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 will name network interfaces differently than in previous
7 versions for virtual network interfaces created with SR-IOV and NPAR
8 and for devices where the PCI network controller device does not have
9 a slot number associated.
11 SR-IOV virtual devices are now named based on the name of the parent
12 interface, with a suffix of "v<N>", where <N> is the virtual device
13 number. Previously those virtual devices were named as if completely
16 The ninth and later NPAR virtual devices will be named following the
17 scheme used for the first eight NPAR partitions. Previously those
18 devices were not renamed and the kernel default (eth<n>) was used.
20 "net_id" will also generate names for PCI devices where the PCI
21 network controller device does not have an associated slot number
22 itself, but one of its parents does. Previously those devices were
23 not renamed and the kernel default (eth<n>) was used.
25 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
26 elogind.service. Since v235, IPAddressDeny=any has been set to
27 the unit. So, it is expected that the default behavior of
28 elogind is not changed. However, if distribution packagers or
29 administrators disabled or modified IPAddressDeny= setting by a
30 drop-in config file, then it may be necessary to update the file to
31 re-enable AF_INET and AF_INET6 to support network user name services,
34 * When the RestrictNamespaces= unit property is specified multiple
35 times, then the specified types are merged now. Previously, only the
36 last assignment was used. So, if distribution packagers or
37 administrators modified the setting by a drop-in config file, then it
38 may be necessary to update the file.
40 * When OnFailure= is used in combination with Restart= on a service
41 unit, then the specified units will no longer be triggered on
42 failures that result in restarting. Previously, the specified units
43 would be activated each time the unit failed, even when the unit was
44 going to be restarted automatically. This behaviour contradicted the
45 documentation. With this release the code is adjusted to match the
48 * elogind-tmpfiles will now print a notice whenever it encounters
49 tmpfiles.d/ lines referencing the /var/run/ directory. It will
50 recommend reworking them to use the /run/ directory instead (for
51 which /var/run/ is simply a symlinked compatibility alias). This way
52 elogind-tmpfiles can properly detect line conflicts and merge lines
53 referencing the same file by two paths, without having to access
56 * systemctl disable/unmask/preset/preset-all cannot be used with
57 --runtime. Previously this was allowed, but resulted in unintuitive
58 behaviour that wasn't useful. systemctl disable/unmask will now undo
59 both runtime and persistent enablement/masking, i.e. it will remove
60 any relevant symlinks both in /run and /etc.
62 * sd-boot acquired new loader configuration settings to optionally turn
63 off Windows and MacOS boot partition discovery as well as
64 reboot-into-firmware menu items. It is also able to pick a better
65 screen resolution for HiDPI systems, and now provides loader
66 configuration settings to change the resolution explicitly.
68 * The elogind-resolve tool has been renamed to resolvectl (it also
69 * elogind-resolved now supports DNS-over-TLS. It's still
70 turned off by default, use DNSOverTLS=opportunistic to turn it on in
71 resolved.conf. We intend to make this the default as soon as couple
72 of additional techniques for optimizing the initial latency caused by
73 establishing a TLS/TCP connection are implemented.
75 * elogind-resolved.service and elogind-networkd.service now set
76 DynamicUser=yes. The users elogind-resolve and elogind-network are
77 not created by elogind-sysusers.
79 remains available under the old name, for compatibility), and its
80 interface is now verb-based, similar in style to the other <xyz>ctl
81 tools, such as systemctl or loginctl.
83 * The resolvectl/elogind-resolve tool also provides 'resolvconf'
84 compatibility. It may be symlinked under the 'resolvconf' name, in
85 which case it will take arguments and input compatible with the
86 Debian and FreeBSD resolvconf tool.
88 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
89 where the system initially suspends, and after a time-out resumes and
92 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
93 set the client will only send a DUID as client identifier.
95 * The nss-elogind glibc NSS module will now enumerate dynamic users and
96 groups in effect. Previously, it could resolve UIDs/GIDs to user
97 names/groups and vice versa, but did not support enumeration.
99 * journald's Compress= configuration setting now optionally accepts a
100 byte threshold value. All journal objects larger than this threshold
101 will be compressed, smaller ones will not. Previously this threshold
102 was not configurable and set to 512.
104 * A new system.conf setting NoNewPrivileges= is now available which may
105 be used to turn off acquisition of new privileges system-wide
106 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
107 for all its children). Note that turning this option on means setuid
108 binaries and file system capabilities lose their special powers.
109 While turning on this option is a big step towards a more secure
110 system, doing so is likely to break numerous pre-existing UNIX tools,
111 in particular su and sudo.
113 * A new service elogind-time-sync-wait.service has been added. If
114 enabled it will delay the time-sync.target unit at boot until time
115 synchronization has been received from the network. This
116 functionality is useful on systems lacking a local RTC or where it is
117 acceptable that the boot process shall be delayed by external network
120 * When hibernating, elogind will now inform the kernel of the image
121 write offset, on kernels new enough to support this. This means swap
122 files should work for hibernation now.
