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
15 elogind is not changed. However, if distribution packagers or
16 administrators disabled or modified IPAddressDeny= setting by a
17 drop-in config file, then it may be necessary to update the file to
18 re-enable AF_INET and AF_INET6 to support network user name services,
21 * When the RestrictNamespaces= unit property is specified multiple
22 times, then the specified types are merged now. Previously, only the
23 last assignment was used. So, if distribution packagers or
24 administrators modified the setting by a drop-in config file, then it
25 may be necessary to update the file.
27 * When OnFailure= is used in combination with Restart= on a service
28 unit, then the specified units will no longer be triggered on
29 failures that result in restarting. Previously, the specified units
30 would be activated each time the unit failed, even when the unit was
31 going to be restarted automatically. This behaviour contradicted the
32 documentation. With this release the code is adjusted to match the
35 * elogind-tmpfiles will now print a notice whenever it encounters
36 tmpfiles.d/ lines referencing the /var/run/ directory. It will
37 recommend reworking them to use the /run/ directory instead (for
38 which /var/run/ is simply a symlinked compatibility alias). This way
39 elogind-tmpfiles can properly detect line conflicts and merge lines
40 referencing the same file by two paths, without having to access
43 * systemctl disable/unmask/preset/preset-all cannot be used with
44 --runtime. Previously this was allowed, but resulted in unintuitive
45 behaviour that wasn't useful. systemctl disable/unmask will now undo
46 both runtime and persistent enablement/masking, i.e. it will remove
47 any relevant symlinks both in /run and /etc.
49 * sd-boot acquired new loader configuration settings to optionally turn
50 off Windows and MacOS boot partition discovery as well as
51 reboot-into-firmware menu items. It is also able to pick a better
52 screen resolution for HiDPI systems, and now provides loader
53 configuration settings to change the resolution explicitly.
55 * The elogind-resolve tool has been renamed to resolvectl (it also
56 * elogind-resolved now supports DNS-over-TLS ("PrivateDNS"). It's still
57 turned off by default, use PrivateDNS=opportunistic to turn it on in
58 resolved.conf. We intend to make this the default as soon as couple
59 of additional techniques for optimizing the initial latency caused by
60 establishing a TLS/TCP connection are implemented.
62 * elogind-resolved.service and elogind-networkd.service now set
63 DynamicUser=yes. The users elogind-resolve and elogind-network are
64 not created by elogind-sysusers.
66 remains available under the old name, for compatibility), and its
67 interface is now verb-based, similar in style to the other <xyz>ctl
68 tools, such as systemctl or loginctl.
70 * The resolvectl/elogind-resolve tool also provides 'resolveconf'
71 compatibility. It may be symlinked under the 'resolveconf' name, in
72 which case it will take arguments and input compatible with the
73 Debian and FreeBSD resolvconf tool.
75 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
76 where the system initially suspends, and after a time-out resumes and
79 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
80 set the client will only send a DUID as client identifier.
82 * The nss-elogind glibc NSS module will now enumerate dynamic users and
83 groups in effect. Previously, it could resolve UIDs/GIDs to user
84 names/groups and vice versa, but did not support enumeration.
86 * journald's Compress= configuration setting now optionally accepts a
87 byte threshold value. All journal objects larger than this threshold
88 will be compressed, smaller ones will not. Previously this threshold
89 was not configurable and set to 512.
91 * A new system.conf setting NoNewPrivileges= is now available which may
92 be used to turn off acquisition of new privileges system-wide
93 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
94 for all its children). Note that turning this option on means setuid
95 binaries and file system capabilities lose their special powers.
96 While turning on this option is a big step towards a more secure
97 system, doing so is likely to break numerous pre-existing UNIX tools,
98 in particular su and sudo.
100 * A new service elogind-time-sync-wait.service has been added. If
101 enabled it will delay the time-sync.target unit at boot until time
102 synchronization has been received from the network. This
103 functionality is useful on systems lacking a local RTC or where it is
104 acceptable that the boot process shall be delayed by external network
107 * When hibernating, elogind will now inform the kernel of the image
108 write offset, on kernels new enough to support this. This means swap
109 files should work for hibernation now.
111 * When loading unit files, elogind will now look for drop-in unit files
112 extensions in additional places. Previously, for a unit file name
113 "foo-bar-baz.service" it would look for dropin files in
114 "foo-bar-baz.service.d/*.conf". Now, it will also look in
115 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
116 service name truncated after all inner dashes. This scheme allows
117 writing drop-ins easily that apply to a whole set of unit files at
118 once. It's particularly useful for mount and slice units (as their
119 naming is prefix based), but is also useful for service and other
120 units, for packages that install multiple unit files at once,
121 following a strict naming regime of beginning the unit file name with
122 the package's name. Two new specifiers are now supported in unit
123 files to match this: %j and %J are replaced by the part of the unit
124 name following the last dash.
126 * Unit files and other configuration files that support specifier
127 expansion now understand another two new specifiers: %T and %V will
128 resolve to /tmp and /var/tmp respectively, or whatever temporary
129 directory has been set for the calling user.
131 * The ExecStart= lines of unit files are no longer required to
132 reference absolute paths. If non-absolute paths are specified the
133 specified binary name is searched within the service manager's
134 built-in $PATH, which may be queried with 'elogind-path
135 search-binaries-default'. It's generally recommended to continue to
136 use absolute paths for all binaries specified in unit files.
138 * Units gained a new load state "bad-setting", which is used when a
139 unit file was loaded, but contained fatal errors which prevent it
140 from being started (for example, an ExecStart= path which references
141 a non-existent executable).
143 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
144 support alternative debuggers, for example lldb. The old name
145 continues to be available however, for compatibility reasons. Use the
146 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
147 to pick an alternative debugger instead of the default gdb.
149 * systemctl and the other tools will now output escape sequences that
150 generate proper clickable hyperlinks in various terminal emulators
151 where useful (for example, in the "systemctl status" output you can
152 now click on the unit file name to quickly open it in the
153 editor/viewer of your choice). Note that not all terminal emulators
154 support this functionality yet, but many do. Unfortunately, the
155 "less" pager doesn't support this yet, hence this functionality is
156 currently automatically turned off when a pager is started (which
157 happens quite often due to auto-paging). We hope to remove this
158 limitation as soon as "less" learns these escape sequences. This new
159 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
160 environment variable. For details on these escape sequences see:
161 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
163 * networkd's .network files now support a new IPv6MTUBytes= option for
164 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
165 option in the [Route] section to configure the MTU to use for
166 specific routes. It also gained support for configuration of the DHCP
167 "UserClass" option through the new UserClass= setting. It gained
168 three new options in the new [CAN] section for configuring CAN
169 networks. The MULTICAST and ALLMULTI interface flags may now be
170 controlled explicitly with the new Multicast= and AllMulticast=
173 * networkd will now automatically make use of the kernel's route
174 expiration feature, if it is available.
176 * udevd's .link files now support setting the number of receive and
177 transmit channels, using the RxChannels=, TxChannels=,
178 OtherChannels=, CombinedChannels= settings.
180 * Support for UDPSegmentationOffload= has been removed, given its
181 limited support in hardware, and waning software support.
183 * networkd's .netdev files now support creating "netdevsim" interfaces.
185 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
186 to query the unit belonging to a specific kernel control group.
188 * elogind-analyze gained a new verb "cat-config", which may be used to
189 dump the contents of any configuration file, with all its matching
190 drop-in files added in, and honouring the usual search and masking
191 logic applied to elogind configuration files. For example use
192 "elogind-analyze cat-config elogind/system.conf" to get the complete
193 system configuration file of elogind how it would be loaded by PID 1
194 itself. Similar to this, various tools such as elogind-tmpfiles or
195 elogind-sysusers, gained a new option "--cat-config", which does the
196 corresponding operation for their own configuration settings. For
197 example, "elogind-tmpfiles --cat-config" will now output the full
198 list of tmpfiles.d/ lines in place.
200 * timedatectl gained two new verbs "timesync-status" (to show the
201 current NTP synchronization state of elogind-timesyncd) and
202 "show-timesync" (to show bus properties of elogind-timesyncd).
204 * elogind-timesyncd gained a bus interface on which it exposes details
207 * elogind-nspawn gained a new --rlimit= switch for setting initial
208 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
209 understood by elogind-timedated. It takes a colon-separated list of
210 unit names of NTP client services. The list is used by
211 "timedatectl set-ntp".
213 resource limits for the container payload. There's a new switch
214 --hostname= to explicitly override the container's hostname. A new
215 --no-new-privileges= switch may be used to control the
216 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
217 --oom-score-adjust= switch controls the OOM scoring adjustment value
218 for the payload. The new --cpu-affinity= switch controls the CPU
219 affinity of the container payload. The new --resolv-conf= switch
220 allows more detailed control of /etc/resolv.conf handling of the
221 container. Similarly, the new --timezone= switch allows more detailed
222 control of /etc/localtime handling of the container.
224 * elogind-detect-virt gained a new --list switch, which will print a
225 list of all currently known VM and container environments.
227 * Support for "Portable Services" has been added, see
228 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
229 experimental, but this is expected to change soon. Reflecting this
230 experimental state, the "portablectl" binary is not installed into
231 /usr/bin yet. The binary has to be called with the full path
232 /usr/lib/elogind/portablectl instead.
234 * journalctl's and systemctl's -o switch now knows a new log output
235 mode "with-unit". The output it generates is very similar to the
236 regular "short" mode, but displays the unit name instead of the
237 syslog tag for each log line. Also, the date is shown with timezone
238 information. This mode is probably more useful than the classic
239 "short" output mode for most purposes, except where pixel-perfect
240 compatibility with classic /var/log/messages formatting is required.
242 * A new --dump-bus-properties switch has been added to the elogind
243 binary, which may be used to dump all supported D-Bus properties.
244 (Options which are still supported, but are deprecated, are *not*
247 * sd-bus gained a set of new calls:
248 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
249 enable/disable the "floating" state of a bus slot object,
250 i.e. whether the slot object pins the bus it is allocated for into
251 memory or if the bus slot object gets disconnected when the bus goes
252 away. sd_bus_open_with_description(),
253 sd_bus_open_user_with_description(),
254 sd_bus_open_system_with_description() may be used to allocate bus
255 objects and set their description string already during allocation.
257 * sd-event gained support for watching inotify events from the event
258 loop, in an efficient way, sharing inotify handles between multiple
259 users. For this a new function sd_event_add_inotify() has been added.
261 * sd-event and sd-bus gained support for calling special user-supplied
262 destructor functions for userdata pointers associated with
263 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
264 functions sd_bus_slot_set_destroy_callback,
265 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
266 sd_bus_track_get_destroy_callback,
267 sd_event_source_set_destroy_callback,
268 sd_event_source_get_destroy_callback have been added.
270 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
272 * PID 1 will now automatically reschedule .timer units whenever the
273 local timezone changes. (They previously got rescheduled
274 automatically when the system clock changed.)
276 * New documentation has been added to document cgroups delegation,
277 portable services and the various code quality tools we have set up:
279 https://github.com/systemd/systemd/blob/master/doc/CGROUP_DELEGATION.md
280 https://github.com/systemd/systemd/blob/master/doc/PORTABLE_SERVICES.md
281 https://github.com/systemd/systemd/blob/master/doc/CODE_QUALITY.md
283 * pam_elogind will now look for PAM userdata fields elogind.memory_max,
284 elogind.tasks_max, elogind.cpu_weight, elogind.io_weight set by
285 earlier PAM modules. The data in these fields is used to initialize
286 the session scope's resource properties. Thus external PAM modules
287 may now configure per-session limits, for example sourced from
288 external user databases.
290 * socket units with Accept=yes will now maintain a "refused" counter in
291 addition to the existing "accepted" counter, counting connections
292 refused due to the enforced limits.
294 * The "elogind-path search-binaries-default" command may now be use to
295 query the default, built-in $PATH PID 1 will pass to the services it
298 * A new unit file setting PrivateMounts= has been added. It's a boolean
299 option. If enabled the unit's processes are invoked in their own file
300 system namespace. Note that this behaviour is also implied if any
301 other file system namespacing options (such as PrivateTmp=,
302 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
303 primarily useful for services that do not use any of the other file
304 system namespacing options. One such service is elogind-udevd.service
305 wher this is now used by default.
307 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
308 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
309 J. Murrell, Bruno Vernay, Chris Lesiak, Christian Brauner, Christian
310 Hesse, Daniel Dao, Daniel Lin, Danylo Korostil, Davide Cavalca, David
311 Tardon, Dimitri John Ledkov, Dmitriy Geels, Douglas Christman, Elia
312 Geretto, emelenas, Evegeny Vereshchagin, Evgeny Vereshchagin, Felipe
313 Sateler, Feng Sun, Filipe Brandenburger, Franck Bui, futpib, Giuseppe
314 Scrivano, Guillem Jover, guixxx, Hans de Goede, Henrique Dante de
315 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
316 James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir, João Paulo
317 Rechi Vita, Joost Heitbrink, Jui-Chi Ricky Liang, Jürg Billeter,
318 Kai-Heng Feng, Karol Augustin, Krzysztof Nowicki, Lauri Tirkkonen,
319 Lennart Poettering, Leonard König, Long Li, Luca Boccassi, Lucas
320 Werkmeister, Marcel Hoppe, Marc Kleine-Budde, Mario Limonciello, Martin
321 Jansa, Martin Wilck, Mathieu Malaterre, Matteo F. Vescovi, Matthew
322 McGinn, Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný,
323 Michal Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan
324 Pässler, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride Legovini,
325 Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot, Peter
326 Hutterer, Peter Jones, Philip Sequeira, Philip Withnall, Piotr Drąg,
327 Radostin Stoyanov, Ricardo Salveti de Araujo, Rosen Penev, Rubén Suárez
328 Alvarez, Ryan Gonzalez, Salvo Tomaselli, Sebastian Reichel, Sergio
329 Lindo Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes,
330 Susant Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias
331 Jungel, Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van
332 Mourik, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
338 * The MemoryAccounting= unit property now defaults to on. After
339 discussions with the upstream control group maintainers we learnt
340 that the negative impact of cgroup memory accounting on current
341 kernels is finally relatively minimal, so that it should be safe to
342 enable this by default without affecting system performance. Besides
343 memory accounting only task accounting is turned on by default, all
344 other forms of resource accounting (CPU, IO, IP) remain off for now,
345 because it's not clear yet that their impact is small enough to move
346 from opt-in to opt-out. We recommend downstreams to leave memory
347 accounting on by default if kernel 4.14 or higher is primarily
348 used. On very resource constrained systems or when support for old
349 kernels is a necessity, -Dmemory-accounting-default=false can be used
350 to revert this change.
352 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
353 %udev_rules_update) and the journal catalog (%journal_catalog_update)
354 from the upgrade scriptlets of individual packages now do nothing.
355 Transfiletriggers have been added which will perform those updates
356 once at the end of the transaction.
358 Similar transfiletriggers have been added to execute any sysctl.d
359 and binfmt.d rules. Thus, it should be unnecessary to provide any
360 scriptlets to execute this configuration from package installation
363 * elogind-sysusers gained a mode where the configuration to execute is
364 specified on the command line, but this configuration is not executed
365 directly, but instead it is merged with the configuration on disk,
366 and the result is executed. This is useful for package installation
367 scripts which want to create the user before installing any files on
368 disk (in case some of those files are owned by that user), while
369 still allowing local admin overrides.
371 This functionality is exposed to rpm scriptlets through a new
372 %sysusers_create_package macro. Old %sysusers_create and
373 %sysusers_create_inline macros are deprecated.
375 A transfiletrigger for sysusers.d configuration is now installed,
376 which means that it should be unnecessary to call elogind-sysusers from
377 package installation scripts, unless the package installs any files
378 owned by those newly-created users, in which case
379 %sysusers_create_package should be used.
381 * Analogous change has been done for elogind-tmpfiles: it gained a mode
382 where the command-line configuration is merged with the configuration
383 on disk. This is exposed as the new %tmpfiles_create_package macro,
384 and %tmpfiles_create is deprecated. A transfiletrigger is installed
385 for tmpfiles.d, hence it should be unnecessary to call elogind-tmpfiles
386 from package installation scripts.
388 * sysusers.d configuration for a user may now also specify the group
389 number, in addition to the user number ("u username 123:456"), or
390 without the user number ("u username -:456").
392 * Configution items for elogind-sysusers can now be specified as
393 positional arguments when the new --inline switch is used.
395 * The login shell of users created through sysusers.d may now be
396 specified (previously, it was always /bin/sh for root and
397 /sbin/nologin for other users).
399 * elogind-analyze gained a new --global switch to look at global user
400 configuration. It also gained a unit-paths verb to list the unit load
401 paths that are compiled into elogind (which can be used with
402 --elogind, --user, or --global).
404 * udevadm trigger gained a new --settle/-w option to wait for any
405 triggered events to finish (but just those, and not any other events
406 which are triggered meanwhile).
408 * The action that elogind takes when the lid is closed and the
409 machine is connected to external power can now be configured using
410 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
411 was determined by HandleLidSwitch=, and, for backwards compatibility,
412 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
414 * journalctl will periodically call sd_journal_process() to make it
415 resilient against inotify queue overruns when journal files are
416 rotated very quickly.
418 * Two new functions in libelogind — sd_bus_get_n_queued_read and
419 sd_bus_get_n_queued_write — may be used to check the number of
420 pending bus messages.
422 * elogind gained a new
423 org.freedesktop.elogind1.Manager.AttachProcessesToUnit dbus call
424 which can be used to migrate foreign processes to scope and service
425 units. The primary user for this new API is elogind itself: the
426 elogind --user instance uses this call of the elogind --system
427 instance to migrate processes if it itself gets the request to
428 migrate processes and the kernel refuses this due to access
429 restrictions. Thanks to this "elogind-run --scope --user …" works
430 again in pure cgroups v2 environments when invoked from the user
433 * A new TemporaryFileSystem= setting can be used to mask out part of
434 the real file system tree with tmpfs mounts. This may be combined
435 with BindPaths= and BindReadOnlyPaths= to hide files or directories
436 not relevant to the unit, while still allowing some paths lower in
437 the tree to be accessed.
439 ProtectHome=tmpfs may now be used to hide user home and runtime
440 directories from units, in a way that is mostly equivalent to
441 "TemporaryFileSystem=/home /run/user /root".
443 * Non-service units are now started with KeyringMode=shared by default.
444 This means that mount and swapon and other mount tools have access
445 to keys in the main keyring.
447 * /sys/fs/bpf is now mounted automatically.
449 * QNX virtualization is now detected by elogind-detect-virt and may
450 be used in ConditionVirtualization=.
452 * IPAccounting= may now be enabled also for slice units.
454 * A new -Dsplit-bin= build configuration switch may be used to specify
455 whether bin and sbin directories are merged, or if they should be
456 included separately in $PATH and various listings of executable
457 directories. The build configuration scripts will try to autodetect
458 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
459 system, but distributions are encouraged to configure this
462 * A new -Dok-color= build configuration switch may be used to change
463 the colour of "OK" status messages.
465 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
466 PrivateNetwork=yes was buggy in previous versions of elogind. This
467 means that after the upgrade and daemon-reexec, any such units must
470 * INCOMPATIBILITY: as announced in the NEWS for 237, elogind-tmpfiles
471 will not exclude read-only files owned by root from cleanup.
473 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
474 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
475 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
476 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
477 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
478 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
479 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
480 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
481 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
482 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
483 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
484 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
485 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
486 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
487 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
488 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
494 * Some keyboards come with a zoom see-saw or rocker which until now got
495 mapped to the Linux "zoomin/out" keys in hwdb. However, these
496 keycodes are not recognized by any major desktop. They now produce
497 Up/Down key events so that they can be used for scrolling.
499 * INCOMPATIBILITY: elogind-tmpfiles' "f" lines changed behaviour
500 slightly: previously, if an argument was specified for lines of this
501 type (i.e. the right-most column was set) this string was appended to
502 existing files each time elogind-tmpfiles was run. This behaviour was
503 different from what the documentation said, and not particularly
504 useful, as repeated elogind-tmpfiles invocations would not be
505 idempotent and grow such files without bounds. With this release
506 behaviour has been altered slightly, to match what the documentation
507 says: lines of this type only have an effect if the indicated files
508 don't exist yet, and only then the argument string is written to the
511 * FUTURE INCOMPATIBILITY: In elogind v238 we intend to slightly change
512 elogind-tmpfiles behaviour: previously, read-only files owned by root
513 were always excluded from the file "aging" algorithm (i.e. the
514 automatic clean-up of directories like /tmp based on
515 atime/mtime/ctime). We intend to drop this restriction, and age files
516 by default even when owned by root and read-only. This behaviour was
517 inherited from older tools, but there have been requests to remove
518 it, and it's not obvious why this restriction was made in the first
519 place. Please speak up now, if you are aware of software that reqires
520 this behaviour, otherwise we'll remove the restriction in v238.
522 * A new environment variable $SYSTEMD_OFFLINE is now understood by
523 systemctl. It takes a boolean argument. If on, systemctl assumes it
524 operates on an "offline" OS tree, and will not attempt to talk to the
525 service manager. Previously, this mode was implicitly enabled if a
526 chroot() environment was detected, and this new environment variable
527 now provides explicit control.
529 * .path and .socket units may now be created transiently, too.
530 Previously only service, mount, automount and timer units were
531 supported as transient units. The elogind-run tool has been updated
532 to expose this new functionality, you may hence use it now to bind
533 arbitrary commands to path or socket activation on-the-fly from the
534 command line. Moreover, almost all properties are now exposed for the
535 unit types that already supported transient operation.
537 * The elogind-mount command gained support for a new --owner= parameter
538 which takes a user name, which is then resolved and included in uid=
539 and gid= mount options string of the file system to mount.
541 * A new unit condition ConditionControlGroupController= has been added
542 that checks whether a specific cgroup controller is available.
544 * Unit files, udev's .link files, and elogind-networkd's .netdev and
545 .network files all gained support for a new condition
546 ConditionKernelVersion= for checking against specific kernel
549 * In elogind-networkd, the [IPVLAN] section in .netdev files gained
550 support for configuring device flags in the Flags= setting. In the
551 same files, the [Tunnel] section gained support for configuring
552 AllowLocalRemote=. The [Route] section in .network files gained
553 support for configuring InitialCongestionWindow=,
554 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
555 understands RapidCommit=.
557 * elogind-networkd's DHCPv6 support gained support for Prefix
560 * sd-bus gained support for a new "watch-bind" feature. When this
561 feature is enabled, an sd_bus connection may be set up to connect to
562 an AF_UNIX socket in the file system as soon as it is created. This
563 functionality is useful for writing early-boot services that
564 automatically connect to the system bus as soon as it is started,
565 without ugly time-based polling. elogind-networkd and
566 elogind-resolved have been updated to make use of this
567 functionality. busctl exposes this functionality in a new
568 --watch-bind= command line switch.
570 * sd-bus will now optionally synthesize a local "Connected" signal as
571 soon as a D-Bus connection is set up fully. This message mirrors the
572 already existing "Disconnected" signal which is synthesized when the
573 connection is terminated. This signal is generally useful but
574 particularly handy in combination with the "watch-bind" feature
575 described above. Synthesizing of this message has to be requested
576 explicitly through the new API call sd_bus_set_connected_signal(). In
577 addition a new call sd_bus_is_ready() has been added that checks
578 whether a connection is fully set up (i.e. between the "Connected" and
579 "Disconnected" signals).
581 * sd-bus gained two new calls sd_bus_request_name_async() and
582 sd_bus_release_name_async() for asynchronously registering bus
583 names. Similar, there is now sd_bus_add_match_async() for installing
584 a signal match asynchronously. All of elogind's own services have
585 been updated to make use of these calls. Doing these operations
586 asynchronously has two benefits: it reduces the risk of deadlocks in
587 case of cyclic dependencies between bus services, and it speeds up
588 service initialization since synchronization points for bus
589 round-trips are removed.
591 * sd-bus gained two new calls sd_bus_match_signal() and
592 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
593 and sd_bus_add_match_async() but instead of taking a D-Bus match
594 string take match fields as normal function parameters.
596 * sd-bus gained two new calls sd_bus_set_sender() and
597 sd_bus_message_set_sender() for setting the sender name of outgoing
598 messages (either for all outgoing messages or for just one specific
599 one). These calls are only useful in direct connections as on
600 brokered connections the broker fills in the sender anyway,
601 overwriting whatever the client filled in.
603 * sd-event gained a new pseudo-handle that may be specified on all API
604 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
605 used this refers to the default event loop object of the calling
606 thread. Note however that this does not implicitly allocate one —
607 which has to be done prior by using sd_event_default(). Similarly
608 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
609 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
610 to the default bus of the specified type of the calling thread. Here
611 too this does not implicitly allocate bus connection objects, this
612 has to be done prior with sd_bus_default() and friends.
614 * sd-event gained a new call pair
615 sd_event_source_{get|set}_io_fd_own(). This may be used to request
616 automatic closure of the file descriptor an IO event source watches
617 when the event source is destroyed.
619 * elogind-networkd gained support for natively configuring WireGuard
622 * In previous versions elogind synthesized user records both for the
623 "nobody" (UID 65534) and "root" (UID 0) users in nss-elogind and
624 internally. In order to simplify distribution-wide renames of the
625 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
626 new transitional flag file has been added: if
627 /etc/elogind/dont-synthesize-nobody exists synthesizing of the 65534
628 user and group record within the elogind codebase is disabled.
630 * elogind-notify gained a new --uid= option for selecting the source
631 user/UID to use for notification messages sent to the service
634 * journalctl gained a new --grep= option to list only entries in which
635 the message matches a certain pattern. By default matching is case
636 insensitive if the pattern is lowercase, and case sensitive
637 otherwise. Option --case-sensitive=yes|no can be used to override
638 this an specify case sensitivity or case insensitivity.
640 * There's now a "elogind-analyze service-watchdogs" command for printing
641 the current state of the service runtime watchdog, and optionally
642 enabling or disabling the per-service watchdogs system-wide if given a
643 boolean argument (i.e. the concept you configure in WatchdogSec=), for
644 debugging purposes. There's also a kernel command line option
645 elogind.service_watchdogs= for controlling the same.
647 * Two new "log-level" and "log-target" options for elogind-analyze were
648 added that merge the now deprecated get-log-level, set-log-level and
649 get-log-target, set-log-target pairs. The deprecated options are still
650 understood for backwards compatibility. The two new options print the
651 current value when no arguments are given, and set them when a
652 level/target is given as an argument.
654 * sysusers.d's "u" lines now optionally accept both a UID and a GID
655 specification, separated by a ":" character, in order to create users
656 where UID and GID do not match.
658 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
659 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
660 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
661 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
662 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
663 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
664 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
665 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
666 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
667 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
668 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
669 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
670 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
671 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
672 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
673 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
674 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
675 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
676 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
677 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
684 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
685 in v235 has been extended to also set the dummy.ko module option
686 numdummies=0, preventing the kernel from automatically creating
687 dummy0. All dummy interfaces must now be explicitly created.
689 * Unknown '%' specifiers in configuration files are now rejected. This
690 applies to units and tmpfiles.d configuration. Any percent characters
691 that are followed by a letter or digit that are not supposed to be
692 interpreted as the beginning of a specifier should be escaped by
693 doubling ("%%"). (So "size=5%" is still accepted, as well as
694 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
695 valid specifiers today.)
697 * systemd-resolved now maintains a new dynamic
698 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
699 recommended to make /etc/resolv.conf a symlink to it. This file
700 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
701 includes dynamically acquired search domains, achieving more correct
702 DNS resolution by software that bypasses local DNS APIs such as NSS.
704 * The "uaccess" udev tag has been dropped from /dev/kvm and
705 /dev/dri/renderD*. These devices now have the 0666 permissions by
706 default (but this may be changed at build-time). /dev/dri/renderD*
707 will now be owned by the "render" group along with /dev/kfd.
709 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
710 systemd-journal-gatewayd.service and
711 systemd-journal-upload.service. This means "nss-systemd" must be
712 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
713 services are resolved properly.
715 * In /etc/fstab two new mount options are now understood:
716 x-systemd.makefs and x-systemd.growfs. The former has the effect that
717 the configured file system is formatted before it is mounted, the
718 latter that the file system is resized to the full block device size
719 after it is mounted (i.e. if the file system is smaller than the
720 partition it resides on, it's grown). This is similar to the fsck
721 logic in /etc/fstab, and pulls in systemd-makefs@.service and
722 systemd-growfs@.service as necessary, similar to
723 systemd-fsck@.service. Resizing is currently only supported on ext4
726 * In systemd-networkd, the IPv6 RA logic now optionally may announce
727 DNS server and domain information.
729 * Support for the LUKS2 on-disk format for encrypted partitions has
730 been added. This requires libcryptsetup2 during compilation and
733 * The systemd --user instance will now signal "readiness" when its
734 basic.target unit has been reached, instead of when the run queue ran
735 empty for the first time.
737 * Tmpfiles.d with user configuration are now also supported.
738 systemd-tmpfiles gained a new --user switch, and snippets placed in
739 ~/.config/user-tmpfiles.d/ and corresponding directories will be
740 executed by systemd-tmpfiles --user running in the new
741 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
742 running in the user session.
744 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
745 %S resolves to the top-level state directory (/var/lib for the system
746 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
747 top-level cache directory (/var/cache for the system instance,
748 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
749 logs directory (/var/log for the system instance,
750 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
751 existing %t specifier, that resolves to the top-level runtime
752 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
755 * journalctl learnt a new parameter --output-fields= for limiting the
756 set of journal fields to output in verbose and JSON output modes.
758 * systemd-timesyncd's configuration file gained a new option
759 RootDistanceMaxSec= for setting the maximum root distance of servers
760 it'll use, as well as the new options PollIntervalMinSec= and
761 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
763 * bootctl gained a new command "list" for listing all available boot
764 menu items on systems that follow the boot loader specification.
766 * systemctl gained a new --dry-run switch that shows what would be done
767 instead of doing it, and is currently supported by the shutdown and
770 * ConditionSecurity= can now detect the TOMOYO security module.
772 * Unit file [Install] sections are now also respected in unit drop-in
773 files. This is intended to be used by drop-ins under /usr/lib/.
775 * systemd-firstboot may now also set the initial keyboard mapping.
777 * Udev "changed" events for devices which are exposed as systemd
778 .device units are now propagated to units specified in
779 ReloadPropagatedFrom= as reload requests.
781 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
782 unit template name (i.e. a name in the form of 'foobar@.service',
783 without the instance component between the '@' and - the '.'), then
784 the escaped sysfs path of the device is automatically used as the
787 * SystemCallFilter= in unit files has been extended so that an "errno"
788 can be specified individually for each system call. Example:
789 SystemCallFilter=~uname:EILSEQ.
791 * The cgroup delegation logic has been substantially updated. Delegate=
792 now optionally takes a list of controllers (instead of a boolean, as
793 before), which lists the controllers to delegate at least.
795 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
797 * A new LogLevelMax= setting configures the maximum log level any
798 process of the service may log at (i.e. anything with a lesser
799 priority than what is specified is automatically dropped). A new
800 LogExtraFields= setting allows configuration of additional journal
801 fields to attach to all log records generated by any of the unit's
804 * New StandardInputData= and StandardInputText= settings along with the
805 new option StandardInput=data may be used to configure textual or
806 binary data that shall be passed to the executed service process via
807 standard input, encoded in-line in the unit file.
809 * StandardInput=, StandardOutput= and StandardError= may now be used to
810 connect stdin/stdout/stderr of executed processes directly with a
811 file or AF_UNIX socket in the file system, using the new "file:" option.
813 * A new unit file option CollectMode= has been added, that allows
814 tweaking the garbage collection logic for units. It may be used to
815 tell systemd to garbage collect units that have failed automatically
816 (normally it only GCs units that exited successfully). systemd-run
817 and systemd-mount expose this new functionality with a new -G option.
819 * "machinectl bind" may now be used to bind mount non-directories
820 (i.e. regularfiles, devices, fifos, sockets).
822 * systemd-analyze gained a new verb "calendar" for validating and
823 testing calendar time specifications to use for OnCalendar= in timer
824 units. Besides validating the expression it will calculate the next
825 time the specified expression would elapse.
827 * In addition to the pre-existing FailureAction= unit file setting
828 there's now SuccessAction=, for configuring a shutdown action to
829 execute when a unit completes successfully. This is useful in
830 particular inside containers that shall terminate after some workload
831 has been completed. Also, both options are now supported for all unit
832 types, not just services.
834 * networkds's IP rule support gained two new options
835 IncomingInterface= and OutgoingInterface= for configuring the incoming
836 and outgoing interfaces of configured rules. systemd-networkd also
837 gained support for "vxcan" network devices.
839 * networkd gained a new setting RequiredForOnline=, taking a
840 boolean. If set, systemd-wait-online will take it into consideration
841 when determining that the system is up, otherwise it will ignore the
842 interface for this purpose.
844 * The sd_notify() protocol gained support for a new operation: with
845 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
846 store again, ahead of POLLHUP or POLLERR when they are removed
849 * A new document doc/UIDS-GIDS.md has been added to the source tree,
850 that documents the UID/GID range and assignment assumptions and
851 requirements of systemd.
853 * The watchdog device PID 1 will ping may now be configured through the
854 WatchdogDevice= configuration file setting, or by setting the
855 systemd.watchdog_service= kernel commandline option.
857 * systemd-resolved's gained support for registering DNS-SD services on
858 the local network using MulticastDNS. Services may either be
859 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
860 the same dir below /run, /usr/lib), or through its D-Bus API.
862 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
863 extend the effective start, runtime, and stop time. The service must
864 continue to send EXTEND_TIMEOUT_USEC within the period specified to
865 prevent the service manager from making the service as timedout.
867 * elogind-resolved's DNSSEC support gained support for RFC 8080
868 (Ed25519 keys and signatures).
870 * The elogind-resolve command line tool gained a new set of options
871 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
872 --set-nta= and --revert to configure per-interface DNS configuration
873 dynamically during runtime. It's useful for pushing DNS information
874 into elogind-resolved from DNS hook scripts that various interface
875 managing software supports (such as pppd).
877 * elogind-nspawn gained a new --network-namespace-path= command line
878 option, which may be used to make a container join an existing
879 network namespace, by specifying a path to a "netns" file.
881 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
882 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
883 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
884 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
885 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
886 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
887 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
888 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
889 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
890 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
891 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
892 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
893 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
894 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
895 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
896 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
897 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
898 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
899 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
900 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
901 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
902 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
903 Jędrzejewski-Szmek, Zeal Jagannatha
909 * INCOMPATIBILITY: systemd-logind.service and other long-running
910 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
911 communication with the outside. This generally improves security of
912 the system, and is in almost all cases a safe and good choice, as
913 these services do not and should not provide any network-facing
914 functionality. However, systemd-logind uses the glibc NSS API to
915 query the user database. This creates problems on systems where NSS
916 is set up to directly consult network services for user database
917 lookups. In particular, this creates incompatibilities with the
918 "nss-nis" module, which attempts to directly contact the NIS/YP
919 network servers it is configured for, and will now consistently
920 fail. In such cases, it is possible to turn off IP sandboxing for
921 systemd-logind.service (set IPAddressDeny= in its [Service] section
922 to the empty string, via a .d/ unit file drop-in). Downstream
923 distributions might want to update their nss-nis packaging to include
924 such a drop-in snippet, accordingly, to hide this incompatibility
925 from the user. Another option is to make use of glibc's nscd service
926 to proxy such network requests through a privilege-separated, minimal
927 local caching daemon, or to switch to more modern technologies such
928 sssd, whose NSS hook-ups generally do not involve direct network
929 access. In general, we think it's definitely time to question the
930 implementation choices of nss-nis, i.e. whether it's a good idea
931 today to embed a network-facing loadable module into all local
932 processes that need to query the user database, including the most
933 trivial and benign ones, such as "ls". For more details about
934 IPAddressDeny= see below.
936 * A new modprobe.d drop-in is now shipped by default that sets the
937 bonding module option max_bonds=0. This overrides the kernel default,
938 to avoid conflicts and ambiguity as to whether or not bond0 should be
939 managed by systemd-networkd or not. This resolves multiple issues
940 with bond0 properties not being applied, when bond0 is configured
941 with systemd-networkd. Distributors may choose to not package this,
942 however in that case users will be prevented from correctly managing
943 bond0 interface using systemd-networkd.
945 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
946 which print the logging level and target of the system manager. They
947 complement the existing "set-log-level" and "set-log-target" verbs
948 used to change those values.
950 * journald.conf gained a new boolean setting ReadKMsg= which defaults
951 to on. If turned off kernel log messages will not be read by
952 systemd-journald or included in the logs. It also gained a new
953 setting LineMax= for configuring the maximum line length in
954 STDOUT/STDERR log streams. The new default for this value is 48K, up
955 from the previous hardcoded 2048.
957 * A new unit setting RuntimeDirectoryPreserve= has been added, which
958 allows more detailed control of what to do with a runtime directory
959 configured with RuntimeDirectory= (i.e. a directory below /run or
960 $XDG_RUNTIME_DIR) after a unit is stopped.
962 * The RuntimeDirectory= setting for units gained support for creating
963 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
964 one top-level directory.
966 * Units gained new options StateDirectory=, CacheDirectory=,
967 LogsDirectory= and ConfigurationDirectory= which are closely related
968 to RuntimeDirectory= but manage per-service directories below
969 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
970 possible to write unit files which when activated automatically gain
971 properly owned service specific directories in these locations, thus
972 making unit files self-contained and increasing compatibility with
973 stateless systems and factory reset where /etc or /var are
974 unpopulated at boot. Matching these new settings there's also
975 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
976 ConfigurationDirectoryMode= for configuring the access mode of these
977 directories. These settings are particularly useful in combination
978 with DynamicUser=yes as they provide secure, properly-owned,
979 writable, and stateful locations for storage, excluded from the
980 sandbox that such services live in otherwise.
982 * Automake support has been removed from this release. systemd is now
985 * systemd-journald will now aggressively cache client metadata during
986 runtime, speeding up log write performance under pressure. This comes
987 at a small price though: as much of the metadata is read
988 asynchronously from /proc/ (and isn't implicitly attached to log
989 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
990 metadata stored alongside a log entry might be slightly
991 out-of-date. Previously it could only be slightly newer than the log
992 message. The time window is small however, and given that the kernel
993 is unlikely to be improved anytime soon in this regard, this appears
996 * nss-myhostname/systemd-resolved will now by default synthesize an
997 A/AAAA resource record for the "_gateway" hostname, pointing to the
998 current default IP gateway. Previously it did that for the "gateway"
999 name, hampering adoption, as some distributions wanted to leave that
1000 host name open for local use. The old behaviour may still be
1001 requested at build time.
