7375 accordingly. journalctl gained a new switch -F to list all
7376 values a certain field takes in the journal database.
7378 * More service events are now written as structured messages
7379 to the journal, and made recognizable via message IDs.
7381 * The timedated, localed and hostnamed mini-services which
7382 previously only provided support for changing time, locale
7383 and hostname settings from graphical DEs such as GNOME now
7384 also have a minimal (but very useful) text-based client
7385 utility each. This is probably the nicest way to changing
7386 these settings from the command line now, especially since
7387 it lists available options and is fully integrated with bash
7390 * There's now a new tool "systemd-coredumpctl" to list and
7391 extract coredumps from the journal.
7393 * We now install a README each in /var/log/ and
7394 /etc/rc.d/init.d explaining where the system logs and init
7395 scripts went. This hopefully should help folks who go to
7396 that dirs and look into the otherwise now empty void and
7397 scratch their heads.
7399 * When user-services are invoked (by systemd --user) the
7400 $MANAGERPID env var is set to the PID of systemd.
7402 * SIGRTMIN+24 when sent to a --user instance will now result
7403 in immediate termination of systemd.
7405 * gatewayd received numerous feature additions such as a
7406 "follow" mode, for live syncing and filtering.
7408 * browse.html now allows filtering and showing detailed
7409 information on specific entries. Keyboard navigation and
7410 mouse screen support has been added.
7412 * gatewayd/journalctl now supports HTML5/JSON
7413 Server-Sent-Events as output.
7415 * The SysV init script compatibility logic will now
7416 heuristically determine whether a script supports the
7417 "reload" verb, and only then make this available as
7420 * "systemctl status --follow" has been removed, use "journalctl
7423 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
7424 have been removed since they are hardly useful to be
7427 * And I'd like to take the opportunity to specifically mention
7428 Zbigniew for his great contributions. Zbigniew, you rock!
7430 Contributions from: Andrew Eikum, Christian Hesse, Colin
7431 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
7432 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
7433 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
7434 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
7435 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
7436 Jędrzejewski-Szmek, Сковорода Никита Андреевич
7440 * If /etc/vconsole.conf is non-existent or empty we will no
7441 longer load any console font or key map at boot by
7442 default. Instead the kernel defaults will be left
7443 intact. This is definitely the right thing to do, as no
7444 configuration should mean no configuration, and hard-coding
7445 font names that are different on all archs is probably a bad
7446 idea. Also, the kernel default key map and font should be
7447 good enough for most cases anyway, and mostly identical to
7448 the userspace fonts/key maps we previously overloaded them
7449 with. If distributions want to continue to default to a
7450 non-kernel font or key map they should ship a default
7451 /etc/vconsole.conf with the appropriate contents.
7453 Contributions from: Colin Walters, Daniel J Walsh, Dave
7454 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
7455 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
7459 * journalctl gained a new --cursor= switch to show entries
7460 starting from the specified location in the journal.
7462 * We now enforce a size limit on journal entry fields exported
7463 with "-o json" in journalctl. Fields larger than 4K will be
7464 assigned null. This can be turned off with --all.
7466 * An (optional) journal gateway daemon is now available as
7467 "systemd-journal-gatewayd.service". This service provides
7468 access to the journal via HTTP and JSON. This functionality
7469 will be used to implement live log synchronization in both
7470 pull and push modes, but has various other users too, such
7471 as easy log access for debugging of embedded devices. Right
7472 now it is already useful to retrieve the journal via HTTP:
7474 # systemctl start systemd-journal-gatewayd.service
7475 # wget http://localhost:19531/entries
7477 This will download the journal contents in a
7478 /var/log/messages compatible format. The same as JSON:
7480 # curl -H"Accept: application/json" http://localhost:19531/entries
7482 This service is also accessible via a web browser where a
7483 single static HTML5 app is served that uses the JSON logic
7484 to enable the user to do some basic browsing of the
7485 journal. This will be extended later on. Here's an example
7486 screenshot of this app in its current state:
7488 http://0pointer.de/public/journal-gatewayd
7490 Contributions from: Kay Sievers, Lennart Poettering, Robert
7491 Milasan, Tom Gundersen
7495 * The bash completion logic is now available for journalctl
7498 * We do not mount the "cpuset" controller anymore together with
7499 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
7500 started if no parameters are assigned to it. "cpuset" hence
7501 broke code that assumed it could create "cpu" groups and
7504 * journalctl -f will now subscribe to terminal size changes,
7505 and line break accordingly.