124 * When loading unit files, elogind will now look for drop-in unit files
125 extensions in additional places. Previously, for a unit file name
126 "foo-bar-baz.service" it would look for dropin files in
127 "foo-bar-baz.service.d/*.conf". Now, it will also look in
128 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
129 service name truncated after all inner dashes. This scheme allows
130 writing drop-ins easily that apply to a whole set of unit files at
131 once. It's particularly useful for mount and slice units (as their
132 naming is prefix based), but is also useful for service and other
133 units, for packages that install multiple unit files at once,
134 following a strict naming regime of beginning the unit file name with
135 the package's name. Two new specifiers are now supported in unit
136 files to match this: %j and %J are replaced by the part of the unit
137 name following the last dash.
139 * Unit files and other configuration files that support specifier
140 expansion now understand another two new specifiers: %T and %V will
141 resolve to /tmp and /var/tmp respectively, or whatever temporary
142 directory has been set for the calling user.
144 * The ExecStart= lines of unit files are no longer required to
145 reference absolute paths. If non-absolute paths are specified the
146 specified binary name is searched within the service manager's
147 built-in $PATH, which may be queried with 'elogind-path
148 search-binaries-default'. It's generally recommended to continue to
149 use absolute paths for all binaries specified in unit files.
151 * Units gained a new load state "bad-setting", which is used when a
152 unit file was loaded, but contained fatal errors which prevent it
153 from being started (for example, a service unit has been defined
154 lacking both ExecStart= and ExecStop= lines).
156 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
157 support alternative debuggers, for example lldb. The old name
158 continues to be available however, for compatibility reasons. Use the
159 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
160 to pick an alternative debugger instead of the default gdb.
162 * systemctl and the other tools will now output escape sequences that
163 generate proper clickable hyperlinks in various terminal emulators
164 where useful (for example, in the "systemctl status" output you can
165 now click on the unit file name to quickly open it in the
166 editor/viewer of your choice). Note that not all terminal emulators
167 support this functionality yet, but many do. Unfortunately, the
168 "less" pager doesn't support this yet, hence this functionality is
169 currently automatically turned off when a pager is started (which
170 happens quite often due to auto-paging). We hope to remove this
171 limitation as soon as "less" learns these escape sequences. This new
172 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
173 environment variable. For details on these escape sequences see:
174 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
176 * networkd's .network files now support a new IPv6MTUBytes= option for
177 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
178 option in the [Route] section to configure the MTU to use for
179 specific routes. It also gained support for configuration of the DHCP
180 "UserClass" option through the new UserClass= setting. It gained
181 three new options in the new [CAN] section for configuring CAN
182 networks. The MULTICAST and ALLMULTI interface flags may now be
183 controlled explicitly with the new Multicast= and AllMulticast=
186 * networkd will now automatically make use of the kernel's route
187 expiration feature, if it is available.
189 * udevd's .link files now support setting the number of receive and
190 transmit channels, using the RxChannels=, TxChannels=,
191 OtherChannels=, CombinedChannels= settings.
193 * Support for UDPSegmentationOffload= has been removed, given its
194 limited support in hardware, and waning software support.
196 * networkd's .netdev files now support creating "netdevsim" interfaces.
198 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
199 to query the unit belonging to a specific kernel control group.
201 * elogind-analyze gained a new verb "cat-config", which may be used to
202 dump the contents of any configuration file, with all its matching
203 drop-in files added in, and honouring the usual search and masking
204 logic applied to elogind configuration files. For example use
205 "elogind-analyze cat-config elogind/system.conf" to get the complete
206 system configuration file of elogind how it would be loaded by PID 1
207 itself. Similar to this, various tools such as elogind-tmpfiles or
208 elogind-sysusers, gained a new option "--cat-config", which does the
209 corresponding operation for their own configuration settings. For
210 example, "elogind-tmpfiles --cat-config" will now output the full
211 list of tmpfiles.d/ lines in place.
213 * timedatectl gained two new verbs "timesync-status" (to show the
214 current NTP synchronization state of elogind-timesyncd) and
215 "show-timesync" (to show bus properties of elogind-timesyncd).
217 * elogind-timesyncd gained a bus interface on which it exposes details
220 * elogind-nspawn gained a new --rlimit= switch for setting initial
221 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
222 understood by elogind-timedated. It takes a colon-separated list of
223 unit names of NTP client services. The list is used by
224 "timedatectl set-ntp".
226 resource limits for the container payload. There's a new switch
227 --hostname= to explicitly override the container's hostname. A new
228 --no-new-privileges= switch may be used to control the
229 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
230 --oom-score-adjust= switch controls the OOM scoring adjustment value
231 for the payload. The new --cpu-affinity= switch controls the CPU
232 affinity of the container payload. The new --resolv-conf= switch
233 allows more detailed control of /etc/resolv.conf handling of the
234 container. Similarly, the new --timezone= switch allows more detailed
235 control of /etc/localtime handling of the container.