1003 * systemd-networkd's [Address] section in .network files gained a new
1004 Scope= setting for configuring the IP address scope. The [Network]
1005 section gained a new boolean setting ConfigureWithoutCarrier= that
1006 tells systemd-networkd to ignore link sensing when configuring the
1007 device. The [DHCP] section gained a new Anonymize= boolean option for
1008 turning on a number of options suggested in RFC 7844. A new
1009 [RoutingPolicyRule] section has been added for configuring the IP
1010 routing policy. The [Route] section has gained support for a new
1011 Type= setting which permits configuring
1012 blackhole/unreachable/prohibit routes.
1014 * The [VRF] section in .netdev files gained a new Table= setting for
1015 configuring the routing table to use. The [Tunnel] section gained a
1016 new Independent= boolean field for configuring tunnels independent of
1017 an underlying network interface. The [Bridge] section gained a new
1018 GroupForwardMask= option for configuration of propagation of link
1019 local frames between bridge ports.
1021 * The WakeOnLan= setting in .link files gained support for a number of
1022 new modes. A new TCP6SegmentationOffload= setting has been added for
1023 configuring TCP/IPv6 hardware segmentation offload.
1025 * The IPv6 RA sender implementation may now optionally send out RDNSS
1026 and RDNSSL records to supply DNS configuration to peers.
1028 * systemd-nspawn gained support for a new --system-call-filter= command
1029 line option for adding and removing entries in the default system
1030 call filter it applies. Moreover systemd-nspawn has been changed to
1031 implement a system call whitelist instead of a blacklist.
1033 * systemd-run gained support for a new --pipe command line option. If
1034 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
1035 are directly passed on to the activated transient service
1036 executable. This allows invoking arbitrary processes as systemd
1037 services (for example to take benefit of dependency management,
1038 accounting management, resource management or log management that is
1039 done automatically for services) — while still allowing them to be
1040 integrated in a classic UNIX shell pipeline.
1042 * When a service sends RELOAD=1 via sd_notify() and reload propagation
1043 using ReloadPropagationTo= is configured, a reload is now propagated
1044 to configured units. (Previously this was only done on explicitly
1045 requested reloads, using "systemctl reload" or an equivalent
1048 * For each service unit a restart counter is now kept: it is increased
1049 each time the service is restarted due to Restart=, and may be
1050 queried using "systemctl show -p NRestarts …".
1052 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
1053 @signal and @timer have been added, for usage with SystemCallFilter=
1054 in unit files and the new --system-call-filter= command line option
1055 of systemd-nspawn (see above).
1057 * ExecStart= lines in unit files gained two new modifiers: when a
1058 command line is prefixed with "!" the command will be executed as
1059 configured, except for the credentials applied by
1060 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
1061 "+", but does still apply namespacing options unlike "+". There's
1062 also "!!" now, which is mostly identical, but becomes a NOP on
1063 systems that support ambient capabilities. This is useful to write
1064 unit files that work with ambient capabilities where possible but
1065 automatically fall back to traditional privilege dropping mechanisms
1066 on systems where this is not supported.
1068 * ListenNetlink= settings in socket units now support RDMA netlink
1071 * A new unit file setting LockPersonality= has been added which permits
1072 locking down the chosen execution domain ("personality") of a service
1075 * A new special target "getty-pre.target" has been added, which is
1076 ordered before all text logins, and may be used to order services
1077 before textual logins acquire access to the console.
1079 * systemd will now attempt to load the virtio-rng.ko kernel module very
1080 early on if a VM environment supporting this is detected. This should
1081 improve entropy during early boot in virtualized environments.
1083 * A _netdev option is now supported in /etc/crypttab that operates in a
1084 similar way as the same option in /etc/fstab: it permits configuring
1085 encrypted devices that need to be ordered after the network is up.
1086 Following this logic, two new special targets
1087 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
1088 added that are to cryptsetup.target what remote-fs.target and
1089 remote-fs-pre.target are to local-fs.target.
1091 * Service units gained a new UnsetEnvironment= setting which permits
1092 unsetting specific environment variables for services that are
1093 normally passed to it (for example in order to mask out locale
1094 settings for specific services that can't deal with it).
1096 * Units acquired a new boolean option IPAccounting=. When turned on, IP
1097 traffic accounting (packet count as well as byte count) is done for
1098 the service, and shown as part of "systemctl status" or "systemd-run
1101 * Service units acquired two new options IPAddressAllow= and
1102 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
1103 for configuring a simple IP access control list for all sockets of
1104 the unit. These options are available also on .slice and .socket
1105 units, permitting flexible access list configuration for individual
1106 services as well as groups of services (as defined by a slice unit),
1107 including system-wide. Note that IP ACLs configured this way are
1108 enforced on every single IPv4 and IPv6 socket created by any process
1109 of the service unit, and apply to ingress as well as egress traffic.
1111 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
1112 structured log message is generated each time the unit is stopped,
1113 containing information about the consumed resources of this
1116 * A new setting KeyringMode= has been added to unit files, which may be
1117 used to control how the kernel keyring is set up for executed
1120 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
1121 "systemctl kexec" and "systemctl exit" are now always asynchronous in
1122 behaviour (that is: these commands return immediately after the
1123 operation was enqueued instead of waiting for the operation to
1124 complete). Previously, "systemctl poweroff" and "systemctl reboot"
1125 were asynchronous on systems using systemd-logind (i.e. almost
1126 always, and like they were on sysvinit), and the other three commands
1127 were unconditionally synchronous. With this release this is cleaned
1128 up, and callers will see the same asynchronous behaviour on all
1129 systems for all five operations.
1131 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
1134 * .timer units now accept calendar specifications in other timezones
1135 than UTC or the local timezone.
1137 * The tmpfiles snippet var.conf has been changed to create
1138 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
1139 the "utmp" group already, and it appears to be generally understood
1140 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
1141 databases. Previously this was implemented correctly for all these
1142 databases excepts btmp, which has been opened up like this now
1143 too. Note that while the other databases are world-readable
1144 (i.e. 0644), btmp is not and remains more restrictive.
1146 * The systemd-resolve tool gained a new --reset-server-features
1147 switch. When invoked like this systemd-resolved will forget
1148 everything it learnt about the features supported by the configured
1149 upstream DNS servers, and restarts the feature probing logic on the
1150 next resolver look-up for them at the highest feature level
1153 * The status dump systemd-resolved sends to the logs upon receiving
1154 SIGUSR1 now also includes information about all DNS servers it is
1155 configured to use, and the features levels it probed for them.
1157 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
1158 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
1159 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
1160 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
1161 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
1162 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
1163 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
1164 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
1165 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
1166 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
1167 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
1168 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
1169 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
1170 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
1171 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
1172 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
1173 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
1174 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
1175 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
1176 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1178 — Berlin, 2017-10-06
1182 * Meson is now supported as build system in addition to Automake. It is
1183 our plan to remove Automake in one of our next releases, so that
1184 Meson becomes our exclusive build system. Hence, please start using
1185 the Meson build system in your downstream packaging. There's plenty
1186 of documentation around how to use Meson, the extremely brief
1189 ./autogen.sh && ./configure && make && sudo make install
1193 meson build && ninja -C build && sudo ninja -C build install
1195 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
1196 which permits configuring a timeout on the time a job is
1197 running. This is particularly useful for setting timeouts on jobs for
1200 * Unit files gained two new options ConditionUser= and ConditionGroup=
1201 for conditionalizing units based on the identity of the user/group
1202 running a systemd user instance.
1204 * systemd-networkd now understands a new FlowLabel= setting in the
1205 [VXLAN] section of .network files, as well as a Priority= in
1206 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
1207 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
1208 gained support for configuration of GENEVE links, and IPv6 address
1209 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
1211 * .link files now understand a new Port= setting.
1213 * systemd-networkd's DHCP support gained support for DHCP option 119
1214 (domain search list).
1216 * systemd-networkd gained support for serving IPv6 address ranges using
1217 the Router Advertisement protocol. The new .network configuration
1218 section [IPv6Prefix] may be used to configure the ranges to
1219 serve. This is implemented based on a new, minimal, native server
1220 implementation of RA.
1222 * journalctl's --output= switch gained support for a new parameter
1223 "short-iso-precise" for a mode where timestamps are shown as precise
1226 * systemd-udevd's "net_id" builtin may now generate stable network
1227 interface names from IBM PowerVM VIO devices as well as ACPI platform
1230 * MulticastDNS support in systemd-resolved may now be explicitly
1231 enabled/disabled using the new MulticastDNS= configuration file
1234 * systemd-resolved may now optionally use libidn2 instead of the libidn
1235 for processing internationalized domain names. Support for libidn2
1236 should be considered experimental and should not be enabled by
1239 * "machinectl pull-tar" and related call may now do verification of
1240 downloaded images using SUSE-style .sha256 checksum files in addition
1241 to the already existing support for validating using Ubuntu-style
1244 * sd-bus gained support for a new sd_bus_message_appendv() call which
1245 is va_list equivalent of sd_bus_message_append().
1247 * sd-boot gained support for validating images using SHIM/MOK.
1249 * The SMACK code learnt support for "onlycap".
1251 * systemd-mount --umount is now much smarter in figuring out how to
1252 properly unmount a device given its mount or device path.
1254 * The code to call libnss_dns as a fallback from libnss_resolve when
1255 the communication with systemd-resolved fails was removed. This
1256 fallback was redundant and interfered with the [!UNAVAIL=return]
1257 suffix. See nss-resolve(8) for the recommended configuration.
1259 * systemd-logind may now be restarted without losing state. It stores
1260 the file descriptors for devices it manages in the system manager
1261 using the FDSTORE= mechanism. Please note that further changes in
1262 other components may be required to make use of this (for example
1263 Xorg has code to listen for stops of systemd-logind and terminate
1264 itself when logind is stopped or restarted, in order to avoid using
1265 stale file descriptors for graphical devices, which is now
1266 counterproductive and must be reverted in order for restarts of
1267 systemd-logind to be safe. See
1268 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
1270 * All kernel install plugins are called with the environment variable
1271 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
1272 /etc/machine-id. If the file is missing or empty, the variable is
1273 empty and BOOT_DIR_ABS is the path of a temporary directory which is
1274 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
1275 is empty, all plugins should not put anything in BOOT_DIR_ABS.
1277 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
1278 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
1279 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
1280 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
1281 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
1282 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
1283 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
1284 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
1285 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
1286 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
1287 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
1288 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
1289 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
1290 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
1291 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
1292 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
1293 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
1294 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
1295 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
1296 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
1297 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
1298 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
1299 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
1300 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
1301 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
1302 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
1303 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
1304 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
1305 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
1308 — Berlin, 2017-07-12
1312 * The "hybrid" control group mode has been modified to improve
1313 compatibility with "legacy" cgroups-v1 setups. Specifically, the
1314 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
1315 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
1316 cgroups-v1 hierarchy), the only externally visible change being that
1317 the cgroups-v2 hierarchy is also mounted, to
1318 /sys/fs/cgroup/unified. This should provide a large degree of
1319 compatibility with "legacy" cgroups-v1, while taking benefit of the
1320 better management capabilities of cgroups-v2.
1322 * The default control group setup mode may be selected both a boot-time
1323 via a set of kernel command line parameters (specifically:
1324 systemd.unified_cgroup_hierarchy= and
1325 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
1326 default selected on the configure command line
1327 (--with-default-hierarchy=). The upstream default is "hybrid"
1328 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
1329 this will change in a future systemd version to be "unified" (pure
1330 cgroups-v2 mode). The third option for the compile time option is
1331 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
1332 distributions to default to "hybrid" mode for release distributions,
1333 starting with v233. We recommend "unified" for development
1334 distributions (specifically: distributions such as Fedora's rawhide)
1335 as that's where things are headed in the long run. Use "legacy" for
1336 greatest stability and compatibility only.
1338 * Note one current limitation of "unified" and "hybrid" control group
1339 setup modes: the kernel currently does not permit the systemd --user
1340 instance (i.e. unprivileged code) to migrate processes between two
1341 disconnected cgroup subtrees, even if both are managed and owned by
1342 the user. This effectively means "systemd-run --user --scope" doesn't
1343 work when invoked from outside of any "systemd --user" service or
1344 scope. Specifically, it is not supported from session scopes. We are
1345 working on fixing this in a future systemd version. (See #3388 for
1346 further details about this.)
1348 * DBus policy files are now installed into /usr rather than /etc. Make
1349 sure your system has dbus >= 1.9.18 running before upgrading to this
1350 version, or override the install path with --with-dbuspolicydir= .
1352 * All python scripts shipped with systemd (specifically: the various
1353 tests written in Python) now require Python 3.
1355 * systemd unit tests can now run standalone (without the source or
1356 build directories), and can be installed into /usr/lib/systemd/tests/
1357 with 'make install-tests'.
1359 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
1360 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
1363 * Support for the %c, %r, %R specifiers in unit files has been
1364 removed. Specifiers are not supposed to be dependent on configuration
1365 in the unit file itself (so that they resolve the same regardless
1366 where used in the unit files), but these specifiers were influenced
1367 by the Slice= option.
1369 * The shell invoked by debug-shell.service now defaults to /bin/sh in
1370 all cases. If distributions want to use a different shell for this
1371 purpose (for example Fedora's /sbin/sushell) they need to specify
1372 this explicitly at configure time using --with-debug-shell=.
1374 * The confirmation spawn prompt has been reworked to offer the
1377 (c)ontinue, proceed without asking anymore
1378 (D)ump, show the state of the unit
1379 (f)ail, don't execute the command and pretend it failed
1381 (i)nfo, show a short summary of the unit
1382 (j)obs, show jobs that are in progress
1383 (s)kip, don't execute the command and pretend it succeeded
1384 (y)es, execute the command
1386 The 'n' choice for the confirmation spawn prompt has been removed,
1387 because its meaning was confusing.
1389 The prompt may now also be redirected to an alternative console by
1390 specifying the console as parameter to systemd.confirm_spawn=.
1392 * Services of Type=notify require a READY=1 notification to be sent
1393 during startup. If no such message is sent, the service now fails,
1394 even if the main process exited with a successful exit code.
1396 * Services that fail to start up correctly now always have their
1397 ExecStopPost= commands executed. Previously, they'd enter "failed"
1398 state directly, without executing these commands.
1400 * The option MulticastDNS= of network configuration files has acquired
1401 an actual implementation. With MulticastDNS=yes a host can resolve
1402 names of remote hosts and reply to mDNS A and AAAA requests.
1404 * When units are about to be started an additional check is now done to
1405 ensure that all dependencies of type BindsTo= (when used in
1406 combination with After=) have been started.
1408 * systemd-analyze gained a new verb "syscall-filter" which shows which
1409 system call groups are defined for the SystemCallFilter= unit file
1410 setting, and which system calls they contain.
1412 * A new system call filter group "@filesystem" has been added,
1413 consisting of various file system related system calls. Group
1414 "@reboot" has been added, covering reboot, kexec and shutdown related
1415 calls. Finally, group "@swap" has been added covering swap
1416 configuration related calls.
1418 * A new unit file option RestrictNamespaces= has been added that may be
1419 used to restrict access to the various process namespace types the
1420 Linux kernel provides. Specifically, it may be used to take away the
1421 right for a service unit to create additional file system, network,
1422 user, and other namespaces. This sandboxing option is particularly
1423 relevant due to the high amount of recently discovered namespacing
1424 related vulnerabilities in the kernel.
1426 * systemd-udev's .link files gained support for a new AutoNegotiation=
1427 setting for configuring Ethernet auto-negotiation.
1429 * systemd-networkd's .network files gained support for a new
1430 ListenPort= setting in the [DHCP] section to explicitly configure the
1431 UDP client port the DHCP client shall listen on.
1433 * .network files gained a new Unmanaged= boolean setting for explicitly
1434 excluding one or more interfaces from management by systemd-networkd.
1436 * The systemd-networkd ProxyARP= option has been renamed to
1437 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
1438 renamed to ReduceARPProxy=. The old names continue to be available
1441 * systemd-networkd gained support for configuring IPv6 Proxy NDP
1442 addresses via the new IPv6ProxyNDPAddress= .network file setting.
1444 * systemd-networkd's bonding device support gained support for two new
1445 configuration options ActiveSlave= and PrimarySlave=.
1447 * The various options in the [Match] section of .network files gained
1448 support for negative matching.
1450 * New systemd-specific mount options are now understood in /etc/fstab:
1452 x-systemd.mount-timeout= may be used to configure the maximum
1453 permitted runtime of the mount command.
1455 x-systemd.device-bound may be set to bind a mount point to its
1456 backing device unit, in order to automatically remove a mount point
1457 if its backing device is unplugged. This option may also be
1458 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
1459 on the block device, which is now automatically set for all CDROM
1460 drives, so that mounted CDs are automatically unmounted when they are
1461 removed from the drive.
1463 x-systemd.after= and x-systemd.before= may be used to explicitly
1464 order a mount after or before another unit or mount point.
1466 * Enqueued start jobs for device units are now automatically garbage
1467 collected if there are no jobs waiting for them anymore.
1469 * systemctl list-jobs gained two new switches: with --after, for every
1470 queued job the jobs it's waiting for are shown; with --before the
1471 jobs which it's blocking are shown.
1473 * systemd-nspawn gained support for ephemeral boots from disk images
1474 (or in other words: --ephemeral and --image= may now be
1475 combined). Moreover, ephemeral boots are now supported for normal
1476 directories, even if the backing file system is not btrfs. Of course,
1477 if the file system does not support file system snapshots or
1478 reflinks, the initial copy operation will be relatively expensive, but
1479 this should still be suitable for many use cases.
1481 * Calendar time specifications in .timer units now support
1482 specifications relative to the end of a month by using "~" instead of
1483 "-" as separator between month and day. For example, "*-02~03" means
1484 "the third last day in February". In addition a new syntax for
1485 repeated events has been added using the "/" character. For example,
1486 "9..17/2:00" means "every two hours from 9am to 5pm".
1488 * systemd-socket-proxyd gained a new parameter --connections-max= for
1489 configuring the maximum number of concurrent connections.
1491 * sd-id128 gained a new API for generating unique IDs for the host in a
1492 way that does not leak the machine ID. Specifically,
1493 sd_id128_get_machine_app_specific() derives an ID based on the
1494 machine ID a in well-defined, non-reversible, stable way. This is
1495 useful whenever an identifier for the host is needed but where the
1496 identifier shall not be useful to identify the system beyond the
1497 scope of the application itself. (Internally this uses HMAC-SHA256 as
1498 keyed hash function using the machine ID as input.)
1500 * NotifyAccess= gained a new supported value "exec". When set
1501 notifications are accepted from all processes systemd itself invoked,
1502 including all control processes.
1504 * .nspawn files gained support for defining overlay mounts using the
1505 Overlay= and OverlayReadOnly= options. Previously this functionality
1506 was only available on the systemd-nspawn command line.
1508 * systemd-nspawn's --bind= and --overlay= options gained support for
1509 bind/overlay mounts whose source lies within the container tree by
1510 prefixing the source path with "+".
1512 * systemd-nspawn's --bind= and --overlay= options gained support for
1513 automatically allocating a temporary source directory in /var/tmp
1514 that is removed when the container dies. Specifically, if the source
1515 directory is specified as empty string this mechanism is selected. An
1516 example usage is --overlay=+/var::/var, which creates an overlay
1517 mount based on the original /var contained in the image, overlayed
1518 with a temporary directory in the host's /var/tmp. This way changes
1519 to /var are automatically flushed when the container shuts down.
1521 * systemd-nspawn --image= option does now permit raw file system block
1522 devices (in addition to images containing partition tables, as
1525 * The disk image dissection logic in systemd-nspawn gained support for
1526 automatically setting up LUKS encrypted as well as Verity protected
1527 partitions. When a container is booted from an encrypted image the
1528 passphrase is queried at start-up time. When a container with Verity
1529 data is started, the root hash is search in a ".roothash" file
1530 accompanying the disk image (alternatively, pass the root hash via
1531 the new --root-hash= command line option).
1533 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
1534 be used to dissect disk images the same way as systemd-nspawn does
1535 it, following the Bootable Partition Specification. It may even be
1536 used to mount disk images with complex partition setups (including
1537 LUKS and Verity partitions) to a local host directory, in order to
1538 inspect them. This tool is not considered public API (yet), and is
1539 thus not installed into /usr/bin. Please do not rely on its
1540 existence, since it might go away or be changed in later systemd
1543 * A new generator "systemd-verity-generator" has been added, similar in
1544 style to "systemd-cryptsetup-generator", permitting automatic setup of
1545 Verity root partitions when systemd boots up. In order to make use of
1546 this your partition setup should follow the Discoverable Partitions
1547 Specification, and the GPT partition ID of the root file system
1548 partition should be identical to the upper 128bit of the Verity root
1549 hash. The GPT partition ID of the Verity partition protecting it
1550 should be the lower 128bit of the Verity root hash. If the partition
1551 image follows this model it is sufficient to specify a single
1552 "roothash=" kernel command line argument to both configure which root
1553 image and verity partition to use as well as the root hash for
1554 it. Note that systemd-nspawn's Verity support follows the same
1555 semantics, meaning that disk images with proper Verity data in place
1556 may be booted in containers with systemd-nspawn as well as on
1557 physical systems via the verity generator. Also note that the "mkosi"
1558 tool available at https://github.com/systemd/mkosi has been updated
1559 to generate Verity protected disk images following this scheme. In
1560 fact, it has been updated to generate disk images that optionally
1561 implement a complete UEFI SecureBoot trust chain, involving a signed
1562 kernel and initrd image that incorporates such a root hash as well as
1563 a Verity-enabled root partition.
1565 * The hardware database (hwdb) udev supports has been updated to carry
1566 accelerometer quirks.
1568 * All system services are now run with a fresh kernel keyring set up
1569 for them. The invocation ID is stored by default in it, thus
1570 providing a safe, non-overridable way to determine the invocation
1573 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
1574 options for bind mounting arbitrary paths in a service-specific
1575 way. When these options are used, arbitrary host or service files and
1576 directories may be mounted to arbitrary locations in the service's
1579 * Documentation has been added that lists all of systemd's low-level
1580 environment variables:
1582 https://github.com/systemd/systemd/blob/master/doc/ENVIRONMENT.md
1584 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
1585 whether a specific socket file descriptor matches a specified socket
1588 * systemd-firstboot has been updated to check for the
1589 systemd.firstboot= kernel command line option. It accepts a boolean
1590 and when set to false the first boot questions are skipped.
1592 * systemd-fstab-generator has been updated to check for the
1593 systemd.volatile= kernel command line option, which either takes an
1594 optional boolean parameter or the special value "state". If used the
1595 system may be booted in a "volatile" boot mode. Specifically,
1596 "systemd.volatile" is used, the root directory will be mounted as
1597 tmpfs, and only /usr is mounted from the actual root file system. If
1598 "systemd.volatile=state" is used, the root directory will be mounted
1599 as usual, but /var is mounted as tmpfs. This concept provides similar
1600 functionality as systemd-nspawn's --volatile= option, but provides it
1601 on physical boots. Use this option for implementing stateless
1602 systems, or testing systems with all state and/or configuration reset
1603 to the defaults. (Note though that many distributions are not
1604 prepared to boot up without a populated /etc or /var, though.)
1606 * systemd-gpt-auto-generator gained support for LUKS encrypted root
1607 partitions. Previously it only supported LUKS encrypted partitions
1608 for all other uses, except for the root partition itself.
1610 * Socket units gained support for listening on AF_VSOCK sockets for
1611 communication in virtualized QEMU environments.
1613 * The "configure" script gained a new option --with-fallback-hostname=
1614 for specifying the fallback hostname to use if none is configured in
1615 /etc/hostname. For example, by specifying
1616 --with-fallback-hostname=fedora it is possible to default to a
1617 hostname of "fedora" on pristine installations.
1619 * systemd-cgls gained support for a new --unit= switch for listing only
1620 the control groups of a specific unit. Similar --user-unit= has been
1621 added for listing only the control groups of a specific user unit.
1623 * systemd-mount gained a new --umount switch for unmounting a mount or
1624 automount point (and all mount/automount points below it).
1626 * systemd will now refuse full configuration reloads (via systemctl
1627 daemon-reload and related calls) unless at least 16MiB of free space
1628 are available in /run. This is a safety precaution in order to ensure
1629 that generators can safely operate after the reload completed.
1631 * A new unit file option RootImage= has been added, which has a similar
1632 effect as RootDirectory= but mounts the service's root directory from
1633 a disk image instead of plain directory. This logic reuses the same
1634 image dissection and mount logic that systemd-nspawn already uses,
1635 and hence supports any disk images systemd-nspawn supports, including
1636 those following the Discoverable Partition Specification, as well as
1637 Verity enabled images. This option enables systemd to run system
1638 services directly off disk images acting as resource bundles,
1639 possibly even including full integrity data.
1641 * A new MountAPIVFS= unit file option has been added, taking a boolean
1642 argument. If enabled /proc, /sys and /dev (collectively called the
1643 "API VFS") will be mounted for the service. This is only relevant if
1644 RootDirectory= or RootImage= is used for the service, as these mounts
1645 are of course in place in the host mount namespace anyway.
1647 * systemd-nspawn gained support for a new --pivot-root= switch. If
1648 specified the root directory within the container image is pivoted to
1649 the specified mount point, while the original root disk is moved to a
1650 different place. This option enables booting of ostree images
1651 directly with systemd-nspawn.
1653 * The systemd build scripts will no longer complain if the NTP server
1654 addresses are not changed from the defaults. Google now supports
1655 these NTP servers officially. We still recommend downstreams to
1656 properly register an NTP pool with the NTP pool project though.
1658 * coredumpctl gained a new "--reverse" option for printing the list
1659 of coredumps in reverse order.
1661 * coredumpctl will now show additional information about truncated and
1662 inaccessible coredumps, as well as coredumps that are still being
1663 processed. It also gained a new --quiet switch for suppressing
1664 additional informational message in its output.
1666 * coredumpctl gained support for only showing coredumps newer and/or
1667 older than specific timestamps, using the new --since= and --until=
1668 options, reminiscent of journalctl's options by the same name.
1670 * The systemd-coredump logic has been improved so that it may be reused
1671 to collect backtraces in non-compiled languages, for example in
1672 scripting languages such as Python.
1674 * machinectl will now show the UID shift of local containers, if user
1675 namespacing is enabled for them.
1677 * systemd will now optionally run "environment generator" binaries at
1678 configuration load time. They may be used to add environment
1679 variables to the environment block passed to services invoked. One
1680 user environment generator is shipped by default that sets up
1681 environment variables based on files dropped into /etc/environment.d
1682 and ~/.config/environment.d/.
1684 * systemd-resolved now includes the new, recently published 2017 DNSSEC
1687 * hostnamed has been updated to report a new chassis type of
1688 "convertible" to cover "foldable" laptops that can both act as a
1689 tablet and as a laptop, such as various Lenovo Yoga devices.
1691 Contributions from: Adrián López, Alexander Galanin, Alexander
1692 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
1693 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
1694 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
1695 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
1696 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
1697 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
1698 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
1699 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
1700 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
1701 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
1702 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
1703 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
1704 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
1705 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
1706 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
1707 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
1708 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
1709 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
1710 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
1711 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
1712 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
1713 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
1714 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
1715 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
1716 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
1717 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
1718 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
1721 — Berlin, 2017-03-01
1725 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
1726 RestrictAddressFamilies= enabled. These sandboxing options should
1727 generally be compatible with the various external udev call-out
1728 binaries we are aware of, however there may be exceptions, in
1729 particular when exotic languages for these call-outs are used. In
1730 this case, consider turning off these settings locally.
1732 * The new RemoveIPC= option can be used to remove IPC objects owned by
1733 the user or group of a service when that service exits.
1735 * The new ProtectKernelModules= option can be used to disable explicit
1736 load and unload operations of kernel modules by a service. In
1737 addition access to /usr/lib/modules is removed if this option is set.
1739 * ProtectSystem= option gained a new value "strict", which causes the
1740 whole file system tree with the exception of /dev, /proc, and /sys,
1741 to be remounted read-only for a service.
1743 * The new ProtectKernelTunables= option can be used to disable
1744 modification of configuration files in /sys and /proc by a service.
1745 Various directories and files are remounted read-only, so access is
1746 restricted even if the file permissions would allow it.
1748 * The new ProtectControlGroups= option can be used to disable write
1749 access by a service to /sys/fs/cgroup.
1751 * Various systemd services have been hardened with
1752 ProtectKernelTunables=yes, ProtectControlGroups=yes,
1753 RestrictAddressFamilies=.
1755 * Support for dynamically creating users for the lifetime of a service
1756 has been added. If DynamicUser=yes is specified, user and group IDs
1757 will be allocated from the range 61184..65519 for the lifetime of the
1758 service. They can be resolved using the new nss-systemd.so NSS
1759 module. The module must be enabled in /etc/nsswitch.conf. Services
1760 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
1761 any resources allocated by the service will be cleaned up when the
1762 service exits. They also have ProtectHome=read-only and
1763 ProtectSystem=strict enabled, so they are not able to make any
1764 permanent modifications to the system.
1766 * The nss-systemd module also always resolves root and nobody, making
1767 it possible to have no /etc/passwd or /etc/group files in minimal
1768 container or chroot environments.
1770 * Services may be started with their own user namespace using the new
1771 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
1772 under which the service is running are mapped. All other users are
1775 * Support for the cgroup namespace has been added to systemd-nspawn. If
1776 supported by kernel, the container system started by systemd-nspawn
1777 will have its own view of the cgroup hierarchy. This new behaviour
1778 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
1780 * The new MemorySwapMax= option can be used to limit the maximum swap
1781 usage under the unified cgroup hierarchy.
1783 * Support for the CPU controller in the unified cgroup hierarchy has
1784 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
1785 options. This controller requires out-of-tree patches for the kernel
1786 and the support is provisional.
1788 * Mount and automount units may now be created transiently
1789 (i.e. dynamically at runtime via the bus API, instead of requiring
1790 unit files in the file system).
1792 * systemd-mount is a new tool which may mount file systems – much like
1793 mount(8), optionally pulling in additional dependencies through
1794 transient .mount and .automount units. For example, this tool
1795 automatically runs fsck on a backing block device before mounting,
1796 and allows the automount logic to be used dynamically from the
1797 command line for establishing mount points. This tool is particularly
1798 useful when dealing with removable media, as it will ensure fsck is
1799 run – if necessary – before the first access and that the file system
1800 is quickly unmounted after each access by utilizing the automount
1801 logic. This maximizes the chance that the file system on the
1802 removable media stays in a clean state, and if it isn't in a clean
1803 state is fixed automatically.
1805 * LazyUnmount=yes option for mount units has been added to expose the
1806 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
1809 * /efi will be used as the mount point of the EFI boot partition, if
1810 the directory is present, and the mount point was not configured
1811 through other means (e.g. fstab). If /efi directory does not exist,
1812 /boot will be used as before. This makes it easier to automatically
1813 mount the EFI partition on systems where /boot is used for something
1816 * When operating on GPT disk images for containers, systemd-nspawn will
1817 now mount the ESP to /boot or /efi according to the same rules as PID
1818 1 running on a host. This allows tools like "bootctl" to operate
1819 correctly within such containers, in order to make container images
1820 bootable on physical systems.
1822 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
1824 * Two new user session targets have been added to support running
1825 graphical sessions under the systemd --user instance:
1826 graphical-session.target and graphical-session-pre.target. See
1827 systemd.special(7) for a description of how those targets should be
1830 * The vconsole initialization code has been significantly reworked to
1831 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
1832 support unicode keymaps. Font and keymap configuration will now be
1833 copied to all allocated virtual consoles.
1835 * FreeBSD's bhyve virtualization is now detected.
1837 * Information recorded in the journal for core dumps now includes the
1838 contents of /proc/mountinfo and the command line of the process at
1839 the top of the process hierarchy (which is usually the init process
1842 * systemd-journal-gatewayd learned the --directory= option to serve
1843 files from the specified location.
1845 * journalctl --root=… can be used to peruse the journal in the
1846 /var/log/ directories inside of a container tree. This is similar to
1847 the existing --machine= option, but does not require the container to
1850 * The hardware database has been extended to support
1851 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
1854 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
1855 specify the click rate for mice which include a horizontal wheel with
1856 a click rate that is different than the one for the vertical wheel.
1858 * systemd-run gained a new --wait option that makes service execution
1859 synchronous. (Specifically, the command will not return until the
1860 specified service binary exited.)
1862 * systemctl gained a new --wait option that causes the start command to
1863 wait until the units being started have terminated again.
1865 * A new journal output mode "short-full" has been added which displays
1866 timestamps with abbreviated English day names and adds a timezone
1867 suffix. Those timestamps include more information than the default
1868 "short" output mode, and can be passed directly to journalctl's
1869 --since= and --until= options.
1871 * /etc/resolv.conf will be bind-mounted into containers started by
1872 systemd-nspawn, if possible, so any changes to resolv.conf contents
1873 are automatically propagated to the container.
1875 * The number of instances for socket-activated services originating
1876 from a single IP address can be limited with
1877 MaxConnectionsPerSource=, extending the existing setting of
1880 * systemd-networkd gained support for vcan ("Virtual CAN") interface
1883 * .netdev and .network configuration can now be extended through
1886 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
1887 Segmentation Offload, Generic Receive Offload, Large Receive Offload
1888 can be enabled and disabled using the new UDPSegmentationOffload=,
1889 TCPSegmentationOffload=, GenericSegmentationOffload=,
1890 GenericReceiveOffload=, LargeReceiveOffload= options in the
1891 [Link] section of .link files.
1893 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
1894 Port VLAN ID can be configured for bridge devices using the new STP=,
1895 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
1896 section of .netdev files.
1898 * The route table to which routes received over DHCP or RA should be
1899 added can be configured with the new RouteTable= option in the [DHCP]
1900 and [IPv6AcceptRA] sections of .network files.
1902 * The Address Resolution Protocol can be disabled on links managed by
1903 systemd-networkd using the ARP=no setting in the [Link] section of
1906 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
1907 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
1908 encode information about the result and exit codes of the current
1909 service runtime cycle.
1911 * systemd-sysctl will now configure kernel parameters in the order
1912 they occur in the configuration files. This matches what sysctl
1913 has been traditionally doing.
1915 * kernel-install "plugins" that are executed to perform various
1916 tasks after a new kernel is added and before an old one is removed
1917 can now return a special value to terminate the procedure and
1918 prevent any later plugins from running.
1920 * Journald's SplitMode=login setting has been deprecated. It has been
1921 removed from documentation, and its use is discouraged. In a future
1922 release it will be completely removed, and made equivalent to current
1923 default of SplitMode=uid.
1925 * Storage=both option setting in /etc/systemd/coredump.conf has been
1926 removed. With fast LZ4 compression storing the core dump twice is not
1929 * The --share-system systemd-nspawn option has been replaced with an
1930 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
1931 this functionality is discouraged. In addition the variables
1932 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
1933 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
1934 individual namespaces.
1936 * "machinectl list" now shows the IP address of running containers in
1937 the output, as well as OS release information.
1939 * "loginctl list" now shows the TTY of each session in the output.
1941 * sd-bus gained new API calls sd_bus_track_set_recursive(),
1942 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
1943 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
1944 tracking objects in a "recursive" mode, where a single client can be
1945 counted multiple times, if it takes multiple references.
1947 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
1948 sd_bus_get_exit_on_disconnect(). They may be used to make a
1949 process using sd-bus automatically exit if the bus connection is
1952 * Bus clients of the service manager may now "pin" loaded units into
1953 memory, by taking an explicit reference on them. This is useful to
1954 ensure the client can retrieve runtime data about the service even
1955 after the service completed execution. Taking such a reference is
1956 available only for privileged clients and should be helpful to watch
1957 running services in a race-free manner, and in particular collect
1958 information about exit statuses and results.
1960 * The nss-resolve module has been changed to strictly return UNAVAIL
1961 when communication via D-Bus with resolved failed, and NOTFOUND when
1962 a lookup completed but was negative. This means it is now possible to
1963 neatly configure fallbacks using nsswitch.conf result checking
1964 expressions. Taking benefit of this, the new recommended
1965 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
1967 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
1969 * A new setting CtrlAltDelBurstAction= has been added to
1970 /etc/systemd/system.conf which may be used to configure the precise
1971 behaviour if the user on the console presses Ctrl-Alt-Del more often
1972 than 7 times in 2s. Previously this would unconditionally result in
1973 an expedited, immediate reboot. With this new setting the precise
1974 operation may be configured in more detail, and also turned off
1977 * In .netdev files two new settings RemoteChecksumTx= and
1978 RemoteChecksumRx= are now understood that permit configuring the
1979 remote checksumming logic for VXLAN networks.
1981 * The service manager learnt a new "invocation ID" concept for invoked
1982 services. Each runtime cycle of a service will get a new invocation
1983 ID (a 128bit random UUID) assigned that identifies the current
1984 run of the service uniquely and globally. A new invocation ID
1985 is generated each time a service starts up. The journal will store
1986 the invocation ID of a service along with any logged messages, thus
1987 making the invocation ID useful for matching the online runtime of a
1988 service with the offline log data it generated in a safe way without
1989 relying on synchronized timestamps. In many ways this new service
1990 invocation ID concept is similar to the kernel's boot ID concept that
1991 uniquely and globally identifies the runtime of each boot. The
1992 invocation ID of a service is passed to the service itself via an
1993 environment variable ($INVOCATION_ID). A new bus call
1994 GetUnitByInvocationID() has been added that is similar to GetUnit()
1995 but instead of retrieving the bus path for a unit by its name
1996 retrieves it by its invocation ID. The returned path is valid only as
1997 long as the passed invocation ID is current.
1999 * systemd-resolved gained a new "DNSStubListener" setting in
2000 resolved.conf. It either takes a boolean value or the special values
2001 "udp" and "tcp", and configures whether to enable the stub DNS
2002 listener on 127.0.0.53:53.