7507 Contributions from: Dave Reisner, Kay Sievers, Lennart
7508 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
7512 * nspawn will now create a symlink /etc/localtime in the
7513 container environment, copying the host's timezone
7514 setting. Previously this has been done via a bind mount, but
7515 since symlinks cannot be bind mounted this has now been
7516 changed to create/update the appropriate symlink.
7518 * journalctl -n's line number argument is now optional, and
7519 will default to 10 if omitted.
7521 * journald will now log the maximum size the journal files may
7522 take up on disk. This is particularly useful if the default
7523 built-in logic of determining this parameter from the file
7524 system size is used. Use "systemctl status
7525 systemd-journald.service" to see this information.
7527 * The multi-seat X wrapper tool has been stripped down. As X
7528 is now capable of enumerating graphics devices via udev in a
7529 seat-aware way the wrapper is not strictly necessary
7530 anymore. A stripped down temporary stop-gap is still shipped
7531 until the upstream display managers have been updated to
7532 fully support the new X logic. Expect this wrapper to be
7533 removed entirely in one of the next releases.
7535 * HandleSleepKey= in logind.conf has been split up into
7536 HandleSuspendKey= and HandleHibernateKey=. The old setting
7537 is not available anymore. X11 and the kernel are
7538 distinguishing between these keys and we should too. This
7539 also means the inhibition lock for these keys has been split
7542 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
7543 Poettering, Lukas Nykryn, Václav Pavlín
7547 * Whenever a unit changes state we will now log this to the
7548 journal and show along the unit's own log output in
7551 * ConditionPathIsMountPoint= can now properly detect bind
7552 mount points too. (Previously, a bind mount of one file
7553 system to another place in the same file system could not be
7554 detected as mount, since they shared struct stat's st_dev
7557 * We will now mount the cgroup controllers cpu, cpuacct,
7558 cpuset and the controllers net_cls, net_prio together by
7561 * nspawn containers will now have a virtualized boot
7562 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
7563 over with a randomized ID at container initialization). This
7564 has the effect of making "journalctl -b" do the right thing
7567 * The JSON output journal serialization has been updated not
7568 to generate "endless" list objects anymore, but rather one
7569 JSON object per line. This is more in line how most JSON
7570 parsers expect JSON objects. The new output mode
7571 "json-pretty" has been added to provide similar output, but
7572 neatly aligned for readability by humans.
7574 * We dropped all explicit sync() invocations in the shutdown
7575 code. The kernel does this implicitly anyway in the kernel
7576 reboot() syscall. halt(8)'s -n option is now a compatibility
7579 * We now support virtualized reboot() in containers, as
7580 supported by newer kernels. We will fall back to exit() if
7581 CAP_SYS_REBOOT is not available to the container. Also,
7582 nspawn makes use of this now and will actually reboot the
7583 container if the containerized OS asks for that.
7585 * journalctl will only show local log output by default
7586 now. Use --merge (-m) to show remote log output, too.
7588 * libsystemd-journal gained the new sd_journal_get_usage()
7589 call to determine the current disk usage of all journal
7590 files. This is exposed in the new "journalctl --disk-usage"
7593 * journald gained a new configuration setting SplitMode= in
7594 journald.conf which may be used to control how user journals
7595 are split off. See journald.conf(5) for details.
7597 * A new condition type ConditionFileNotEmpty= has been added.
7599 * tmpfiles' "w" lines now support file globbing, to write
7600 multiple files at once.
7602 * We added Python bindings for the journal submission
7603 APIs. More Python APIs for a number of selected APIs will
7604 likely follow. Note that we intend to add native bindings
7605 only for the Python language, as we consider it common
7606 enough to deserve bindings shipped within systemd. There are
7607 various projects outside of systemd that provide bindings
7608 for languages such as PHP or Lua.
7610 * Many conditions will now resolve specifiers such as %i. In
7611 addition, PathChanged= and related directives of .path units
7612 now support specifiers as well.
7614 * There's now a new RPM macro definition for the system preset
7617 * journald will now warn if it ca not forward a message to the
7618 syslog daemon because its socket is full.
7620 * timedated will no longer write or process /etc/timezone,
7621 except on Debian. As we do not support late mounted /usr
7622 anymore /etc/localtime always being a symlink is now safe,
7623 and hence the information in /etc/timezone is not necessary
7626 * logind will now always reserve one VT for a text getty (VT6
7627 by default). Previously if more than 6 X sessions where
7628 started they took up all the VTs with auto-spawned gettys,
7629 so that no text gettys were available anymore.