237 * elogind-detect-virt gained a new --list switch, which will print a
238 list of all currently known VM and container environments.
240 * Support for "Portable Services" has been added, see
241 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
242 experimental, but this is expected to change soon. Reflecting this
243 experimental state, the "portablectl" binary is not installed into
244 /usr/bin yet. The binary has to be called with the full path
245 /usr/lib/elogind/portablectl instead.
247 * journalctl's and systemctl's -o switch now knows a new log output
248 mode "with-unit". The output it generates is very similar to the
249 regular "short" mode, but displays the unit name instead of the
250 syslog tag for each log line. Also, the date is shown with timezone
251 information. This mode is probably more useful than the classic
252 "short" output mode for most purposes, except where pixel-perfect
253 compatibility with classic /var/log/messages formatting is required.
255 * A new --dump-bus-properties switch has been added to the elogind
256 binary, which may be used to dump all supported D-Bus properties.
257 (Options which are still supported, but are deprecated, are *not*
260 * sd-bus gained a set of new calls:
261 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
262 enable/disable the "floating" state of a bus slot object,
263 i.e. whether the slot object pins the bus it is allocated for into
264 memory or if the bus slot object gets disconnected when the bus goes
265 away. sd_bus_open_with_description(),
266 sd_bus_open_user_with_description(),
267 sd_bus_open_system_with_description() may be used to allocate bus
268 objects and set their description string already during allocation.
270 * sd-event gained support for watching inotify events from the event
271 loop, in an efficient way, sharing inotify handles between multiple
272 users. For this a new function sd_event_add_inotify() has been added.
274 * sd-event and sd-bus gained support for calling special user-supplied
275 destructor functions for userdata pointers associated with
276 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
277 functions sd_bus_slot_set_destroy_callback,
278 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
279 sd_bus_track_get_destroy_callback,
280 sd_event_source_set_destroy_callback,
281 sd_event_source_get_destroy_callback have been added.
283 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
285 * PID 1 will now automatically reschedule .timer units whenever the
286 local timezone changes. (They previously got rescheduled
287 automatically when the system clock changed.)
289 * New documentation has been added to document cgroups delegation,
290 portable services and the various code quality tools we have set up:
292 https://github.com/systemd/systemd/blob/master/doc/CGROUP_DELEGATION.md
293 https://github.com/systemd/systemd/blob/master/doc/PORTABLE_SERVICES.md
294 https://github.com/systemd/systemd/blob/master/doc/CODE_QUALITY.md
296 * pam_elogind will now look for PAM userdata fields elogind.memory_max,
297 elogind.tasks_max, elogind.cpu_weight, elogind.io_weight set by
298 earlier PAM modules. The data in these fields is used to initialize
299 the session scope's resource properties. Thus external PAM modules
300 may now configure per-session limits, for example sourced from
301 external user databases.
303 * socket units with Accept=yes will now maintain a "refused" counter in
304 addition to the existing "accepted" counter, counting connections
305 refused due to the enforced limits.
307 * The "elogind-path search-binaries-default" command may now be use to
308 query the default, built-in $PATH PID 1 will pass to the services it
311 * A new unit file setting PrivateMounts= has been added. It's a boolean
312 option. If enabled the unit's processes are invoked in their own file
313 system namespace. Note that this behaviour is also implied if any
314 other file system namespacing options (such as PrivateTmp=,
315 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
316 primarily useful for services that do not use any of the other file
317 system namespacing options. One such service is elogind-udevd.service
318 wher this is now used by default.
320 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
321 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
322 J. Murrell, Bruno Vernay, Chris Lesiak, Christian Brauner, Christian
323 Hesse, Daniel Dao, Daniel Lin, Danylo Korostil, Davide Cavalca, David
324 Tardon, Dimitri John Ledkov, Dmitriy Geels, Douglas Christman, Elia
325 Geretto, emelenas, Emil Velikov, Evgeny Vereshchagin, Felipe Sateler,
326 Feng Sun, Filipe Brandenburger, Franck Bui, futpib, Giuseppe Scrivano,
327 Guillem Jover, guixxx, Hans de Goede, Henrique Dante de Almeida, Hiram
328 van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov, Iwan Timmer,
329 James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir, João Paulo
330 Rechi Vita, Joost Heitbrink, Jui-Chi Ricky Liang, Jürg Billeter,
331 Kai-Heng Feng, Karol Augustin, Krzysztof Nowicki, Lauri Tirkkonen,
332 Lennart Poettering, Leonard König, Long Li, Luca Boccassi, Lucas
333 Werkmeister, Marcel Hoppe, Marc Kleine-Budde, Mario Limonciello, Martin
334 Jansa, Martin Wilck, Mathieu Malaterre, Matteo F. Vescovi, Matthew
335 McGinn, Matthias-Christian Ott, Michael Biebl, Michael Olbrich, Michael
336 Prokop, Michal Koutný, Michal Sekletar, Mike Gilbert, Mikhail Kasimov,
337 Milan Broz, Milan Pässler, Muhammet Kara, Nicolas Boichat, Omer Katz,
338 Paride Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter
339 A. Bigot, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
340 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
341 Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez, Salvo Tomaselli,
342 Sebastian Reichel, Sergio Lindo Mansilla, Stefan Schweter, Stephen
343 Hemminger, Stuart Hayes, Susant Sahani, Sylvain Plantefève, Thomas
344 H. P. Andersen, Tobias Jungel, Tomasz Torcz, Vito Caputo, Will Dietz,
345 Will Thompson, Wim van Mourik, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
351 * The MemoryAccounting= unit property now defaults to on. After
352 discussions with the upstream control group maintainers we learnt
353 that the negative impact of cgroup memory accounting on current
354 kernels is finally relatively minimal, so that it should be safe to
355 enable this by default without affecting system performance. Besides
356 memory accounting only task accounting is turned on by default, all
357 other forms of resource accounting (CPU, IO, IP) remain off for now,
358 because it's not clear yet that their impact is small enough to move
359 from opt-in to opt-out. We recommend downstreams to leave memory
360 accounting on by default if kernel 4.14 or higher is primarily
361 used. On very resource constrained systems or when support for old
362 kernels is a necessity, -Dmemory-accounting-default=false can be used
363 to revert this change.