2004 * IP addresses configured via networkd may now carry additional
2005 configuration settings supported by the kernel. New options include:
2006 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
2007 PrefixRoute=, AutoJoin=.
2009 * The PAM configuration fragment file for "user@.service" shipped with
2010 systemd (i.e. the --user instance of systemd) has been stripped to
2011 the minimum necessary to make the system boot. Previously, it
2012 contained Fedora-specific stanzas that did not apply to other
2013 distributions. It is expected that downstream distributions add
2014 additional configuration lines, matching their needs to this file,
2015 using it only as rough template of what systemd itself needs. Note
2016 that this reduced fragment does not even include an invocation of
2017 pam_limits which most distributions probably want to add, even though
2018 systemd itself does not need it. (There's also the new build time
2019 option --with-pamconfdir=no to disable installation of the PAM
2022 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
2023 capability is now also dropped from its set (in addition to
2024 CAP_SYS_MKNOD as before).
2026 * In service unit files it is now possible to connect a specific named
2027 file descriptor with stdin/stdout/stdout of an executed service. The
2028 name may be specified in matching .socket units using the
2029 FileDescriptorName= setting.
2031 * A number of journal settings may now be configured on the kernel
2032 command line. Specifically, the following options are now understood:
2033 systemd.journald.max_level_console=,
2034 systemd.journald.max_level_store=,
2035 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
2036 systemd.journald.max_level_wall=.
2038 * "systemctl is-enabled --full" will now show by which symlinks a unit
2039 file is enabled in the unit dependency tree.
2041 * Support for VeraCrypt encrypted partitions has been added to the
2042 "cryptsetup" logic and /etc/crypttab.
2044 * systemd-detect-virt gained support for a new --private-users switch
2045 that checks whether the invoking processes are running inside a user
2046 namespace. Similar, a new special value "private-users" for the
2047 existing ConditionVirtualization= setting has been added, permitting
2048 skipping of specific units in user namespace environments.
2050 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
2051 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
2052 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
2053 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
2054 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
2055 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
2056 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
2057 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
2058 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
2059 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
2060 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
2061 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
2062 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
2063 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
2064 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
2065 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
2066 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
2067 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
2068 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
2069 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
2070 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
2071 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
2072 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
2073 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
2074 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
2075 Jędrzejewski-Szmek, Zeal Jagannatha
2077 — Santa Fe, 2016-11-03
2081 * In service units the various ExecXYZ= settings have been extended
2082 with an additional special character as first argument of the
2083 assigned value: if the character '+' is used the specified command
2084 line it will be run with full privileges, regardless of User=,
2085 Group=, CapabilityBoundingSet= and similar options. The effect is
2086 similar to the existing PermissionsStartOnly= option, but allows
2087 configuration of this concept for each executed command line
2090 * Services may now alter the service watchdog timeout at runtime by
2091 sending a WATCHDOG_USEC= message via sd_notify().
2093 * MemoryLimit= and related unit settings now optionally take percentage
2094 specifications. The percentage is taken relative to the amount of
2095 physical memory in the system (or in case of containers, the assigned
2096 amount of memory). This allows scaling service resources neatly with
2097 the amount of RAM available on the system. Similarly, systemd-logind's
2098 RuntimeDirectorySize= option now also optionally takes percentage
2101 * In similar fashion TasksMax= takes percentage values now, too. The
2102 value is taken relative to the configured maximum number of processes
2103 on the system. The per-service task maximum has been changed to 15%
2104 using this functionality. (Effectively this is an increase of 512 →
2105 4915 for service units, given the kernel's default pid_max setting.)
2107 * Calendar time specifications in .timer units now understand a ".."
2108 syntax for time ranges. Example: "4..7:10" may now be used for
2109 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
2112 * The InaccessableDirectories=, ReadOnlyDirectories= and
2113 ReadWriteDirectories= unit file settings have been renamed to
2114 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
2115 applied to all kinds of file nodes, and not just directories, with
2116 the exception of symlinks. Specifically these settings may now be
2117 used on block and character device nodes, UNIX sockets and FIFOS as
2118 well as regular files. The old names of these settings remain
2119 available for compatibility.
2121 * systemd will now log about all service processes it kills forcibly
2122 (using SIGKILL) because they remained after the clean shutdown phase
2123 of the service completed. This should help identifying services that
2124 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
2125 systemd-logind's configuration a similar log message is generated for
2126 processes killed at the end of each session due to this setting.
2128 * systemd will now set the $JOURNAL_STREAM environment variable for all
2129 services whose stdout/stderr are connected to the Journal (which
2130 effectively means by default: all services). The variable contains
2131 the device and inode number of the file descriptor used for
2132 stdout/stderr. This may be used by invoked programs to detect whether
2133 their stdout/stderr is connected to the Journal, in which case they
2134 can switch over to direct Journal communication, thus being able to
2135 pass extended, structured metadata along with their log messages. As
2136 one example, this is now used by glib's logging primitives.
2138 * When using systemd's default tmp.mount unit for /tmp, the mount point
2139 will now be established with the "nosuid" and "nodev" options. This
2140 avoids privilege escalation attacks that put traps and exploits into
2141 /tmp. However, this might cause problems if you e. g. put container
2142 images or overlays into /tmp; if you need this, override tmp.mount's
2143 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
2146 * systemd now supports the "memory" cgroup controller also on
2149 * The systemd-cgtop tool now optionally takes a control group path as
2150 command line argument. If specified, the control group list shown is
2151 limited to subgroups of that group.
2153 * The SystemCallFilter= unit file setting gained support for
2154 pre-defined, named system call filter sets. For example
2155 SystemCallFilter=@clock is now an effective way to make all clock
2156 changing-related system calls unavailable to a service. A number of
2157 similar pre-defined groups are defined. Writing system call filters
2158 for system services is simplified substantially with this new
2159 concept. Accordingly, all of systemd's own, long-running services now
2160 enable system call filtering based on this, by default.
2162 * A new service setting MemoryDenyWriteExecute= has been added, taking
2163 a boolean value. If turned on, a service may no longer create memory
2164 mappings that are writable and executable at the same time. This
2165 enhances security for services where this is enabled as it becomes
2166 harder to dynamically write and then execute memory in exploited
2167 service processes. This option has been enabled for all of systemd's
2168 own long-running services.
2170 * A new RestrictRealtime= service setting has been added, taking a
2171 boolean argument. If set the service's processes may no longer
2172 acquire realtime scheduling. This improves security as realtime
2173 scheduling may otherwise be used to easily freeze the system.
2175 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
2176 value. This may be used for requesting that the system manager inside
2177 of the container reports start-up completion to nspawn which then
2178 propagates this notification further to the service manager
2179 supervising nspawn itself. A related option NotifyReady= in .nspawn
2180 files has been added too. This functionality allows ordering of the
2181 start-up of multiple containers using the usual systemd ordering
2184 * machinectl gained a new command "stop" that is an alias for
2187 * systemd-resolved gained support for contacting DNS servers on
2188 link-local IPv6 addresses.
2190 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
2191 its caches. A method call for requesting the same operation has been
2192 added to the bus API too, and is made available via "systemd-resolve
2195 * systemd-resolve gained a new --status switch. If passed a brief
2196 summary of the used DNS configuration with per-interface information
2199 * resolved.conf gained a new Cache= boolean option, defaulting to
2200 on. If turned off local DNS caching is disabled. This comes with a
2201 performance penalty in particular when DNSSEC is enabled. Note that
2202 resolved disables its internal caching implicitly anyway, when the
2203 configured DNS server is on a host-local IP address such as ::1 or
2204 127.0.0.1, thus automatically avoiding double local caching.
2206 * systemd-resolved now listens on the local IP address 127.0.0.53:53
2207 for DNS requests. This improves compatibility with local programs
2208 that do not use the libc NSS or systemd-resolved's bus APIs for name
2209 resolution. This minimal DNS service is only available to local
2210 programs and does not implement the full DNS protocol, but enough to
2211 cover local DNS clients. A new, static resolv.conf file, listing just
2212 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
2213 now recommended to make /etc/resolv.conf a symlink to this file in
2214 order to route all DNS lookups to systemd-resolved, regardless if
2215 done via NSS, the bus API or raw DNS packets. Note that this local
2216 DNS service is not as fully featured as the libc NSS or
2217 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
2218 used to deliver link-local address information (as this implies
2219 sending a local interface index along), LLMNR/mDNS support via this
2220 interface is severely restricted. It is thus strongly recommended for
2221 all applications to use the libc NSS API or native systemd-resolved
2224 * systemd-networkd's bridge support learned a new setting
2225 VLANFiltering= for controlling VLAN filtering. Moreover a new section
2226 in .network files has been added for configuring VLAN bridging in
2227 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
2229 * systemd-networkd's IPv6 Router Advertisement code now makes use of
2230 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
2231 now be acquired without relying on DHCPv6. Two new options
2232 UseDomains= and UseDNS= have been added to configure this behaviour.
2234 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
2235 renamed IPv6AcceptRA=, without altering its behaviour. The old
2236 setting name remains available for compatibility reasons.
2238 * The systemd-networkd VTI/VTI6 tunneling support gained new options
2239 Key=, InputKey= and OutputKey=.
2241 * systemd-networkd gained support for VRF ("Virtual Routing Function")
2242 interface configuration.
2244 * "systemctl edit" may now be used to create new unit files by
2245 specifying the --force switch.
2247 * sd-event gained a new function sd_event_get_iteration() for
2248 requesting the current iteration counter of the event loop. It starts
2249 at zero and is increased by one with each event loop iteration.
2251 * A new rpm macro %systemd_ordering is provided by the macros.systemd
2252 file. It can be used in lieu of %systemd_requires in packages which
2253 don't use any systemd functionality and are intended to be installed
2254 in minimal containers without systemd present. This macro provides
2255 ordering dependencies to ensure that if the package is installed in
2256 the same rpm transaction as systemd, systemd will be installed before
2257 the scriptlets for the package are executed, allowing unit presets
2260 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
2261 been added to simplify packaging of generators.
2263 * The os-release file gained VERSION_CODENAME field for the
2264 distribution nickname (e.g. VERSION_CODENAME=woody).
2266 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
2267 can be set to disable parsing of metadata and the creation
2268 of persistent symlinks for that device.
2270 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
2271 to make them available to logged-in users has been reverted.
2273 * Much of the common code of the various systemd components is now
2274 built into an internal shared library libsystemd-shared-231.so
2275 (incorporating the systemd version number in the name, to be updated
2276 with future releases) that the components link to. This should
2277 decrease systemd footprint both in memory during runtime and on
2278 disk. Note that the shared library is not for public use, and is
2279 neither API nor ABI stable, but is likely to change with every new
2280 released update. Packagers need to make sure that binaries
2281 linking to libsystemd-shared.so are updated in step with the
2284 * Configuration for "mkosi" is now part of the systemd
2285 repository. mkosi is a tool to easily build legacy-free OS images,
2286 and is available on github: https://github.com/systemd/mkosi. If
2287 "mkosi" is invoked in the build tree a new raw OS image is generated
2288 incorporating the systemd sources currently being worked on and a
2289 clean, fresh distribution installation. The generated OS image may be
2290 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
2291 UEFI PC. This functionality is particularly useful to easily test
2292 local changes made to systemd in a pristine, defined environment. See
2293 doc/HACKING for details.
2295 * configure learned the --with-support-url= option to specify the
2296 distribution's bugtracker.
2298 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
2299 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
2300 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
2301 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
2302 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
2303 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
2304 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
2305 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
2306 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
2307 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
2308 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
2309 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
2310 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
2311 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
2312 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
2313 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
2314 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
2315 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
2316 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
2318 — Berlin, 2016-07-25
2322 * DNSSEC is now turned on by default in systemd-resolved (in
2323 "allow-downgrade" mode), but may be turned off during compile time by
2324 passing "--with-default-dnssec=no" to "configure" (and of course,
2325 during runtime with DNSSEC= in resolved.conf). We recommend
2326 downstreams to leave this on at least during development cycles and
2327 report any issues with the DNSSEC logic upstream. We are very
2328 interested in collecting feedback about the DNSSEC validator and its
2329 limitations in the wild. Note however, that DNSSEC support is
2330 probably nothing downstreams should turn on in stable distros just
2331 yet, as it might create incompatibilities with a few DNS servers and
2332 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
2333 automatically whenever we detect such incompatible setups, but there
2334 might be systems we do not cover yet. Hence: please help us testing
2335 the DNSSEC code, leave this on where you can, report back, but then
2336 again don't consider turning this on in your stable, LTS or
2337 production release just yet. (Note that you have to enable
2338 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
2339 and its DNSSEC mode for host name resolution from local
2342 * systemd-resolve conveniently resolves DANE records with the --tlsa
2343 option and OPENPGPKEY records with the --openpgp option. It also
2344 supports dumping raw DNS record data via the new --raw= switch.
2346 * systemd-logind will now by default terminate user processes that are
2347 part of the user session scope unit (session-XX.scope) when the user
2348 logs out. This behavior is controlled by the KillUserProcesses=
2349 setting in logind.conf, and the previous default of "no" is now
2350 changed to "yes". This means that user sessions will be properly
2351 cleaned up after, but additional steps are necessary to allow
2352 intentionally long-running processes to survive logout.
2354 While the user is logged in at least once, user@.service is running,
2355 and any service that should survive the end of any individual login
2356 session can be started at a user service or scope using systemd-run.
2357 systemd-run(1) man page has been extended with an example which shows
2358 how to run screen in a scope unit underneath user@.service. The same
2359 command works for tmux.
2361 After the user logs out of all sessions, user@.service will be
2362 terminated too, by default, unless the user has "lingering" enabled.
2363 To effectively allow users to run long-term tasks even if they are
2364 logged out, lingering must be enabled for them. See loginctl(1) for
2365 details. The default polkit policy was modified to allow users to
2366 set lingering for themselves without authentication.
2368 Previous defaults can be restored at compile time by the
2369 --without-kill-user-processes option to "configure".
2371 * systemd-logind gained new configuration settings SessionsMax= and
2372 InhibitorsMax=, both with a default of 8192. It will not register new
2373 user sessions or inhibitors above this limit.
2375 * systemd-logind will now reload configuration on SIGHUP.
2377 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
2378 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
2379 enable. Also, support for the "io" cgroup controller in the unified
2380 hierarchy has been added, so that the "memory", "pids" and "io" are
2381 now the controllers that are supported on the unified hierarchy.
2383 WARNING: it is not possible to use previous systemd versions with
2384 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
2385 is necessary to also update systemd in the initramfs if using the
2386 unified hierarchy. An updated SELinux policy is also required.
2388 * LLDP support has been extended, and both passive (receive-only) and
2389 active (sender) modes are supported. Passive mode ("routers-only") is
2390 enabled by default in systemd-networkd. Active LLDP mode is enabled
2391 by default for containers on the internal network. The "networkctl
2392 lldp" command may be used to list information gathered. "networkctl
2393 status" will also show basic LLDP information on connected peers now.
2395 * The IAID and DUID unique identifier sent in DHCP requests may now be
2396 configured for the system and each .network file managed by
2397 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
2399 * systemd-networkd gained support for configuring proxy ARP support for
2400 each interface, via the ProxyArp= setting in .network files. It also
2401 gained support for configuring the multicast querier feature of
2402 bridge devices, via the new MulticastQuerier= setting in .netdev
2403 files. Similarly, snooping on the IGMP traffic can be controlled
2404 via the new setting MulticastSnooping=.
2406 A new setting PreferredLifetime= has been added for addresses
2407 configured in .network file to configure the lifetime intended for an
2410 The systemd-networkd DHCP server gained the option EmitRouter=, which
2411 defaults to yes, to configure whether the DHCP Option 3 (Router)
2414 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
2415 systemd-socket-activate and installed into /usr/bin. It is now fully
2418 * systemd-journald now uses separate threads to flush changes to disk
2419 when closing journal files, thus reducing impact of slow disk I/O on
2420 logging performance.
2422 * The sd-journal API gained two new calls
2423 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
2424 can be used to open journal files using file descriptors instead of
2425 file or directory paths. sd_journal_open_container() has been
2426 deprecated, sd_journal_open_directory_fd() should be used instead
2427 with the flag SD_JOURNAL_OS_ROOT.
2429 * journalctl learned a new output mode "-o short-unix" that outputs log
2430 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
2431 UTC). It also gained support for a new --no-hostname setting to
2432 suppress the hostname column in the family of "short" output modes.
2434 * systemd-ask-password now optionally skips printing of the password to
2435 stdout with --no-output which can be useful in scripts.
2437 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
2438 (devices tagged with ID_MAKER_TOOL) are now tagged with
2439 "uaccess" and are available to logged in users.
2441 * The DeviceAllow= unit setting now supports specifiers (with "%").
2443 * "systemctl show" gained a new --value switch, which allows print a
2444 only the contents of a specific unit property, without also printing
2445 the property's name. Similar support was added to "show*" verbs
2446 of loginctl and machinectl that output "key=value" lists.
2448 * A new unit type "generated" was added for files dynamically generated
2449 by generator tools. Similarly, a new unit type "transient" is used
2450 for unit files created using the runtime API. "systemctl enable" will
2451 refuse to operate on such files.
2453 * A new command "systemctl revert" has been added that may be used to
2454 revert to the vendor version of a unit file, in case local changes
2455 have been made by adding drop-ins or overriding the unit file.
2457 * "machinectl clean" gained a new verb to automatically remove all or
2458 just hidden container images.
2460 * systemd-tmpfiles gained support for a new line type "e" for emptying
2461 directories, if they exist, without creating them if they don't.
2463 * systemd-nspawn gained support for automatically patching the UID/GIDs
2464 of the owners and the ACLs of all files and directories in a
2465 container tree to match the UID/GID user namespacing range selected
2466 for the container invocation. This mode is enabled via the new
2467 --private-users-chown switch. It also gained support for
2468 automatically choosing a free, previously unused UID/GID range when
2469 starting a container, via the new --private-users=pick setting (which
2470 implies --private-users-chown). Together, these options for the first
2471 time make user namespacing for nspawn containers fully automatic and
2472 thus deployable. The systemd-nspawn@.service template unit file has
2473 been changed to use this functionality by default.
2475 * systemd-nspawn gained a new --network-zone= switch, that allows
2476 creating ad-hoc virtual Ethernet links between multiple containers,
2477 that only exist as long as at least one container referencing them is
2478 running. This allows easy connecting of multiple containers with a
2479 common link that implements an Ethernet broadcast domain. Each of
2480 these network "zones" may be named relatively freely by the user, and
2481 may be referenced by any number of containers, but each container may
2482 only reference one of these "zones". On the lower level, this is
2483 implemented by an automatically managed bridge network interface for
2484 each zone, that is created when the first container referencing its
2485 zone is created and removed when the last one referencing its zone
2488 * The default start timeout may now be configured on the kernel command
2489 line via systemd.default_timeout_start_sec=. It was already
2490 configurable via the DefaultTimeoutStartSec= option in
2491 /etc/systemd/system.conf.
2493 * Socket units gained a new TriggerLimitIntervalSec= and
2494 TriggerLimitBurst= setting to configure a limit on the activation
2495 rate of the socket unit.
2497 * The LimitNICE= setting now optionally takes normal UNIX nice values
2498 in addition to the raw integer limit value. If the specified
2499 parameter is prefixed with "+" or "-" and is in the range -20..19 the
2500 value is understood as UNIX nice value. If not prefixed like this it
2501 is understood as raw RLIMIT_NICE limit.
2503 * Note that the effect of the PrivateDevices= unit file setting changed
2504 slightly with this release: the per-device /dev file system will be
2505 mounted read-only from this version on, and will have "noexec"
2506 set. This (minor) change of behavior might cause some (exceptional)
2507 legacy software to break, when PrivateDevices=yes is set for its
2508 service. Please leave PrivateDevices= off if you run into problems
2511 * systemd-bootchart has been split out to a separate repository:
2512 https://github.com/systemd/systemd-bootchart
2514 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
2515 merged into the kernel in its current form.
2517 * The compatibility libraries libsystemd-daemon.so,
2518 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
2519 which have been deprecated since systemd-209 have been removed along
2520 with the corresponding pkg-config files. All symbols provided by
2521 those libraries are provided by libsystemd.so.
2523 * The Capabilities= unit file setting has been removed (it is ignored
2524 for backwards compatibility). AmbientCapabilities= and
2525 CapabilityBoundingSet= should be used instead.
2527 * A new special target has been added, initrd-root-device.target,
2528 which creates a synchronization point for dependencies of the root
2529 device in early userspace. Initramfs builders must ensure that this
2530 target is now included in early userspace.
2532 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
2533 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
2534 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
2535 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
2536 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
2537 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
2538 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
2539 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
2540 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
2541 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
2542 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
2543 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
2544 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
2545 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
2546 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
2547 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
2548 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
2549 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
2550 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
2551 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
2552 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
2553 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
2554 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
2555 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
2556 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
2559 — Fairfax, 2016-05-21
2563 * The systemd-resolved DNS resolver service has gained a substantial
2564 set of new features, most prominently it may now act as a DNSSEC
2565 validating stub resolver. DNSSEC mode is currently turned off by
2566 default, but is expected to be turned on by default in one of the
2567 next releases. For now, we invite everybody to test the DNSSEC logic
2568 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
2569 service also gained a full set of D-Bus interfaces, including calls
2570 to configure DNS and DNSSEC settings per link (for use by external
2571 network management software). systemd-resolved and systemd-networkd
2572 now distinguish between "search" and "routing" domains. The former
2573 are used to qualify single-label names, the latter are used purely
2574 for routing lookups within certain domains to specific links.
2575 resolved now also synthesizes RRs for all entries from /etc/hosts.
2577 * The systemd-resolve tool (which is a client utility for
2578 systemd-resolved) has been improved considerably and is now fully
2579 supported and documented. Hence it has moved from /usr/lib/systemd to
2582 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
2585 * The coredump collection logic has been reworked: when a coredump is
2586 collected it is now written to disk, compressed and processed
2587 (including stacktrace extraction) from a new instantiated service
2588 systemd-coredump@.service, instead of directly from the
2589 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
2590 processing large coredumps can take up a substantial amount of
2591 resources and time, and this previously happened entirely outside of
2592 systemd's service supervision. With the new logic the core_pattern
2593 hook only does minimal metadata collection before passing off control
2594 to the new instantiated service, which is configured with a time
2595 limit, a nice level and other settings to minimize negative impact on
2596 the rest of the system. Also note that the new logic will honour the
2597 RLIMIT_CORE setting of the crashed process, which now allows users
2598 and processes to turn off coredumping for their processes by setting
2601 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
2602 and all forked processes by default. Previously, PID 1 would leave
2603 the setting at "0" for all processes, as set by the kernel. Note that
2604 the resource limit traditionally has no effect on the generated
2605 coredumps on the system if the /proc/sys/kernel/core_pattern hook
2606 logic is used. Since the limit is now honoured (see above) its
2607 default has been changed so that the coredumping logic is enabled by
2608 default for all processes, while allowing specific opt-out.
2610 * When the stacktrace is extracted from processes of system users, this
2611 is now done as "systemd-coredump" user, in order to sandbox this
2612 potentially security sensitive parsing operation. (Note that when
2613 processing coredumps of normal users this is done under the user ID
2614 of process that crashed, as before.) Packagers should take notice
2615 that it is now necessary to create the "systemd-coredump" system user
2616 and group at package installation time.
2618 * The systemd-activate socket activation testing tool gained support
2619 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
2620 and --seqpacket switches. It also has been extended to support both
2621 new-style and inetd-style file descriptor passing. Use the new
2622 --inetd switch to request inetd-style file descriptor passing.
2624 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
2625 variable, which takes a boolean value. If set to false, ANSI color
2626 output is disabled in the tools even when run on a terminal that
2629 * The VXLAN support in networkd now supports two new settings
2630 DestinationPort= and PortRange=.
2632 * A new systemd.machine_id= kernel command line switch has been added,
2633 that may be used to set the machine ID in /etc/machine-id if it is
2634 not initialized yet. This command line option has no effect if the
2635 file is already initialized.
2637 * systemd-nspawn gained a new --as-pid2 switch that invokes any
2638 specified command line as PID 2 rather than PID 1 in the
2639 container. In this mode PID 1 is a minimal stub init process that
2640 implements the special POSIX and Linux semantics of PID 1 regarding
2641 signal and child process management. Note that this stub init process
2642 is implemented in nspawn itself and requires no support from the
2643 container image. This new logic is useful to support running
2644 arbitrary commands in the container, as normal processes are
2645 generally not prepared to run as PID 1.
2647 * systemd-nspawn gained a new --chdir= switch for setting the current
2648 working directory for the process started in the container.
2650 * "journalctl /dev/sda" will now output all kernel log messages for
2651 specified device from the current boot, in addition to all devices
2652 that are parents of it. This should make log output about devices
2653 pretty useful, as long as kernel drivers attach enough metadata to
2654 the log messages. (The usual SATA drivers do.)
2656 * The sd-journal API gained two new calls
2657 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
2658 that report whether log data from /run or /var has been found.
2660 * journalctl gained a new switch "--fields" that prints all journal
2661 record field names currently in use in the journal. This is backed
2662 by two new sd-journal API calls sd_journal_enumerate_fields() and
2663 sd_journal_restart_fields().
2665 * Most configurable timeouts in systemd now expect an argument of
2666 "infinity" to turn them off, instead of "0" as before. The semantics
2667 from now on is that a timeout of "0" means "now", and "infinity"
2668 means "never". To maintain backwards compatibility, "0" continues to
2669 turn off previously existing timeout settings.
2671 * "systemctl reload-or-try-restart" has been renamed to "systemctl
2672 try-reload-or-restart" to clarify what it actually does: the "try"
2673 logic applies to both reloading and restarting, not just restarting.
2674 The old name continues to be accepted for compatibility.
2676 * On boot-up, when PID 1 detects that the system clock is behind the
2677 release date of the systemd version in use, the clock is now set
2678 to the latter. Previously, this was already done in timesyncd, in order
2679 to avoid running with clocks set to the various clock epochs such as
2680 1902, 1938 or 1970. With this change the logic is now done in PID 1
2681 in addition to timesyncd during early boot-up, so that it is enforced
2682 before the first process is spawned by systemd. Note that the logic
2683 in timesyncd remains, as it is more comprehensive and ensures
2684 clock monotonicity by maintaining a persistent timestamp file in
2685 /var. Since /var is generally not available in earliest boot or the
2686 initrd, this part of the logic remains in timesyncd, and is not done
2689 * Support for tweaking details in net_cls.class_id through the
2690 NetClass= configuration directive has been removed, as the kernel
2691 people have decided to deprecate that controller in cgroup v2.
2692 Userspace tools such as nftables are moving over to setting rules
2693 that are specific to the full cgroup path of a task, which obsoletes
2694 these controllers anyway. The NetClass= directive is kept around for
2695 legacy compatibility reasons. For a more in-depth description of the
2696 kernel change, please refer to the respective upstream commit:
2698 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
2700 * A new service setting RuntimeMaxSec= has been added that may be used
2701 to specify a maximum runtime for a service. If the timeout is hit, the
2702 service is terminated and put into a failure state.
2704 * A new service setting AmbientCapabilities= has been added. It allows
2705 configuration of additional Linux process capabilities that are
2706 passed to the activated processes. This is only available on very
2709 * The process resource limit settings in service units may now be used
2710 to configure hard and soft limits individually.
2712 * The various libsystemd APIs such as sd-bus or sd-event now publicly
2713 expose support for gcc's __attribute__((cleanup())) C extension.
2714 Specifically, for many object destructor functions alternative
2715 versions have been added that have names suffixed with "p" and take a
2716 pointer to a pointer to the object to destroy, instead of just a
2717 pointer to the object itself. This is useful because these destructor
2718 functions may be used directly as parameters to the cleanup
2719 construct. Internally, systemd has been a heavy user of this GCC
2720 extension for a long time, and with this change similar support is
2721 now available to consumers of the library outside of systemd. Note
2722 that by using this extension in your sources compatibility with old
2723 and strictly ANSI compatible C compilers is lost. However, all gcc or
2724 LLVM versions of recent years support this extension.
2726 * Timer units gained support for a new setting RandomizedDelaySec= that
2727 allows configuring some additional randomized delay to the configured
2728 time. This is useful to spread out timer events to avoid load peaks in
2729 clusters or larger setups.
2731 * Calendar time specifications now support sub-second accuracy.
2733 * Socket units now support listening on SCTP and UDP-lite protocol
2736 * The sd-event API now comes with a full set of man pages.
2738 * Older versions of systemd contained experimental support for
2739 compressing journal files and coredumps with the LZ4 compressor that
2740 was not compatible with the lz4 binary (due to API limitations of the
2741 lz4 library). This support has been removed; only support for files
2742 compatible with the lz4 binary remains. This LZ4 logic is now
2743 officially supported and no longer considered experimental.
2745 * The dkr image import logic has been removed again from importd. dkr's
2746 micro-services focus doesn't fit into the machine image focus of
2747 importd, and quickly got out of date with the upstream dkr API.
2749 * Creation of the /run/lock/lockdev/ directory was dropped from
2750 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
2751 been available for many years. If you still need this, you need to
2752 create your own tmpfiles.d config file with:
2754 d /run/lock/lockdev 0775 root lock -
2756 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
2757 and RebootArgument= have been moved from the [Service] section of
2758 unit files to [Unit], and they are now supported on all unit types,
2759 not just service units. Of course, elogind will continue to
2760 understand these settings also at the old location, in order to
2761 maintain compatibility.
2763 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
2764 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
2765 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
2766 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
2767 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
2768 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
2769 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
2770 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
2771 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
2772 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
2773 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
2774 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
2775 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
2776 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
2777 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
2778 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
2779 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
2780 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
2781 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2783 — Berlin, 2016-02-11
2787 * A number of properties previously only settable in unit
2788 files are now also available as properties to set when
2789 creating transient units programmatically via the bus, as it
2790 is exposed with systemd-run's --property=
2791 setting. Specifically, these are: SyslogIdentifier=,
2792 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
2793 EnvironmentFile=, ReadWriteDirectories=,
2794 ReadOnlyDirectories=, InaccessibleDirectories=,
2795 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
2797 * When creating transient services via the bus API it is now
2798 possible to pass in a set of file descriptors to use as
2799 STDIN/STDOUT/STDERR for the invoked process.
2801 * Slice units may now be created transiently via the bus APIs,
2802 similar to the way service and scope units may already be
2803 created transiently.
2805 * Wherever systemd expects a calendar timestamp specification
2806 (like in journalctl's --since= and --until= switches) UTC
2807 timestamps are now supported. Timestamps suffixed with "UTC"
2808 are now considered to be in Universal Time Coordinated
2809 instead of the local timezone. Also, timestamps may now
2810 optionally be specified with sub-second accuracy. Both of
2811 these additions also apply to recurring calendar event
2812 specification, such as OnCalendar= in timer units.
2814 * journalctl gained a new "--sync" switch that asks the
2815 journal daemon to write all so far unwritten log messages to
2816 disk and sync the files, before returning.
2818 * systemd-tmpfiles learned two new line types "q" and "Q" that
2819 operate like "v", but also set up a basic btrfs quota
2820 hierarchy when used on a btrfs file system with quota
2823 * tmpfiles' "v", "q" and "Q" will now create a plain directory
2824 instead of a subvolume (even on a btrfs file system) if the
2825 root directory is a plain directory, and not a
2826 subvolume. This should simplify things with certain chroot()
2827 environments which are not aware of the concept of btrfs
2830 * systemd-detect-virt gained a new --chroot switch to detect
2831 whether execution takes place in a chroot() environment.
2833 * CPUAffinity= now takes CPU index ranges in addition to
2836 * The various memory-related resource limit settings (such as
2837 LimitAS=) now understand the usual K, M, G, ... suffixes to
2838 the base of 1024 (IEC). Similar, the time-related resource
2839 limit settings understand the usual min, h, day, ...
2842 * There's a new system.conf setting DefaultTasksMax= to
2843 control the default TasksMax= setting for services and
2844 scopes running on the system. (TasksMax= is the primary
2845 setting that exposes the "pids" cgroup controller on systemd
2846 and was introduced in the previous systemd release.) The
2847 setting now defaults to 512, which means services that are
2848 not explicitly configured otherwise will only be able to
2849 create 512 processes or threads at maximum, from this
2850 version on. Note that this means that thread- or
2851 process-heavy services might need to be reconfigured to set
2852 TasksMax= to a higher value. It is sufficient to set
2853 TasksMax= in these specific unit files to a higher value, or
2854 even "infinity". Similar, there's now a logind.conf setting
2855 UserTasksMax= that defaults to 4096 and limits the total
2856 number of processes or tasks each user may own
2857 concurrently. nspawn containers also have the TasksMax=
2858 value set by default now, to 8192. Note that all of this
2859 only has an effect if the "pids" cgroup controller is
2860 enabled in the kernel. The general benefit of these changes
2861 should be a more robust and safer system, that provides a
2862 certain amount of per-service fork() bomb protection.
2864 * systemd-nspawn gained the new --network-veth-extra= switch
2865 to define additional and arbitrarily-named virtual Ethernet
2866 links between the host and the container.
2868 * A new service execution setting PassEnvironment= has been
2869 added that allows importing select environment variables
2870 from PID1's environment block into the environment block of
2873 * Timer units gained support for a new RemainAfterElapse=
2874 setting which takes a boolean argument. It defaults to on,
2875 exposing behaviour unchanged to previous releases. If set to
2876 off, timer units are unloaded after they elapsed if they
2877 cannot elapse again. This is particularly useful for
2878 transient timer units, which shall not stay around longer
2879 than until they first elapse.
2881 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
2882 default now (the kernel default is 16). This is beneficial
2883 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
2884 allows substantially larger numbers of queued
2885 datagrams. This should increase the capability of systemd to
2886 parallelize boot-up, as logging and sd_notify() are unlikely
2887 to stall execution anymore. If you need to change the value
2888 from the new defaults, use the usual sysctl.d/ snippets.
2890 * The compression framing format used by the journal or
2891 coredump processing has changed to be in line with what the
2892 official LZ4 tools generate. LZ4 compression support in
2893 systemd was considered unsupported previously, as the format
2894 was not compatible with the normal tools. With this release
2895 this has changed now, and it is hence safe for downstream
2896 distributions to turn it on. While not compressing as well
2897 as the XZ, LZ4 is substantially faster, which makes
2898 it a good default choice for the compression logic in the
2899 journal and in coredump handling.
2901 * Any reference to /etc/mtab has been dropped from
2902 systemd. The file has been obsolete since a while, but
2903 systemd refused to work on systems where it was incorrectly
2904 set up (it should be a symlink or non-existent). Please make
2905 sure to update to util-linux 2.27.1 or newer in conjunction
2906 with this systemd release, which also drops any reference to
2907 /etc/mtab. If you maintain a distribution make sure that no
2908 software you package still references it, as this is a
2909 likely source of bugs. There's also a glibc bug pending,
2910 asking for removal of any reference to this obsolete file:
2912 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
2914 Note that only util-linux versions built with
2915 --enable-libmount-force-mountinfo are supported.
2917 * Support for the ".snapshot" unit type has been removed. This
2918 feature turned out to be little useful and little used, and
2919 has now been removed from the core and from systemctl.
2921 * The dependency types RequiresOverridable= and
2922 RequisiteOverridable= have been removed from systemd. They
2923 have been used only very sparingly to our knowledge and
2924 other options that provide a similar effect (such as
2925 systemctl --mode=ignore-dependencies) are much more useful
2926 and commonly used. Moreover, they were only half-way
2927 implemented as the option to control behaviour regarding
2928 these dependencies was never added to systemctl. By removing
2929 these dependency types the execution engine becomes a bit
2930 simpler. Unit files that use these dependencies should be
2931 changed to use the non-Overridable dependency types
2932 instead. In fact, when parsing unit files with these
2933 options, that's what systemd will automatically convert them
2934 too, but it will also warn, asking users to fix the unit
2935 files accordingly. Removal of these dependency types should
2936 only affect a negligible number of unit files in the wild.
2938 * Behaviour of networkd's IPForward= option changed
2939 (again). It will no longer maintain a per-interface setting,
2940 but propagate one way from interfaces where this is enabled
2941 to the global kernel setting. The global setting will be
2942 enabled when requested by a network that is set up, but
2943 never be disabled again. This change was made to make sure
2944 IPv4 and IPv6 behaviour regarding packet forwarding is
2945 similar (as the Linux IPv6 stack does not support
2946 per-interface control of this setting) and to minimize
2949 * In unit files the behaviour of %u, %U, %h, %s has
2950 changed. These specifiers will now unconditionally resolve
2951 to the various user database fields of the user that the
2952 systemd instance is running as, instead of the user
2953 configured in the specific unit via User=. Note that this
2954 effectively doesn't change much, as resolving of these
2955 specifiers was already turned off in the --system instance
2956 of systemd, as we cannot do NSS lookups from PID 1. In the
2957 --user instance of systemd these specifiers where correctly
2958 resolved, but hardly made any sense, since the user instance
2959 lacks privileges to do user switches anyway, and User= is
2960 hence useless. Moreover, even in the --user instance of
2961 systemd behaviour was awkward as it would only take settings
2962 from User= assignment placed before the specifier into
2963 account. In order to unify and simplify the logic around
2964 this the specifiers will now always resolve to the
2965 credentials of the user invoking the manager (which in case
2966 of PID 1 is the root user).
2968 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
2969 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
2970 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
2971 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
2972 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
2973 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
2974 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
2975 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
2976 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
2977 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
2978 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
2979 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
2980 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
2981 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
2984 — Berlin, 2015-11-18
2988 * systemd now depends on util-linux v2.27. More specifically,
2989 the newly added mount monitor feature in libmount now
2990 replaces systemd's former own implementation.
2992 * libmount mandates /etc/mtab not to be regular file, and
2993 systemd now enforces this condition at early boot.
2994 /etc/mtab has been deprecated and warned about for a very
2995 long time, so systems running systemd should already have
2996 stopped having this file around as anything else than a
2997 symlink to /proc/self/mounts.
2999 * Support for the "pids" cgroup controller has been added. It
3000 allows accounting the number of tasks in a cgroup and
3001 enforcing limits on it. This adds two new setting
3002 TasksAccounting= and TasksMax= to each unit, as well as a
3003 global option DefaultTasksAccounting=.
3005 * Support for the "net_cls" cgroup controller has been added.
3006 It allows assigning a net class ID to each task in the
3007 cgroup, which can then be used in firewall rules and traffic
3008 shaping configurations. Note that the kernel netfilter net
3009 class code does not currently work reliably for ingress
3010 packets on unestablished sockets.