7631 * udev will now automatically inform the btrfs kernel logic
7632 about btrfs RAID components showing up. This should make
7633 simple hotplug based btrfs RAID assembly work.
7635 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
7636 (but not for its children which will stay at the kernel
7637 default). This should allow setups with a lot more listening
7640 * systemd will now always pass the configured timezone to the
7641 kernel at boot. timedated will do the same when the timezone
7644 * logind's inhibition logic has been updated. By default,
7645 logind will now handle the lid switch, the power and sleep
7646 keys all the time, even in graphical sessions. If DEs want
7647 to handle these events on their own they should take the new
7648 handle-power-key, handle-sleep-key and handle-lid-switch
7649 inhibitors during their runtime. A simple way to achieve
7650 that is to invoke the DE wrapped in an invocation of:
7652 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
7654 * Access to unit operations is now checked via SELinux taking
7655 the unit file label and client process label into account.
7657 * systemd will now notify the administrator in the journal
7658 when he over-mounts a non-empty directory.
7660 * There are new specifiers that are resolved in unit files,
7661 for the host name (%H), the machine ID (%m) and the boot ID
7664 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
7665 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
7666 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
7667 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
7668 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
7669 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
7670 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
7674 * Support for reading structured kernel messages from
7675 /dev/kmsg has now been added and is enabled by default.
7677 * Support for reading kernel messages from /proc/kmsg has now
7678 been removed. If you want kernel messages in the journal
7679 make sure to run a recent kernel (>= 3.5) that supports
7680 reading structured messages from /dev/kmsg (see
7681 above). /proc/kmsg is now exclusive property of classic
7682 syslog daemons again.
7684 * The libudev API gained the new
7685 udev_device_new_from_device_id() call.
7687 * The logic for file system namespace (ReadOnlyDirectory=,
7688 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
7689 require pivot_root() anymore. This means fewer temporary
7690 directories are created below /tmp for this feature.
7692 * nspawn containers will now see and receive all submounts
7693 made on the host OS below the root file system of the
7696 * Forward Secure Sealing is now supported for Journal files,
7697 which provide cryptographical sealing of journal files so
7698 that attackers cannot alter log history anymore without this
7699 being detectable. Lennart will soon post a blog story about
7700 this explaining it in more detail.
7702 * There are two new service settings RestartPreventExitStatus=
7703 and SuccessExitStatus= which allow configuration of exit
7704 status (exit code or signal) which will be excepted from the
7705 restart logic, resp. consider successful.
7707 * journalctl gained the new --verify switch that can be used
7708 to check the integrity of the structure of journal files and
7709 (if Forward Secure Sealing is enabled) the contents of
7712 * nspawn containers will now be run with /dev/stdin, /dev/fd/
7713 and similar symlinks pre-created. This makes running shells
7714 as container init process a lot more fun.
7716 * The fstab support can now handle PARTUUID= and PARTLABEL=
7719 * A new ConditionHost= condition has been added to match
7720 against the hostname (with globs) and machine ID. This is
7721 useful for clusters where a single OS image is used to
7722 provision a large number of hosts which shall run slightly
7723 different sets of services.
7725 * Services which hit the restart limit will now be placed in a
7728 Contributions from: Bertram Poettering, Dave Reisner, Huang
7729 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
7730 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
7734 * When running in --user mode systemd will now become a
7735 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
7736 tree a lot more organized.
7738 * A new PartOf= unit dependency type has been introduced that
7739 may be used to group services in a natural way.
7741 * "systemctl enable" may now be used to enable instances of
7744 * journalctl now prints error log levels in red, and
7745 warning/notice log levels in bright white. It also supports
7746 filtering by log level now.
7748 * cgtop gained a new -n switch (similar to top), to configure
7749 the maximum number of iterations to run for. It also gained
7750 -b, to run in batch mode (accepting no input).
7752 * The suffix ".service" may now be omitted on most systemctl
7753 command lines involving service unit names.
7755 * There's a new bus call in logind to lock all sessions, as
7756 well as a loginctl verb for it "lock-sessions".
7758 * libsystemd-logind.so gained a new call sd_journal_perror()
7759 that works similar to libc perror() but logs to the journal
7760 and encodes structured information about the error number.
7762 * /etc/crypttab entries now understand the new keyfile-size=
7765 * shutdown(8) now can send a (configurable) wall message when
7766 a shutdown is cancelled.