365 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
366 %udev_rules_update) and the journal catalog (%journal_catalog_update)
367 from the upgrade scriptlets of individual packages now do nothing.
368 Transfiletriggers have been added which will perform those updates
369 once at the end of the transaction.
371 Similar transfiletriggers have been added to execute any sysctl.d
372 and binfmt.d rules. Thus, it should be unnecessary to provide any
373 scriptlets to execute this configuration from package installation
376 * elogind-sysusers gained a mode where the configuration to execute is
377 specified on the command line, but this configuration is not executed
378 directly, but instead it is merged with the configuration on disk,
379 and the result is executed. This is useful for package installation
380 scripts which want to create the user before installing any files on
381 disk (in case some of those files are owned by that user), while
382 still allowing local admin overrides.
384 This functionality is exposed to rpm scriptlets through a new
385 %sysusers_create_package macro. Old %sysusers_create and
386 %sysusers_create_inline macros are deprecated.
388 A transfiletrigger for sysusers.d configuration is now installed,
389 which means that it should be unnecessary to call elogind-sysusers from
390 package installation scripts, unless the package installs any files
391 owned by those newly-created users, in which case
392 %sysusers_create_package should be used.
394 * Analogous change has been done for elogind-tmpfiles: it gained a mode
395 where the command-line configuration is merged with the configuration
396 on disk. This is exposed as the new %tmpfiles_create_package macro,
397 and %tmpfiles_create is deprecated. A transfiletrigger is installed
398 for tmpfiles.d, hence it should be unnecessary to call elogind-tmpfiles
399 from package installation scripts.
401 * sysusers.d configuration for a user may now also specify the group
402 number, in addition to the user number ("u username 123:456"), or
403 without the user number ("u username -:456").
405 * Configution items for elogind-sysusers can now be specified as
406 positional arguments when the new --inline switch is used.
408 * The login shell of users created through sysusers.d may now be
409 specified (previously, it was always /bin/sh for root and
410 /sbin/nologin for other users).
412 * elogind-analyze gained a new --global switch to look at global user
413 configuration. It also gained a unit-paths verb to list the unit load
414 paths that are compiled into elogind (which can be used with
415 --elogind, --user, or --global).
417 * udevadm trigger gained a new --settle/-w option to wait for any
418 triggered events to finish (but just those, and not any other events
419 which are triggered meanwhile).
421 * The action that elogind takes when the lid is closed and the
422 machine is connected to external power can now be configured using
423 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
424 was determined by HandleLidSwitch=, and, for backwards compatibility,
425 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
427 * journalctl will periodically call sd_journal_process() to make it
428 resilient against inotify queue overruns when journal files are
429 rotated very quickly.
431 * Two new functions in libelogind — sd_bus_get_n_queued_read and
432 sd_bus_get_n_queued_write — may be used to check the number of
433 pending bus messages.
435 * elogind gained a new
436 org.freedesktop.elogind1.Manager.AttachProcessesToUnit dbus call
437 which can be used to migrate foreign processes to scope and service
438 units. The primary user for this new API is elogind itself: the
439 elogind --user instance uses this call of the elogind --system
440 instance to migrate processes if it itself gets the request to
441 migrate processes and the kernel refuses this due to access
442 restrictions. Thanks to this "elogind-run --scope --user …" works
443 again in pure cgroups v2 environments when invoked from the user
446 * A new TemporaryFileSystem= setting can be used to mask out part of
447 the real file system tree with tmpfs mounts. This may be combined
448 with BindPaths= and BindReadOnlyPaths= to hide files or directories
449 not relevant to the unit, while still allowing some paths lower in
450 the tree to be accessed.
452 ProtectHome=tmpfs may now be used to hide user home and runtime
453 directories from units, in a way that is mostly equivalent to
454 "TemporaryFileSystem=/home /run/user /root".