3012 This adds a new config directive called NetClass= to CGroup
3013 enabled units. Allowed values are positive numbers for fixed
3014 assignments and "auto" for picking a free value
3017 * 'systemctl is-system-running' now returns 'offline' if the
3018 system is not booted with systemd. This command can now be
3019 used as a substitute for 'systemd-notify --booted'.
3021 * Watchdog timeouts have been increased to 3 minutes for all
3022 in-tree service files. Apparently, disk IO issues are more
3023 frequent than we hoped, and user reported >1 minute waiting
3026 * 'machine-id-commit' functionality has been merged into
3027 'machine-id-setup --commit'. The separate binary has been
3030 * The WorkingDirectory= directive in unit files may now be set
3031 to the special value '~'. In this case, the working
3032 directory is set to the home directory of the user
3033 configured in User=.
3035 * "machinectl shell" will now open the shell in the home
3036 directory of the selected user by default.
3038 * The CrashChVT= configuration file setting is renamed to
3039 CrashChangeVT=, following our usual logic of not
3040 abbreviating unnecessarily. The old directive is still
3041 supported for compat reasons. Also, this directive now takes
3042 an integer value between 1 and 63, or a boolean value. The
3043 formerly supported '-1' value for disabling stays around for
3046 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
3047 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
3048 RootDirectory= properties can now be set for transient
3051 * The systemd-analyze tool gained a new "set-log-target" verb
3052 to change the logging target the system manager logs to
3053 dynamically during runtime. This is similar to how
3054 "systemd-analyze set-log-level" already changes the log
3057 * In nspawn /sys is now mounted as tmpfs, with only a selected
3058 set of subdirectories mounted in from the real sysfs. This
3059 enhances security slightly, and is useful for ensuring user
3060 namespaces work correctly.
3062 * Support for USB FunctionFS activation has been added. This
3063 allows implementation of USB gadget services that are
3064 activated as soon as they are requested, so that they don't
3065 have to run continuously, similar to classic socket
3068 * The "systemctl exit" command now optionally takes an
3069 additional parameter that sets the exit code to return from
3070 the systemd manager when exiting. This is only relevant when
3071 running the systemd user instance, or when running the
3072 system instance in a container.
3074 * sd-bus gained the new API calls sd_bus_path_encode_many()
3075 and sd_bus_path_decode_many() that allow easy encoding and
3076 decoding of multiple identifier strings inside a D-Bus
3077 object path. Another new call sd_bus_default_flush_close()
3078 has been added to flush and close per-thread default
3081 * systemd-cgtop gained support for a -M/--machine= switch to
3082 show the control groups within a certain container only.
3084 * "systemctl kill" gained support for an optional --fail
3085 switch. If specified the requested operation will fail of no
3086 processes have been killed, because the unit had no
3087 processes attached, or similar.
3089 * A new systemd.crash_reboot=1 kernel command line option has
3090 been added that triggers a reboot after crashing. This can
3091 also be set through CrashReboot= in systemd.conf.
3093 * The RuntimeDirectory= setting now understands unit
3094 specifiers like %i or %f.
3096 * A new (still internal) library API sd-ipv4acd has been added,
3097 that implements address conflict detection for IPv4. It's
3098 based on code from sd-ipv4ll, and will be useful for
3099 detecting DHCP address conflicts.
3101 * File descriptors passed during socket activation may now be
3102 named. A new API sd_listen_fds_with_names() is added to
3103 access the names. The default names may be overridden,
3104 either in the .socket file using the FileDescriptorName=
3105 parameter, or by passing FDNAME= when storing the file
3106 descriptors using sd_notify().
3108 * systemd-networkd gained support for:
3110 - Setting the IPv6 Router Advertisement settings via
3111 IPv6AcceptRouterAdvertisements= in .network files.
3113 - Configuring the HelloTimeSec=, MaxAgeSec= and
3114 ForwardDelaySec= bridge parameters in .netdev files.
3116 - Configuring PreferredSource= for static routes in
3119 * The "ask-password" framework used to query for LUKS harddisk
3120 passwords or SSL passwords during boot gained support for
3121 caching passwords in the kernel keyring, if it is
3122 available. This makes sure that the user only has to type in
3123 a passphrase once if there are multiple objects to unlock
3124 with the same one. Previously, such password caching was
3125 available only when Plymouth was used; this moves the
3126 caching logic into the systemd codebase itself. The
3127 "systemd-ask-password" utility gained a new --keyname=
3128 switch to control which kernel keyring key to use for
3129 caching a password in. This functionality is also useful for
3130 enabling display managers such as gdm to automatically
3131 unlock the user's GNOME keyring if its passphrase, the
3132 user's password and the harddisk password are the same, if
3133 gdm-autologin is used.
3135 * When downloading tar or raw images using "machinectl
3136 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
3137 file is now also downloaded, if it is available and stored
3138 next to the image file.
3140 * Units of type ".socket" gained a new boolean setting
3141 Writable= which is only useful in conjunction with
3142 ListenSpecial=. If true, enables opening the specified
3143 special file in O_RDWR mode rather than O_RDONLY mode.
3145 * systemd-rfkill has been reworked to become a singleton
3146 service that is activated through /dev/rfkill on each rfkill
3147 state change and saves the settings to disk. This way,
3148 systemd-rfkill is now compatible with devices that exist
3149 only intermittendly, and even restores state if the previous
3150 system shutdown was abrupt rather than clean.
3152 * The journal daemon gained support for vacuuming old journal
3153 files controlled by the number of files that shall remain,
3154 in addition to the already existing control by size and by
3155 date. This is useful as journal interleaving performance
3156 degrades with too many separate journal files, and allows
3157 putting an effective limit on them. The new setting defaults
3158 to 100, but this may be changed by setting SystemMaxFiles=
3159 and RuntimeMaxFiles= in journald.conf. Also, the
3160 "journalctl" tool gained the new --vacuum-files= switch to
3161 manually vacuum journal files to leave only the specified
3162 number of files in place.
3164 * udev will now create /dev/disk/by-path links for ATA devices
3165 on kernels where that is supported.
3167 * Galician, Serbian, Turkish and Korean translations were added.
3169 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
3170 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
3171 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
3172 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
3173 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
3174 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
3175 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
3176 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
3177 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
3178 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
3179 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
3180 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
3181 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
3182 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
3183 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
3184 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
3185 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
3186 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
3188 — Berlin, 2015-10-07
3192 * The DHCP implementation of systemd-networkd gained a set of
3195 - The DHCP server now supports emitting DNS and NTP
3196 information. It may be enabled and configured via
3197 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
3198 and NTP information is enabled, but no servers are
3199 configured, the corresponding uplink information (if there
3200 is any) is propagated.
3202 - Server and client now support transmission and reception
3203 of timezone information. It can be configured via the
3204 newly introduced network options UseTimezone=,
3205 EmitTimezone=, and Timezone=. Transmission of timezone
3206 information is enabled between host and containers by
3207 default now: the container will change its local timezone
3208 to what the host has set.
3210 - Lease timeouts can now be configured via
3211 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
3213 - The DHCP server improved on the stability of
3214 leases. Clients are more likely to get the same lease
3215 information back, even if the server loses state.
3217 - The DHCP server supports two new configuration options to
3218 control the lease address pool metrics, PoolOffset= and
3221 * The encapsulation limit of tunnels in systemd-networkd may
3222 now be configured via 'EncapsulationLimit='. It allows
3223 modifying the maximum additional levels of encapsulation
3224 that are permitted to be prepended to a packet.
3226 * systemd now supports the concept of user buses replacing
3227 session buses, if used with dbus-1.10 (and enabled via dbus
3228 --enable-user-session). It previously only supported this on
3229 kdbus-enabled systems, and this release expands this to
3230 'dbus-daemon' systems.
3232 * systemd-networkd now supports predictable interface names
3235 * systemd now optionally supports the new Linux kernel
3236 "unified" control group hierarchy. If enabled via the kernel
3237 command-line option 'systemd.unified_cgroup_hierarchy=1',
3238 systemd will try to mount the unified cgroup hierarchy
3239 directly on /sys/fs/cgroup. If not enabled, or not
3240 available, systemd will fall back to the legacy cgroup
3241 hierarchy setup, as before. Host system and containers can
3242 mix and match legacy and unified hierarchies as they
3243 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
3244 environment variable to individually select the hierarchy to
3245 use for executed containers. By default, nspawn will use the
3246 unified hierarchy for the containers if the host uses the
3247 unified hierarchy, and the legacy hierarchy otherwise.
3248 Please note that at this point the unified hierarchy is an
3249 experimental kernel feature and is likely to change in one
3250 of the next kernel releases. Therefore, it should not be
3251 enabled by default in downstream distributions yet. The
3252 minimum required kernel version for the unified hierarchy to
3253 work is 4.2. Note that when the unified hierarchy is used
3254 for the first time delegated access to controllers is
3255 safe. Because of this systemd-nspawn containers will get
3256 access to controllers now, as will systemd user
3257 sessions. This means containers and user sessions may now
3258 manage their own resources, partitioning up what the system
3261 * A new special scope unit "init.scope" has been introduced
3262 that encapsulates PID 1 of the system. It may be used to
3263 determine resource usage and enforce resource limits on PID
3264 1 itself. PID 1 hence moved out of the root of the control
3267 * The cgtop tool gained support for filtering out kernel
3268 threads when counting tasks in a control group. Also, the
3269 count of processes is now recursively summed up by
3270 default. Two options -k and --recursive= have been added to
3271 revert to old behaviour. The tool has also been updated to
3272 work correctly in containers now.
3274 * systemd-nspawn's --bind= and --bind-ro= options have been
3275 extended to allow creation of non-recursive bind mounts.
3277 * libsystemd gained two new calls sd_pid_get_cgroup() and
3278 sd_peer_get_cgroup() which return the control group path of
3279 a process or peer of a connected AF_UNIX socket. This
3280 function call is particularly useful when implementing
3281 delegated subtrees support in the control group hierarchy.
3283 * The "sd-event" event loop API of libsystemd now supports
3284 correct dequeuing of real-time signals, without losing
3287 * When systemd requests a PolicyKit decision when managing
3288 units it will now add additional fields to the request,
3289 including unit name and desired operation. This enables more
3290 powerful PolicyKit policies, that make decisions depending
3291 on these parameters.
3293 * nspawn learnt support for .nspawn settings files, that may
3294 accompany the image files or directories of containers, and
3295 may contain additional settings for the container. This is
3296 an alternative to configuring container parameters via the
3297 nspawn command line.
3299 Contributions from: Cristian Rodríguez, Daniel Mack, David
3300 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
3301 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
3302 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
3303 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
3304 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
3305 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
3306 Andersen, Tom Gundersen, Torstein Husebø
3308 — Berlin, 2015-09-08
3312 * machinectl gained a new verb 'shell' which opens a fresh
3313 shell on the target container or the host. It is similar to
3314 the existing 'login' command of machinectl, but spawns the
3315 shell directly without prompting for username or
3316 password. The pseudo machine '.host' now refers to the local
3317 host and is used by default. Hence, 'machinectl shell' can
3318 be used as replacement for 'su -' which spawns a session as
3319 a fresh systemd unit in a way that is fully isolated from
3320 the originating session.
3322 * systemd-networkd learned to cope with private-zone DHCP
3323 options and allows other programs to query the values.
3325 * SELinux access control when enabling/disabling units is no
3326 longer enforced with this release. The previous
3327 implementation was incorrect, and a new corrected
3328 implementation is not yet available. As unit file operations
3329 are still protected via PolicyKit and D-Bus policy this is
3330 not a security problem. Yet, distributions which care about
3331 optimal SELinux support should probably not stabilize on
3334 * sd-bus gained support for matches of type "arg0has=", that
3335 test for membership of strings in string arrays sent in bus
3338 * systemd-resolved now dumps the contents of its DNS and LLMNR
3339 caches to the logs on reception of the SIGUSR1 signal. This
3340 is useful to debug DNS behaviour.
3342 * The coredumpctl tool gained a new --directory= option to
3343 operate on journal files in a specific directory.
3345 * "systemctl reboot" and related commands gained a new
3346 "--message=" option which may be used to set a free-text
3347 wall message when shutting down or rebooting the
3348 system. This message is also logged, which is useful for
3349 figuring out the reason for a reboot or shutdown a
3352 * The "systemd-resolve-host" tool's -i switch now takes
3353 network interface numbers as alternative to interface names.
3355 * A new unit file setting for services has been introduced:
3356 UtmpMode= allows configuration of how precisely systemd
3357 handles utmp and wtmp entries for the service if this is
3358 enabled. This allows writing services that appear similar to
3359 user sessions in the output of the "w", "who", "last" and
3362 * systemd-resolved will now locally synthesize DNS resource
3363 records for the "localhost" and "gateway" domains as well as
3364 the local hostname. This should ensure that clients querying
3365 RRs via resolved will get similar results as those going via
3366 NSS, if nss-myhostname is enabled.
3368 Contributions from: Alastair Hughes, Alex Crawford, Daniel
3369 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
3370 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
3371 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
3372 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
3373 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
3374 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
3375 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
3376 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
3377 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
3378 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
3379 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3381 — Berlin, 2015-08-27
3385 * The systemd-efi-boot-generator functionality was merged into
3386 systemd-gpt-auto-generator.
3388 * systemd-networkd now supports Group Policy for vxlan
3389 devices. It can be enabled via the new boolean configuration
3390 option called 'GroupPolicyExtension='.
3392 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
3393 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
3394 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
3396 — Berlin, 2015-07-31
3400 * The python-systemd code has been removed from the systemd repository.
3401 A new repository has been created which accommodates the code from
3402 now on, and we kindly ask distributions to create a separate package
3403 for this: https://github.com/systemd/python-systemd
3405 * The systemd daemon will now reload its main configuration
3406 (/etc/systemd/system.conf) on daemon-reload.
3408 * sd-dhcp now exposes vendor specific extensions via
3409 sd_dhcp_lease_get_vendor_specific().
3411 * systemd-networkd gained a number of new configuration options.
3413 - A new boolean configuration option for TAP devices called
3414 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
3415 device, thus allowing to send and receive GSO packets.
3417 - A new tunnel configuration option called 'CopyDSCP='.
3418 If enabled, the DSCP field of ip6 tunnels is copied into the
3419 decapsulated packet.
3421 - A set of boolean bridge configuration options were added.
3422 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
3423 and 'UnicastFlood=' are now parsed by networkd and applied to the
3424 respective bridge link device via the respective IFLA_BRPORT_*
3427 - A new string configuration option to override the hostname sent
3428 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
3429 is true, networkd will use the configured hostname instead of the
3430 system hostname when sending DHCP requests.
3432 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
3433 networkd will configure the IPv6 flow-label of the tunnel device
3434 according to RFC2460.
3436 - The 'macvtap' virtual network devices are now supported, similar to
3437 the already supported 'macvlan' devices.
3439 * systemd-resolved now implements RFC5452 to improve resilience against
3440 cache poisoning. Additionally, source port randomization is enabled
3441 by default to further protect against DNS spoofing attacks.
3443 * nss-mymachines now supports translating UIDs and GIDs of running
3444 containers with user-namespaces enabled. If a container 'foo'
3445 translates a host uid 'UID' to the container uid 'TUID', then
3446 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
3447 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
3448 mapped as 'vg-foo-TGID'.
3450 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
3451 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
3452 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
3453 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
3454 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
3455 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
3456 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
3457 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
3458 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
3459 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
3461 — Berlin, 2015-07-29
3465 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
3466 There are no known issues with current sysfs, and udev does not need
3467 or should be used to work around such bugs.
3469 * udev does no longer enable USB HID power management. Several reports
3470 indicate, that some devices cannot handle that setting.
3472 * The udev accelerometer helper was removed. The functionality
3473 is now fully included in iio-sensor-proxy. But this means,
3474 older iio-sensor-proxy versions will no longer provide
3475 accelerometer/orientation data with this systemd version.
3476 Please upgrade iio-sensor-proxy to version 1.0.
3478 * networkd gained a new configuration option IPv6PrivacyExtensions=
3479 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
3480 for Stateless Address") on selected networks.
3482 * For the sake of fewer build-time dependencies and less code in the
3483 main repository, the python bindings are about to be removed in the
3484 next release. A new repository has been created which accommodates
3485 the code from now on, and we kindly ask distributions to create a
3486 separate package for this. The removal will take place in v223.
3488 https://github.com/systemd/python-systemd
3490 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
3491 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
3492 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
3493 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
3494 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
3495 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
3496 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
3497 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
3498 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3499 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3501 — Berlin, 2015-07-07
3505 * The sd-bus.h and sd-event.h APIs have now been declared
3506 stable and have been added to the official interface of
3507 libsystemd.so. sd-bus implements an alternative D-Bus client
3508 library, that is relatively easy to use, very efficient and
3509 supports both classic D-Bus as well as kdbus as transport
3510 backend. sd-event is a generic event loop abstraction that
3511 is built around Linux epoll, but adds features such as event
3512 prioritization or efficient timer handling. Both APIs are good
3513 choices for C programs looking for a bus and/or event loop
3514 implementation that is minimal and does not have to be
3515 portable to other kernels.
3517 * kdbus support is no longer compile-time optional. It is now
3518 always built-in. However, it can still be disabled at
3519 runtime using the kdbus=0 kernel command line setting, and
3520 that setting may be changed to default to off, by specifying
3521 --disable-kdbus at build-time. Note though that the kernel
3522 command line setting has no effect if the kdbus.ko kernel
3523 module is not installed, in which case kdbus is (obviously)
3524 also disabled. We encourage all downstream distributions to
3525 begin testing kdbus by adding it to the kernel images in the
3526 development distributions, and leaving kdbus support in
3529 * The minimal required util-linux version has been bumped to
3532 * Support for chkconfig (--enable-chkconfig) was removed in
3533 favor of calling an abstraction tool
3534 /lib/systemd/systemd-sysv-install. This needs to be
3535 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
3536 in README for details.
3538 * If there's a systemd unit and a SysV init script for the
3539 same service name, and the user executes "systemctl enable"
3540 for it (or a related call), then this will now enable both
3541 (or execute the related operation on both), not just the
3544 * The libudev API documentation has been converted from gtkdoc
3547 * gudev has been removed from the systemd tree, it is now an
3550 * The systemd-cgtop tool learnt a new --raw switch to generate
3551 "raw" (machine parsable) output.
3553 * networkd's IPForwarding= .network file setting learnt the
3554 new setting "kernel", which ensures that networkd does not
3555 change the IP forwarding sysctl from the default kernel
3558 * The systemd-logind bus API now exposes a new boolean
3559 property "Docked" that reports whether logind considers the
3560 system "docked", i.e. connected to a docking station or not.
3562 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
3563 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
3564 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
3565 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
3566 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
3567 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
3568 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
3569 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
3570 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
3571 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
3572 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
3573 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
3574 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
3575 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3576 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
3577 Fink, Zbigniew Jędrzejewski-Szmek
3579 — Berlin, 2015-06-19
3583 * The gudev library has been extracted into a separate repository
3584 available at: https://git.gnome.org/browse/libgudev/
3585 It is now managed as part of the Gnome project. Distributions
3586 are recommended to pass --disable-gudev to systemd and use
3587 gudev from the Gnome project instead. gudev is still included
3588 in systemd, for now. It will be removed soon, though. Please
3589 also see the announcement-thread on systemd-devel:
3590 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
3592 * systemd now exposes a CPUUsageNSec= property for each
3593 service unit on the bus, that contains the overall consumed
3594 CPU time of a service (the sum of what each process of the
3595 service consumed). This value is only available if
3596 CPUAccounting= is turned on for a service, and is then shown
3597 in the "systemctl status" output.
3599 * Support for configuring alternative mappings of the old SysV
3600 runlevels to systemd targets has been removed. They are now
3601 hardcoded in a way that runlevels 2, 3, 4 all map to
3602 multi-user.target and 5 to graphical.target (which
3603 previously was already the default behaviour).
3605 * The auto-mounter logic gained support for mount point
3606 expiry, using a new TimeoutIdleSec= setting in .automount
3607 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
3609 * The EFI System Partition (ESP) as mounted to /boot by
3610 systemd-efi-boot-generator will now be unmounted
3611 automatically after 2 minutes of not being used. This should
3612 minimize the risk of ESP corruptions.
3614 * New /etc/fstab options x-systemd.requires= and
3615 x-systemd.requires-mounts-for= are now supported to express
3616 additional dependencies for mounts. This is useful for
3617 journalling file systems that support external journal
3618 devices or overlay file systems that require underlying file
3619 systems to be mounted.
3621 * systemd does not support direct live-upgrades (via systemctl
3622 daemon-reexec) from versions older than v44 anymore. As no
3623 distribution we are aware of shipped such old versions in a
3624 stable release this should not be problematic.
3626 * When systemd forks off a new per-connection service instance
3627 it will now set the $REMOTE_ADDR environment variable to the
3628 remote IP address, and $REMOTE_PORT environment variable to
3629 the remote IP port. This behaviour is similar to the
3630 corresponding environment variables defined by CGI.
3632 * systemd-networkd gained support for uplink failure
3633 detection. The BindCarrier= option allows binding interface
3634 configuration dynamically to the link sense of other
3635 interfaces. This is useful to achieve behaviour like in
3638 * systemd-networkd gained support for configuring the DHCP
3639 client identifier to use when requesting leases.
3641 * systemd-networkd now has a per-network UseNTP= option to
3642 configure whether NTP server information acquired via DHCP
3643 is passed on to services like systemd-timesyncd.
3645 * systemd-networkd gained support for vti6 tunnels.
3647 * Note that systemd-networkd manages the sysctl variable
3648 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
3649 it is configured for since v219. The variable controls IP
3650 forwarding, and is a per-interface alternative to the global
3651 /proc/sys/net/ipv[46]/ip_forward. This setting is
3652 configurable in the IPForward= option, which defaults to
3653 "no". This means if networkd is used for an interface it is
3654 no longer sufficient to set the global sysctl option to turn
3655 on IP forwarding! Instead, the .network file option
3656 IPForward= needs to be turned on! Note that the
3657 implementation of this behaviour was broken in v219 and has
3660 * Many bonding and vxlan options are now configurable in
3663 * systemd-nspawn gained a new --property= setting to set unit
3664 properties for the container scope. This is useful for
3665 setting resource parameters (e.g. "CPUShares=500") on
3666 containers started from the command line.
3668 * systemd-nspawn gained a new --private-users= switch to make
3669 use of user namespacing available on recent Linux kernels.
3671 * systemd-nspawn may now be called as part of a shell pipeline
3672 in which case the pipes used for stdin and stdout are passed
3673 directly to the process invoked in the container, without
3674 indirection via a pseudo tty.
3676 * systemd-nspawn gained a new switch to control the UNIX
3677 signal to use when killing the init process of the container
3680 * systemd-nspawn gained a new --overlay= switch for mounting
3681 overlay file systems into the container using the new kernel
3684 * When a container image is imported via systemd-importd and
3685 the host file system is not btrfs, a loopback block device
3686 file is created in /var/lib/machines.raw with a btrfs file
3687 system inside. It is then mounted to /var/lib/machines to
3688 enable btrfs features for container management. The loopback
3689 file and btrfs file system is grown as needed when container
3690 images are imported via systemd-importd.
3692 * systemd-machined/systemd-importd gained support for btrfs
3693 quota, to enforce container disk space limits on disk. This
3694 is exposed in "machinectl set-limit".
3696 * systemd-importd now can import containers from local .tar,
3697 .raw and .qcow2 images, and export them to .tar and .raw. It
3698 can also import dkr v2 images now from the network (on top
3701 * systemd-importd gained support for verifying downloaded
3702 images with gpg2 (previously only gpg1 was supported).
3704 * systemd-machined, systemd-logind, systemd: most bus calls
3705 are now accessible to unprivileged processes via
3706 PolicyKit. Also, systemd-logind will now allow users to kill
3707 their own sessions without further privileges or
3710 * systemd-shutdownd has been removed. This service was
3711 previously responsible for implementing scheduled shutdowns
3712 as exposed in /usr/bin/shutdown's time parameter. This
3713 functionality has now been moved into systemd-logind and is
3714 accessible via a bus interface.
3716 * "systemctl reboot" gained a new switch --firmware-setup that
3717 can be used to reboot into the EFI firmware setup, if that
3718 is available. systemd-logind now exposes an API on the bus
3719 to trigger such reboots, in case graphical desktop UIs want
3720 to cover this functionality.
3722 * "systemctl enable", "systemctl disable" and "systemctl mask"
3723 now support a new "--now" switch. If specified the units
3724 that are enabled will also be started, and the ones
3725 disabled/masked also stopped.
3727 * The Gummiboot EFI boot loader tool has been merged into
3728 systemd, and renamed to "systemd-boot". The bootctl tool has been
3729 updated to support systemd-boot.
3731 * An EFI kernel stub has been added that may be used to create
3732 kernel EFI binaries that contain not only the actual kernel,
3733 but also an initrd, boot splash, command line and OS release
3734 information. This combined binary can then be signed as a
3735 single image, so that the firmware can verify it all in one
3736 step. systemd-boot has special support for EFI binaries created
3737 like this and can extract OS release information from them
3738 and show them in the boot menu. This functionality is useful
3739 to implement cryptographically verified boot schemes.
3741 * Optional support has been added to systemd-fsck to pass
3742 fsck's progress report to an AF_UNIX socket in the file
3745 * udev will no longer create device symlinks for all block
3746 devices by default. A blacklist for excluding special block
3747 devices from this logic has been turned into a whitelist
3748 that requires picking block devices explicitly that require
3751 * A new (currently still internal) API sd-device.h has been
3752 added to libsystemd. This modernized API is supposed to
3753 replace libudev eventually. In fact, already much of libudev
3754 is now just a wrapper around sd-device.h.
3756 * A new hwdb database for storing metadata about pointing
3757 stick devices has been added.
3759 * systemd-tmpfiles gained support for setting file attributes
3760 similar to the "chattr" tool with new 'h' and 'H' lines.
3762 * systemd-journald will no longer unconditionally set the
3763 btrfs NOCOW flag on new journal files. This is instead done
3764 with tmpfiles snippet using the new 'h' line type. This
3765 allows easy disabling of this logic, by masking the
3766 journal-nocow.conf tmpfiles file.
3768 * systemd-journald will now translate audit message types to
3769 human readable identifiers when writing them to the
3770 journal. This should improve readability of audit messages.
3772 * The LUKS logic gained support for the offset= and skip=
3773 options in /etc/crypttab, as previously implemented by
3776 * /usr/lib/os-release gained a new optional field VARIANT= for
3777 distributions that support multiple variants (such as a
3778 desktop edition, a server edition, ...)
3780 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
3781 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
3782 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
3783 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
3784 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
3785 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
3786 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
3787 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
3788 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
3789 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
3790 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
3791 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
3792 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
3793 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
3794 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
3795 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
3796 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
3797 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
3798 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
3799 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
3800 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
3801 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
3802 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
3803 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
3804 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
3805 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
3806 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
3808 — Berlin, 2015-05-22
3812 * Introduce a new API "sd-hwdb.h" for querying the hardware
3813 metadata database. With this minimal interface one can query
3814 and enumerate the udev hwdb, decoupled from the old libudev
3815 library. libudev's interface for this is now only a wrapper
3816 around sd-hwdb. A new tool systemd-hwdb has been added to
3817 interface with and update the database.
3819 * When any of systemd's tools copies files (for example due to
3820 tmpfiles' C lines) a btrfs reflink will attempted first,
3821 before bytewise copying is done.
3823 * systemd-nspawn gained a new --ephemeral switch. When
3824 specified a btrfs snapshot is taken of the container's root
3825 directory, and immediately removed when the container
3826 terminates again. Thus, a container can be started whose
3827 changes never alter the container's root directory, and are
3828 lost on container termination. This switch can also be used
3829 for starting a container off the root file system of the
3830 host without affecting the host OS. This switch is only
3831 available on btrfs file systems.
3833 * systemd-nspawn gained a new --template= switch. It takes the
3834 path to a container tree to use as template for the tree
3835 specified via --directory=, should that directory be
3836 missing. This allows instantiating containers dynamically,
3837 on first run. This switch is only available on btrfs file
3840 * When a .mount unit refers to a mount point on which multiple
3841 mounts are stacked, and the .mount unit is stopped all of
3842 the stacked mount points will now be unmounted until no
3843 mount point remains.
3845 * systemd now has an explicit notion of supported and
3846 unsupported unit types. Jobs enqueued for unsupported unit
3847 types will now fail with an "unsupported" error code. More
3848 specifically .swap, .automount and .device units are not
3849 supported in containers, .busname units are not supported on
3850 non-kdbus systems. .swap and .automount are also not
3851 supported if their respective kernel compile time options
3854 * machinectl gained support for two new "copy-from" and
3855 "copy-to" commands for copying files from a running
3856 container to the host or vice versa.
3858 * machinectl gained support for a new "bind" command to bind
3859 mount host directories into local containers. This is
3860 currently only supported for nspawn containers.
3862 * networkd gained support for configuring bridge forwarding
3863 database entries (fdb) from .network files.
3865 * A new tiny daemon "systemd-importd" has been added that can
3866 download container images in tar, raw, qcow2 or dkr formats,
3867 and make them available locally in /var/lib/machines, so
3868 that they can run as nspawn containers. The daemon can GPG
3869 verify the downloads (not supported for dkr, since it has no
3870 provisions for verifying downloads). It will transparently
3871 decompress bz2, xz, gzip compressed downloads if necessary,
3872 and restore sparse files on disk. The daemon uses privilege
3873 separation to ensure the actual download logic runs with
3874 fewer privileges than the daemon itself. machinectl has
3875 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
3876 make the functionality of importd available to the
3877 user. With this in place the Fedora and Ubuntu "Cloud"
3878 images can be downloaded and booted as containers unmodified
3879 (the Fedora images lack the appropriate GPG signature files
3880 currently, so they cannot be verified, but this will change
3881 soon, hopefully). Note that downloading images is currently
3882 only fully supported on btrfs.
3884 * machinectl is now able to list container images found in
3885 /var/lib/machines, along with some metadata about sizes of
3886 disk and similar. If the directory is located on btrfs and
3887 quota is enabled, this includes quota display. A new command
3888 "image-status" has been added that shows additional
3889 information about images.
3891 * machinectl is now able to clone container images
3892 efficiently, if the underlying file system (btrfs) supports
3893 it, with the new "machinectl clone" command. It also
3894 gained commands for renaming and removing images, as well as
3895 marking them read-only or read-write (supported also on
3896 legacy file systems).
3898 * networkd gained support for collecting LLDP network
3899 announcements, from hardware that supports this. This is
3900 shown in networkctl output.
3902 * systemd-run gained support for a new -t (--pty) switch for
3903 invoking a binary on a pty whose input and output is
3904 connected to the invoking terminal. This allows executing
3905 processes as system services while interactively
3906 communicating with them via the terminal. Most interestingly
3907 this is supported across container boundaries. Invoking
3908 "systemd-run -t /bin/bash" is an alternative to running a
3909 full login session, the difference being that the former
3910 will not register a session, nor go through the PAM session
3913 * tmpfiles gained support for a new "v" line type for creating
3914 btrfs subvolumes. If the underlying file system is a legacy
3915 file system, this automatically degrades to creating a
3916 normal directory. Among others /var/lib/machines is now
3917 created like this at boot, should it be missing.
3919 * The directory /var/lib/containers/ has been deprecated and
3920 been replaced by /var/lib/machines. The term "machines" has
3921 been used in the systemd context as generic term for both
3922 VMs and containers, and hence appears more appropriate for
3923 this, as the directory can also contain raw images bootable
3926 * systemd-nspawn when invoked with -M but without --directory=
3927 or --image= is now capable of searching for the container
3928 root directory, subvolume or disk image automatically, in
3929 /var/lib/machines. systemd-nspawn@.service has been updated
3930 to make use of this, thus allowing it to be used for raw
3933 * A new machines.target unit has been introduced that is
3934 supposed to group all containers/VMs invoked as services on
3935 the system. systemd-nspawn@.service has been updated to
3936 integrate with that.
3938 * machinectl gained a new "start" command, for invoking a
3939 container as a service. "machinectl start foo" is mostly
3940 equivalent to "systemctl start systemd-nspawn@foo.service",
3941 but handles escaping in a nicer way.
3943 * systemd-nspawn will now mount most of the cgroupfs tree
3944 read-only into each container, with the exception of the
3945 container's own subtree in the name=systemd hierarchy.
3947 * journald now sets the special FS_NOCOW file flag for its
3948 journal files. This should improve performance on btrfs, by
3949 avoiding heavy fragmentation when journald's write-pattern
3950 is used on COW file systems. It degrades btrfs' data
3951 integrity guarantees for the files to the same levels as for
3952 ext3/ext4 however. This should be OK though as journald does
3953 its own data integrity checks and all its objects are
3954 checksummed on disk. Also, journald should handle btrfs disk
3955 full events a lot more gracefully now, by processing SIGBUS
3956 errors, and not relying on fallocate() anymore.
3958 * When journald detects that journal files it is writing to
3959 have been deleted it will immediately start new journal
3962 * systemd now provides a way to store file descriptors
3963 per-service in PID 1. This is useful for daemons to ensure
3964 that fds they require are not lost during a daemon
3965 restart. The fds are passed to the daemon on the next
3966 invocation in the same way socket activation fds are
3967 passed. This is now used by journald to ensure that the
3968 various sockets connected to all the system's stdout/stderr
3969 are not lost when journald is restarted. File descriptors
3970 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
3971 an extension to sd_notify(). Note that a limit is enforced
3972 on the number of fds a service can store in PID 1, and it
3973 defaults to 0, so that no fds may be stored, unless this is
3974 explicitly turned on.
3976 * The default TERM variable to use for units connected to a
3977 terminal, when no other value is explicitly is set is now
3978 vt220 rather than vt102. This should be fairly safe still,
3979 but allows PgUp/PgDn work.
3981 * The /etc/crypttab option header= as known from Debian is now
3984 * "loginctl user-status" and "loginctl session-status" will
3985 now show the last 10 lines of log messages of the
3986 user/session following the status output. Similar,
3987 "machinectl status" will show the last 10 log lines
3988 associated with a virtual machine or container
3989 service. (Note that this is usually not the log messages
3990 done in the VM/container itself, but simply what the
3991 container manager logs. For nspawn this includes all console
3994 * "loginctl session-status" without further argument will now
3995 show the status of the session of the caller. Similar,
3996 "lock-session", "unlock-session", "activate",
3997 "enable-linger", "disable-linger" may now be called without
3998 session/user parameter in which case they apply to the
3999 caller's session/user.
4001 * An X11 session scriptlet is now shipped that uploads
4002 $DISPLAY and $XAUTHORITY into the environment of the systemd
4003 --user daemon if a session begins. This should improve
4004 compatibility with X11 enabled applications run as systemd
4007 * Generators are now subject to masking via /etc and /run, the
4008 same way as unit files.
4010 * networkd .network files gained support for configuring
4011 per-link IPv4/IPv6 packet forwarding as well as IPv4
4012 masquerading. This is by default turned on for veth links to
4013 containers, as registered by systemd-nspawn. This means that
4014 nspawn containers run with --network-veth will now get
4015 automatic routed access to the host's networks without any
4016 further configuration or setup, as long as networkd runs on
4019 * systemd-nspawn gained the --port= (-p) switch to expose TCP
4020 or UDP posts of a container on the host. With this in place
4021 it is possible to run containers with private veth links
4022 (--network-veth), and have their functionality exposed on
4023 the host as if their services were running directly on the
4026 * systemd-nspawn's --network-veth switch now gained a short
4027 version "-n", since with the changes above it is now truly
4028 useful out-of-the-box. The systemd-nspawn@.service has been
4029 updated to make use of it too by default.
4031 * systemd-nspawn will now maintain a per-image R/W lock, to
4032 ensure that the same image is not started more than once
4033 writable. (It's OK to run an image multiple times
4034 simultaneously in read-only mode.)
4036 * systemd-nspawn's --image= option is now capable of
4037 dissecting and booting MBR and GPT disk images that contain
4038 only a single active Linux partition. Previously it
4039 supported only GPT disk images with proper GPT type
4040 IDs. This allows running cloud images from major
4041 distributions directly with systemd-nspawn, without
4044 * In addition to collecting mouse dpi data in the udev
4045 hardware database, there's now support for collecting angle
4046 information for mouse scroll wheels. The database is
4047 supposed to guarantee similar scrolling behavior on mice
4048 that it knows about. There's also support for collecting
4049 information about Touchpad types.
4051 * udev's input_id built-in will now also collect touch screen
4052 dimension data and attach it to probed devices.
4054 * /etc/os-release gained support for a Distribution Privacy
4057 * networkd gained support for creating "ipvlan", "gretap",
4058 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
4060 * systemd-tmpfiles gained support for "a" lines for setting
4063 * systemd-nspawn will now mount /tmp in the container to
4064 tmpfs, automatically.
4066 * systemd now exposes the memory.usage_in_bytes cgroup
4067 attribute and shows it for each service in the "systemctl
4068 status" output, if available.
4070 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
4071 immediate reboot is triggered. This useful if shutdown is
4072 hung and is unable to complete, to expedite the
4073 operation. Note that this kind of reboot will still unmount
4074 all file systems, and hence should not result in fsck being
4077 * A .device unit for an optical block device will now be
4078 considered active only when a medium is in the drive. Also,
4079 mount units are now bound to their backing devices thus
4080 triggering automatic unmounting when devices become
4081 unavailable. With this in place systemd will now
4082 automatically unmount left-over mounts when a CD-ROM is
4083 ejected or an USB stick is yanked from the system.
4085 * networkd-wait-online now has support for waiting for
4086 specific interfaces only (with globbing), and for giving up
4087 after a configurable timeout.
4089 * networkd now exits when idle. It will be automatically
4090 restarted as soon as interfaces show up, are removed or
4091 change state. networkd will stay around as long as there is
4092 at least one DHCP state machine or similar around, that keep
4095 * networkd may now configure IPv6 link-local addressing in
4096 addition to IPv4 link-local addressing.
4098 * The IPv6 "token" for use in SLAAC may now be configured for
4099 each .network interface in networkd.
4101 * Routes configured with networkd may now be assigned a scope
4104 * networkd's [Match] sections now support globbing and lists
4105 of multiple space-separated matches per item.