7768 * The mount propagation mode for the root file system will now
7769 default to "shared", which is useful to make containers work
7770 nicely out-of-the-box so that they receive new mounts from
7771 the host. This can be undone locally by running "mount
7772 --make-rprivate /" if needed.
7774 * The prefdm.service file has been removed. Distributions
7775 should maintain this unit downstream if they intend to keep
7776 it around. However, we recommend writing normal unit files
7777 for display managers instead.
7779 * Since systemd is a crucial part of the OS we will now
7780 default to a number of compiler switches that improve
7781 security (hardening) such as read-only relocations, stack
7782 protection, and suchlike.
7784 * The TimeoutSec= setting for services is now split into
7785 TimeoutStartSec= and TimeoutStopSec= to allow configuration
7786 of individual time outs for the start and the stop phase of
7789 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
7790 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
7791 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
7792 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
7793 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
7794 Gundersen, Zbigniew Jędrzejewski-Szmek
7798 * The journal and id128 C APIs are now fully documented as man
7801 * Extra safety checks have been added when transitioning from
7802 the initial RAM disk to the main system to avoid accidental
7805 * /etc/crypttab entries now understand the new keyfile-offset=
7808 * systemctl -t can now be used to filter by unit load state.
7810 * The journal C API gained the new sd_journal_wait() call to
7811 make writing synchronous journal clients easier.
7813 * journalctl gained the new -D switch to show journals from a
7816 * journalctl now displays a special marker between log
7817 messages of two different boots.
7819 * The journal is now explicitly flushed to /var via a service
7820 systemd-journal-flush.service, rather than implicitly simply
7821 by seeing /var/log/journal to be writable.
7823 * journalctl (and the journal C APIs) can now match for much
7824 more complex expressions, with alternatives and
7827 * When transitioning from the initial RAM disk to the main
7828 system we will now kill all processes in a killing spree to
7829 ensure no processes stay around by accident.
7831 * Three new specifiers may be used in unit files: %u, %h, %s
7832 resolve to the user name, user home directory resp. user
7833 shell. This is useful for running systemd user instances.
7835 * We now automatically rotate journal files if their data
7836 object hash table gets a fill level > 75%. We also size the
7837 hash table based on the configured maximum file size. This
7838 together should lower hash collisions drastically and thus
7839 speed things up a bit.
7841 * journalctl gained the new "--header" switch to introspect
7842 header data of journal files.
7844 * A new setting SystemCallFilters= has been added to services
7845 which may be used to apply blacklists or whitelists to
7846 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
7848 * nspawn gained a new --link-journal= switch (and quicker: -j)
7849 to link the container journal with the host. This makes it
7850 very easy to centralize log viewing on the host for all
7851 guests while still keeping the journal files separated.
7853 * Many bugfixes and optimizations
7855 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
7856 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
7857 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
7862 * Several tools now understand kernel command line arguments,
7863 which are only read when run in an initial RAM disk. They
7864 usually follow closely their normal counterparts, but are
7867 * There's a new tool to analyze the readahead files that are
7868 automatically generated at boot. Use:
7870 /usr/lib/systemd/systemd-readahead analyze /.readahead
7872 * We now provide an early debug shell on tty9 if this enabled. Use:
7874 systemctl enable debug-shell.service
7876 * All plymouth related units have been moved into the Plymouth
7877 package. Please make sure to upgrade your Plymouth version
7880 * systemd-tmpfiles now supports getting passed the basename of
7881 a configuration file only, in which case it will look for it
7882 in all appropriate directories automatically.
7884 * udevadm info now takes a /dev or /sys path as argument, and
7885 does the right thing. Example:
7887 udevadm info /dev/sda
7888 udevadm info /sys/class/block/sda
7890 * systemctl now prints a warning if a unit is stopped but a
7891 unit that might trigger it continues to run. Example: a
7892 service is stopped but the socket that activates it is left
7895 * "systemctl status" will now mention if the log output was
7896 shortened due to rotation since a service has been started.
7898 * The journal API now exposes functions to determine the
7899 "cutoff" times due to rotation.
7901 * journald now understands SIGUSR1 and SIGUSR2 for triggering
7902 immediately flushing of runtime logs to /var if possible,
7903 resp. for triggering immediate rotation of the journal
7906 * It is now considered an error if a service is attempted to
7907 be stopped that is not loaded.