456 * Non-service units are now started with KeyringMode=shared by default.
457 This means that mount and swapon and other mount tools have access
458 to keys in the main keyring.
460 * /sys/fs/bpf is now mounted automatically.
462 * QNX virtualization is now detected by elogind-detect-virt and may
463 be used in ConditionVirtualization=.
465 * IPAccounting= may now be enabled also for slice units.
467 * A new -Dsplit-bin= build configuration switch may be used to specify
468 whether bin and sbin directories are merged, or if they should be
469 included separately in $PATH and various listings of executable
470 directories. The build configuration scripts will try to autodetect
471 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
472 system, but distributions are encouraged to configure this
475 * A new -Dok-color= build configuration switch may be used to change
476 the colour of "OK" status messages.
478 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
479 PrivateNetwork=yes was buggy in previous versions of elogind. This
480 means that after the upgrade and daemon-reexec, any such units must
483 * INCOMPATIBILITY: as announced in the NEWS for 237, elogind-tmpfiles
484 will not exclude read-only files owned by root from cleanup.
486 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
487 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
488 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
489 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
490 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
491 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
492 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
493 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
494 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
495 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
496 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
497 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
498 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
499 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
500 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
501 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
507 * Some keyboards come with a zoom see-saw or rocker which until now got
508 mapped to the Linux "zoomin/out" keys in hwdb. However, these
509 keycodes are not recognized by any major desktop. They now produce
510 Up/Down key events so that they can be used for scrolling.
512 * INCOMPATIBILITY: elogind-tmpfiles' "f" lines changed behaviour
513 slightly: previously, if an argument was specified for lines of this
514 type (i.e. the right-most column was set) this string was appended to
515 existing files each time elogind-tmpfiles was run. This behaviour was
516 different from what the documentation said, and not particularly
517 useful, as repeated elogind-tmpfiles invocations would not be
518 idempotent and grow such files without bounds. With this release
519 behaviour has been altered slightly, to match what the documentation
520 says: lines of this type only have an effect if the indicated files
521 don't exist yet, and only then the argument string is written to the
524 * FUTURE INCOMPATIBILITY: In elogind v238 we intend to slightly change
525 elogind-tmpfiles behaviour: previously, read-only files owned by root
526 were always excluded from the file "aging" algorithm (i.e. the
527 automatic clean-up of directories like /tmp based on
528 atime/mtime/ctime). We intend to drop this restriction, and age files
529 by default even when owned by root and read-only. This behaviour was
530 inherited from older tools, but there have been requests to remove
531 it, and it's not obvious why this restriction was made in the first
532 place. Please speak up now, if you are aware of software that reqires
533 this behaviour, otherwise we'll remove the restriction in v238.
535 * A new environment variable $SYSTEMD_OFFLINE is now understood by
536 systemctl. It takes a boolean argument. If on, systemctl assumes it
537 operates on an "offline" OS tree, and will not attempt to talk to the
538 service manager. Previously, this mode was implicitly enabled if a
539 chroot() environment was detected, and this new environment variable
540 now provides explicit control.
542 * .path and .socket units may now be created transiently, too.
543 Previously only service, mount, automount and timer units were
544 supported as transient units. The elogind-run tool has been updated
545 to expose this new functionality, you may hence use it now to bind
546 arbitrary commands to path or socket activation on-the-fly from the
547 command line. Moreover, almost all properties are now exposed for the
548 unit types that already supported transient operation.
550 * The elogind-mount command gained support for a new --owner= parameter
551 which takes a user name, which is then resolved and included in uid=
552 and gid= mount options string of the file system to mount.
554 * A new unit condition ConditionControlGroupController= has been added
555 that checks whether a specific cgroup controller is available.
557 * Unit files, udev's .link files, and elogind-networkd's .netdev and
558 .network files all gained support for a new condition
559 ConditionKernelVersion= for checking against specific kernel
562 * In elogind-networkd, the [IPVLAN] section in .netdev files gained
563 support for configuring device flags in the Flags= setting. In the
564 same files, the [Tunnel] section gained support for configuring
565 AllowLocalRemote=. The [Route] section in .network files gained
566 support for configuring InitialCongestionWindow=,
567 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
568 understands RapidCommit=.
570 * elogind-networkd's DHCPv6 support gained support for Prefix
573 * sd-bus gained support for a new "watch-bind" feature. When this
574 feature is enabled, an sd_bus connection may be set up to connect to
575 an AF_UNIX socket in the file system as soon as it is created. This
576 functionality is useful for writing early-boot services that
577 automatically connect to the system bus as soon as it is started,
578 without ugly time-based polling. elogind-networkd and
579 elogind-resolved have been updated to make use of this
580 functionality. busctl exposes this functionality in a new
581 --watch-bind= command line switch.