4107 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
4108 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
4109 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
4110 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
4111 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
4112 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
4113 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
4114 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
4115 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
4116 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
4117 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
4118 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
4119 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
4120 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
4121 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
4122 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
4123 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
4124 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
4125 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
4126 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
4127 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
4128 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
4129 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
4130 Hoffmann, Zbigniew Jędrzejewski-Szmek
4132 — Berlin, 2015-02-16
4136 * When querying unit file enablement status (for example via
4137 "systemctl is-enabled"), a new state "indirect" is now known
4138 which indicates that a unit might not be enabled itself, but
4139 another unit listed in its Also= setting might be.
4141 * Similar to the various existing ConditionXYZ= settings for
4142 units, there are now matching AssertXYZ= settings. While
4143 failing conditions cause a unit to be skipped, but its job
4144 to succeed, failing assertions declared like this will cause
4145 a unit start operation and its job to fail.
4147 * hostnamed now knows a new chassis type "embedded".
4149 * systemctl gained a new "edit" command. When used on a unit
4150 file, this allows extending unit files with .d/ drop-in
4151 configuration snippets or editing the full file (after
4152 copying it from /usr/lib to /etc). This will invoke the
4153 user's editor (as configured with $EDITOR), and reload the
4154 modified configuration after editing.
4156 * "systemctl status" now shows the suggested enablement state
4157 for a unit, as declared in the (usually vendor-supplied)
4158 system preset files.
4160 * nss-myhostname will now resolve the single-label host name
4161 "gateway" to the locally configured default IP routing
4162 gateways, ordered by their metrics. This assigns a stable
4163 name to the used gateways, regardless which ones are
4164 currently configured. Note that the name will only be
4165 resolved after all other name sources (if nss-myhostname is
4166 configured properly) and should hence not negatively impact
4167 systems that use the single-label host name "gateway" in
4170 * systemd-inhibit now allows filtering by mode when listing
4173 * Scope and service units gained a new "Delegate" boolean
4174 property, which, when set, allows processes running inside the
4175 unit to further partition resources. This is primarily
4176 useful for systemd user instances as well as container
4179 * journald will now pick up audit messages directly from
4180 the kernel, and log them like any other log message. The
4181 audit fields are split up and fully indexed. This means that
4182 journalctl in many ways is now a (nicer!) alternative to
4183 ausearch, the traditional audit client. Note that this
4184 implements only a minimal audit client. If you want the
4185 special audit modes like reboot-on-log-overflow, please use
4186 the traditional auditd instead, which can be used in
4187 parallel to journald.
4189 * The ConditionSecurity= unit file option now understands the
4190 special string "audit" to check whether auditing is
4193 * journalctl gained two new commands --vacuum-size= and
4194 --vacuum-time= to delete old journal files until the
4195 remaining ones take up no more than the specified size on disk,
4196 or are not older than the specified time.
4198 * A new, native PPPoE library has been added to sd-network,
4199 systemd's library of light-weight networking protocols. This
4200 library will be used in a future version of networkd to
4201 enable PPPoE communication without an external pppd daemon.
4203 * The busctl tool now understands a new "capture" verb that
4204 works similar to "monitor", but writes a packet capture
4205 trace to STDOUT that can be redirected to a file which is
4206 compatible with libcap's capture file format. This can then
4207 be loaded in Wireshark and similar tools to inspect bus
4210 * The busctl tool now understands a new "tree" verb that shows
4211 the object trees of a specific service on the bus, or of all
4214 * The busctl tool now understands a new "introspect" verb that
4215 shows all interfaces and members of objects on the bus,
4216 including their signature and values. This is particularly
4217 useful to get more information about bus objects shown by
4218 the new "busctl tree" command.
4220 * The busctl tool now understands new verbs "call",
4221 "set-property" and "get-property" for invoking bus method
4222 calls, setting and getting bus object properties in a
4225 * busctl gained a new --augment-creds= argument that controls
4226 whether the tool shall augment credential information it
4227 gets from the bus with data from /proc, in a possibly
4230 * nspawn's --link-journal= switch gained two new values
4231 "try-guest" and "try-host" that work like "guest" and
4232 "host", but do not fail if the host has no persistent
4233 journalling enabled. -j is now equivalent to
4234 --link-journal=try-guest.
4236 * macvlan network devices created by nspawn will now have
4237 stable MAC addresses.
4239 * A new SmackProcessLabel= unit setting has been added, which
4240 controls the SMACK security label processes forked off by
4241 the respective unit shall use.
4243 * If compiled with --enable-xkbcommon, systemd-localed will
4244 verify x11 keymap settings by compiling the given keymap. It
4245 will spew out warnings if the compilation fails. This
4246 requires libxkbcommon to be installed.
4248 * When a coredump is collected, a larger number of metadata
4249 fields is now collected and included in the journal records
4250 created for it. More specifically, control group membership,
4251 environment variables, memory maps, working directory,
4252 chroot directory, /proc/$PID/status, and a list of open file
4253 descriptors is now stored in the log entry.
4255 * The udev hwdb now contains DPI information for mice. For
4258 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
4260 * All systemd programs that read standalone configuration
4261 files in /etc now also support a corresponding series of
4262 .conf.d configuration directories in /etc/, /run/,
4263 /usr/local/lib/, /usr/lib/, and (if configured with
4264 --enable-split-usr) /lib/. In particular, the following
4265 configuration files now have corresponding configuration
4266 directories: system.conf user.conf, logind.conf,
4267 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
4268 resolved.conf, timesyncd.conf, journal-remote.conf, and
4269 journal-upload.conf. Note that distributions should use the
4270 configuration directories in /usr/lib/; the directories in
4271 /etc/ are reserved for the system administrator.
4273 * systemd-rfkill will no longer take the rfkill device name
4274 into account when storing rfkill state on disk, as the name
4275 might be dynamically assigned and not stable. Instead, the
4276 ID_PATH udev variable combined with the rfkill type (wlan,
4277 bluetooth, ...) is used.
4279 * A new service systemd-machine-id-commit.service has been
4280 added. When used on systems where /etc is read-only during
4281 boot, and /etc/machine-id is not initialized (but an empty
4282 file), this service will copy the temporary machine ID
4283 created as replacement into /etc after the system is fully
4284 booted up. This is useful for systems that are freshly
4285 installed with a non-initialized machine ID, but should get
4286 a fixed machine ID for subsequent boots.
4288 * networkd's .netdev files now provide a large set of
4289 configuration parameters for VXLAN devices. Similarly, the
4290 bridge port cost parameter is now configurable in .network
4291 files. There's also new support for configuring IP source
4292 routing. networkd .link files gained support for a new
4293 OriginalName= match that is useful to match against the
4294 original interface name the kernel assigned. .network files
4295 may include MTU= and MACAddress= fields for altering the MTU
4296 and MAC address while being connected to a specific network
4299 * The LUKS logic gained supported for configuring
4300 UUID-specific key files. There's also new support for naming
4301 LUKS device from the kernel command line, using the new
4302 luks.name= argument.
4304 * Timer units may now be transiently created via the bus API
4305 (this was previously already available for scope and service
4306 units). In addition it is now possible to create multiple
4307 transient units at the same time with a single bus call. The
4308 "systemd-run" tool has been updated to make use of this for
4309 running commands on a specified time, in at(1)-style.
4311 * tmpfiles gained support for "t" lines, for assigning
4312 extended attributes to files. Among other uses this may be
4313 used to assign SMACK labels to files.
4315 Contributions from: Alin Rauta, Alison Chaiken, Andrej
4316 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
4317 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
4318 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
4319 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
4320 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
4321 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
4322 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
4323 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
4324 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
4325 Michael Biebl, Michael Chapman, Michael Marineau, Michal
4326 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
4327 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
4328 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
4329 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
4330 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4331 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
4332 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
4334 — Berlin, 2014-12-10
4338 * journalctl gained the new options -t/--identifier= to match
4339 on the syslog identifier (aka "tag"), as well as --utc to
4340 show log timestamps in the UTC timezone. journalctl now also
4341 accepts -n/--lines=all to disable line capping in a pager.
4343 * journalctl gained a new switch, --flush, that synchronously
4344 flushes logs from /run/log/journal to /var/log/journal if
4345 persistent storage is enabled. systemd-journal-flush.service
4346 now waits until the operation is complete.
4348 * Services can notify the manager before they start a reload
4349 (by sending RELOADING=1) or shutdown (by sending
4350 STOPPING=1). This allows the manager to track and show the
4351 internal state of daemons and closes a race condition when
4352 the process is still running but has closed its D-Bus
4355 * Services with Type=oneshot do not have to have any ExecStart
4358 * User units are now loaded also from
4359 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
4360 /run/systemd/user directory that was already previously
4361 supported, but is under the control of the user.
4363 * Job timeouts (i.e. time-outs on the time a job that is
4364 queued stays in the run queue) can now optionally result in
4365 immediate reboot or power-off actions (JobTimeoutAction= and
4366 JobTimeoutRebootArgument=). This is useful on ".target"
4367 units, to limit the maximum time a target remains
4368 undispatched in the run queue, and to trigger an emergency
4369 operation in such a case. This is now used by default to
4370 turn off the system if boot-up (as defined by everything in
4371 basic.target) hangs and does not complete for at least
4372 15min. Also, if power-off or reboot hang for at least 30min
4373 an immediate power-off/reboot operation is triggered. This
4374 functionality is particularly useful to increase reliability
4375 on embedded devices, but also on laptops which might
4376 accidentally get powered on when carried in a backpack and
4377 whose boot stays stuck in a hard disk encryption passphrase
4380 * systemd-logind can be configured to also handle lid switch
4381 events even when the machine is docked or multiple displays
4382 are attached (HandleLidSwitchDocked= option).
4384 * A helper binary and a service have been added which can be
4385 used to resume from hibernation in the initramfs. A
4386 generator will parse the resume= option on the kernel
4387 command line to trigger resume.
4389 * A user console daemon systemd-consoled has been
4390 added. Currently, it is a preview, and will so far open a
4391 single terminal on each session of the user marked as
4392 Desktop=systemd-console.
4394 * Route metrics can be specified for DHCP routes added by
4397 * The SELinux context of socket-activated services can be set
4398 from the information provided by the networking stack
4399 (SELinuxContextFromNet= option).
4401 * Userspace firmware loading support has been removed and
4402 the minimum supported kernel version is thus bumped to 3.7.
4404 * Timeout for udev workers has been increased from 1 to 3
4405 minutes, but a warning will be printed after 1 minute to
4406 help diagnose kernel modules that take a long time to load.
4408 * Udev rules can now remove tags on devices with TAG-="foobar".
4410 * systemd's readahead implementation has been removed. In many
4411 circumstances it didn't give expected benefits even for
4412 rotational disk drives and was becoming less relevant in the
4413 age of SSDs. As none of the developers has been using
4414 rotating media anymore, and nobody stepped up to actively
4415 maintain this component of systemd it has now been removed.
4417 * Swap units can use Options= to specify discard options.
4418 Discard options specified for swaps in /etc/fstab are now
4421 * Docker containers are now detected as a separate type of
4424 * The Password Agent protocol gained support for queries where
4425 the user input is shown, useful e.g. for user names.
4426 systemd-ask-password gained a new --echo option to turn that
4429 * The default sysctl.d/ snippets will now set:
4431 net.core.default_qdisc = fq_codel
4433 This selects Fair Queuing Controlled Delay as the default
4434 queuing discipline for network interfaces. fq_codel helps
4435 fight the network bufferbloat problem. It is believed to be
4436 a good default with no tuning required for most workloads.
4437 Downstream distributions may override this choice. On 10Gbit
4438 servers that do not do forwarding, "fq" may perform better.
4439 Systems without a good clocksource should use "pfifo_fast".
4441 * If kdbus is enabled during build a new option BusPolicy= is
4442 available for service units, that allows locking all service
4443 processes into a stricter bus policy, in order to limit
4444 access to various bus services, or even hide most of them
4445 from the service's view entirely.
4447 * networkctl will now show the .network and .link file
4448 networkd has applied to a specific interface.
4450 * sd-login gained a new API call sd_session_get_desktop() to
4451 query which desktop environment has been selected for a
4454 * UNIX utmp support is now compile-time optional to support
4455 legacy-free systems.
4457 * systemctl gained two new commands "add-wants" and
4458 "add-requires" for pulling in units from specific targets
4461 * If the word "rescue" is specified on the kernel command line
4462 the system will now boot into rescue mode (aka
4463 rescue.target), which was previously available only by
4464 specifying "1" or "systemd.unit=rescue.target" on the kernel
4465 command line. This new kernel command line option nicely
4466 mirrors the already existing "emergency" kernel command line
4469 * New kernel command line options mount.usr=, mount.usrflags=,
4470 mount.usrfstype= have been added that match root=, rootflags=,
4471 rootfstype= but allow mounting a specific file system to
4474 * The $NOTIFY_SOCKET is now also passed to control processes of
4475 services, not only the main process.
4477 * This version reenables support for fsck's -l switch. This
4478 means at least version v2.25 of util-linux is required for
4479 operation, otherwise dead-locks on device nodes may
4480 occur. Again: you need to update util-linux to at least
4481 v2.25 when updating systemd to v217.
4483 * The "multi-seat-x" tool has been removed from systemd, as
4484 its functionality has been integrated into X servers 1.16,
4485 and the tool is hence redundant. It is recommended to update
4486 display managers invoking this tool to simply invoke X
4487 directly from now on, again.
4489 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
4490 message flag has been added for all of systemd's PolicyKit
4491 authenticated method calls has been added. In particular
4492 this now allows optional interactive authorization via
4493 PolicyKit for many of PID1's privileged operations such as
4494 unit file enabling and disabling.
4496 * "udevadm hwdb --update" learnt a new switch "--usr" for
4497 placing the rebuilt hardware database in /usr instead of
4498 /etc. When used only hardware database entries stored in
4499 /usr will be used, and any user database entries in /etc are
4500 ignored. This functionality is useful for vendors to ship a
4501 pre-built database on systems where local configuration is
4502 unnecessary or unlikely.
4504 * Calendar time specifications in .timer units now also
4505 understand the strings "semi-annually", "quarterly" and
4506 "minutely" as shortcuts (in addition to the preexisting
4507 "anually", "hourly", ...).
4509 * systemd-tmpfiles will now correctly create files in /dev
4510 at boot which are marked for creation only at boot. It is
4511 recommended to always create static device nodes with 'c!'
4512 and 'b!', so that they are created only at boot and not
4513 overwritten at runtime.
4515 * When the watchdog logic is used for a service (WatchdogSec=)
4516 and the watchdog timeout is hit the service will now be
4517 terminated with SIGABRT (instead of just SIGTERM), in order
4518 to make sure a proper coredump and backtrace is
4519 generated. This ensures that hanging services will result in
4520 similar coredump/backtrace behaviour as services that hit a
4523 Contributions from: Andreas Henriksson, Andrei Borzenkov,
4524 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
4525 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
4526 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
4527 Herrmann, David Sommerseth, David Strauss, Emil Renner
4528 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
4529 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
4530 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
4531 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
4532 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
4533 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
4534 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
4535 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
4536 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
4537 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
4538 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
4539 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
4540 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
4541 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
4542 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
4543 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
4546 — Berlin, 2014-10-28
4550 * timedated no longer reads NTP implementation unit names from
4551 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
4552 implementations should add a
4554 Conflicts=systemd-timesyncd.service
4556 to their unit files to take over and replace systemd's NTP
4557 default functionality.
4559 * systemd-sysusers gained a new line type "r" for configuring
4560 which UID/GID ranges to allocate system users/groups
4561 from. Lines of type "u" may now add an additional column
4562 that specifies the home directory for the system user to be
4563 created. Also, systemd-sysusers may now optionally read user
4564 information from STDIN instead of a file. This is useful for
4565 invoking it from RPM preinst scriptlets that need to create
4566 users before the first RPM file is installed since these
4567 files might need to be owned by them. A new
4568 %sysusers_create_inline RPM macro has been introduced to do
4569 just that. systemd-sysusers now updates the shadow files as
4570 well as the user/group databases, which should enhance
4571 compatibility with certain tools like grpck.
4573 * A number of bus APIs of PID 1 now optionally consult
4574 PolicyKit to permit access for otherwise unprivileged
4575 clients under certain conditions. Note that this currently
4576 doesn't support interactive authentication yet, but this is
4577 expected to be added eventually, too.
4579 * /etc/machine-info now has new fields for configuring the
4580 deployment environment of the machine, as well as the
4581 location of the machine. hostnamectl has been updated with
4582 new command to update these fields.
4584 * systemd-timesyncd has been updated to automatically acquire
4585 NTP server information from systemd-networkd, which might
4586 have been discovered via DHCP.
4588 * systemd-resolved now includes a caching DNS stub resolver
4589 and a complete LLMNR name resolution implementation. A new
4590 NSS module "nss-resolve" has been added which can be used
4591 instead of glibc's own "nss-dns" to resolve hostnames via
4592 systemd-resolved. Hostnames, addresses and arbitrary RRs may
4593 be resolved via systemd-resolved D-Bus APIs. In contrast to
4594 the glibc internal resolver systemd-resolved is aware of
4595 multi-homed system, and keeps DNS server and caches separate
4596 and per-interface. Queries are sent simultaneously on all
4597 interfaces that have DNS servers configured, in order to
4598 properly handle VPNs and local LANs which might resolve
4599 separate sets of domain names. systemd-resolved may acquire
4600 DNS server information from systemd-networkd automatically,
4601 which in turn might have discovered them via DHCP. A tool
4602 "systemd-resolve-host" has been added that may be used to
4603 query the DNS logic in resolved. systemd-resolved implements
4604 IDNA and automatically uses IDNA or UTF-8 encoding depending
4605 on whether classic DNS or LLMNR is used as transport. In the
4606 next releases we intend to add a DNSSEC and mDNS/DNS-SD
4607 implementation to systemd-resolved.
4609 * A new NSS module nss-mymachines has been added, that
4610 automatically resolves the names of all local registered
4611 containers to their respective IP addresses.
4613 * A new client tool "networkctl" for systemd-networkd has been
4614 added. It currently is entirely passive and will query
4615 networking configuration from udev, rtnetlink and networkd,
4616 and present it to the user in a very friendly
4617 way. Eventually, we hope to extend it to become a full
4618 control utility for networkd.
4620 * .socket units gained a new DeferAcceptSec= setting that
4621 controls the kernels' TCP_DEFER_ACCEPT sockopt for
4622 TCP. Similarly, support for controlling TCP keep-alive
4623 settings has been added (KeepAliveTimeSec=,
4624 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
4625 turning off Nagle's algorithm on TCP has been added
4628 * logind learned a new session type "web", for use in projects
4629 like Cockpit which register web clients as PAM sessions.
4631 * timer units with at least one OnCalendar= setting will now
4632 be started only after time-sync.target has been
4633 reached. This way they will not elapse before the system
4634 clock has been corrected by a local NTP client or
4635 similar. This is particular useful on RTC-less embedded
4636 machines, that come up with an invalid system clock.
4638 * systemd-nspawn's --network-veth= switch should now result in
4639 stable MAC addresses for both the outer and the inner side
4642 * systemd-nspawn gained a new --volatile= switch for running
4643 container instances with /etc or /var unpopulated.
4645 * The kdbus client code has been updated to use the new Linux
4646 3.17 memfd subsystem instead of the old kdbus-specific one.
4648 * systemd-networkd's DHCP client and server now support
4649 FORCERENEW. There are also new configuration options to
4650 configure the vendor client identifier and broadcast mode
4653 * systemd will no longer inform the kernel about the current
4654 timezone, as this is necessarily incorrect and racy as the
4655 kernel has no understanding of DST and similar
4656 concepts. This hence means FAT timestamps will be always
4657 considered UTC, similar to what Android is already
4658 doing. Also, when the RTC is configured to the local time
4659 (rather than UTC) systemd will never synchronize back to it,
4660 as this might confuse Windows at a later boot.
4662 * systemd-analyze gained a new command "verify" for offline
4663 validation of unit files.
4665 * systemd-networkd gained support for a couple of additional
4666 settings for bonding networking setups. Also, the metric for
4667 statically configured routes may now be configured. For
4668 network interfaces where this is appropriate the peer IP
4669 address may now be configured.
4671 * systemd-networkd's DHCP client will no longer request
4672 broadcasting by default, as this tripped up some networks.
4673 For hardware where broadcast is required the feature should
4674 be switched back on using RequestBroadcast=yes.
4676 * systemd-networkd will now set up IPv4LL addresses (when
4677 enabled) even if DHCP is configured successfully.
4679 * udev will now default to respect network device names given
4680 by the kernel when the kernel indicates that these are
4681 predictable. This behavior can be tweaked by changing
4682 NamePolicy= in the relevant .link file.
4684 * A new library systemd-terminal has been added that
4685 implements full TTY stream parsing and rendering. This
4686 library is supposed to be used later on for implementing a
4687 full userspace VT subsystem, replacing the current kernel
4690 * A new tool systemd-journal-upload has been added to push
4691 journal data to a remote system running
4692 systemd-journal-remote.
4694 * journald will no longer forward all local data to another
4695 running syslog daemon. This change has been made because
4696 rsyslog (which appears to be the most commonly used syslog
4697 implementation these days) no longer makes use of this, and
4698 instead pulls the data out of the journal on its own. Since
4699 forwarding the messages to a non-existent syslog server is
4700 more expensive than we assumed we have now turned this
4701 off. If you run a syslog server that is not a recent rsyslog
4702 version, you have to turn this option on again
4703 (ForwardToSyslog= in journald.conf).
4705 * journald now optionally supports the LZ4 compressor for
4706 larger journal fields. This compressor should perform much
4707 better than XZ which was the previous default.
4709 * machinectl now shows the IP addresses of local containers,
4710 if it knows them, plus the interface name of the container.
4712 * A new tool "systemd-escape" has been added that makes it
4713 easy to escape strings to build unit names and similar.
4715 * sd_notify() messages may now include a new ERRNO= field
4716 which is parsed and collected by systemd and shown among the
4717 "systemctl status" output for a service.
4719 * A new component "systemd-firstboot" has been added that
4720 queries the most basic systemd information (timezone,
4721 hostname, root password) interactively on first
4722 boot. Alternatively it may also be used to provision these
4723 things offline on OS images installed into directories.
4725 * The default sysctl.d/ snippets will now set
4727 net.ipv4.conf.default.promote_secondaries=1
4729 This has the benefit of no flushing secondary IP addresses
4730 when primary addresses are removed.
4732 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
4733 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
4734 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
4735 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
4736 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
4737 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
4738 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
4739 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
4740 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
4741 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
4742 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
4743 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
4744 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
4745 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
4746 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
4748 — Berlin, 2014-08-19
4752 * A new tool systemd-sysusers has been added. This tool
4753 creates system users and groups in /etc/passwd and
4754 /etc/group, based on static declarative system user/group
4755 definitions in /usr/lib/sysusers.d/. This is useful to
4756 enable factory resets and volatile systems that boot up with
4757 an empty /etc directory, and thus need system users and
4758 groups created during early boot. systemd now also ships
4759 with two default sysusers.d/ files for the most basic
4760 users and groups systemd and the core operating system
4763 * A new tmpfiles snippet has been added that rebuilds the
4764 essential files in /etc on boot, should they be missing.
4766 * A directive for ensuring automatic clean-up of
4767 /var/cache/man/ has been removed from the default
4768 configuration. This line should now be shipped by the man
4769 implementation. The necessary change has been made to the
4770 man-db implementation. Note that you need to update your man
4771 implementation to one that ships this line, otherwise no
4772 automatic clean-up of /var/cache/man will take place.
4774 * A new condition ConditionNeedsUpdate= has been added that
4775 may conditionalize services to only run when /etc or /var
4776 are "older" than the vendor operating system resources in
4777 /usr. This is useful for reconstructing or updating /etc
4778 after an offline update of /usr or a factory reset, on the
4779 next reboot. Services that want to run once after such an
4780 update or reset should use this condition and order
4781 themselves before the new systemd-update-done.service, which
4782 will mark the two directories as fully updated. A number of
4783 service files have been added making use of this, to rebuild
4784 the udev hardware database, the journald message catalog and
4785 dynamic loader cache (ldconfig). The systemd-sysusers tool
4786 described above also makes use of this now. With this in
4787 place it is now possible to start up a minimal operating
4788 system with /etc empty cleanly. For more information on the
4789 concepts involved see this recent blog story:
4791 http://0pointer.de/blog/projects/stateless.html
4793 * A new system group "input" has been introduced, and all
4794 input device nodes get this group assigned. This is useful
4795 for system-level software to get access to input devices. It
4796 complements what is already done for "audio" and "video".
4798 * systemd-networkd learnt minimal DHCPv4 server support in
4799 addition to the existing DHCPv4 client support. It also
4800 learnt DHCPv6 client and IPv6 Router Solicitation client
4801 support. The DHCPv4 client gained support for static routes
4802 passed in from the server. Note that the [DHCPv4] section
4803 known in older systemd-networkd versions has been renamed to
4804 [DHCP] and is now also used by the DHCPv6 client. Existing
4805 .network files using settings of this section should be
4806 updated, though compatibility is maintained. Optionally, the
4807 client hostname may now be sent to the DHCP server.
4809 * networkd gained support for vxlan virtual networks as well
4810 as tun/tap and dummy devices.
4812 * networkd gained support for automatic allocation of address
4813 ranges for interfaces from a system-wide pool of
4814 addresses. This is useful for dynamically managing a large
4815 number of interfaces with a single network configuration
4816 file. In particular this is useful to easily assign
4817 appropriate IP addresses to the veth links of a large number
4818 of nspawn instances.
4820 * RPM macros for processing sysusers, sysctl and binfmt
4821 drop-in snippets at package installation time have been
4824 * The /etc/os-release file should now be placed in
4825 /usr/lib/os-release. The old location is automatically
4826 created as symlink. /usr/lib is the more appropriate
4827 location of this file, since it shall actually describe the
4828 vendor operating system shipped in /usr, and not the
4829 configuration stored in /etc.
4831 * .mount units gained a new boolean SloppyOptions= setting
4832 that maps to mount(8)'s -s option which enables permissive
4833 parsing of unknown mount options.
4835 * tmpfiles learnt a new "L+" directive which creates a symlink
4836 but (unlike "L") deletes a pre-existing file first, should
4837 it already exist and not already be the correct
4838 symlink. Similarly, "b+", "c+" and "p+" directives have been
4839 added as well, which create block and character devices, as
4840 well as fifos in the filesystem, possibly removing any
4841 pre-existing files of different types.
4843 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
4844 'argument' field (which so far specified the source to
4845 symlink/copy the files from) is now optional. If omitted the
4846 same file os copied from /usr/share/factory/ suffixed by the
4847 full destination path. This is useful for populating /etc
4848 with essential files, by copying them from vendor defaults
4849 shipped in /usr/share/factory/etc.
4851 * A new command "systemctl preset-all" has been added that
4852 applies the service preset settings to all installed unit
4853 files. A new switch --preset-mode= has been added that
4854 controls whether only enable or only disable operations
4857 * A new command "systemctl is-system-running" has been added
4858 that allows checking the overall state of the system, for
4859 example whether it is fully up and running.
4861 * When the system boots up with an empty /etc, the equivalent
4862 to "systemctl preset-all" is executed during early boot, to
4863 make sure all default services are enabled after a factory
4866 * systemd now contains a minimal preset file that enables the
4867 most basic services systemd ships by default.
4869 * Unit files' [Install] section gained a new DefaultInstance=
4870 field for defining the default instance to create if a
4871 template unit is enabled with no instance specified.
4873 * A new passive target cryptsetup-pre.target has been added
4874 that may be used by services that need to make they run and
4875 finish before the first LUKS cryptographic device is set up.
4877 * The /dev/loop-control and /dev/btrfs-control device nodes
4878 are now owned by the "disk" group by default, opening up
4879 access to this group.
4881 * systemd-coredump will now automatically generate a
4882 stack trace of all core dumps taking place on the system,
4883 based on elfutils' libdw library. This stack trace is logged
4886 * systemd-coredump may now optionally store coredumps directly
4887 on disk (in /var/lib/systemd/coredump, possibly compressed),
4888 instead of storing them unconditionally in the journal. This
4889 mode is the new default. A new configuration file
4890 /etc/systemd/coredump.conf has been added to configure this
4891 and other parameters of systemd-coredump.
4893 * coredumpctl gained a new "info" verb to show details about a
4894 specific coredump. A new switch "-1" has also been added
4895 that makes sure to only show information about the most
4896 recent entry instead of all entries. Also, as the tool is
4897 generally useful now the "systemd-" prefix of the binary
4898 name has been removed. Distributions that want to maintain
4899 compatibility with the old name should add a symlink from
4900 the old name to the new name.
4902 * journald's SplitMode= now defaults to "uid". This makes sure
4903 that unprivileged users can access their own coredumps with
4904 coredumpctl without restrictions.
4906 * New kernel command line options "systemd.wants=" (for
4907 pulling an additional unit during boot), "systemd.mask="
4908 (for masking a specific unit for the boot), and
4909 "systemd.debug-shell" (for enabling the debug shell on tty9)
4910 have been added. This is implemented in the new generator
4911 "systemd-debug-generator".
4913 * systemd-nspawn will now by default filter a couple of
4914 syscalls for containers, among them those required for
4915 kernel module loading, direct x86 IO port access, swap
4916 management, and kexec. Most importantly though
4917 open_by_handle_at() is now prohibited for containers,
4918 closing a hole similar to a recently discussed vulnerability
4919 in docker regarding access to files on file hierarchies the
4920 container should normally not have access to. Note that, for
4921 nspawn, we generally make no security claims anyway (and
4922 this is explicitly documented in the man page), so this is
4923 just a fix for one of the most obvious problems.
4925 * A new man page file-hierarchy(7) has been added that
4926 contains a minimized, modernized version of the file system
4927 layout systemd expects, similar in style to the FHS
4928 specification or hier(5). A new tool systemd-path(1) has
4929 been added to query many of these paths for the local
4932 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
4933 longer done. Since the directory now has a per-user size
4934 limit, and is cleaned on logout this appears unnecessary,
4935 in particular since this now brings the lifecycle of this
4936 directory closer in line with how IPC objects are handled.
4938 * systemd.pc now exports a number of additional directories,
4939 including $libdir (which is useful to identify the library
4940 path for the primary architecture of the system), and a
4941 couple of drop-in directories.
4943 * udev's predictable network interface names now use the dev_port
4944 sysfs attribute, introduced in linux 3.15 instead of dev_id to
4945 distinguish between ports of the same PCI function. dev_id should
4946 only be used for ports using the same HW address, hence the need
4949 * machined has been updated to export the OS version of a
4950 container (read from /etc/os-release and
4951 /usr/lib/os-release) on the bus. This is now shown in
4952 "machinectl status" for a machine.
4954 * A new service setting RestartForceExitStatus= has been
4955 added. If configured to a set of exit signals or process
4956 return values, the service will be restarted when the main
4957 daemon process exits with any of them, regardless of the
4960 * systemctl's -H switch for connecting to remote systemd
4961 machines has been extended so that it may be used to
4962 directly connect to a specific container on the
4963 host. "systemctl -H root@foobar:waldi" will now connect as
4964 user "root" to host "foobar", and then proceed directly to
4965 the container named "waldi". Note that currently you have to
4966 authenticate as user "root" for this to work, as entering
4967 containers is a privileged operation.
4969 Contributions from: Andreas Henriksson, Benjamin Steinwender,
4970 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
4971 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
4972 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
4973 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
4974 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
4975 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
4976 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
4977 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
4978 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
4979 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
4980 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
4982 — Berlin, 2014-07-03
4986 * As an experimental feature, udev now tries to lock the
4987 disk device node (flock(LOCK_SH|LOCK_NB)) while it
4988 executes events for the disk or any of its partitions.
4989 Applications like partitioning programs can lock the
4990 disk device node (flock(LOCK_EX)) and claim temporary
4991 device ownership that way; udev will entirely skip all event
4992 handling for this disk and its partitions. If the disk
4993 was opened for writing, the close will trigger a partition
4994 table rescan in udev's "watch" facility, and if needed
4995 synthesize "change" events for the disk and all its partitions.
4996 This is now unconditionally enabled, and if it turns out to
4997 cause major problems, we might turn it on only for specific
4998 devices, or might need to disable it entirely. Device Mapper
4999 devices are excluded from this logic.
5001 * We temporarily dropped the "-l" switch for fsck invocations,
5002 since they collide with the flock() logic above. util-linux
5003 upstream has been changed already to avoid this conflict,
5004 and we will readd "-l" as soon as util-linux with this
5005 change has been released.
5007 * The dependency on libattr has been removed. Since a long
5008 time, the extended attribute calls have moved to glibc, and
5009 libattr is thus unnecessary.
5011 * Virtualization detection works without privileges now. This
5012 means the systemd-detect-virt binary no longer requires
5013 CAP_SYS_PTRACE file capabilities, and our daemons can run
5014 with fewer privileges.
5016 * systemd-networkd now runs under its own "systemd-network"
5017 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
5018 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
5019 loses the ability to write to files owned by root this way.
5021 * Similarly, systemd-resolved now runs under its own
5022 "systemd-resolve" user with no capabilities remaining.
5024 * Similarly, systemd-bus-proxyd now runs under its own
5025 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
5027 * systemd-networkd gained support for setting up "veth"
5028 virtual Ethernet devices for container connectivity, as well
5029 as GRE and VTI tunnels.
5031 * systemd-networkd will no longer automatically attempt to
5032 manually load kernel modules necessary for certain tunnel
5033 transports. Instead, it is assumed the kernel loads them
5034 automatically when required. This only works correctly on
5035 very new kernels. On older kernels, please consider adding
5036 the kernel modules to /etc/modules-load.d/ as a work-around.
5038 * The resolv.conf file systemd-resolved generates has been
5039 moved to /run/systemd/resolve/. If you have a symlink from
5040 /etc/resolv.conf, it might be necessary to correct it.
5042 * Two new service settings, ProtectHome= and ProtectSystem=,
5043 have been added. When enabled, they will make the user data
5044 (such as /home) inaccessible or read-only and the system
5045 (such as /usr) read-only, for specific services. This allows
5046 very light-weight per-service sandboxing to avoid
5047 modifications of user data or system files from
5048 services. These two new switches have been enabled for all
5049 of systemd's long-running services, where appropriate.
5051 * Socket units gained new SocketUser= and SocketGroup=
5052 settings to set the owner user and group of AF_UNIX sockets
5053 and FIFOs in the file system.
5055 * Socket units gained a new RemoveOnStop= setting. If enabled,
5056 all FIFOS and sockets in the file system will be removed
5057 when the specific socket unit is stopped.
5059 * Socket units gained a new Symlinks= setting. It takes a list
5060 of symlinks to create to file system sockets or FIFOs
5061 created by the specific Unix sockets. This is useful to
5062 manage symlinks to socket nodes with the same life-cycle as
5065 * The /dev/log socket and /dev/initctl FIFO have been moved to
5066 /run, and have been replaced by symlinks. This allows
5067 connecting to these facilities even if PrivateDevices=yes is
5068 used for a service (which makes /dev/log itself unavailable,
5069 but /run is left). This also has the benefit of ensuring
5070 that /dev only contains device nodes, directories and
5071 symlinks, and nothing else.
5073 * sd-daemon gained two new calls sd_pid_notify() and
5074 sd_pid_notifyf(). They are similar to sd_notify() and
5075 sd_notifyf(), but allow overriding of the source PID of
5076 notification messages if permissions permit this. This is
5077 useful to send notify messages on behalf of a different
5078 process (for example, the parent process). The
5079 systemd-notify tool has been updated to make use of this
5080 when sending messages (so that notification messages now
5081 originate from the shell script invoking systemd-notify and
5082 not the systemd-notify process itself. This should minimize
5083 a race where systemd fails to associate notification
5084 messages to services when the originating process already
5087 * A new "on-abnormal" setting for Restart= has been added. If
5088 set, it will result in automatic restarts on all "abnormal"
5089 reasons for a process to exit, which includes unclean
5090 signals, core dumps, timeouts and watchdog timeouts, but
5091 does not include clean and unclean exit codes or clean
5092 signals. Restart=on-abnormal is an alternative for
5093 Restart=on-failure for services that shall be able to
5094 terminate and avoid restarts on certain errors, by
5095 indicating so with an unclean exit code. Restart=on-failure
5096 or Restart=on-abnormal is now the recommended setting for
5097 all long-running services.
5099 * If the InaccessibleDirectories= service setting points to a
5100 mount point (or if there are any submounts contained within
5101 it), it is now attempted to completely unmount it, to make
5102 the file systems truly unavailable for the respective
5105 * The ReadOnlyDirectories= service setting and
5106 systemd-nspawn's --read-only parameter are now recursively
5107 applied to all submounts, too.
5109 * Mount units may now be created transiently via the bus APIs.
5111 * The support for SysV and LSB init scripts has been removed
5112 from the systemd daemon itself. Instead, it is now
5113 implemented as a generator that creates native systemd units
5114 from these scripts when needed. This enables us to remove a
5115 substantial amount of legacy code from PID 1, following the
5116 fact that many distributions only ship a very small number
5117 of LSB/SysV init scripts nowadays.
5119 * Privileged Xen (dom0) domains are not considered
5120 virtualization anymore by the virtualization detection
5121 logic. After all, they generally have unrestricted access to
5122 the hardware and usually are used to manage the unprivileged
5125 * systemd-tmpfiles gained a new "C" line type, for copying
5126 files or entire directories.
5128 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
5129 lines. So far, they have been non-globbing versions of the
5130 latter, and have thus been redundant. In future, it is
5131 recommended to only use "z". "m" has hence been removed
5132 from the documentation, even though it stays supported.
5134 * A tmpfiles snippet to recreate the most basic structure in
5135 /var has been added. This is enough to create the /var/run →
5136 /run symlink and create a couple of structural
5137 directories. This allows systems to boot up with an empty or
5138 volatile /var. Of course, while with this change, the core OS
5139 now is capable with dealing with a volatile /var, not all
5140 user services are ready for it. However, we hope that sooner
5141 or later, many service daemons will be changed upstream so
5142 that they are able to automatically create their necessary
5143 directories in /var at boot, should they be missing. This is
5144 the first step to allow state-less systems that only require
5145 the vendor image for /usr to boot.