7909 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
7911 * systemd-analyze now supports Python 3
7913 * tmpfiles now supports cleaning up directories via aging
7914 where the first level dirs are always kept around but
7915 directories beneath it automatically aged. This is enabled
7916 by prefixing the age field with '~'.
7918 * Seat objects now expose CanGraphical, CanTTY properties
7919 which is required to deal with very fast bootups where the
7920 display manager might be running before the graphics drivers
7921 completed initialization.
7923 * Seat objects now expose a State property.
7925 * We now include RPM macros for service enabling/disabling
7926 based on the preset logic. We recommend RPM based
7927 distributions to make use of these macros if possible. This
7928 makes it simpler to reuse RPM spec files across
7931 * We now make sure that the collected systemd unit name is
7932 always valid when services log to the journal via
7935 * There's a new man page kernel-command-line(7) detailing all
7936 command line options we understand.
7938 * The fstab generator may now be disabled at boot by passing
7939 fstab=0 on the kernel command line.
7941 * A new kernel command line option modules-load= is now understood
7942 to load a specific kernel module statically, early at boot.
7944 * Unit names specified on the systemctl command line are now
7945 automatically escaped as needed. Also, if file system or
7946 device paths are specified they are automatically turned
7947 into the appropriate mount or device unit names. Example:
7949 systemctl status /home
7950 systemctl status /dev/sda
7952 * The SysVConsole= configuration option has been removed from
7953 system.conf parsing.
7955 * The SysV search path is no longer exported on the D-Bus
7958 * The Names= option has been removed from unit file parsing.
7960 * There's a new man page bootup(7) detailing the boot process.
7962 * Every unit and every generator we ship with systemd now
7963 comes with full documentation. The self-explanatory boot is
7966 * A couple of services gained "systemd-" prefixes in their
7967 name if they wrap systemd code, rather than only external
7968 code. Among them fsck@.service which is now
7969 systemd-fsck@.service.
7971 * The HaveWatchdog property has been removed from the D-Bus
7974 * systemd.confirm_spawn= on the kernel command line should now
7977 * There's a new man page crypttab(5) which details all options
7978 we actually understand.
7980 * systemd-nspawn gained a new --capability= switch to pass
7981 additional capabilities to the container.
7983 * timedated will now read known NTP implementation unit names
7984 from /usr/lib/systemd/ntp-units.d/*.list,
7985 systemd-timedated-ntp.target has been removed.
7987 * journalctl gained a new switch "-b" that lists log data of
7988 the current boot only.
7990 * The notify socket is in the abstract namespace again, in
7991 order to support daemons which chroot() at start-up.
7993 * There is a new Storage= configuration option for journald
7994 which allows configuration of where log data should go. This
7995 also provides a way to disable journal logging entirely, so
7996 that data collected is only forwarded to the console, the
7997 kernel log buffer or another syslog implementation.
7999 * Many bugfixes and optimizations
8001 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
8002 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
8003 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
8004 Shawn Landden, Tom Gundersen
8008 * "systemctl help <unit>" now shows the man page if one is
8011 * Several new man pages have been added.
8013 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
8014 MaxLevelConsole= can now be specified in
8015 journald.conf. These options allow reducing the amount of
8016 data stored on disk or forwarded by the log level.
8018 * TimerSlackNSec= can now be specified in system.conf for
8019 PID1. This allows system-wide power savings.
8021 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
8022 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
8027 * logind is now capable of (optionally) handling power and
8028 sleep keys as well as the lid switch.
8030 * journalctl now understands the syntax "journalctl
8031 /usr/bin/avahi-daemon" to get all log output of a specific
8034 * CapabilityBoundingSet= in system.conf now also influences
8035 the capability bound set of usermode helpers of the kernel.
8037 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
8038 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
8039 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
8040 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
8044 * Note that we skipped 139 releases here in order to set the
8045 new version to something that is greater than both udev's
8046 and systemd's most recent version number.
8048 * udev: all udev sources are merged into the systemd source tree now.
8049 All future udev development will happen in the systemd tree. It
8050 is still fully supported to use the udev daemon and tools without
8051 systemd running, like in initramfs or other init systems. Building
8052 udev though, will require the *build* of the systemd tree, but
8053 udev can be properly *run* without systemd.
8055 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
8056 should be used to create dead device nodes as workarounds for broken
8059 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
8060 no longer supported. udev_monitor_new_from_netlink() needs to be
8061 used to subscribe to events.