583 * sd-bus will now optionally synthesize a local "Connected" signal as
584 soon as a D-Bus connection is set up fully. This message mirrors the
585 already existing "Disconnected" signal which is synthesized when the
586 connection is terminated. This signal is generally useful but
587 particularly handy in combination with the "watch-bind" feature
588 described above. Synthesizing of this message has to be requested
589 explicitly through the new API call sd_bus_set_connected_signal(). In
590 addition a new call sd_bus_is_ready() has been added that checks
591 whether a connection is fully set up (i.e. between the "Connected" and
592 "Disconnected" signals).
594 * sd-bus gained two new calls sd_bus_request_name_async() and
595 sd_bus_release_name_async() for asynchronously registering bus
596 names. Similar, there is now sd_bus_add_match_async() for installing
597 a signal match asynchronously. All of elogind's own services have
598 been updated to make use of these calls. Doing these operations
599 asynchronously has two benefits: it reduces the risk of deadlocks in
600 case of cyclic dependencies between bus services, and it speeds up
601 service initialization since synchronization points for bus
602 round-trips are removed.
604 * sd-bus gained two new calls sd_bus_match_signal() and
605 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
606 and sd_bus_add_match_async() but instead of taking a D-Bus match
607 string take match fields as normal function parameters.
609 * sd-bus gained two new calls sd_bus_set_sender() and
610 sd_bus_message_set_sender() for setting the sender name of outgoing
611 messages (either for all outgoing messages or for just one specific
612 one). These calls are only useful in direct connections as on
613 brokered connections the broker fills in the sender anyway,
614 overwriting whatever the client filled in.
616 * sd-event gained a new pseudo-handle that may be specified on all API
617 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
618 used this refers to the default event loop object of the calling
619 thread. Note however that this does not implicitly allocate one —
620 which has to be done prior by using sd_event_default(). Similarly
621 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
622 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
623 to the default bus of the specified type of the calling thread. Here
624 too this does not implicitly allocate bus connection objects, this
625 has to be done prior with sd_bus_default() and friends.
627 * sd-event gained a new call pair
628 sd_event_source_{get|set}_io_fd_own(). This may be used to request
629 automatic closure of the file descriptor an IO event source watches
630 when the event source is destroyed.
632 * elogind-networkd gained support for natively configuring WireGuard
635 * In previous versions elogind synthesized user records both for the
636 "nobody" (UID 65534) and "root" (UID 0) users in nss-elogind and
637 internally. In order to simplify distribution-wide renames of the
638 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
639 new transitional flag file has been added: if
640 /etc/elogind/dont-synthesize-nobody exists synthesizing of the 65534
641 user and group record within the elogind codebase is disabled.
643 * elogind-notify gained a new --uid= option for selecting the source
644 user/UID to use for notification messages sent to the service
647 * journalctl gained a new --grep= option to list only entries in which
648 the message matches a certain pattern. By default matching is case
649 insensitive if the pattern is lowercase, and case sensitive
650 otherwise. Option --case-sensitive=yes|no can be used to override
651 this an specify case sensitivity or case insensitivity.
653 * There's now a "elogind-analyze service-watchdogs" command for printing
654 the current state of the service runtime watchdog, and optionally
655 enabling or disabling the per-service watchdogs system-wide if given a
656 boolean argument (i.e. the concept you configure in WatchdogSec=), for
657 debugging purposes. There's also a kernel command line option
658 elogind.service_watchdogs= for controlling the same.
660 * Two new "log-level" and "log-target" options for elogind-analyze were
661 added that merge the now deprecated get-log-level, set-log-level and
662 get-log-target, set-log-target pairs. The deprecated options are still
663 understood for backwards compatibility. The two new options print the
664 current value when no arguments are given, and set them when a
665 level/target is given as an argument.
667 * sysusers.d's "u" lines now optionally accept both a UID and a GID
668 specification, separated by a ":" character, in order to create users
669 where UID and GID do not match.
671 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
672 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
673 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
674 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
675 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
676 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
677 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
678 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
679 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
680 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
681 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
682 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
683 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
684 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
685 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
686 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
687 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
688 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
689 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
690 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
697 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
698 in v235 has been extended to also set the dummy.ko module option
699 numdummies=0, preventing the kernel from automatically creating
700 dummy0. All dummy interfaces must now be explicitly created.
702 * Unknown '%' specifiers in configuration files are now rejected. This
703 applies to units and tmpfiles.d configuration. Any percent characters
704 that are followed by a letter or digit that are not supposed to be
705 interpreted as the beginning of a specifier should be escaped by
706 doubling ("%%"). (So "size=5%" is still accepted, as well as
707 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
708 valid specifiers today.)
710 * systemd-resolved now maintains a new dynamic
711 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
712 recommended to make /etc/resolv.conf a symlink to it. This file
713 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
714 includes dynamically acquired search domains, achieving more correct
715 DNS resolution by software that bypasses local DNS APIs such as NSS.
717 * The "uaccess" udev tag has been dropped from /dev/kvm and
718 /dev/dri/renderD*. These devices now have the 0666 permissions by
719 default (but this may be changed at build-time). /dev/dri/renderD*
720 will now be owned by the "render" group along with /dev/kfd.