5147 * systemd-nspawn has gained a new --tmpfs= switch to mount an
5148 empty tmpfs instance to a specific directory. This is
5149 particularly useful for making use of the automatic
5150 reconstruction of /var (see above), by passing --tmpfs=/var.
5152 * Access modes specified in tmpfiles snippets may now be
5153 prefixed with "~", which indicates that they shall be masked
5154 by whether the existing file or directory is currently
5155 writable, readable or executable at all. Also, if specified,
5156 the sgid/suid/sticky bits will be masked for all
5159 * A new passive target unit "network-pre.target" has been
5160 added which is useful for services that shall run before any
5161 network is configured, for example firewall scripts.
5163 * The "floppy" group that previously owned the /dev/fd*
5164 devices is no longer used. The "disk" group is now used
5165 instead. Distributions should probably deprecate usage of
5168 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
5169 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
5170 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
5171 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
5172 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
5173 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
5176 — Berlin, 2014-06-11
5180 * A new "systemd-timesyncd" daemon has been added for
5181 synchronizing the system clock across the network. It
5182 implements an SNTP client. In contrast to NTP
5183 implementations such as chrony or the NTP reference server,
5184 this only implements a client side, and does not bother with
5185 the full NTP complexity, focusing only on querying time from
5186 one remote server and synchronizing the local clock to
5187 it. Unless you intend to serve NTP to networked clients or
5188 want to connect to local hardware clocks, this simple NTP
5189 client should be more than appropriate for most
5190 installations. The daemon runs with minimal privileges, and
5191 has been hooked up with networkd to only operate when
5192 network connectivity is available. The daemon saves the
5193 current clock to disk every time a new NTP sync has been
5194 acquired, and uses this to possibly correct the system clock
5195 early at bootup, in order to accommodate for systems that
5196 lack an RTC such as the Raspberry Pi and embedded devices,
5197 and to make sure that time monotonically progresses on these
5198 systems, even if it is not always correct. To make use of
5199 this daemon, a new system user and group "systemd-timesync"
5200 needs to be created on installation of systemd.
5202 * The queue "seqnum" interface of libudev has been disabled, as
5203 it was generally incompatible with device namespacing as
5204 sequence numbers of devices go "missing" if the devices are
5205 part of a different namespace.
5207 * "systemctl list-timers" and "systemctl list-sockets" gained
5208 a --recursive switch for showing units of these types also
5209 for all local containers, similar in style to the already
5210 supported --recursive switch for "systemctl list-units".
5212 * A new RebootArgument= setting has been added for service
5213 units, which may be used to specify a kernel reboot argument
5214 to use when triggering reboots with StartLimitAction=.
5216 * A new FailureAction= setting has been added for service
5217 units which may be used to specify an operation to trigger
5218 when a service fails. This works similarly to
5219 StartLimitAction=, but unlike it, controls what is done
5220 immediately rather than only after several attempts to
5221 restart the service in question.
5223 * hostnamed got updated to also expose the kernel name,
5224 release, and version on the bus. This is useful for
5225 executing commands like hostnamectl with the -H switch.
5226 systemd-analyze makes use of this to properly display
5227 details when running non-locally.
5229 * The bootchart tool can now show cgroup information in the
5230 graphs it generates.
5232 * The CFS CPU quota cgroup attribute is now exposed for
5233 services. The new CPUQuota= switch has been added for this
5234 which takes a percentage value. Setting this will have the
5235 result that a service may never get more CPU time than the
5236 specified percentage, even if the machine is otherwise idle.
5238 * systemd-networkd learned IPIP and SIT tunnel support.
5240 * LSB init scripts exposing a dependency on $network will now
5241 get a dependency on network-online.target rather than simply
5242 network.target. This should bring LSB handling closer to
5243 what it was on SysV systems.
5245 * A new fsck.repair= kernel option has been added to control
5246 how fsck shall deal with unclean file systems at boot.
5248 * The (.ini) configuration file parser will now silently
5249 ignore sections whose name begins with "X-". This may be
5250 used to maintain application-specific extension sections in unit
5253 * machined gained a new API to query the IP addresses of
5254 registered containers. "machinectl status" has been updated
5255 to show these addresses in its output.
5257 * A new call sd_uid_get_display() has been added to the
5258 sd-login APIs for querying the "primary" session of a
5259 user. The "primary" session of the user is elected from the
5260 user's sessions and generally a graphical session is
5261 preferred over a text one.
5263 * A minimal systemd-resolved daemon has been added. It
5264 currently simply acts as a companion to systemd-networkd and
5265 manages resolv.conf based on per-interface DNS
5266 configuration, possibly supplied via DHCP. In the long run
5267 we hope to extend this into a local DNSSEC enabled DNS and
5270 * The systemd-networkd-wait-online tool is now enabled by
5271 default. It will delay network-online.target until a network
5272 connection has been configured. The tool primarily integrates
5273 with networkd, but will also make a best effort to make sense
5274 of network configuration performed in some other way.
5276 * Two new service options StartupCPUShares= and
5277 StartupBlockIOWeight= have been added that work similarly to
5278 CPUShares= and BlockIOWeight= however only apply during
5279 system startup. This is useful to prioritize certain services
5280 differently during bootup than during normal runtime.
5282 * hostnamed has been changed to prefer the statically
5283 configured hostname in /etc/hostname (unless set to
5284 'localhost' or empty) over any dynamic one supplied by
5285 dhcp. With this change, the rules for picking the hostname
5286 match more closely the rules of other configuration settings
5287 where the local administrator's configuration in /etc always
5288 overrides any other settings.
5290 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
5291 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
5292 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
5293 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
5294 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
5295 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
5296 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
5297 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
5298 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
5299 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
5300 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
5301 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
5302 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
5303 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
5304 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
5305 Lindskog, WaLyong Cho, Will Woods, Zbigniew
5308 — Beijing, 2014-05-28
5312 * When restoring the screen brightness at boot, stay away from
5313 the darkest setting or from the lowest 5% of the available
5314 range, depending on which is the larger value of both. This
5315 should effectively protect the user from rebooting into a
5316 black screen, should the brightness have been set to minimum
5319 * sd-login gained a new sd_machine_get_class() call to
5320 determine the class ("vm" or "container") of a machine
5321 registered with machined.
5323 * sd-login gained new calls
5324 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
5325 to query the identity of the peer of a local AF_UNIX
5326 connection. They operate similarly to their sd_pid_get_xyz()
5329 * PID 1 will now maintain a system-wide system state engine
5330 with the states "starting", "running", "degraded",
5331 "maintenance", "stopping". These states are bound to system
5332 startup, normal runtime, runtime with at least one failed
5333 service, rescue/emergency mode and system shutdown. This
5334 state is shown in the "systemctl status" output when no unit
5335 name is passed. It is useful to determine system state, in
5336 particularly when doing so for many systems or containers at
5339 * A new command "list-machines" has been added to "systemctl"
5340 that lists all local OS containers and shows their system
5341 state (see above), if systemd runs inside of them.
5343 * systemctl gained a new "-r" switch to recursively enumerate
5344 units on all local containers, when used with the
5345 "list-unit" command (which is the default one that is
5346 executed when no parameters are specified).
5348 * The GPT automatic partition discovery logic will now honour
5349 two GPT partition flags: one may be set on a partition to
5350 cause it to be mounted read-only, and the other may be set
5351 on a partition to ignore it during automatic discovery.
5353 * Two new GPT type UUIDs have been added for automatic root
5354 partition discovery, for 32-bit and 64-bit ARM. This is not
5355 particularly useful for discovering the root directory on
5356 these architectures during bare-metal boots (since UEFI is
5357 not common there), but still very useful to allow booting of
5358 ARM disk images in nspawn with the -i option.
5360 * MAC addresses of interfaces created with nspawn's
5361 --network-interface= switch will now be generated from the
5362 machine name, and thus be stable between multiple invocations
5365 * logind will now automatically remove all IPC objects owned
5366 by a user if she or he fully logs out. This makes sure that
5367 users who are logged out cannot continue to consume IPC
5368 resources. This covers SysV memory, semaphores and message
5369 queues as well as POSIX shared memory and message
5370 queues. Traditionally, SysV and POSIX IPC had no life-cycle
5371 limits. With this functionality, that is corrected. This may
5372 be turned off by using the RemoveIPC= switch of logind.conf.
5374 * The systemd-machine-id-setup and tmpfiles tools gained a
5375 --root= switch to operate on a specific root directory,
5378 * journald can now forward logged messages to the TTYs of all
5379 logged in users ("wall"). This is the default for all
5380 emergency messages now.
5382 * A new tool systemd-journal-remote has been added to stream
5383 journal log messages across the network.
5385 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
5386 controller trees are mounted into it. Note that the
5387 directories mounted beneath it are not read-only. This is a
5388 security measure and is particularly useful because glibc
5389 actually includes a search logic to pick any tmpfs it can
5390 find to implement shm_open() if /dev/shm is not available
5391 (which it might very well be in namespaced setups).
5393 * machinectl gained a new "poweroff" command to cleanly power
5394 down a local OS container.
5396 * The PrivateDevices= unit file setting will now also drop the
5397 CAP_MKNOD capability from the capability bound set, and
5398 imply DevicePolicy=closed.
5400 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
5401 comprehensively on all long-running systemd services where
5402 this is appropriate.
5404 * systemd-udevd will now run in a disassociated mount
5405 namespace. To mount directories from udev rules, make sure to
5406 pull in mount units via SYSTEMD_WANTS properties.
5408 * The kdbus support gained support for uploading policy into
5409 the kernel. sd-bus gained support for creating "monitoring"
5410 connections that can eavesdrop into all bus communication
5411 for debugging purposes.
5413 * Timestamps may now be specified in seconds since the UNIX
5414 epoch Jan 1st, 1970 by specifying "@" followed by the value
5417 * Native tcpwrap support in systemd has been removed. tcpwrap
5418 is old code, not really maintained anymore and has serious
5419 shortcomings, and better options such as firewalls
5420 exist. For setups that require tcpwrap usage, please
5421 consider invoking your socket-activated service via tcpd,
5422 like on traditional inetd.
5424 * A new system.conf configuration option
5425 DefaultTimerAccuracySec= has been added that controls the
5426 default AccuracySec= setting of .timer units.
5428 * Timer units gained a new WakeSystem= switch. If enabled,
5429 timers configured this way will cause the system to resume
5430 from system suspend (if the system supports that, which most
5433 * Timer units gained a new Persistent= switch. If enabled,
5434 timers configured this way will save to disk when they have
5435 been last triggered. This information is then used on next
5436 reboot to possible execute overdue timer events, that
5437 could not take place because the system was powered off.
5438 This enables simple anacron-like behaviour for timer units.
5440 * systemctl's "list-timers" will now also list the time a
5441 timer unit was last triggered in addition to the next time
5442 it will be triggered.
5444 * systemd-networkd will now assign predictable IPv4LL
5445 addresses to its local interfaces.
5447 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
5448 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
5449 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
5450 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
5451 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
5452 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
5453 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
5454 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
5457 — Berlin, 2014-03-25
5461 * A new unit file setting RestrictAddressFamilies= has been
5462 added to restrict which socket address families unit
5463 processes gain access to. This takes address family names
5464 like "AF_INET" or "AF_UNIX", and is useful to minimize the
5465 attack surface of services via exotic protocol stacks. This
5466 is built on seccomp system call filters.
5468 * Two new unit file settings RuntimeDirectory= and
5469 RuntimeDirectoryMode= have been added that may be used to
5470 manage a per-daemon runtime directories below /run. This is
5471 an alternative for setting up directory permissions with
5472 tmpfiles snippets, and has the advantage that the runtime
5473 directory's lifetime is bound to the daemon runtime and that
5474 the daemon starts up with an empty directory each time. This
5475 is particularly useful when writing services that drop
5476 privileges using the User= or Group= setting.
5478 * The DeviceAllow= unit setting now supports globbing for
5479 matching against device group names.
5481 * The systemd configuration file system.conf gained new
5482 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
5483 DefaultMemoryAccounting= to globally turn on/off accounting
5484 for specific resources (cgroups) for all units. These
5485 settings may still be overridden individually in each unit
5488 * systemd-gpt-auto-generator is now able to discover /srv and
5489 root partitions in addition to /home and swap partitions. It
5490 also supports LUKS-encrypted partitions now. With this in
5491 place, automatic discovery of partitions to mount following
5492 the Discoverable Partitions Specification
5493 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
5494 is now a lot more complete. This allows booting without
5495 /etc/fstab and without root= on the kernel command line on
5496 systems prepared appropriately.
5498 * systemd-nspawn gained a new --image= switch which allows
5499 booting up disk images and Linux installations on any block
5500 device that follow the Discoverable Partitions Specification
5501 (see above). This means that installations made with
5502 appropriately updated installers may now be started and
5503 deployed using container managers, completely
5504 unmodified. (We hope that libvirt-lxc will add support for
5505 this feature soon, too.)
5507 * systemd-nspawn gained a new --network-macvlan= setting to
5508 set up a private macvlan interface for the
5509 container. Similarly, systemd-networkd gained a new
5510 Kind=macvlan setting in .netdev files.
5512 * systemd-networkd now supports configuring local addresses
5515 * A new tool systemd-network-wait-online has been added to
5516 synchronously wait for network connectivity using
5519 * The sd-bus.h bus API gained a new sd_bus_track object for
5520 tracking the life-cycle of bus peers. Note that sd-bus.h is
5521 still not a public API though (unless you specify
5522 --enable-kdbus on the configure command line, which however
5523 voids your warranty and you get no API stability guarantee).
5525 * The $XDG_RUNTIME_DIR runtime directories for each user are
5526 now individual tmpfs instances, which has the benefit of
5527 introducing separate pools for each user, with individual
5528 size limits, and thus making sure that unprivileged clients
5529 can no longer negatively impact the system or other users by
5530 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
5531 RuntimeDirectorySize= has been introduced that allows
5532 controlling the default size limit for all users. It
5533 defaults to 10% of the available physical memory. This is no
5534 replacement for quotas on tmpfs though (which the kernel
5535 still does not support), as /dev/shm and /tmp are still
5536 shared resources used by both the system and unprivileged
5539 * logind will now automatically turn off automatic suspending
5540 on laptop lid close when more than one display is
5541 connected. This was previously expected to be implemented
5542 individually in desktop environments (such as GNOME),
5543 however has been added to logind now, in order to fix a
5544 boot-time race where a desktop environment might not have
5545 been started yet and thus not been able to take an inhibitor
5546 lock at the time where logind already suspends the system
5547 due to a closed lid.
5549 * logind will now wait at least 30s after each system
5550 suspend/resume cycle, and 3min after system boot before
5551 suspending the system due to a closed laptop lid. This
5552 should give USB docking stations and similar enough time to
5553 be probed and configured after system resume and boot in
5554 order to then act as suspend blocker.
5556 * systemd-run gained a new --property= setting which allows
5557 initialization of resource control properties (and others)
5558 for the created scope or service unit. Example: "systemd-run
5559 --property=BlockIOWeight=10 updatedb" may be used to run
5560 updatedb at a low block IO scheduling weight.
5562 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
5563 now also work in --scope mode.
5565 * When systemd is compiled with kdbus support, basic support
5566 for enforced policies is now in place. (Note that enabling
5567 kdbus still voids your warranty and no API compatibility
5570 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
5571 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5572 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
5573 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
5574 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
5575 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
5576 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
5577 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
5578 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
5579 Zbigniew Jędrzejewski-Szmek
5581 — Berlin, 2014-03-12
5585 * systemd will now relabel /dev after loading the SMACK policy
5586 according to SMACK rules.
5588 * A new unit file option AppArmorProfile= has been added to
5589 set the AppArmor profile for the processes of a unit.
5591 * A new condition check ConditionArchitecture= has been added
5592 to conditionalize units based on the system architecture, as
5593 reported by uname()'s "machine" field.
5595 * systemd-networkd now supports matching on the system
5596 virtualization, architecture, kernel command line, host name
5599 * logind is now a lot more aggressive when suspending the
5600 machine due to a closed laptop lid. Instead of acting only
5601 on the lid close action, it will continuously watch the lid
5602 status and act on it. This is useful for laptops where the
5603 power button is on the outside of the chassis so that it can
5604 be reached without opening the lid (such as the Lenovo
5605 Yoga). On those machines, logind will now immediately
5606 re-suspend the machine if the power button has been
5607 accidentally pressed while the laptop was suspended and in a
5608 backpack or similar.
5610 * logind will now watch SW_DOCK switches and inhibit reaction
5611 to the lid switch if it is pressed. This means that logind
5612 will not suspend the machine anymore if the lid is closed
5613 and the system is docked, if the laptop supports SW_DOCK
5614 notifications via the input layer. Note that ACPI docking
5615 stations do not generate this currently. Also note that this
5616 logic is usually not fully sufficient and Desktop
5617 Environments should take a lid switch inhibitor lock when an
5618 external display is connected, as systemd will not watch
5621 * nspawn will now make use of the devices cgroup controller by
5622 default, and only permit creation of and access to the usual
5623 API device nodes like /dev/null or /dev/random, as well as
5624 access to (but not creation of) the pty devices.
5626 * We will now ship a default .network file for
5627 systemd-networkd that automatically configures DHCP for
5628 network interfaces created by nspawn's --network-veth or
5629 --network-bridge= switches.
5631 * systemd will now understand the usual M, K, G, T suffixes
5632 according to SI conventions (i.e. to the base 1000) when
5633 referring to throughput and hardware metrics. It will stay
5634 with IEC conventions (i.e. to the base 1024) for software
5635 metrics, according to what is customary according to
5636 Wikipedia. We explicitly document which base applies for
5637 each configuration option.
5639 * The DeviceAllow= setting in unit files now supports a syntax
5640 to whitelist an entire group of devices node majors at once,
5641 based on the /proc/devices listing. For example, with the
5642 string "char-pts", it is now possible to whitelist all
5643 current and future pseudo-TTYs at once.
5645 * sd-event learned a new "post" event source. Event sources of
5646 this type are triggered by the dispatching of any event
5647 source of a type that is not "post". This is useful for
5648 implementing clean-up and check event sources that are
5649 triggered by other work being done in the program.
5651 * systemd-networkd is no longer statically enabled, but uses
5652 the usual [Install] sections so that it can be
5653 enabled/disabled using systemctl. It still is enabled by
5656 * When creating a veth interface pair with systemd-nspawn, the
5657 host side will now be prefixed with "vb-" if
5658 --network-bridge= is used, and with "ve-" if --network-veth
5659 is used. This way, it is easy to distinguish these cases on
5660 the host, for example to apply different configuration to
5661 them with systemd-networkd.
5663 * The compatibility libraries for libsystemd-journal.so,
5664 libsystem-id128.so, libsystemd-login.so and
5665 libsystemd-daemon.so do not make use of IFUNC
5666 anymore. Instead, we now build libsystemd.so multiple times
5667 under these alternative names. This means that the footprint
5668 is drastically increased, but given that these are
5669 transitional compatibility libraries, this should not matter
5670 much. This change has been made necessary to support the ARM
5671 platform for these compatibility libraries, as the ARM
5672 toolchain is not really at the same level as the toolchain
5673 for other architectures like x86 and does not support
5674 IFUNC. Please make sure to use --enable-compat-libs only
5675 during a transitional period!
5677 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
5678 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5679 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
5680 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
5681 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
5682 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
5683 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
5684 Zbigniew Jędrzejewski-Szmek
5686 — Berlin, 2014-02-24
5690 * A new component "systemd-networkd" has been added that can
5691 be used to configure local network interfaces statically or
5692 via DHCP. It is capable of bringing up bridges, VLANs, and
5693 bonding. Currently, no hook-ups for interactive network
5694 configuration are provided. Use this for your initrd,
5695 container, embedded, or server setup if you need a simple,
5696 yet powerful, network configuration solution. This
5697 configuration subsystem is quite nifty, as it allows wildcard
5698 hotplug matching in interfaces. For example, with a single
5699 configuration snippet, you can configure that all Ethernet
5700 interfaces showing up are automatically added to a bridge,
5701 or similar. It supports link-sensing and more.
5703 * A new tool "systemd-socket-proxyd" has been added which can
5704 act as a bidirectional proxy for TCP sockets. This is
5705 useful for adding socket activation support to services that
5706 do not actually support socket activation, including virtual
5707 machines and the like.
5709 * Add a new tool to save/restore rfkill state on
5712 * Save/restore state of keyboard backlights in addition to
5713 display backlights on shutdown/boot.
5715 * udev learned a new SECLABEL{} construct to label device
5716 nodes with a specific security label when they appear. For
5717 now, only SECLABEL{selinux} is supported, but the syntax is
5718 prepared for additional security frameworks.
5720 * udev gained a new scheme to configure link-level attributes
5721 from files in /etc/systemd/network/*.link. These files can
5722 match against MAC address, device path, driver name and type,
5723 and will apply attributes like the naming policy, link speed,
5724 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
5725 address assignment policy (randomized, ...).
5727 * The configuration of network interface naming rules for
5728 "permanent interface names" has changed: a new NamePolicy=
5729 setting in the [Link] section of .link files determines the
5730 priority of possible naming schemes (onboard, slot, MAC,
5731 path). The default value of this setting is determined by
5732 /usr/lib/net/links/99-default.link. Old
5733 80-net-name-slot.rules udev configuration file has been
5734 removed, so local configuration overriding this file should
5735 be adapted to override 99-default.link instead.
5737 * When the User= switch is used in a unit file, also
5738 initialize $SHELL= based on the user database entry.
5740 * systemd no longer depends on libdbus. All communication is
5741 now done with sd-bus, systemd's low-level bus library
5744 * kdbus support has been added to PID 1 itself. When kdbus is
5745 enabled, this causes PID 1 to set up the system bus and
5746 enable support for a new ".busname" unit type that
5747 encapsulates bus name activation on kdbus. It works a little
5748 bit like ".socket" units, except for bus names. A new
5749 generator has been added that converts classic dbus1 service
5750 activation files automatically into native systemd .busname
5753 * sd-bus: add a light-weight vtable implementation that allows
5754 defining objects on the bus with a simple static const
5755 vtable array of its methods, signals and properties.
5757 * systemd will not generate or install static dbus
5758 introspection data anymore to /usr/share/dbus-1/interfaces,
5759 as the precise format of these files is unclear, and
5760 nothing makes use of it.
5762 * A proxy daemon is now provided to proxy clients connecting
5763 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
5764 compatibility with classic D-Bus.
5766 * A bus driver implementation has been added that supports the
5767 classic D-Bus bus driver calls on kdbus, also for
5768 compatibility purposes.
5770 * A new API "sd-event.h" has been added that implements a
5771 minimal event loop API built around epoll. It provides a
5772 couple of features that direct epoll usage is lacking:
5773 prioritization of events, scales to large numbers of timer
5774 events, per-event timer slack (accuracy), system-wide
5775 coalescing of timer events, exit handlers, watchdog
5776 supervision support using systemd's sd_notify() API, child
5779 * A new API "sd-rntl.h" has been added that provides an API
5780 around the route netlink interface of the kernel, similar in
5781 style to "sd-bus.h".
5783 * A new API "sd-dhcp-client.h" has been added that provides a
5784 small DHCPv4 client-side implementation. This is used by
5787 * There is a new kernel command line option
5788 "systemd.restore_state=0|1". When set to "0", none of the
5789 systemd tools will restore saved runtime state to hardware
5790 devices. More specifically, the rfkill and backlight states
5793 * The FsckPassNo= compatibility option in mount/service units
5794 has been removed. The fstab generator will now add the
5795 necessary dependencies automatically, and does not require
5796 PID1's support for that anymore.
5798 * journalctl gained a new switch, --list-boots, that lists
5799 recent boots with their times and boot IDs.
5801 * The various tools like systemctl, loginctl, timedatectl,
5802 busctl, systemd-run, ... have gained a new switch "-M" to
5803 connect to a specific, local OS container (as direct
5804 connection, without requiring SSH). This works on any
5805 container that is registered with machined, such as those
5806 created by libvirt-lxc or nspawn.
5808 * systemd-run and systemd-analyze also gained support for "-H"
5809 to connect to remote hosts via SSH. This is particularly
5810 useful for systemd-run because it enables queuing of jobs
5811 onto remote systems.
5813 * machinectl gained a new command "login" to open a getty
5814 login in any local container. This works with any container
5815 that is registered with machined (such as those created by
5816 libvirt-lxc or nspawn), and which runs systemd inside.
5818 * machinectl gained a new "reboot" command that may be used to
5819 trigger a reboot on a specific container that is registered
5820 with machined. This works on any container that runs an init
5821 system of some kind.
5823 * systemctl gained a new "list-timers" command to print a nice
5824 listing of installed timer units with the times they elapse
5827 * Alternative reboot() parameters may now be specified on the
5828 "systemctl reboot" command line and are passed to the
5829 reboot() system call.
5831 * systemctl gained a new --job-mode= switch to configure the
5832 mode to queue a job with. This is a more generic version of
5833 --fail, --irreversible, and --ignore-dependencies, which are
5834 still available but not advertised anymore.
5836 * /etc/systemd/system.conf gained new settings to configure
5837 various default timeouts of units, as well as the default
5838 start limit interval and burst. These may still be overridden
5841 * PID1 will now export on the bus profile data of the security
5842 policy upload process (such as the SELinux policy upload to
5845 * journald: when forwarding logs to the console, include
5846 timestamps (following the setting in
5847 /sys/module/printk/parameters/time).
5849 * OnCalendar= in timer units now understands the special
5850 strings "yearly" and "annually". (Both are equivalent)
5852 * The accuracy of timer units is now configurable with the new
5853 AccuracySec= setting. It defaults to 1min.
5855 * A new dependency type JoinsNamespaceOf= has been added that
5856 allows running two services within the same /tmp and network
5857 namespace, if PrivateNetwork= or PrivateTmp= are used.
5859 * A new command "cat" has been added to systemctl. It outputs
5860 the original unit file of a unit, and concatenates the
5861 contents of additional "drop-in" unit file snippets, so that
5862 the full configuration is shown.
5864 * systemctl now supports globbing on the various "list-xyz"
5865 commands, like "list-units" or "list-sockets", as well as on
5866 those commands which take multiple unit names.
5868 * journalctl's --unit= switch gained support for globbing.
5870 * All systemd daemons now make use of the watchdog logic so
5871 that systemd automatically notices when they hang.
5873 * If the $container_ttys environment variable is set,
5874 getty-generator will automatically spawn a getty for each
5875 listed tty. This is useful for container managers to request
5876 login gettys to be spawned on as many ttys as needed.
5878 * %h, %s, %U specifier support is not available anymore when
5879 used in unit files for PID 1. This is because NSS calls are
5880 not safe from PID 1. They stay available for --user
5881 instances of systemd, and as special case for the root user.
5883 * loginctl gained a new "--no-legend" switch to turn off output
5886 * The "sd-login.h" API gained three new calls:
5887 sd_session_is_remote(), sd_session_get_remote_user(),
5888 sd_session_get_remote_host() to query information about
5891 * The udev hardware database now also carries vendor/product
5892 information of SDIO devices.
5894 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
5895 determine whether watchdog notifications are requested by
5898 * Socket-activated per-connection services now include a
5899 short description of the connection parameters in the
5902 * tmpfiles gained a new "--boot" option. When this is not used,
5903 only lines where the command character is not suffixed with
5904 "!" are executed. When this option is specified, those
5905 options are executed too. This partitions tmpfiles
5906 directives into those that can be safely executed at any
5907 time, and those which should be run only at boot (for
5908 example, a line that creates /run/nologin).
5910 * A new API "sd-resolve.h" has been added which provides a simple
5911 asynchronous wrapper around glibc NSS host name resolution
5912 calls, such as getaddrinfo(). In contrast to glibc's
5913 getaddrinfo_a(), it does not use signals. In contrast to most
5914 other asynchronous name resolution libraries, this one does
5915 not reimplement DNS, but reuses NSS, so that alternate
5916 host name resolution systems continue to work, such as mDNS,
5917 LDAP, etc. This API is based on libasyncns, but it has been
5918 cleaned up for inclusion in systemd.
5920 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
5921 "sd-daemon.h" are no longer found in individual libraries
5922 libsystemd-journal.so, libsystemd-login.so,
5923 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
5924 merged them into a single library, libsystemd.so, which
5925 provides all symbols. The reason for this is cyclic
5926 dependencies, as these libraries tend to use each other's
5927 symbols. So far, we have managed to workaround that by linking
5928 a copy of a good part of our code into each of these
5929 libraries again and again, which, however, makes certain
5930 things hard to do, like sharing static variables. Also, it
5931 substantially increases footprint. With this change, there
5932 is only one library for the basic APIs systemd
5933 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
5934 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
5935 library as well, however are subject to the --enable-kdbus
5936 switch (see below). Note that "sd-dhcp-client.h" is not part
5937 of this library (this is because it only consumes, never
5938 provides, services of/to other APIs). To make the transition
5939 easy from the separate libraries to the unified one, we
5940 provide the --enable-compat-libs compile-time switch which
5941 will generate stub libraries that are compatible with the
5942 old ones but redirect all calls to the new one.
5944 * All of the kdbus logic and the new APIs "sd-bus.h",
5945 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
5946 and "sd-utf8.h" are compile-time optional via the
5947 "--enable-kdbus" switch, and they are not compiled in by
5948 default. To make use of kdbus, you have to explicitly enable
5949 the switch. Note however, that neither the kernel nor the
5950 userspace API for all of this is considered stable yet. We
5951 want to maintain the freedom to still change the APIs for
5952 now. By specifying this build-time switch, you acknowledge
5953 that you are aware of the instability of the current
5956 * Also, note that while kdbus is pretty much complete,
5957 it lacks one thing: proper policy support. This means you
5958 can build a fully working system with all features; however,
5959 it will be highly insecure. Policy support will be added in
5960 one of the next releases, at the same time that we will
5961 declare the APIs stable.
5963 * When the kernel command line argument "kdbus" is specified,
5964 systemd will automatically load the kdbus.ko kernel module. At
5965 this stage of development, it is only useful for testing kdbus
5966 and should not be used in production. Note: if "--enable-kdbus"
5967 is specified, and the kdbus.ko kernel module is available, and
5968 "kdbus" is added to the kernel command line, the entire system
5969 runs with kdbus instead of dbus-daemon, with the above mentioned
5970 problem of missing the system policy enforcement. Also a future
5971 version of kdbus.ko or a newer systemd will not be compatible with
5972 each other, and will unlikely be able to boot the machine if only
5973 one of them is updated.
5975 * systemctl gained a new "import-environment" command which
5976 uploads the caller's environment (or parts thereof) into the
5977 service manager so that it is inherited by services started
5978 by the manager. This is useful to upload variables like
5979 $DISPLAY into the user service manager.
5981 * A new PrivateDevices= switch has been added to service units
5982 which allows running a service with a namespaced /dev
5983 directory that does not contain any device nodes for
5984 physical devices. More specifically, it only includes devices
5985 such as /dev/null, /dev/urandom, and /dev/zero which are API
5988 * logind has been extended to support behaviour like VT
5989 switching on seats that do not support a VT. This makes
5990 multi-session available on seats that are not the first seat
5991 (seat0), and on systems where kernel support for VTs has
5992 been disabled at compile-time.
5994 * If a process holds a delay lock for system sleep or shutdown
5995 and fails to release it in time, we will now log its
5996 identity. This makes it easier to identify processes that
5997 cause slow suspends or power-offs.
5999 * When parsing /etc/crypttab, support for a new key-slot=
6000 option as supported by Debian is added. It allows indicating
6001 which LUKS slot to use on disk, speeding up key loading.
6003 * The sd_journal_sendv() API call has been checked and
6004 officially declared to be async-signal-safe so that it may
6005 be invoked from signal handlers for logging purposes.
6007 * Boot-time status output is now enabled automatically after a
6008 short timeout if boot does not progress, in order to give
6009 the user an indication what she or he is waiting for.
6011 * The boot-time output has been improved to show how much time
6012 remains until jobs expire.
6014 * The KillMode= switch in service units gained a new possible
6015 value "mixed". If set, and the unit is shut down, then the
6016 initial SIGTERM signal is sent only to the main daemon
6017 process, while the following SIGKILL signal is sent to
6018 all remaining processes of the service.
6020 * When a scope unit is registered, a new property "Controller"
6021 may be set. If set to a valid bus name, systemd will send a
6022 RequestStop() signal to this name when it would like to shut
6023 down the scope. This may be used to hook manager logic into
6024 the shutdown logic of scope units. Also, scope units may now
6025 be put in a special "abandoned" state, in which case the
6026 manager process which created them takes no further
6027 responsibilities for it.
6029 * When reading unit files, systemd will now verify
6030 the access mode of these files, and warn about certain
6031 suspicious combinations. This has been added to make it
6032 easier to track down packaging bugs where unit files are
6033 marked executable or world-writable.
6035 * systemd-nspawn gained a new "--setenv=" switch to set
6036 container-wide environment variables. The similar option in
6037 systemd-activate was renamed from "--environment=" to
6038 "--setenv=" for consistency.
6040 * systemd-nspawn has been updated to create a new kdbus domain
6041 for each container that is invoked, thus allowing each
6042 container to have its own set of system and user buses,
6043 independent of the host.
6045 * systemd-nspawn gained a new --drop-capability= switch to run
6046 the container with less capabilities than the default. Both
6047 --drop-capability= and --capability= now take the special
6048 string "all" for dropping or keeping all capabilities.
6050 * systemd-nspawn gained new switches for executing containers
6051 with specific SELinux labels set.
6053 * systemd-nspawn gained a new --quiet switch to not generate
6054 any additional output but the container's own console
6057 * systemd-nspawn gained a new --share-system switch to run a
6058 container without PID namespacing enabled.
6060 * systemd-nspawn gained a new --register= switch to control
6061 whether the container is registered with systemd-machined or
6062 not. This is useful for containers that do not run full
6063 OS images, but only specific apps.
6065 * systemd-nspawn gained a new --keep-unit which may be used
6066 when invoked as the only program from a service unit, and
6067 results in registration of the unit service itself in
6068 systemd-machined, instead of a newly opened scope unit.
6070 * systemd-nspawn gained a new --network-interface= switch for
6071 moving arbitrary interfaces to the container. The new
6072 --network-veth switch creates a virtual Ethernet connection
6073 between host and container. The new --network-bridge=
6074 switch then allows assigning the host side of this virtual
6075 Ethernet connection to a bridge device.
6077 * systemd-nspawn gained a new --personality= switch for
6078 setting the kernel personality for the container. This is
6079 useful when running a 32-bit container on a 64-bit host. A
6080 similar option Personality= is now also available for service
6083 * logind will now also track a "Desktop" identifier for each
6084 session which encodes the desktop environment of it. This is
6085 useful for desktop environments that want to identify
6086 multiple running sessions of itself easily.
6088 * A new SELinuxContext= setting for service units has been
6089 added that allows setting a specific SELinux execution
6090 context for a service.
6092 * Most systemd client tools will now honour $SYSTEMD_LESS for
6093 settings of the "less" pager. By default, these tools will
6094 override $LESS to allow certain operations to work, such as
6095 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
6096 influence this logic.
6098 * systemd's "seccomp" hook-up has been changed to make use of
6099 the libseccomp library instead of using its own
6100 implementation. This has benefits for portability among
6103 * For usage together with SystemCallFilter=, a new
6104 SystemCallErrorNumber= setting has been introduced that
6105 allows configuration of a system error number to be returned
6106 on filtered system calls, instead of immediately killing the
6107 process. Also, SystemCallArchitectures= has been added to
6108 limit access to system calls of a particular architecture
6109 (in order to turn off support for unused secondary
6110 architectures). There is also a global
6111 SystemCallArchitectures= setting in system.conf now to turn
6112 off support for non-native system calls system-wide.
6114 * systemd requires a kernel with a working name_to_handle_at(),
6115 please see the kernel config requirements in the README file.
6117 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
6118 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
6119 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
6120 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
6121 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
6122 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
6123 Elia Pinto, Florian Weimer, George McCollister, Goffredo
6124 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
6125 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
6126 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
6127 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
6128 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
6129 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
6130 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
6131 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
6132 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
6133 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
6134 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
6135 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
6136 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
6137 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
6138 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
6139 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
6140 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
6142 — Berlin, 2014-02-20
6146 * logind has gained support for facilitating privileged input
6147 and drm device access for unprivileged clients. This work is
6148 useful to allow Wayland display servers (and similar
6149 programs, such as kmscon) to run under the user's ID and
6150 access input and drm devices which are normally
6151 protected. When this is used (and the kernel is new enough)
6152 logind will "mute" IO on the file descriptors passed to
6153 Wayland as long as it is in the background and "unmute" it
6154 if it returns into the foreground. This allows secure
6155 session switching without allowing background sessions to
6156 eavesdrop on input and display data. This also introduces
6157 session switching support if VT support is turned off in the
6158 kernel, and on seats that are not seat0.
6160 * A new kernel command line option luks.options= is understood
6161 now which allows specifying LUKS options for usage for LUKS
6162 encrypted partitions specified with luks.uuid=.
6164 * tmpfiles.d(5) snippets may now use specifier expansion in
6165 path names. More specifically %m, %b, %H, %v, are now
6166 replaced by the local machine id, boot id, hostname, and
6167 kernel version number.
6169 * A new tmpfiles.d(5) command "m" has been introduced which
6170 may be used to change the owner/group/access mode of a file
6171 or directory if it exists, but do nothing if it does not.
6173 * This release removes high-level support for the
6174 MemorySoftLimit= cgroup setting. The underlying kernel
6175 cgroup attribute memory.soft_limit= is currently badly
6176 designed and likely to be removed from the kernel API in its
6177 current form, hence we should not expose it for now.
6179 * The memory.use_hierarchy cgroup attribute is now enabled for
6180 all cgroups systemd creates in the memory cgroup
6181 hierarchy. This option is likely to be come the built-in
6182 default in the kernel anyway, and the non-hierarchical mode
6183 never made much sense in the intrinsically hierarchical
6186 * A new field _SYSTEMD_SLICE= is logged along with all journal
6187 messages containing the slice a message was generated
6188 from. This is useful to allow easy per-customer filtering of
6189 logs among other things.