8063 * udev: when udevd is started by systemd, processes which are left
8064 behind by forking them off of udev rules, are unconditionally cleaned
8065 up and killed now after the event handling has finished. Services or
8066 daemons must be started as systemd services. Services can be
8067 pulled-in by udev to get started, but they can no longer be directly
8068 forked by udev rules.
8070 * udev: the daemon binary is called systemd-udevd now and installed
8071 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
8072 to adapt to that, create symlink, or rename the binary after building
8075 * libudev no longer provides these symbols:
8076 udev_monitor_from_socket()
8077 udev_queue_get_failed_list_entry()
8078 udev_get_{dev,sys,run}_path()
8079 The versions number was bumped and symbol versioning introduced.
8081 * systemd-loginctl and systemd-journalctl have been renamed
8082 to loginctl and journalctl to match systemctl.
8084 * The config files: /etc/systemd/systemd-logind.conf and
8085 /etc/systemd/systemd-journald.conf have been renamed to
8086 logind.conf and journald.conf. Package updates should rename
8087 the files to the new names on upgrade.
8089 * For almost all files the license is now LGPL2.1+, changed
8090 from the previous GPL2.0+. Exceptions are some minor stuff
8091 of udev (which will be changed to LGPL2.1 eventually, too),
8092 and the MIT licensed sd-daemon.[ch] library that is suitable
8093 to be used as drop-in files.
8095 * systemd and logind now handle system sleep states, in
8096 particular suspending and hibernating.
8098 * logind now implements a sleep/shutdown/idle inhibiting logic
8099 suitable for a variety of uses. Soonishly Lennart will blog
8100 about this in more detail.
8102 * var-run.mount and var-lock.mount are no longer provided
8103 (which previously bind mounted these directories to their new
8104 places). Distributions which have not converted these
8105 directories to symlinks should consider stealing these files
8106 from git history and add them downstream.
8108 * We introduced the Documentation= field for units and added
8109 this to all our shipped units. This is useful to make it
8110 easier to explore the boot and the purpose of the various
8113 * All smaller setup units (such as
8114 systemd-vconsole-setup.service) now detect properly if they
8115 are run in a container and are skipped when
8116 appropriate. This guarantees an entirely noise-free boot in
8117 Linux container environments such as systemd-nspawn.
8119 * A framework for implementing offline system updates is now
8120 integrated, for details see:
8121 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
8123 * A new service type Type=idle is available now which helps us
8124 avoiding ugly interleaving of getty output and boot status
8127 * There's now a system-wide CapabilityBoundingSet= option to
8128 globally reduce the set of capabilities for the
8129 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
8130 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
8131 even CAP_NET_ADMIN system-wide for secure systems.
8133 * There are now system-wide DefaultLimitXXX= options to
8134 globally change the defaults of the various resource limits
8135 for all units started by PID 1.
8137 * Harald Hoyer's systemd test suite has been integrated into
8138 systemd which allows easy testing of systemd builds in qemu
8139 and nspawn. (This is really awesome! Ask us for details!)
8141 * The fstab parser is now implemented as generator, not inside
8144 * systemctl will now warn you if .mount units generated from
8145 /etc/fstab are out of date due to changes in fstab that
8146 have not been read by systemd yet.
8148 * systemd is now suitable for usage in initrds. Dracut has
8149 already been updated to make use of this. With this in place
8150 initrds get a slight bit faster but primarily are much
8151 easier to introspect and debug since "systemctl status" in
8152 the host system can be used to introspect initrd services,
8153 and the journal from the initrd is kept around too.
8155 * systemd-delta has been added, a tool to explore differences
8156 between user/admin configuration and vendor defaults.
8158 * PrivateTmp= now affects both /tmp and /var/tmp.
8160 * Boot time status messages are now much prettier and feature
8161 proper english language. Booting up systemd has never been
8164 * Read-ahead pack files now include the inode number of all
8165 files to pre-cache. When the inode changes the pre-caching
8166 is not attempted. This should be nicer to deal with updated
8167 packages which might result in changes of read-ahead
8170 * We now temporaritly lower the kernel's read_ahead_kb variable
8171 when collecting read-ahead data to ensure the kernel's
8172 built-in read-ahead does not add noise to our measurements
8173 of necessary blocks to pre-cache.
8175 * There's now RequiresMountsFor= to add automatic dependencies
8176 for all mounts necessary for a specific file system path.