722 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
723 systemd-journal-gatewayd.service and
724 systemd-journal-upload.service. This means "nss-systemd" must be
725 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
726 services are resolved properly.
728 * In /etc/fstab two new mount options are now understood:
729 x-systemd.makefs and x-systemd.growfs. The former has the effect that
730 the configured file system is formatted before it is mounted, the
731 latter that the file system is resized to the full block device size
732 after it is mounted (i.e. if the file system is smaller than the
733 partition it resides on, it's grown). This is similar to the fsck
734 logic in /etc/fstab, and pulls in systemd-makefs@.service and
735 systemd-growfs@.service as necessary, similar to
736 systemd-fsck@.service. Resizing is currently only supported on ext4
739 * In systemd-networkd, the IPv6 RA logic now optionally may announce
740 DNS server and domain information.
742 * Support for the LUKS2 on-disk format for encrypted partitions has
743 been added. This requires libcryptsetup2 during compilation and
746 * The systemd --user instance will now signal "readiness" when its
747 basic.target unit has been reached, instead of when the run queue ran
748 empty for the first time.
750 * Tmpfiles.d with user configuration are now also supported.
751 systemd-tmpfiles gained a new --user switch, and snippets placed in
752 ~/.config/user-tmpfiles.d/ and corresponding directories will be
753 executed by systemd-tmpfiles --user running in the new
754 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
755 running in the user session.
757 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
758 %S resolves to the top-level state directory (/var/lib for the system
759 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
760 top-level cache directory (/var/cache for the system instance,
761 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
762 logs directory (/var/log for the system instance,
763 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
764 existing %t specifier, that resolves to the top-level runtime
765 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
768 * journalctl learnt a new parameter --output-fields= for limiting the
769 set of journal fields to output in verbose and JSON output modes.
771 * systemd-timesyncd's configuration file gained a new option
772 RootDistanceMaxSec= for setting the maximum root distance of servers
773 it'll use, as well as the new options PollIntervalMinSec= and
774 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
776 * bootctl gained a new command "list" for listing all available boot
777 menu items on systems that follow the boot loader specification.
779 * systemctl gained a new --dry-run switch that shows what would be done
780 instead of doing it, and is currently supported by the shutdown and
783 * ConditionSecurity= can now detect the TOMOYO security module.
785 * Unit file [Install] sections are now also respected in unit drop-in
786 files. This is intended to be used by drop-ins under /usr/lib/.
788 * systemd-firstboot may now also set the initial keyboard mapping.
790 * Udev "changed" events for devices which are exposed as systemd
791 .device units are now propagated to units specified in
792 ReloadPropagatedFrom= as reload requests.
794 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
795 unit template name (i.e. a name in the form of 'foobar@.service',
796 without the instance component between the '@' and - the '.'), then
797 the escaped sysfs path of the device is automatically used as the
800 * SystemCallFilter= in unit files has been extended so that an "errno"
801 can be specified individually for each system call. Example:
802 SystemCallFilter=~uname:EILSEQ.
804 * The cgroup delegation logic has been substantially updated. Delegate=
805 now optionally takes a list of controllers (instead of a boolean, as
806 before), which lists the controllers to delegate at least.
808 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
810 * A new LogLevelMax= setting configures the maximum log level any
811 process of the service may log at (i.e. anything with a lesser
812 priority than what is specified is automatically dropped). A new
813 LogExtraFields= setting allows configuration of additional journal
814 fields to attach to all log records generated by any of the unit's
817 * New StandardInputData= and StandardInputText= settings along with the
818 new option StandardInput=data may be used to configure textual or
819 binary data that shall be passed to the executed service process via
820 standard input, encoded in-line in the unit file.
822 * StandardInput=, StandardOutput= and StandardError= may now be used to
823 connect stdin/stdout/stderr of executed processes directly with a
824 file or AF_UNIX socket in the file system, using the new "file:" option.
826 * A new unit file option CollectMode= has been added, that allows
827 tweaking the garbage collection logic for units. It may be used to
828 tell systemd to garbage collect units that have failed automatically
829 (normally it only GCs units that exited successfully). systemd-run
830 and systemd-mount expose this new functionality with a new -G option.
832 * "machinectl bind" may now be used to bind mount non-directories
833 (i.e. regularfiles, devices, fifos, sockets).
835 * systemd-analyze gained a new verb "calendar" for validating and
836 testing calendar time specifications to use for OnCalendar= in timer
837 units. Besides validating the expression it will calculate the next
838 time the specified expression would elapse.
840 * In addition to the pre-existing FailureAction= unit file setting
841 there's now SuccessAction=, for configuring a shutdown action to
842 execute when a unit completes successfully. This is useful in
843 particular inside containers that shall terminate after some workload
844 has been completed. Also, both options are now supported for all unit
845 types, not just services.
847 * networkds's IP rule support gained two new options
848 IncomingInterface= and OutgoingInterface= for configuring the incoming
849 and outgoing interfaces of configured rules. systemd-networkd also
850 gained support for "vxcan" network devices.