6191 * systemd-journald will no longer adjust the group of journal
6192 files it creates to the "systemd-journal" group. Instead we
6193 rely on the journal directory to be owned by the
6194 "systemd-journal" group, and its setgid bit set, so that the
6195 kernel file system layer will automatically enforce that
6196 journal files inherit this group assignment. The reason for
6197 this change is that we cannot allow NSS look-ups from
6198 journald which would be necessary to resolve
6199 "systemd-journal" to a numeric GID, because this might
6200 create deadlocks if NSS involves synchronous queries to
6201 other daemons (such as nscd, or sssd) which in turn are
6202 logging clients of journald and might block on it, which
6203 would then dead lock. A tmpfiles.d(5) snippet included in
6204 systemd will make sure the setgid bit and group are
6205 properly set on the journal directory if it exists on every
6206 boot. However, we recommend adjusting it manually after
6207 upgrades too (or from RPM scriptlets), so that the change is
6208 not delayed until next reboot.
6210 * Backlight and random seed files in /var/lib/ have moved into
6211 the /var/lib/systemd/ directory, in order to centralize all
6212 systemd generated files in one directory.
6214 * Boot time performance measurements (as displayed by
6215 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
6216 performance information if that's available to determine how
6217 much time BIOS and boot loader initialization required. With
6218 a sufficiently new BIOS you hence no longer need to boot
6219 with Gummiboot to get access to such information.
6221 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
6222 Cristian Rodríguez, Dave Reisner, David Herrmann, David
6223 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
6224 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
6225 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
6226 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
6227 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6229 — Berlin, 2013-10-02
6233 * The Restart= option for services now understands a new
6234 on-watchdog setting, which will restart the service
6235 automatically if the service stops sending out watchdog keep
6236 alive messages (as configured with WatchdogSec=).
6238 * The getty generator (which is responsible for bringing up a
6239 getty on configured serial consoles) will no longer only
6240 start a getty on the primary kernel console but on all
6241 others, too. This makes the order in which console= is
6242 specified on the kernel command line less important.
6244 * libsystemd-logind gained a new sd_session_get_vt() call to
6245 retrieve the VT number of a session.
6247 * If the option "tries=0" is set for an entry of /etc/crypttab
6248 its passphrase is queried indefinitely instead of any
6249 maximum number of tries.
6251 * If a service with a configure PID file terminates its PID
6252 file will now be removed automatically if it still exists
6253 afterwards. This should put an end to stale PID files.
6255 * systemd-run will now also take relative binary path names
6256 for execution and no longer insists on absolute paths.
6258 * InaccessibleDirectories= and ReadOnlyDirectories= now take
6259 paths that are optionally prefixed with "-" to indicate that
6260 it should not be considered a failure if they do not exist.
6262 * journalctl -o (and similar commands) now understands a new
6263 output mode "short-precise", it is similar to "short" but
6264 shows timestamps with usec accuracy.
6266 * The option "discard" (as known from Debian) is now
6267 synonymous to "allow-discards" in /etc/crypttab. In fact,
6268 "discard" is preferred now (since it is easier to remember
6271 * Some licensing clean-ups were made, so that more code is now
6272 LGPL-2.1 licensed than before.
6274 * A minimal tool to save/restore the display backlight
6275 brightness across reboots has been added. It will store the
6276 backlight setting as late as possible at shutdown, and
6277 restore it as early as possible during reboot.
6279 * A logic to automatically discover and enable home and swap
6280 partitions on GPT disks has been added. With this in place
6281 /etc/fstab becomes optional for many setups as systemd can
6282 discover certain partitions located on the root disk
6283 automatically. Home partitions are recognized under their
6284 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
6285 partitions are recognized under their GPT type ID
6286 0657fd6da4ab43c484e50933c84b4f4f.
6288 * systemd will no longer pass any environment from the kernel
6289 or initrd to system services. If you want to set an
6290 environment for all services, do so via the kernel command
6291 line systemd.setenv= assignment.
6293 * The systemd-sysctl tool no longer natively reads the file
6294 /etc/sysctl.conf. If desired, the file should be symlinked
6295 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
6296 legacy support by a symlink rather than built-in code, it
6297 also makes the otherwise hidden order of application of the
6298 different files visible. (Note that this partly reverts to a
6299 pre-198 application order of sysctl knobs!)
6301 * The "systemctl set-log-level" and "systemctl dump" commands
6302 have been moved to systemd-analyze.
6304 * systemd-run learned the new --remain-after-exit switch,
6305 which causes the scope unit not to be cleaned up
6306 automatically after the process terminated.
6308 * tmpfiles learned a new --exclude-prefix= switch to exclude
6309 certain paths from operation.
6311 * journald will now automatically flush all messages to disk
6312 as soon as a message at the log level CRIT, ALERT or EMERG
6315 Contributions from: Andrew Cook, Brandon Philips, Christian
6316 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
6317 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
6318 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
6319 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
6320 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
6321 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
6322 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
6323 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
6324 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
6325 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
6326 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
6327 William Giokas, Zbigniew Jędrzejewski-Szmek
6329 — Berlin, 2013-09-13
6333 * The documentation has been updated to cover the various new
6334 concepts introduced with 205.
6336 * Unit files now understand the new %v specifier which
6337 resolves to the kernel version string as returned by "uname
6340 * systemctl now supports filtering the unit list output by
6341 load state, active state and sub state, using the new
6344 * "systemctl status" will now show the results of the
6345 condition checks (like ConditionPathExists= and similar) of
6346 the last start attempts of the unit. They are also logged to
6349 * "journalctl -b" may now be used to look for boot output of a
6350 specific boot. Try "journalctl -b -1" for the previous boot,
6351 but the syntax is substantially more powerful.
6353 * "journalctl --show-cursor" has been added which prints the
6354 cursor string the last shown log line. This may then be used
6355 with the new "journalctl --after-cursor=" switch to continue
6356 browsing logs from that point on.
6358 * "journalctl --force" may now be used to force regeneration
6361 * Creation of "dead" device nodes has been moved from udev
6362 into kmod and tmpfiles. Previously, udev would read the kmod
6363 databases to pre-generate dead device nodes based on meta
6364 information contained in kernel modules, so that these would
6365 be auto-loaded on access rather then at boot. As this
6366 does not really have much to do with the exposing actual
6367 kernel devices to userspace this has always been slightly
6368 alien in the udev codebase. Following the new scheme kmod
6369 will now generate a runtime snippet for tmpfiles from the
6370 module meta information and it now is tmpfiles' job to the
6371 create the nodes. This also allows overriding access and
6372 other parameters for the nodes using the usual tmpfiles
6373 facilities. As side effect this allows us to remove the
6374 CAP_SYS_MKNOD capability bit from udevd entirely.
6376 * logind's device ACLs may now be applied to these "dead"
6377 devices nodes too, thus finally allowing managed access to
6378 devices such as /dev/snd/sequencer without loading the
6379 backing module right-away.
6381 * A new RPM macro has been added that may be used to apply
6382 tmpfiles configuration during package installation.
6384 * systemd-detect-virt and ConditionVirtualization= now can
6385 detect User-Mode-Linux machines (UML).
6387 * journald will now implicitly log the effective capabilities
6388 set of processes in the message metadata.
6390 * systemd-cryptsetup has gained support for TrueCrypt volumes.
6392 * The initrd interface has been simplified (more specifically,
6393 support for passing performance data via environment
6394 variables and fsck results via files in /run has been
6395 removed). These features were non-essential, and are
6396 nowadays available in a much nicer way by having systemd in
6397 the initrd serialize its state and have the hosts systemd
6398 deserialize it again.
6400 * The udev "keymap" data files and tools to apply keyboard
6401 specific mappings of scan to key codes, and force-release
6402 scan code lists have been entirely replaced by a udev
6403 "keyboard" builtin and a hwdb data file.
6405 * systemd will now honour the kernel's "quiet" command line
6406 argument also during late shutdown, resulting in a
6407 completely silent shutdown when used.
6409 * There's now an option to control the SO_REUSEPORT socket
6410 option in .socket units.
6412 * Instance units will now automatically get a per-template
6413 subslice of system.slice unless something else is explicitly
6414 configured. For example, instances of sshd@.service will now
6415 implicitly be placed in system-sshd.slice rather than
6416 system.slice as before.
6418 * Test coverage support may now be enabled at build time.
6420 Contributions from: Dave Reisner, Frederic Crozat, Harald
6421 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
6422 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
6423 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
6424 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
6425 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
6426 Giokas, Zbigniew Jędrzejewski-Szmek
6428 — Berlin, 2013-07-23
6432 * Two new unit types have been introduced:
6434 Scope units are very similar to service units, however, are
6435 created out of pre-existing processes — instead of PID 1
6436 forking off the processes. By using scope units it is
6437 possible for system services and applications to group their
6438 own child processes (worker processes) in a powerful way
6439 which then maybe used to organize them, or kill them
6440 together, or apply resource limits on them.
6442 Slice units may be used to partition system resources in an
6443 hierarchical fashion and then assign other units to them. By
6444 default there are now three slices: system.slice (for all
6445 system services), user.slice (for all user sessions),
6446 machine.slice (for VMs and containers).
6448 Slices and scopes have been introduced primarily in
6449 context of the work to move cgroup handling to a
6450 single-writer scheme, where only PID 1
6451 creates/removes/manages cgroups.
6453 * There's a new concept of "transient" units. In contrast to
6454 normal units these units are created via an API at runtime,
6455 not from configuration from disk. More specifically this
6456 means it is now possible to run arbitrary programs as
6457 independent services, with all execution parameters passed
6458 in via bus APIs rather than read from disk. Transient units
6459 make systemd substantially more dynamic then it ever was,
6460 and useful as a general batch manager.
6462 * logind has been updated to make use of scope and slice units
6463 for managing user sessions. As a user logs in he will get
6464 his own private slice unit, to which all sessions are added
6465 as scope units. We also added support for automatically
6466 adding an instance of user@.service for the user into the
6467 slice. Effectively logind will no longer create cgroup
6468 hierarchies on its own now, it will defer entirely to PID 1
6469 for this by means of scope, service and slice units. Since
6470 user sessions this way become entities managed by PID 1
6471 the output of "systemctl" is now a lot more comprehensive.
6473 * A new mini-daemon "systemd-machined" has been added which
6474 may be used by virtualization managers to register local
6475 VMs/containers. nspawn has been updated accordingly, and
6476 libvirt will be updated shortly. machined will collect a bit
6477 of meta information about the VMs/containers, and assign
6478 them their own scope unit (see above). The collected
6479 meta-data is then made available via the "machinectl" tool,
6480 and exposed in "ps" and similar tools. machined/machinectl
6481 is compile-time optional.
6483 * As discussed earlier, the low-level cgroup configuration
6484 options ControlGroup=, ControlGroupModify=,
6485 ControlGroupPersistent=, ControlGroupAttribute= have been
6486 removed. Please use high-level attribute settings instead as
6487 well as slice units.
6489 * A new bus call SetUnitProperties() has been added to alter
6490 various runtime parameters of a unit. This is primarily
6491 useful to alter cgroup parameters dynamically in a nice way,
6492 but will be extended later on to make more properties
6493 modifiable at runtime. systemctl gained a new set-properties
6494 command that wraps this call.
6496 * A new tool "systemd-run" has been added which can be used to
6497 run arbitrary command lines as transient services or scopes,
6498 while configuring a number of settings via the command
6499 line. This tool is currently very basic, however already
6500 very useful. We plan to extend this tool to even allow
6501 queuing of execution jobs with time triggers from the
6502 command line, similar in fashion to "at".
6504 * nspawn will now inform the user explicitly that kernels with
6505 audit enabled break containers, and suggest the user to turn
6508 * Support for detecting the IMA and AppArmor security
6509 frameworks with ConditionSecurity= has been added.
6511 * journalctl gained a new "-k" switch for showing only kernel
6512 messages, mimicking dmesg output; in addition to "--user"
6513 and "--system" switches for showing only user's own logs
6516 * systemd-delta can now show information about drop-in
6517 snippets extending unit files.
6519 * libsystemd-bus has been substantially updated but is still
6520 not available as public API.
6522 * systemd will now look for the "debug" argument on the kernel
6523 command line and enable debug logging, similar to what
6524 "systemd.log_level=debug" already did before.
6526 * "systemctl set-default", "systemctl get-default" has been
6527 added to configure the default.target symlink, which
6528 controls what to boot into by default.
6530 * "systemctl set-log-level" has been added as a convenient
6531 way to raise and lower systemd logging threshold.
6533 * "systemd-analyze plot" will now show the time the various
6534 generators needed for execution, as well as information
6535 about the unit file loading.
6537 * libsystemd-journal gained a new sd_journal_open_files() call
6538 for opening specific journal files. journactl also gained a
6539 new switch to expose this new functionality. Previously we
6540 only supported opening all files from a directory, or all
6541 files from the system, as opening individual files only is
6542 racy due to journal file rotation.
6544 * systemd gained the new DefaultEnvironment= setting in
6545 /etc/systemd/system.conf to set environment variables for
6548 * If a privileged process logs a journal message with the
6549 OBJECT_PID= field set, then journald will automatically
6550 augment this with additional OBJECT_UID=, OBJECT_GID=,
6551 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
6552 system services want to log events about specific client
6553 processes. journactl/systemctl has been updated to make use
6554 of this information if all log messages regarding a specific
6557 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
6558 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
6559 Reisner, David Coppa, David King, David Strauss, Eelco
6560 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
6561 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
6562 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
6563 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
6564 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
6565 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
6566 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
6567 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
6568 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
6569 Łukasz Stelmach, 장동준
6573 * The Python bindings gained some minimal support for the APIs
6574 exposed by libsystemd-logind.
6576 * ConditionSecurity= gained support for detecting SMACK. Since
6577 this condition already supports SELinux and AppArmor we only
6578 miss IMA for this. Patches welcome!
6580 Contributions from: Karol Lewandowski, Lennart Poettering,
6581 Zbigniew Jędrzejewski-Szmek
6585 * systemd-nspawn will now create /etc/resolv.conf if
6586 necessary, before bind-mounting the host's file onto it.
6588 * systemd-nspawn will now store meta information about a
6589 container on the container's cgroup as extended attribute
6590 fields, including the root directory.
6592 * The cgroup hierarchy has been reworked in many ways. All
6593 objects any of the components systemd creates in the cgroup
6594 tree are now suffixed. More specifically, user sessions are
6595 now placed in cgroups suffixed with ".session", users in
6596 cgroups suffixed with ".user", and nspawn containers in
6597 cgroups suffixed with ".nspawn". Furthermore, all cgroup
6598 names are now escaped in a simple scheme to avoid collision
6599 of userspace object names with kernel filenames. This work
6600 is preparation for making these objects relocatable in the
6601 cgroup tree, in order to allow easy resource partitioning of
6602 these objects without causing naming conflicts.
6604 * systemctl list-dependencies gained the new switches
6605 --plain, --reverse, --after and --before.
6607 * systemd-inhibit now shows the process name of processes that
6608 have taken an inhibitor lock.
6610 * nss-myhostname will now also resolve "localhost"
6611 implicitly. This makes /etc/hosts an optional file and
6612 nicely handles that on IPv6 ::1 maps to both "localhost" and
6615 * libsystemd-logind.so gained a new call
6616 sd_get_machine_names() to enumerate running containers and
6617 VMs (currently only supported by very new libvirt and
6618 nspawn). sd_login_monitor can now be used to watch
6619 VMs/containers coming and going.
6621 * .include is not allowed recursively anymore, and only in
6622 unit files. Usually it is better to use drop-in snippets in
6623 .d/*.conf anyway, as introduced with systemd 198.
6625 * systemd-analyze gained a new "critical-chain" command that
6626 determines the slowest chain of units run during system
6627 boot-up. It is very useful for tracking down where
6628 optimizing boot time is the most beneficial.
6630 * systemd will no longer allow manipulating service paths in
6631 the name=systemd:/system cgroup tree using ControlGroup= in
6632 units. (But is still fine with it in all other dirs.)
6634 * There's a new systemd-nspawn@.service service file that may
6635 be used to easily run nspawn containers as system
6636 services. With the container's root directory in
6637 /var/lib/container/foobar it is now sufficient to run
6638 "systemctl start systemd-nspawn@foobar.service" to boot it.
6640 * systemd-cgls gained a new parameter "--machine" to list only
6641 the processes within a certain container.
6643 * ConditionSecurity= now can check for "apparmor". We still
6644 are lacking checks for SMACK and IMA for this condition
6645 check though. Patches welcome!
6647 * A new configuration file /etc/systemd/sleep.conf has been
6648 added that may be used to configure which kernel operation
6649 systemd is supposed to execute when "suspend", "hibernate"
6650 or "hybrid-sleep" is requested. This makes the new kernel
6651 "freeze" state accessible to the user.
6653 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
6654 the passed argument if applicable.
6656 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
6657 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
6658 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
6659 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
6660 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
6661 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
6662 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
6667 * The output of 'systemctl list-jobs' got some polishing. The
6668 '--type=' argument may now be passed more than once. A new
6669 command 'systemctl list-sockets' has been added which shows
6670 a list of kernel sockets systemd is listening on with the
6671 socket units they belong to, plus the units these socket
6674 * The experimental libsystemd-bus library got substantial
6675 updates to work in conjunction with the (also experimental)
6676 kdbus kernel project. It works well enough to exchange
6677 messages with some sophistication. Note that kdbus is not
6678 ready yet, and the library is mostly an elaborate test case
6679 for now, and not installable.
6681 * systemd gained a new unit 'systemd-static-nodes.service'
6682 that generates static device nodes earlier during boot, and
6683 can run in conjunction with udev.
6685 * libsystemd-login gained a new call sd_pid_get_user_unit()
6686 to retrieve the user systemd unit a process is running
6687 in. This is useful for systems where systemd is used as
6690 * systemd-nspawn now places all containers in the new /machine
6691 top-level cgroup directory in the name=systemd
6692 hierarchy. libvirt will soon do the same, so that we get a
6693 uniform separation of /system, /user and /machine for system
6694 services, user processes and containers/virtual
6695 machines. This new cgroup hierarchy is also useful to stick
6696 stable names to specific container instances, which can be
6697 recognized later this way (this name may be controlled
6698 via systemd-nspawn's new -M switch). libsystemd-login also
6699 gained a new call sd_pid_get_machine_name() to retrieve the
6700 name of the container/VM a specific process belongs to.
6702 * bootchart can now store its data in the journal.
6704 * libsystemd-journal gained a new call
6705 sd_journal_add_conjunction() for AND expressions to the
6706 matching logic. This can be used to express more complex
6707 logical expressions.
6709 * journactl can now take multiple --unit= and --user-unit=
6712 * The cryptsetup logic now understands the "luks.key=" kernel
6713 command line switch for specifying a file to read the
6714 decryption key from. Also, if a configured key file is not
6715 found the tool will now automatically fall back to prompting
6718 * Python systemd.journal module was updated to wrap recently
6719 added functions from libsystemd-journal. The interface was
6720 changed to bring the low level interface in s.j._Reader
6721 closer to the C API, and the high level interface in
6722 s.j.Reader was updated to wrap and convert all data about
6725 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
6726 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
6727 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
6728 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
6729 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
6730 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6734 * journalctl --update-catalog now understands a new --root=
6735 option to operate on catalogs found in a different root
6738 * During shutdown after systemd has terminated all running
6739 services a final killing loop kills all remaining left-over
6740 processes. We will now print the name of these processes
6741 when we send SIGKILL to them, since this usually indicates a
6744 * If /etc/crypttab refers to password files stored on
6745 configured mount points automatic dependencies will now be
6746 generated to ensure the specific mount is established first
6747 before the key file is attempted to be read.
6749 * 'systemctl status' will now show information about the
6750 network sockets a socket unit is listening on.
6752 * 'systemctl status' will also shown information about any
6753 drop-in configuration file for units. (Drop-In configuration
6754 files in this context are files such as
6755 /etc/systemd/systemd/foobar.service.d/*.conf)
6757 * systemd-cgtop now optionally shows summed up CPU times of
6758 cgroups. Press '%' while running cgtop to switch between
6759 percentage and absolute mode. This is useful to determine
6760 which cgroups use up the most CPU time over the entire
6761 runtime of the system. systemd-cgtop has also been updated
6762 to be 'pipeable' for processing with further shell tools.
6764 * 'hostnamectl set-hostname' will now allow setting of FQDN
6767 * The formatting and parsing of time span values has been
6768 changed. The parser now understands fractional expressions
6769 such as "5.5h". The formatter will now output fractional
6770 expressions for all time spans under 1min, i.e. "5.123456s"
6771 rather than "5s 123ms 456us". For time spans under 1s
6772 millisecond values are shown, for those under 1ms
6773 microsecond values are shown. This should greatly improve
6774 all time-related output of systemd.
6776 * libsystemd-login and libsystemd-journal gained new
6777 functions for querying the poll() events mask and poll()
6778 timeout value for integration into arbitrary event
6781 * localectl gained the ability to list available X11 keymaps
6782 (models, layouts, variants, options).
6784 * 'systemd-analyze dot' gained the ability to filter for
6785 specific units via shell-style globs, to create smaller,
6786 more useful graphs. I.e. it is now possible to create simple
6787 graphs of all the dependencies between only target units, or
6788 of all units that Avahi has dependencies with.
6790 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
6791 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
6792 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
6793 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
6794 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
6795 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
6796 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
6800 * The boot-time readahead implementation for rotating media
6801 will now read the read-ahead data in multiple passes which
6802 consist of all read requests made in equidistant time
6803 intervals. This means instead of strictly reading read-ahead
6804 data in its physical order on disk we now try to find a
6805 middle ground between physical and access time order.
6807 * /etc/os-release files gained a new BUILD_ID= field for usage
6808 on operating systems that provide continuous builds of OS
6811 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
6812 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
6813 William Douglas, Zbigniew Jędrzejewski-Szmek
6817 * systemd-python gained an API exposing libsystemd-daemon.
6819 * The SMACK setup logic gained support for uploading CIPSO
6822 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
6823 ReadOnlyDirectories= and InaccessibleDirectories= has
6824 changed. The private /tmp and /var/tmp directories are now
6825 shared by all processes of a service (which means
6826 ExecStartPre= may now leave data in /tmp that ExecStart= of
6827 the same service can still access). When a service is
6828 stopped its temporary directories are immediately deleted
6829 (normal clean-up with tmpfiles is still done in addition to
6832 * By default, systemd will now set a couple of sysctl
6833 variables in the kernel: the safe sysrq options are turned
6834 on, IP route verification is turned on, and source routing
6835 disabled. The recently added hardlink and softlink
6836 protection of the kernel is turned on. These settings should
6837 be reasonably safe, and good defaults for all new systems.
6839 * The predictable network naming logic may now be turned off
6840 with a new kernel command line switch: net.ifnames=0.
6842 * A new libsystemd-bus module has been added that implements a
6843 pretty complete D-Bus client library. For details see:
6845 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
6847 * journald will now explicitly flush the journal files to disk
6848 at the latest 5min after each write. The file will then also
6849 be marked offline until the next write. This should increase
6850 reliability in case of a crash. The synchronization delay
6851 can be configured via SyncIntervalSec= in journald.conf.
6853 * There's a new remote-fs-setup.target unit that can be used
6854 to pull in specific services when at least one remote file
6855 system is to be mounted.
6857 * There are new targets timers.target and paths.target as
6858 canonical targets to pull user timer and path units in
6859 from. This complements sockets.target with a similar
6860 purpose for socket units.
6862 * libudev gained a new call udev_device_set_attribute_value()
6863 to set sysfs attributes of a device.
6865 * The udev daemon now sets the default number of worker
6866 processes executed in parallel based on the number of available
6867 CPUs instead of the amount of available RAM. This is supposed
6868 to provide a more reliable default and limit a too aggressive
6869 parallelism for setups with 1000s of devices connected.
6871 Contributions from: Auke Kok, Colin Walters, Cristian
6872 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
6873 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
6874 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
6875 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
6876 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
6877 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
6878 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
6879 Zbigniew Jędrzejewski-Szmek
6883 * Configuration of unit files may now be extended via drop-in
6884 files without having to edit/override the unit files
6885 themselves. More specifically, if the administrator wants to
6886 change one value for a service file foobar.service he can
6887 now do so by dropping in a configuration snippet into
6888 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
6889 will load all these snippets and apply them on top of the
6890 main unit configuration file, possibly extending or
6891 overriding its settings. Using these drop-in snippets is
6892 generally nicer than the two earlier options for changing
6893 unit files locally: copying the files from
6894 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
6895 them there; or creating a new file in /etc/systemd/system/
6896 that incorporates the original one via ".include". Drop-in
6897 snippets into these .d/ directories can be placed in any
6898 directory systemd looks for units in, and the usual
6899 overriding semantics between /usr/lib, /etc and /run apply
6902 * Most unit file settings which take lists of items can now be
6903 reset by assigning the empty string to them. For example,
6904 normally, settings such as Environment=FOO=BAR append a new
6905 environment variable assignment to the environment block,
6906 each time they are used. By assigning Environment= the empty
6907 string the environment block can be reset to empty. This is
6908 particularly useful with the .d/*.conf drop-in snippets
6909 mentioned above, since this adds the ability to reset list
6910 settings from vendor unit files via these drop-ins.
6912 * systemctl gained a new "list-dependencies" command for
6913 listing the dependencies of a unit recursively.
6915 * Inhibitors are now honored and listed by "systemctl
6916 suspend", "systemctl poweroff" (and similar) too, not only
6917 GNOME. These commands will also list active sessions by
6920 * Resource limits (as exposed by the various control group
6921 controllers) can now be controlled dynamically at runtime
6922 for all units. More specifically, you can now use a command
6923 like "systemctl set-cgroup-attr foobar.service cpu.shares
6924 2000" to alter the CPU shares a specific service gets. These
6925 settings are stored persistently on disk, and thus allow the
6926 administrator to easily adjust the resource usage of
6927 services with a few simple commands. This dynamic resource
6928 management logic is also available to other programs via the
6929 bus. Almost any kernel cgroup attribute and controller is
6932 * systemd-vconsole-setup will now copy all font settings to
6933 all allocated VTs, where it previously applied them only to
6936 * libsystemd-login gained the new sd_session_get_tty() API
6939 * This release drops support for a few legacy or
6940 distribution-specific LSB facility names when parsing init
6941 scripts: $x-display-manager, $mail-transfer-agent,
6942 $mail-transport-agent, $mail-transfer-agent, $smtp,
6943 $null. Also, the mail-transfer-agent.target unit backing
6944 this has been removed. Distributions which want to retain
6945 compatibility with this should carry the burden for
6946 supporting this themselves and patch support for these back
6947 in, if they really need to. Also, the facilities $syslog and
6948 $local_fs are now ignored, since systemd does not support
6949 early-boot LSB init scripts anymore, and these facilities
6950 are implied anyway for normal services. syslog.target has
6953 * There are new bus calls on PID1's Manager object for
6954 cancelling jobs, and removing snapshot units. Previously,
6955 both calls were only available on the Job and Snapshot
6958 * systemd-journal-gatewayd gained SSL support.
6960 * The various "environment" files, such as /etc/locale.conf
6961 now support continuation lines with a backslash ("\") as
6962 last character in the line, similarly in style (but different)
6963 to how this is supported in shells.
6965 * For normal user processes the _SYSTEMD_USER_UNIT= field is
6966 now implicitly appended to every log entry logged. systemctl
6967 has been updated to filter by this field when operating on a
6968 user systemd instance.
6970 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
6971 CAP_AUDIT_CONTROL capabilities to the capabilities set for
6972 the container. This makes it easier to boot unmodified
6973 Fedora systems in a container, which however still requires
6974 audit=0 to be passed on the kernel command line. Auditing in
6975 kernel and userspace is unfortunately still too broken in
6976 context of containers, hence we recommend compiling it out
6977 of the kernel or using audit=0. Hopefully this will be fixed
6978 one day for good in the kernel.
6980 * nspawn gained the new --bind= and --bind-ro= parameters to
6981 bind mount specific directories from the host into the
6984 * nspawn will now mount its own devpts file system instance
6985 into the container, in order not to leak pty devices from
6986 the host into the container.
6988 * systemd will now read the firmware boot time performance
6989 information from the EFI variables, if the used boot loader
6990 supports this, and takes it into account for boot performance
6991 analysis via "systemd-analyze". This is currently supported
6992 only in conjunction with Gummiboot, but could be supported
6993 by other boot loaders too. For details see:
6995 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
6997 * A new generator has been added that automatically mounts the
6998 EFI System Partition (ESP) to /boot, if that directory
6999 exists, is empty, and no other file system has been
7000 configured to be mounted there.
7002 * logind will now send out PrepareForSleep(false) out
7003 unconditionally, after coming back from suspend. This may be
7004 used by applications as asynchronous notification for
7005 system resume events.
7007 * "systemctl unlock-sessions" has been added, that allows
7008 unlocking the screens of all user sessions at once, similar
7009 to how "systemctl lock-sessions" already locked all users
7010 sessions. This is backed by a new D-Bus call UnlockSessions().
7012 * "loginctl seat-status" will now show the master device of a
7013 seat. (i.e. the device of a seat that needs to be around for
7014 the seat to be considered available, usually the graphics
7017 * tmpfiles gained a new "X" line type, that allows
7018 configuration of files and directories (with wildcards) that
7019 shall be excluded from automatic cleanup ("aging").
7021 * udev default rules set the device node permissions now only
7022 at "add" events, and do not change them any longer with a
7023 later "change" event.
7025 * The log messages for lid events and power/sleep keypresses
7026 now carry a message ID.
7028 * We now have a substantially larger unit test suite, but this
7029 continues to be work in progress.
7031 * udevadm hwdb gained a new --root= parameter to change the
7032 root directory to operate relative to.
7034 * logind will now issue a background sync() request to the kernel
7035 early at shutdown, so that dirty buffers are flushed to disk early
7036 instead of at the last moment, in order to optimize shutdown
7039 * A new bootctl tool has been added that is an interface for
7040 certain boot loader operations. This is currently a preview
7041 and is likely to be extended into a small mechanism daemon
7042 like timedated, localed, hostnamed, and can be used by
7043 graphical UIs to enumerate available boot options, and
7044 request boot into firmware operations.
7046 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
7047 the rest of the package. It also has been updated to work
7048 correctly in initrds.
7050 * Policykit previously has been runtime optional, and is now
7051 also compile time optional via a configure switch.
7053 * systemd-analyze has been reimplemented in C. Also "systemctl
7054 dot" has moved into systemd-analyze.
7056 * "systemctl status" with no further parameters will now print
7057 the status of all active or failed units.
7059 * Operations such as "systemctl start" can now be executed
7060 with a new mode "--irreversible" which may be used to queue
7061 operations that cannot accidentally be reversed by a later
7062 job queuing. This is by default used to make shutdown
7063 requests more robust.
7065 * The Python API of systemd now gained a new module for
7066 reading journal files.
7068 * A new tool kernel-install has been added that can install
7069 kernel images according to the Boot Loader Specification:
7071 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
7073 * Boot time console output has been improved to provide
7074 animated boot time output for hanging jobs.
7076 * A new tool systemd-activate has been added which can be used
7077 to test socket activation with, directly from the command
7078 line. This should make it much easier to test and debug
7079 socket activation in daemons.
7081 * journalctl gained a new "--reverse" (or -r) option to show
7082 journal output in reverse order (i.e. newest line first).
7084 * journalctl gained a new "--pager-end" (or -e) option to jump
7085 to immediately jump to the end of the journal in the
7086 pager. This is only supported in conjunction with "less".
7088 * journalctl gained a new "--user-unit=" option, that works
7089 similarly to "--unit=" but filters for user units rather than
7092 * A number of unit files to ease adoption of systemd in
7093 initrds has been added. This moves some minimal logic from
7094 the various initrd implementations into systemd proper.
7096 * The journal files are now owned by a new group
7097 "systemd-journal", which exists specifically to allow access
7098 to the journal, and nothing else. Previously, we used the
7099 "adm" group for that, which however possibly covers more
7100 than just journal/log file access. This new group is now
7101 already used by systemd-journal-gatewayd to ensure this
7102 daemon gets access to the journal files and as little else
7103 as possible. Note that "make install" will also set FS ACLs
7104 up for /var/log/journal to give "adm" and "wheel" read
7105 access to it, in addition to "systemd-journal" which owns
7106 the journal files. We recommend that packaging scripts also
7107 add read access to "adm" + "wheel" to /var/log/journal, and
7108 all existing/future journal files. To normal users and
7109 administrators little changes, however packagers need to
7110 ensure to create the "systemd-journal" system group at
7111 package installation time.
7113 * The systemd-journal-gatewayd now runs as unprivileged user
7114 systemd-journal-gateway:systemd-journal-gateway. Packaging
7115 scripts need to create these system user/group at
7118 * timedated now exposes a new boolean property CanNTP that
7119 indicates whether a local NTP service is available or not.
7121 * systemd-detect-virt will now also detect xen PVs
7123 * The pstore file system is now mounted by default, if it is
7126 * In addition to the SELinux and IMA policies we will now also
7127 load SMACK policies at early boot.
7129 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
7130 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
7131 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
7132 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
7133 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
7134 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
7135 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
7136 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
7137 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
7138 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
7139 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
7140 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
7141 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
7142 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
7146 * Timer units now support calendar time events in addition to
7147 monotonic time events. That means you can now trigger a unit
7148 based on a calendar time specification such as "Thu,Fri
7149 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
7150 or fifth day of any month of the year 2013, given that it is
7151 a thursday or friday. This brings timer event support
7152 considerably closer to cron's capabilities. For details on
7153 the supported calendar time specification language see
7156 * udev now supports a number of different naming policies for
7157 network interfaces for predictable names, and a combination
7158 of these policies is now the default. Please see this wiki
7159 document for details:
7161 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
7163 * Auke Kok's bootchart implementation has been added to the
7164 systemd tree. It is an optional component that can graph the
7165 boot in quite some detail. It is one of the best bootchart
7166 implementations around and minimal in its code and
7169 * nss-myhostname has been integrated into the systemd source
7170 tree. nss-myhostname guarantees that the local hostname
7171 always stays resolvable via NSS. It has been a weak
7172 requirement of systemd-hostnamed since a long time, and
7173 since its code is actually trivial we decided to just
7174 include it in systemd's source tree. It can be turned off
7175 with a configure switch.
7177 * The read-ahead logic is now capable of properly detecting
7178 whether a btrfs file system is on SSD or rotating media, in
7179 order to optimize the read-ahead scheme. Previously, it was
7180 only capable of detecting this on traditional file systems
7183 * In udev, additional device properties are now read from the
7184 IAB in addition to the OUI database. Also, Bluetooth company
7185 identities are attached to the devices as well.
7187 * In service files %U may be used as specifier that is
7188 replaced by the configured user name of the service.
7190 * nspawn may now be invoked without a controlling TTY. This
7191 makes it suitable for invocation as its own service. This
7192 may be used to set up a simple containerized server system
7193 using only core OS tools.
7195 * systemd and nspawn can now accept socket file descriptors
7196 when they are started for socket activation. This enables
7197 implementation of socket activated nspawn
7198 containers. i.e. think about autospawning an entire OS image
7199 when the first SSH or HTTP connection is received. We expect
7200 that similar functionality will also be added to libvirt-lxc
7203 * journalctl will now suppress ANSI color codes when
7204 presenting log data.
7206 * systemctl will no longer show control group information for
7207 a unit if the control group is empty anyway.
7209 * logind can now automatically suspend/hibernate/shutdown the
7212 * /etc/machine-info and hostnamed now also expose the chassis
7213 type of the system. This can be used to determine whether
7214 the local system is a laptop, desktop, handset or
7215 tablet. This information may either be configured by the
7216 user/vendor or is automatically determined from ACPI and DMI
7217 information if possible.
7219 * A number of PolicyKit actions are now bound together with
7220 "imply" rules. This should simplify creating UIs because
7221 many actions will now authenticate similar ones as well.
7223 * Unit files learnt a new condition ConditionACPower= which
7224 may be used to conditionalize a unit depending on whether an
7225 AC power source is connected or not, of whether the system
7226 is running on battery power.
7228 * systemctl gained a new "is-failed" verb that may be used in
7229 shell scripts and suchlike to check whether a specific unit
7230 is in the "failed" state.
7232 * The EnvironmentFile= setting in unit files now supports file
7233 globbing, and can hence be used to easily read a number of
7234 environment files at once.
7236 * systemd will no longer detect and recognize specific
7237 distributions. All distribution-specific #ifdeffery has been
7238 removed, systemd is now fully generic and
7239 distribution-agnostic. Effectively, not too much is lost as
7240 a lot of the code is still accessible via explicit configure
7241 switches. However, support for some distribution specific
7242 legacy configuration file formats has been dropped. We
7243 recommend distributions to simply adopt the configuration
7244 files everybody else uses now and convert the old
7245 configuration from packaging scripts. Most distributions
7246 already did that. If that's not possible or desirable,
7247 distributions are welcome to forward port the specific
7248 pieces of code locally from the git history.
7250 * When logging a message about a unit systemd will now always
7251 log the unit name in the message meta data.
7253 * localectl will now also discover system locale data that is
7254 not stored in locale archives, but directly unpacked.
7256 * logind will no longer unconditionally use framebuffer
7257 devices as seat masters, i.e. as devices that are required
7258 to be existing before a seat is considered preset. Instead,
7259 it will now look for all devices that are tagged as
7260 "seat-master" in udev. By default, framebuffer devices will
7261 be marked as such, but depending on local systems, other
7262 devices might be marked as well. This may be used to
7263 integrate graphics cards using closed source drivers (such
7264 as NVidia ones) more nicely into logind. Note however, that
7265 we recommend using the open source NVidia drivers instead,
7266 and no udev rules for the closed-source drivers will be
7267 shipped from us upstream.
7269 Contributions from: Adam Williamson, Alessandro Crismani, Auke
7270 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
7271 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
7272 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
7273 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
7274 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
7275 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
7276 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
7277 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
7278 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
7279 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
7280 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
7285 * udev gained support for loading additional device properties
7286 from an indexed database that is keyed by vendor/product IDs
7287 and similar device identifiers. For the beginning this
7288 "hwdb" is populated with data from the well-known PCI and
7289 USB database, but also includes PNP, ACPI and OID data. In
7290 the longer run this indexed database shall grow into
7291 becoming the one central database for non-essential
7292 userspace device metadata. Previously, data from the PCI/USB
7293 database was only attached to select devices, since the
7294 lookup was a relatively expensive operation due to O(n) time
7295 complexity (with n being the number of entries in the
7296 database). Since this is now O(1), we decided to add in this
7297 data for all devices where this is available, by
7298 default. Note that the indexed database needs to be rebuilt
7299 when new data files are installed. To achieve this you need
7300 to update your packaging scripts to invoke "udevadm hwdb
7301 --update" after installation of hwdb data files. For
7302 RPM-based distributions we introduced the new
7303 %udev_hwdb_update macro for this purpose.