8178 * MountAuto= and SwapAuto= have been removed from
8179 system.conf. Mounting file systems at boot has to take place
8182 * nspawn now learned a new switch --uuid= to set the machine
8183 ID on the command line.
8185 * nspawn now learned the -b switch to automatically search
8188 * vt102 is now the default TERM for serial TTYs, upgraded from
8191 * systemd-logind now works on VT-less systems.
8193 * The build tree has been reorganized. The individual
8194 components now have directories of their own.
8196 * A new condition type ConditionPathIsReadWrite= is now available.
8198 * nspawn learned the new -C switch to create cgroups for the
8199 container in other hierarchies.
8201 * We now have support for hardware watchdogs, configurable in
8204 * The scheduled shutdown logic now has a public API.
8206 * We now mount /tmp as tmpfs by default, but this can be
8207 masked and /etc/fstab can override it.
8209 * Since udisks does not make use of /media anymore we are not
8210 mounting a tmpfs on it anymore.
8212 * journalctl gained a new --local switch to only interleave
8213 locally generated journal files.
8215 * We can now load the IMA policy at boot automatically.
8217 * The GTK tools have been split off into a systemd-ui.
8219 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
8220 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
8221 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
8222 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
8223 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
8224 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
8225 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
8226 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
8227 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
8232 * This is mostly a bugfix release
8234 * Support optional initialization of the machine ID from the
8235 KVM or container configured UUID.
8237 * Support immediate reboots with "systemctl reboot -ff"
8239 * Show /etc/os-release data in systemd-analyze output
8241 * Many bugfixes for the journal, including endianness fixes and
8242 ensuring that disk space enforcement works
8244 * sd-login.h is C++ compatible again
8246 * Extend the /etc/os-release format on request of the Debian
8249 * We now refuse non-UTF8 strings used in various configuration
8250 and unit files. This is done to ensure we do not pass invalid
8251 data over D-Bus or expose it elsewhere.
8253 * Register Mimo USB Screens as suitable for automatic seat
8256 * Read SELinux client context from journal clients in a race
8259 * Reorder configuration file lookup order. /etc now always
8260 overrides /run in order to allow the administrator to always
8261 and unconditionally override vendor-supplied or
8262 automatically generated data.
8264 * The various user visible bits of the journal now have man
8265 pages. We still lack man pages for the journal API calls
8268 * We now ship all man pages in HTML format again in the
8271 Contributions from: Dave Reisner, Dirk Eibach, Frederic
8272 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
8273 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
8278 * This is mostly a bugfix release
8280 * systems lacking /etc/os-release are no longer supported.
8282 * Various functionality updates to libsystemd-login.so
8284 * Track class of PAM logins to distinguish greeters from
8287 Contributions from: Kay Sievers, Lennart Poettering, Michael
8292 * This is an important bugfix release for v41.
8294 * Building man pages is now optional which should be useful
8295 for those building systemd from git but unwilling to install
8298 * Watchdog support for supervising services is now usable. In
8299 a future release support for hardware watchdogs
8300 (i.e. /dev/watchdog) will be added building on this.
8302 * Service start rate limiting is now configurable and can be
8303 turned off per service. When a start rate limit is hit a
8304 reboot can automatically be triggered.
8306 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
8308 Contributions from: Benjamin Franzke, Bill Nottingham,
8309 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
8310 Schmidt, Michał Górny, Piotr Drąg
8314 * The systemd binary is installed /usr/lib/systemd/systemd now;
8315 An existing /sbin/init symlink needs to be adapted with the
8318 * The code that loads kernel modules has been ported to invoke
8319 libkmod directly, instead of modprobe. This means we do not
8320 support systems with module-init-tools anymore.
8322 * Watchdog support is now already useful, but still not
8325 * A new kernel command line option systemd.setenv= is
8326 understood to set system wide environment variables
8327 dynamically at boot.
8329 * We now limit the set of capabilities of systemd-journald.
8331 * We now set SIGPIPE to ignore by default, since it only is
8332 useful in shell pipelines, and has little use in general
8333 code. This can be disabled with IgnoreSIPIPE=no in unit
8336 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
8337 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
8342 * This is mostly a bugfix release
8344 * We now expose the reason why a service failed in the
8345 "Result" D-Bus property.
8347 * Rudimentary service watchdog support (will be completed over
8348 the next few releases.)
8350 * When systemd forks off in order execute some service we will
8351 now immediately changes its argv[0] to reflect which process
8352 it will execute. This is useful to minimize the time window
8353 with a generic argv[0], which makes bootcharts more useful
8355 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
8356 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
8357 Mike Kazantsev, Ray Strode
8361 * This is mostly a test release, but incorporates many
8364 * New systemd-cgtop tool to show control groups by their
8367 * Linking against libacl for ACLs is optional again. If
8368 disabled, support tracking device access for active logins
8369 goes becomes unavailable, and so does access to the user
8370 journals by the respective users.
8372 * If a group "adm" exists, journal files are automatically
8373 owned by them, thus allow members of this group full access
8374 to the system journal as well as all user journals.
8376 * The journal now stores the SELinux context of the logging
8377 client for all entries.
8379 * Add C++ inclusion guards to all public headers
8381 * New output mode "cat" in the journal to print only text
8382 messages, without any meta data like date or time.
8384 * Include tiny X server wrapper as a temporary stop-gap to
8385 teach XOrg udev display enumeration. This is used by display
8386 managers such as gdm, and will go away as soon as XOrg
8387 learned native udev hotplugging for display devices.
8389 * Add new systemd-cat tool for executing arbitrary programs
8390 with STDERR/STDOUT connected to the journal. Can also act as
8391 BSD logger replacement, and does so by default.
8393 * Optionally store all locally generated coredumps in the
8394 journal along with meta data.
8396 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
8397 writing short strings to files (for usage for /sys), and for
8398 creating symlinks, character and block device nodes.
8400 * New unit file option ControlGroupPersistent= to make cgroups
8401 persistent, following the mechanisms outlined in
8402 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
8404 * Support multiple local RTCs in a sane way
8406 * No longer monopolize IO when replaying readahead data on
8407 rotating disks, since we might starve non-file-system IO to
8408 death, since fanotify() will not see accesses done by blkid,
8411 * Do not show kernel threads in systemd-cgls anymore, unless
8412 requested with new -k switch.
8414 Contributions from: Dan Horák, Kay Sievers, Lennart
8415 Poettering, Michal Schmidt
8419 * This is mostly a test release, but incorporates many
8422 * The git repository moved to:
8423 git://anongit.freedesktop.org/systemd/systemd
8424 ssh://git.freedesktop.org/git/systemd/systemd
8426 * First release with the journal
8427 http://0pointer.de/blog/projects/the-journal.html
8429 * The journal replaces both systemd-kmsg-syslogd and
8430 systemd-stdout-bridge.
8432 * New sd_pid_get_unit() API call in libsystemd-logind
8434 * Many systemadm clean-ups
8436 * Introduce remote-fs-pre.target which is ordered before all
8437 remote mounts and may be used to start services before all
8440 * Added Mageia support
8442 * Add bash completion for systemd-loginctl
8444 * Actively monitor PID file creation for daemons which exit in
8445 the parent process before having finished writing the PID
8446 file in the daemon process. Daemons which do this need to be
8447 fixed (i.e. PID file creation must have finished before the
8448 parent exits), but we now react a bit more gracefully to them.
8450 * Add colourful boot output, mimicking the well-known output
8451 of existing distributions.
8453 * New option PassCredentials= for socket units, for
8454 compatibility with a recent kernel ABI breakage.
8456 * /etc/rc.local is now hooked in via a generator binary, and
8457 thus will no longer act as synchronization point during
8460 * systemctl list-unit-files now supports --root=.
8462 * systemd-tmpfiles now understands two new commands: z, Z for
8463 relabelling files according to the SELinux database. This is
8464 useful to apply SELinux labels to specific files in /sys,
8467 * Output of SysV services is now forwarded to both the console
8468 and the journal by default, not only just the console.
8470 * New man pages for all APIs from libsystemd-login.
8472 * The build tree got reorganized and the build system is a
8473 lot more modular allowing embedded setups to specifically
8474 select the components of systemd they are interested in.
8476 * Support for Linux systems lacking the kernel VT subsystem is
8479 * configure's --with-rootdir= got renamed to
8480 --with-rootprefix= to follow the naming used by udev and
8483 * Unless specified otherwise we will now install to /usr instead
8484 of /usr/local by default.
8486 * Processes with '@' in argv[0][0] are now excluded from the
8487 final shut-down killing spree, following the logic explained
8489 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
8491 * All processes remaining in a service cgroup when we enter
8492 the START or START_PRE states are now killed with
8493 SIGKILL. That means it is no longer possible to spawn
8494 background processes from ExecStart= lines (which was never
8495 supported anyway, and bad style).
8497 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
8498 reloading of units together.
8500 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
8501 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
8502 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
8503 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
8504 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
~ianmdlvl / elogind.git / blob