7305 * The Journal gained support for the "Message Catalog", an
7306 indexed database to link up additional information with
7307 journal entries. For further details please check:
7309 https://www.freedesktop.org/wiki/Software/systemd/catalog
7311 The indexed message catalog database also needs to be
7312 rebuilt after installation of message catalog files. Use
7313 "journalctl --update-catalog" for this. For RPM-based
7314 distributions we introduced the %journal_catalog_update
7315 macro for this purpose.
7317 * The Python Journal bindings gained support for the standard
7318 Python logging framework.
7320 * The Journal API gained new functions for checking whether
7321 the underlying file system of a journal file is capable of
7322 properly reporting file change notifications, or whether
7323 applications that want to reflect journal changes "live"
7324 need to recheck journal files continuously in appropriate
7327 * It is now possible to set the "age" field for tmpfiles
7328 entries to 0, indicating that files matching this entry
7329 shall always be removed when the directories are cleaned up.
7331 * coredumpctl gained a new "gdb" verb which invokes gdb
7332 right-away on the selected coredump.
7334 * There's now support for "hybrid sleep" on kernels that
7335 support this, in addition to "suspend" and "hibernate". Use
7336 "systemctl hybrid-sleep" to make use of this.
7338 * logind's HandleSuspendKey= setting (and related settings)
7339 now gained support for a new "lock" setting to simply
7340 request the screen lock on all local sessions, instead of
7341 actually executing a suspend or hibernation.
7343 * systemd will now mount the EFI variables file system by
7346 * Socket units now gained support for configuration of the
7347 SMACK security label.
7349 * timedatectl will now output the time of the last and next
7350 daylight saving change.
7352 * We dropped support for various legacy and distro-specific
7353 concepts, such as insserv, early-boot SysV services
7354 (i.e. those for non-standard runlevels such as 'b' or 'S')
7355 or ArchLinux /etc/rc.conf support. We recommend the
7356 distributions who still need support this to either continue
7357 to maintain the necessary patches downstream, or find a
7358 different solution. (Talk to us if you have questions!)
7360 * Various systemd components will now bypass PolicyKit checks
7361 for root and otherwise handle properly if PolicyKit is not
7362 found to be around. This should fix most issues for
7363 PolicyKit-less systems. Quite frankly this should have been
7364 this way since day one. It is absolutely our intention to
7365 make systemd work fine on PolicyKit-less systems, and we
7366 consider it a bug if something does not work as it should if
7367 PolicyKit is not around.
7369 * For embedded systems it is now possible to build udev and
7370 systemd without blkid and/or kmod support.
7372 * "systemctl switch-root" is now capable of switching root
7373 more than once. I.e. in addition to transitions from the
7374 initrd to the host OS it is now possible to transition to
7375 further OS images from the host. This is useful to implement
7376 offline updating tools.
7378 * Various other additions have been made to the RPM macros
7379 shipped with systemd. Use %udev_rules_update() after
7380 installing new udev rules files. %_udevhwdbdir,
7381 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
7382 %_sysctldir are now available which resolve to the right
7383 directories for packages to place various data files in.
7385 * journalctl gained the new --full switch (in addition to
7386 --all, to disable ellipsation for long messages.
7388 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
7389 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
7390 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
7391 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
7392 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
7393 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
7394 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
7395 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
7396 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
7400 * journalctl gained new --since= and --until= switches to
7401 filter by time. It also now supports nice filtering for
7402 units via --unit=/-u.
7404 * Type=oneshot services may use ExecReload= and do the
7407 * The journal daemon now supports time-based rotation and
7408 vacuuming, in addition to the usual disk-space based
7411 * The journal will now index the available field values for
7412 each field name. This enables clients to show pretty drop
7413 downs of available match values when filtering. The bash
7414 completion of journalctl has been updated
7415 accordingly. journalctl gained a new switch -F to list all
7416 values a certain field takes in the journal database.
7418 * More service events are now written as structured messages
7419 to the journal, and made recognizable via message IDs.
7421 * The timedated, localed and hostnamed mini-services which
7422 previously only provided support for changing time, locale
7423 and hostname settings from graphical DEs such as GNOME now
7424 also have a minimal (but very useful) text-based client
7425 utility each. This is probably the nicest way to changing
7426 these settings from the command line now, especially since
7427 it lists available options and is fully integrated with bash
7430 * There's now a new tool "systemd-coredumpctl" to list and
7431 extract coredumps from the journal.
7433 * We now install a README each in /var/log/ and
7434 /etc/rc.d/init.d explaining where the system logs and init
7435 scripts went. This hopefully should help folks who go to
7436 that dirs and look into the otherwise now empty void and
7437 scratch their heads.
7439 * When user-services are invoked (by systemd --user) the
7440 $MANAGERPID env var is set to the PID of systemd.
7442 * SIGRTMIN+24 when sent to a --user instance will now result
7443 in immediate termination of systemd.
7445 * gatewayd received numerous feature additions such as a
7446 "follow" mode, for live syncing and filtering.
7448 * browse.html now allows filtering and showing detailed
7449 information on specific entries. Keyboard navigation and
7450 mouse screen support has been added.
7452 * gatewayd/journalctl now supports HTML5/JSON
7453 Server-Sent-Events as output.
7455 * The SysV init script compatibility logic will now
7456 heuristically determine whether a script supports the
7457 "reload" verb, and only then make this available as
7460 * "systemctl status --follow" has been removed, use "journalctl
7463 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
7464 have been removed since they are hardly useful to be
7467 * And I'd like to take the opportunity to specifically mention
7468 Zbigniew for his great contributions. Zbigniew, you rock!
7470 Contributions from: Andrew Eikum, Christian Hesse, Colin
7471 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
7472 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
7473 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
7474 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
7475 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
7476 Jędrzejewski-Szmek, Сковорода Никита Андреевич
7480 * If /etc/vconsole.conf is non-existent or empty we will no
7481 longer load any console font or key map at boot by
7482 default. Instead the kernel defaults will be left
7483 intact. This is definitely the right thing to do, as no
7484 configuration should mean no configuration, and hard-coding
7485 font names that are different on all archs is probably a bad
7486 idea. Also, the kernel default key map and font should be
7487 good enough for most cases anyway, and mostly identical to
7488 the userspace fonts/key maps we previously overloaded them
7489 with. If distributions want to continue to default to a
7490 non-kernel font or key map they should ship a default
7491 /etc/vconsole.conf with the appropriate contents.
7493 Contributions from: Colin Walters, Daniel J Walsh, Dave
7494 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
7495 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
7499 * journalctl gained a new --cursor= switch to show entries
7500 starting from the specified location in the journal.
7502 * We now enforce a size limit on journal entry fields exported
7503 with "-o json" in journalctl. Fields larger than 4K will be
7504 assigned null. This can be turned off with --all.
7506 * An (optional) journal gateway daemon is now available as
7507 "systemd-journal-gatewayd.service". This service provides
7508 access to the journal via HTTP and JSON. This functionality
7509 will be used to implement live log synchronization in both
7510 pull and push modes, but has various other users too, such
7511 as easy log access for debugging of embedded devices. Right
7512 now it is already useful to retrieve the journal via HTTP:
7514 # systemctl start systemd-journal-gatewayd.service
7515 # wget http://localhost:19531/entries
7517 This will download the journal contents in a
7518 /var/log/messages compatible format. The same as JSON:
7520 # curl -H"Accept: application/json" http://localhost:19531/entries
7522 This service is also accessible via a web browser where a
7523 single static HTML5 app is served that uses the JSON logic
7524 to enable the user to do some basic browsing of the
7525 journal. This will be extended later on. Here's an example
7526 screenshot of this app in its current state:
7528 http://0pointer.de/public/journal-gatewayd
7530 Contributions from: Kay Sievers, Lennart Poettering, Robert
7531 Milasan, Tom Gundersen
7535 * The bash completion logic is now available for journalctl
7538 * We do not mount the "cpuset" controller anymore together with
7539 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
7540 started if no parameters are assigned to it. "cpuset" hence
7541 broke code that assumed it could create "cpu" groups and
7544 * journalctl -f will now subscribe to terminal size changes,
7545 and line break accordingly.
7547 Contributions from: Dave Reisner, Kay Sievers, Lennart
7548 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
7552 * nspawn will now create a symlink /etc/localtime in the
7553 container environment, copying the host's timezone
7554 setting. Previously this has been done via a bind mount, but
7555 since symlinks cannot be bind mounted this has now been
7556 changed to create/update the appropriate symlink.
7558 * journalctl -n's line number argument is now optional, and
7559 will default to 10 if omitted.
7561 * journald will now log the maximum size the journal files may
7562 take up on disk. This is particularly useful if the default
7563 built-in logic of determining this parameter from the file
7564 system size is used. Use "systemctl status
7565 systemd-journald.service" to see this information.
7567 * The multi-seat X wrapper tool has been stripped down. As X
7568 is now capable of enumerating graphics devices via udev in a
7569 seat-aware way the wrapper is not strictly necessary
7570 anymore. A stripped down temporary stop-gap is still shipped
7571 until the upstream display managers have been updated to
7572 fully support the new X logic. Expect this wrapper to be
7573 removed entirely in one of the next releases.
7575 * HandleSleepKey= in logind.conf has been split up into
7576 HandleSuspendKey= and HandleHibernateKey=. The old setting
7577 is not available anymore. X11 and the kernel are
7578 distinguishing between these keys and we should too. This
7579 also means the inhibition lock for these keys has been split
7582 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
7583 Poettering, Lukas Nykryn, Václav Pavlín
7587 * Whenever a unit changes state we will now log this to the
7588 journal and show along the unit's own log output in
7591 * ConditionPathIsMountPoint= can now properly detect bind
7592 mount points too. (Previously, a bind mount of one file
7593 system to another place in the same file system could not be
7594 detected as mount, since they shared struct stat's st_dev
7597 * We will now mount the cgroup controllers cpu, cpuacct,
7598 cpuset and the controllers net_cls, net_prio together by
7601 * nspawn containers will now have a virtualized boot
7602 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
7603 over with a randomized ID at container initialization). This
7604 has the effect of making "journalctl -b" do the right thing
7607 * The JSON output journal serialization has been updated not
7608 to generate "endless" list objects anymore, but rather one
7609 JSON object per line. This is more in line how most JSON
7610 parsers expect JSON objects. The new output mode
7611 "json-pretty" has been added to provide similar output, but
7612 neatly aligned for readability by humans.
7614 * We dropped all explicit sync() invocations in the shutdown
7615 code. The kernel does this implicitly anyway in the kernel
7616 reboot() syscall. halt(8)'s -n option is now a compatibility
7619 * We now support virtualized reboot() in containers, as
7620 supported by newer kernels. We will fall back to exit() if
7621 CAP_SYS_REBOOT is not available to the container. Also,
7622 nspawn makes use of this now and will actually reboot the
7623 container if the containerized OS asks for that.
7625 * journalctl will only show local log output by default
7626 now. Use --merge (-m) to show remote log output, too.
7628 * libsystemd-journal gained the new sd_journal_get_usage()
7629 call to determine the current disk usage of all journal
7630 files. This is exposed in the new "journalctl --disk-usage"
7633 * journald gained a new configuration setting SplitMode= in
7634 journald.conf which may be used to control how user journals
7635 are split off. See journald.conf(5) for details.
7637 * A new condition type ConditionFileNotEmpty= has been added.
7639 * tmpfiles' "w" lines now support file globbing, to write
7640 multiple files at once.
7642 * We added Python bindings for the journal submission
7643 APIs. More Python APIs for a number of selected APIs will
7644 likely follow. Note that we intend to add native bindings
7645 only for the Python language, as we consider it common
7646 enough to deserve bindings shipped within systemd. There are
7647 various projects outside of systemd that provide bindings
7648 for languages such as PHP or Lua.
7650 * Many conditions will now resolve specifiers such as %i. In
7651 addition, PathChanged= and related directives of .path units
7652 now support specifiers as well.
7654 * There's now a new RPM macro definition for the system preset
7657 * journald will now warn if it ca not forward a message to the
7658 syslog daemon because its socket is full.
7660 * timedated will no longer write or process /etc/timezone,
7661 except on Debian. As we do not support late mounted /usr
7662 anymore /etc/localtime always being a symlink is now safe,
7663 and hence the information in /etc/timezone is not necessary
7666 * logind will now always reserve one VT for a text getty (VT6
7667 by default). Previously if more than 6 X sessions where
7668 started they took up all the VTs with auto-spawned gettys,
7669 so that no text gettys were available anymore.
7671 * udev will now automatically inform the btrfs kernel logic
7672 about btrfs RAID components showing up. This should make
7673 simple hotplug based btrfs RAID assembly work.
7675 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
7676 (but not for its children which will stay at the kernel
7677 default). This should allow setups with a lot more listening
7680 * systemd will now always pass the configured timezone to the
7681 kernel at boot. timedated will do the same when the timezone
7684 * logind's inhibition logic has been updated. By default,
7685 logind will now handle the lid switch, the power and sleep
7686 keys all the time, even in graphical sessions. If DEs want
7687 to handle these events on their own they should take the new
7688 handle-power-key, handle-sleep-key and handle-lid-switch
7689 inhibitors during their runtime. A simple way to achieve
7690 that is to invoke the DE wrapped in an invocation of:
7692 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
7694 * Access to unit operations is now checked via SELinux taking
7695 the unit file label and client process label into account.
7697 * systemd will now notify the administrator in the journal
7698 when he over-mounts a non-empty directory.
7700 * There are new specifiers that are resolved in unit files,
7701 for the host name (%H), the machine ID (%m) and the boot ID
7704 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
7705 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
7706 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
7707 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
7708 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
7709 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
7710 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
7714 * Support for reading structured kernel messages from
7715 /dev/kmsg has now been added and is enabled by default.
7717 * Support for reading kernel messages from /proc/kmsg has now
7718 been removed. If you want kernel messages in the journal
7719 make sure to run a recent kernel (>= 3.5) that supports
7720 reading structured messages from /dev/kmsg (see
7721 above). /proc/kmsg is now exclusive property of classic
7722 syslog daemons again.
7724 * The libudev API gained the new
7725 udev_device_new_from_device_id() call.
7727 * The logic for file system namespace (ReadOnlyDirectory=,
7728 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
7729 require pivot_root() anymore. This means fewer temporary
7730 directories are created below /tmp for this feature.
7732 * nspawn containers will now see and receive all submounts
7733 made on the host OS below the root file system of the
7736 * Forward Secure Sealing is now supported for Journal files,
7737 which provide cryptographical sealing of journal files so
7738 that attackers cannot alter log history anymore without this
7739 being detectable. Lennart will soon post a blog story about
7740 this explaining it in more detail.
7742 * There are two new service settings RestartPreventExitStatus=
7743 and SuccessExitStatus= which allow configuration of exit
7744 status (exit code or signal) which will be excepted from the
7745 restart logic, resp. consider successful.
7747 * journalctl gained the new --verify switch that can be used
7748 to check the integrity of the structure of journal files and
7749 (if Forward Secure Sealing is enabled) the contents of
7752 * nspawn containers will now be run with /dev/stdin, /dev/fd/
7753 and similar symlinks pre-created. This makes running shells
7754 as container init process a lot more fun.
7756 * The fstab support can now handle PARTUUID= and PARTLABEL=
7759 * A new ConditionHost= condition has been added to match
7760 against the hostname (with globs) and machine ID. This is
7761 useful for clusters where a single OS image is used to
7762 provision a large number of hosts which shall run slightly
7763 different sets of services.
7765 * Services which hit the restart limit will now be placed in a
7768 Contributions from: Bertram Poettering, Dave Reisner, Huang
7769 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
7770 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
7774 * When running in --user mode systemd will now become a
7775 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
7776 tree a lot more organized.
7778 * A new PartOf= unit dependency type has been introduced that
7779 may be used to group services in a natural way.
7781 * "systemctl enable" may now be used to enable instances of
7784 * journalctl now prints error log levels in red, and
7785 warning/notice log levels in bright white. It also supports
7786 filtering by log level now.
7788 * cgtop gained a new -n switch (similar to top), to configure
7789 the maximum number of iterations to run for. It also gained
7790 -b, to run in batch mode (accepting no input).
7792 * The suffix ".service" may now be omitted on most systemctl
7793 command lines involving service unit names.
7795 * There's a new bus call in logind to lock all sessions, as
7796 well as a loginctl verb for it "lock-sessions".
7798 * libsystemd-logind.so gained a new call sd_journal_perror()
7799 that works similar to libc perror() but logs to the journal
7800 and encodes structured information about the error number.
7802 * /etc/crypttab entries now understand the new keyfile-size=
7805 * shutdown(8) now can send a (configurable) wall message when
7806 a shutdown is cancelled.
7808 * The mount propagation mode for the root file system will now
7809 default to "shared", which is useful to make containers work
7810 nicely out-of-the-box so that they receive new mounts from
7811 the host. This can be undone locally by running "mount
7812 --make-rprivate /" if needed.
7814 * The prefdm.service file has been removed. Distributions
7815 should maintain this unit downstream if they intend to keep
7816 it around. However, we recommend writing normal unit files
7817 for display managers instead.
7819 * Since systemd is a crucial part of the OS we will now
7820 default to a number of compiler switches that improve
7821 security (hardening) such as read-only relocations, stack
7822 protection, and suchlike.
7824 * The TimeoutSec= setting for services is now split into
7825 TimeoutStartSec= and TimeoutStopSec= to allow configuration
7826 of individual time outs for the start and the stop phase of
7829 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
7830 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
7831 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
7832 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
7833 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
7834 Gundersen, Zbigniew Jędrzejewski-Szmek
7838 * The journal and id128 C APIs are now fully documented as man
7841 * Extra safety checks have been added when transitioning from
7842 the initial RAM disk to the main system to avoid accidental
7845 * /etc/crypttab entries now understand the new keyfile-offset=
7848 * systemctl -t can now be used to filter by unit load state.
7850 * The journal C API gained the new sd_journal_wait() call to
7851 make writing synchronous journal clients easier.
7853 * journalctl gained the new -D switch to show journals from a
7856 * journalctl now displays a special marker between log
7857 messages of two different boots.
7859 * The journal is now explicitly flushed to /var via a service
7860 systemd-journal-flush.service, rather than implicitly simply
7861 by seeing /var/log/journal to be writable.
7863 * journalctl (and the journal C APIs) can now match for much
7864 more complex expressions, with alternatives and
7867 * When transitioning from the initial RAM disk to the main
7868 system we will now kill all processes in a killing spree to
7869 ensure no processes stay around by accident.
7871 * Three new specifiers may be used in unit files: %u, %h, %s
7872 resolve to the user name, user home directory resp. user
7873 shell. This is useful for running systemd user instances.
7875 * We now automatically rotate journal files if their data
7876 object hash table gets a fill level > 75%. We also size the
7877 hash table based on the configured maximum file size. This
7878 together should lower hash collisions drastically and thus
7879 speed things up a bit.
7881 * journalctl gained the new "--header" switch to introspect
7882 header data of journal files.
7884 * A new setting SystemCallFilters= has been added to services
7885 which may be used to apply blacklists or whitelists to
7886 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
7888 * nspawn gained a new --link-journal= switch (and quicker: -j)
7889 to link the container journal with the host. This makes it
7890 very easy to centralize log viewing on the host for all
7891 guests while still keeping the journal files separated.
7893 * Many bugfixes and optimizations
7895 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
7896 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
7897 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
7902 * Several tools now understand kernel command line arguments,
7903 which are only read when run in an initial RAM disk. They
7904 usually follow closely their normal counterparts, but are
7907 * There's a new tool to analyze the readahead files that are
7908 automatically generated at boot. Use:
7910 /usr/lib/systemd/systemd-readahead analyze /.readahead
7912 * We now provide an early debug shell on tty9 if this enabled. Use:
7914 systemctl enable debug-shell.service
7916 * All plymouth related units have been moved into the Plymouth
7917 package. Please make sure to upgrade your Plymouth version
7920 * systemd-tmpfiles now supports getting passed the basename of
7921 a configuration file only, in which case it will look for it
7922 in all appropriate directories automatically.
7924 * udevadm info now takes a /dev or /sys path as argument, and
7925 does the right thing. Example:
7927 udevadm info /dev/sda
7928 udevadm info /sys/class/block/sda
7930 * systemctl now prints a warning if a unit is stopped but a
7931 unit that might trigger it continues to run. Example: a
7932 service is stopped but the socket that activates it is left
7935 * "systemctl status" will now mention if the log output was
7936 shortened due to rotation since a service has been started.
7938 * The journal API now exposes functions to determine the
7939 "cutoff" times due to rotation.
7941 * journald now understands SIGUSR1 and SIGUSR2 for triggering
7942 immediately flushing of runtime logs to /var if possible,
7943 resp. for triggering immediate rotation of the journal
7946 * It is now considered an error if a service is attempted to
7947 be stopped that is not loaded.
7949 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
7951 * systemd-analyze now supports Python 3
7953 * tmpfiles now supports cleaning up directories via aging
7954 where the first level dirs are always kept around but
7955 directories beneath it automatically aged. This is enabled
7956 by prefixing the age field with '~'.
7958 * Seat objects now expose CanGraphical, CanTTY properties
7959 which is required to deal with very fast bootups where the
7960 display manager might be running before the graphics drivers
7961 completed initialization.
7963 * Seat objects now expose a State property.
7965 * We now include RPM macros for service enabling/disabling
7966 based on the preset logic. We recommend RPM based
7967 distributions to make use of these macros if possible. This
7968 makes it simpler to reuse RPM spec files across
7971 * We now make sure that the collected systemd unit name is
7972 always valid when services log to the journal via
7975 * There's a new man page kernel-command-line(7) detailing all
7976 command line options we understand.
7978 * The fstab generator may now be disabled at boot by passing
7979 fstab=0 on the kernel command line.
7981 * A new kernel command line option modules-load= is now understood
7982 to load a specific kernel module statically, early at boot.
7984 * Unit names specified on the systemctl command line are now
7985 automatically escaped as needed. Also, if file system or
7986 device paths are specified they are automatically turned
7987 into the appropriate mount or device unit names. Example:
7989 systemctl status /home
7990 systemctl status /dev/sda
7992 * The SysVConsole= configuration option has been removed from
7993 system.conf parsing.
7995 * The SysV search path is no longer exported on the D-Bus
7998 * The Names= option has been removed from unit file parsing.
8000 * There's a new man page bootup(7) detailing the boot process.
8002 * Every unit and every generator we ship with systemd now
8003 comes with full documentation. The self-explanatory boot is
8006 * A couple of services gained "systemd-" prefixes in their
8007 name if they wrap systemd code, rather than only external
8008 code. Among them fsck@.service which is now
8009 systemd-fsck@.service.
8011 * The HaveWatchdog property has been removed from the D-Bus
8014 * systemd.confirm_spawn= on the kernel command line should now
8017 * There's a new man page crypttab(5) which details all options
8018 we actually understand.
8020 * systemd-nspawn gained a new --capability= switch to pass
8021 additional capabilities to the container.
8023 * timedated will now read known NTP implementation unit names
8024 from /usr/lib/systemd/ntp-units.d/*.list,
8025 systemd-timedated-ntp.target has been removed.
8027 * journalctl gained a new switch "-b" that lists log data of
8028 the current boot only.
8030 * The notify socket is in the abstract namespace again, in
8031 order to support daemons which chroot() at start-up.
8033 * There is a new Storage= configuration option for journald
8034 which allows configuration of where log data should go. This
8035 also provides a way to disable journal logging entirely, so
8036 that data collected is only forwarded to the console, the
8037 kernel log buffer or another syslog implementation.
8039 * Many bugfixes and optimizations
8041 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
8042 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
8043 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
8044 Shawn Landden, Tom Gundersen
8048 * "systemctl help <unit>" now shows the man page if one is
8051 * Several new man pages have been added.
8053 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
8054 MaxLevelConsole= can now be specified in
8055 journald.conf. These options allow reducing the amount of
8056 data stored on disk or forwarded by the log level.
8058 * TimerSlackNSec= can now be specified in system.conf for
8059 PID1. This allows system-wide power savings.
8061 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
8062 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
8067 * logind is now capable of (optionally) handling power and
8068 sleep keys as well as the lid switch.
8070 * journalctl now understands the syntax "journalctl
8071 /usr/bin/avahi-daemon" to get all log output of a specific
8074 * CapabilityBoundingSet= in system.conf now also influences
8075 the capability bound set of usermode helpers of the kernel.
8077 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
8078 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
8079 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
8080 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
8084 * Note that we skipped 139 releases here in order to set the
8085 new version to something that is greater than both udev's
8086 and systemd's most recent version number.
8088 * udev: all udev sources are merged into the systemd source tree now.
8089 All future udev development will happen in the systemd tree. It
8090 is still fully supported to use the udev daemon and tools without
8091 systemd running, like in initramfs or other init systems. Building
8092 udev though, will require the *build* of the systemd tree, but
8093 udev can be properly *run* without systemd.
8095 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
8096 should be used to create dead device nodes as workarounds for broken
8099 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
8100 no longer supported. udev_monitor_new_from_netlink() needs to be
8101 used to subscribe to events.
8103 * udev: when udevd is started by systemd, processes which are left
8104 behind by forking them off of udev rules, are unconditionally cleaned
8105 up and killed now after the event handling has finished. Services or
8106 daemons must be started as systemd services. Services can be
8107 pulled-in by udev to get started, but they can no longer be directly
8108 forked by udev rules.
8110 * udev: the daemon binary is called systemd-udevd now and installed
8111 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
8112 to adapt to that, create symlink, or rename the binary after building
8115 * libudev no longer provides these symbols:
8116 udev_monitor_from_socket()
8117 udev_queue_get_failed_list_entry()
8118 udev_get_{dev,sys,run}_path()
8119 The versions number was bumped and symbol versioning introduced.
8121 * systemd-loginctl and systemd-journalctl have been renamed
8122 to loginctl and journalctl to match systemctl.
8124 * The config files: /etc/systemd/systemd-logind.conf and
8125 /etc/systemd/systemd-journald.conf have been renamed to
8126 logind.conf and journald.conf. Package updates should rename
8127 the files to the new names on upgrade.
8129 * For almost all files the license is now LGPL2.1+, changed
8130 from the previous GPL2.0+. Exceptions are some minor stuff
8131 of udev (which will be changed to LGPL2.1 eventually, too),
8132 and the MIT licensed sd-daemon.[ch] library that is suitable
8133 to be used as drop-in files.
8135 * systemd and logind now handle system sleep states, in
8136 particular suspending and hibernating.
8138 * logind now implements a sleep/shutdown/idle inhibiting logic
8139 suitable for a variety of uses. Soonishly Lennart will blog
8140 about this in more detail.
8142 * var-run.mount and var-lock.mount are no longer provided
8143 (which previously bind mounted these directories to their new
8144 places). Distributions which have not converted these
8145 directories to symlinks should consider stealing these files
8146 from git history and add them downstream.
8148 * We introduced the Documentation= field for units and added
8149 this to all our shipped units. This is useful to make it
8150 easier to explore the boot and the purpose of the various
8153 * All smaller setup units (such as
8154 systemd-vconsole-setup.service) now detect properly if they
8155 are run in a container and are skipped when
8156 appropriate. This guarantees an entirely noise-free boot in
8157 Linux container environments such as systemd-nspawn.
8159 * A framework for implementing offline system updates is now
8160 integrated, for details see:
8161 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
8163 * A new service type Type=idle is available now which helps us
8164 avoiding ugly interleaving of getty output and boot status
8167 * There's now a system-wide CapabilityBoundingSet= option to
8168 globally reduce the set of capabilities for the
8169 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
8170 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
8171 even CAP_NET_ADMIN system-wide for secure systems.
8173 * There are now system-wide DefaultLimitXXX= options to
8174 globally change the defaults of the various resource limits
8175 for all units started by PID 1.
8177 * Harald Hoyer's systemd test suite has been integrated into
8178 systemd which allows easy testing of systemd builds in qemu
8179 and nspawn. (This is really awesome! Ask us for details!)
8181 * The fstab parser is now implemented as generator, not inside
8184 * systemctl will now warn you if .mount units generated from
8185 /etc/fstab are out of date due to changes in fstab that
8186 have not been read by systemd yet.
8188 * systemd is now suitable for usage in initrds. Dracut has
8189 already been updated to make use of this. With this in place
8190 initrds get a slight bit faster but primarily are much
8191 easier to introspect and debug since "systemctl status" in
8192 the host system can be used to introspect initrd services,
8193 and the journal from the initrd is kept around too.
8195 * systemd-delta has been added, a tool to explore differences
8196 between user/admin configuration and vendor defaults.
8198 * PrivateTmp= now affects both /tmp and /var/tmp.
8200 * Boot time status messages are now much prettier and feature
8201 proper english language. Booting up systemd has never been
8204 * Read-ahead pack files now include the inode number of all
8205 files to pre-cache. When the inode changes the pre-caching
8206 is not attempted. This should be nicer to deal with updated
8207 packages which might result in changes of read-ahead
8210 * We now temporaritly lower the kernel's read_ahead_kb variable
8211 when collecting read-ahead data to ensure the kernel's
8212 built-in read-ahead does not add noise to our measurements
8213 of necessary blocks to pre-cache.
8215 * There's now RequiresMountsFor= to add automatic dependencies
8216 for all mounts necessary for a specific file system path.
8218 * MountAuto= and SwapAuto= have been removed from
8219 system.conf. Mounting file systems at boot has to take place
8222 * nspawn now learned a new switch --uuid= to set the machine
8223 ID on the command line.
8225 * nspawn now learned the -b switch to automatically search
8228 * vt102 is now the default TERM for serial TTYs, upgraded from
8231 * systemd-logind now works on VT-less systems.
8233 * The build tree has been reorganized. The individual
8234 components now have directories of their own.
8236 * A new condition type ConditionPathIsReadWrite= is now available.
8238 * nspawn learned the new -C switch to create cgroups for the
8239 container in other hierarchies.
8241 * We now have support for hardware watchdogs, configurable in
8244 * The scheduled shutdown logic now has a public API.
8246 * We now mount /tmp as tmpfs by default, but this can be
8247 masked and /etc/fstab can override it.
8249 * Since udisks does not make use of /media anymore we are not
8250 mounting a tmpfs on it anymore.
8252 * journalctl gained a new --local switch to only interleave
8253 locally generated journal files.
8255 * We can now load the IMA policy at boot automatically.
8257 * The GTK tools have been split off into a systemd-ui.
8259 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
8260 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
8261 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
8262 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
8263 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
8264 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
8265 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
8266 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
8267 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
8272 * This is mostly a bugfix release
8274 * Support optional initialization of the machine ID from the
8275 KVM or container configured UUID.
8277 * Support immediate reboots with "systemctl reboot -ff"
8279 * Show /etc/os-release data in systemd-analyze output
8281 * Many bugfixes for the journal, including endianness fixes and
8282 ensuring that disk space enforcement works
8284 * sd-login.h is C++ compatible again
8286 * Extend the /etc/os-release format on request of the Debian
8289 * We now refuse non-UTF8 strings used in various configuration
8290 and unit files. This is done to ensure we do not pass invalid
8291 data over D-Bus or expose it elsewhere.
8293 * Register Mimo USB Screens as suitable for automatic seat
8296 * Read SELinux client context from journal clients in a race
8299 * Reorder configuration file lookup order. /etc now always
8300 overrides /run in order to allow the administrator to always
8301 and unconditionally override vendor-supplied or
8302 automatically generated data.
8304 * The various user visible bits of the journal now have man
8305 pages. We still lack man pages for the journal API calls
8308 * We now ship all man pages in HTML format again in the
8311 Contributions from: Dave Reisner, Dirk Eibach, Frederic
8312 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
8313 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
8318 * This is mostly a bugfix release
8320 * systems lacking /etc/os-release are no longer supported.
8322 * Various functionality updates to libsystemd-login.so
8324 * Track class of PAM logins to distinguish greeters from
8327 Contributions from: Kay Sievers, Lennart Poettering, Michael
8332 * This is an important bugfix release for v41.
8334 * Building man pages is now optional which should be useful
8335 for those building systemd from git but unwilling to install
8338 * Watchdog support for supervising services is now usable. In
8339 a future release support for hardware watchdogs
8340 (i.e. /dev/watchdog) will be added building on this.
8342 * Service start rate limiting is now configurable and can be
8343 turned off per service. When a start rate limit is hit a
8344 reboot can automatically be triggered.
8346 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
8348 Contributions from: Benjamin Franzke, Bill Nottingham,
8349 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
8350 Schmidt, Michał Górny, Piotr Drąg
8354 * The systemd binary is installed /usr/lib/systemd/systemd now;
8355 An existing /sbin/init symlink needs to be adapted with the
8358 * The code that loads kernel modules has been ported to invoke
8359 libkmod directly, instead of modprobe. This means we do not
8360 support systems with module-init-tools anymore.
8362 * Watchdog support is now already useful, but still not
8365 * A new kernel command line option systemd.setenv= is
8366 understood to set system wide environment variables
8367 dynamically at boot.
8369 * We now limit the set of capabilities of systemd-journald.
8371 * We now set SIGPIPE to ignore by default, since it only is
8372 useful in shell pipelines, and has little use in general
8373 code. This can be disabled with IgnoreSIPIPE=no in unit
8376 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
8377 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
8382 * This is mostly a bugfix release
8384 * We now expose the reason why a service failed in the
8385 "Result" D-Bus property.
8387 * Rudimentary service watchdog support (will be completed over
8388 the next few releases.)
8390 * When systemd forks off in order execute some service we will
8391 now immediately changes its argv[0] to reflect which process
8392 it will execute. This is useful to minimize the time window
8393 with a generic argv[0], which makes bootcharts more useful
8395 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
8396 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
8397 Mike Kazantsev, Ray Strode
8401 * This is mostly a test release, but incorporates many
8404 * New systemd-cgtop tool to show control groups by their
8407 * Linking against libacl for ACLs is optional again. If
8408 disabled, support tracking device access for active logins
8409 goes becomes unavailable, and so does access to the user
8410 journals by the respective users.
8412 * If a group "adm" exists, journal files are automatically
8413 owned by them, thus allow members of this group full access
8414 to the system journal as well as all user journals.
8416 * The journal now stores the SELinux context of the logging
8417 client for all entries.
8419 * Add C++ inclusion guards to all public headers
8421 * New output mode "cat" in the journal to print only text
8422 messages, without any meta data like date or time.
8424 * Include tiny X server wrapper as a temporary stop-gap to
8425 teach XOrg udev display enumeration. This is used by display
8426 managers such as gdm, and will go away as soon as XOrg
8427 learned native udev hotplugging for display devices.
8429 * Add new systemd-cat tool for executing arbitrary programs
8430 with STDERR/STDOUT connected to the journal. Can also act as
8431 BSD logger replacement, and does so by default.
8433 * Optionally store all locally generated coredumps in the
8434 journal along with meta data.
8436 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
8437 writing short strings to files (for usage for /sys), and for
8438 creating symlinks, character and block device nodes.
8440 * New unit file option ControlGroupPersistent= to make cgroups
8441 persistent, following the mechanisms outlined in
8442 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
8444 * Support multiple local RTCs in a sane way
8446 * No longer monopolize IO when replaying readahead data on
8447 rotating disks, since we might starve non-file-system IO to
8448 death, since fanotify() will not see accesses done by blkid,
8451 * Do not show kernel threads in systemd-cgls anymore, unless
8452 requested with new -k switch.
8454 Contributions from: Dan Horák, Kay Sievers, Lennart
8455 Poettering, Michal Schmidt
8459 * This is mostly a test release, but incorporates many
8462 * The git repository moved to:
8463 git://anongit.freedesktop.org/systemd/systemd
8464 ssh://git.freedesktop.org/git/systemd/systemd
8466 * First release with the journal
8467 http://0pointer.de/blog/projects/the-journal.html
8469 * The journal replaces both systemd-kmsg-syslogd and
8470 systemd-stdout-bridge.
8472 * New sd_pid_get_unit() API call in libsystemd-logind
8474 * Many systemadm clean-ups
8476 * Introduce remote-fs-pre.target which is ordered before all
8477 remote mounts and may be used to start services before all
8480 * Added Mageia support
8482 * Add bash completion for systemd-loginctl
8484 * Actively monitor PID file creation for daemons which exit in
8485 the parent process before having finished writing the PID
8486 file in the daemon process. Daemons which do this need to be
8487 fixed (i.e. PID file creation must have finished before the
8488 parent exits), but we now react a bit more gracefully to them.
8490 * Add colourful boot output, mimicking the well-known output
8491 of existing distributions.
8493 * New option PassCredentials= for socket units, for
8494 compatibility with a recent kernel ABI breakage.
8496 * /etc/rc.local is now hooked in via a generator binary, and
8497 thus will no longer act as synchronization point during
8500 * systemctl list-unit-files now supports --root=.
8502 * systemd-tmpfiles now understands two new commands: z, Z for
8503 relabelling files according to the SELinux database. This is
8504 useful to apply SELinux labels to specific files in /sys,
8507 * Output of SysV services is now forwarded to both the console
8508 and the journal by default, not only just the console.
8510 * New man pages for all APIs from libsystemd-login.
8512 * The build tree got reorganized and the build system is a
8513 lot more modular allowing embedded setups to specifically
8514 select the components of systemd they are interested in.
8516 * Support for Linux systems lacking the kernel VT subsystem is
8519 * configure's --with-rootdir= got renamed to
8520 --with-rootprefix= to follow the naming used by udev and
8523 * Unless specified otherwise we will now install to /usr instead
8524 of /usr/local by default.
8526 * Processes with '@' in argv[0][0] are now excluded from the
8527 final shut-down killing spree, following the logic explained
8529 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
8531 * All processes remaining in a service cgroup when we enter
8532 the START or START_PRE states are now killed with
8533 SIGKILL. That means it is no longer possible to spawn
8534 background processes from ExecStart= lines (which was never
8535 supported anyway, and bad style).
8537 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
8538 reloading of units together.
8540 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
8541 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
8542 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
8543 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
8544 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
~ianmdlvl / elogind.git / blob