852 * networkd gained a new setting RequiredForOnline=, taking a
853 boolean. If set, systemd-wait-online will take it into consideration
854 when determining that the system is up, otherwise it will ignore the
855 interface for this purpose.
857 * The sd_notify() protocol gained support for a new operation: with
858 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
859 store again, ahead of POLLHUP or POLLERR when they are removed
862 * A new document doc/UIDS-GIDS.md has been added to the source tree,
863 that documents the UID/GID range and assignment assumptions and
864 requirements of systemd.
866 * The watchdog device PID 1 will ping may now be configured through the
867 WatchdogDevice= configuration file setting, or by setting the
868 systemd.watchdog_service= kernel commandline option.
870 * systemd-resolved's gained support for registering DNS-SD services on
871 the local network using MulticastDNS. Services may either be
872 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
873 the same dir below /run, /usr/lib), or through its D-Bus API.
875 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
876 extend the effective start, runtime, and stop time. The service must
877 continue to send EXTEND_TIMEOUT_USEC within the period specified to
878 prevent the service manager from making the service as timedout.
880 * elogind-resolved's DNSSEC support gained support for RFC 8080
881 (Ed25519 keys and signatures).
883 * The elogind-resolve command line tool gained a new set of options
884 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
885 --set-nta= and --revert to configure per-interface DNS configuration
886 dynamically during runtime. It's useful for pushing DNS information
887 into elogind-resolved from DNS hook scripts that various interface
888 managing software supports (such as pppd).
890 * elogind-nspawn gained a new --network-namespace-path= command line
891 option, which may be used to make a container join an existing
892 network namespace, by specifying a path to a "netns" file.
894 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
895 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
896 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
897 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
898 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
899 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
900 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
901 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
902 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
903 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
904 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
905 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
906 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
907 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
908 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
909 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
910 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
911 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
912 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
913 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
914 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
915 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
916 Jędrzejewski-Szmek, Zeal Jagannatha
922 * INCOMPATIBILITY: systemd-logind.service and other long-running
923 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
924 communication with the outside. This generally improves security of
925 the system, and is in almost all cases a safe and good choice, as
926 these services do not and should not provide any network-facing
927 functionality. However, systemd-logind uses the glibc NSS API to
928 query the user database. This creates problems on systems where NSS
929 is set up to directly consult network services for user database
930 lookups. In particular, this creates incompatibilities with the
931 "nss-nis" module, which attempts to directly contact the NIS/YP
932 network servers it is configured for, and will now consistently
933 fail. In such cases, it is possible to turn off IP sandboxing for
934 systemd-logind.service (set IPAddressDeny= in its [Service] section
935 to the empty string, via a .d/ unit file drop-in). Downstream
936 distributions might want to update their nss-nis packaging to include
937 such a drop-in snippet, accordingly, to hide this incompatibility
938 from the user. Another option is to make use of glibc's nscd service
939 to proxy such network requests through a privilege-separated, minimal
940 local caching daemon, or to switch to more modern technologies such
941 sssd, whose NSS hook-ups generally do not involve direct network
942 access. In general, we think it's definitely time to question the
943 implementation choices of nss-nis, i.e. whether it's a good idea
944 today to embed a network-facing loadable module into all local
945 processes that need to query the user database, including the most
946 trivial and benign ones, such as "ls". For more details about
947 IPAddressDeny= see below.
949 * A new modprobe.d drop-in is now shipped by default that sets the
950 bonding module option max_bonds=0. This overrides the kernel default,
951 to avoid conflicts and ambiguity as to whether or not bond0 should be
952 managed by systemd-networkd or not. This resolves multiple issues
953 with bond0 properties not being applied, when bond0 is configured
954 with systemd-networkd. Distributors may choose to not package this,
955 however in that case users will be prevented from correctly managing
956 bond0 interface using systemd-networkd.
958 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
959 which print the logging level and target of the system manager. They
960 complement the existing "set-log-level" and "set-log-target" verbs
961 used to change those values.
963 * journald.conf gained a new boolean setting ReadKMsg= which defaults
964 to on. If turned off kernel log messages will not be read by
965 systemd-journald or included in the logs. It also gained a new
966 setting LineMax= for configuring the maximum line length in
967 STDOUT/STDERR log streams. The new default for this value is 48K, up
968 from the previous hardcoded 2048.
970 * A new unit setting RuntimeDirectoryPreserve= has been added, which
971 allows more detailed control of what to do with a runtime directory
972 configured with RuntimeDirectory= (i.e. a directory below /run or
973 $XDG_RUNTIME_DIR) after a unit is stopped.
975 * The RuntimeDirectory= setting for units gained support for creating
976 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
977 one top-level directory.
979 * Units gained new options StateDirectory=, CacheDirectory=,
980 LogsDirectory= and ConfigurationDirectory= which are closely related
981 to RuntimeDirectory= but manage per-service directories below
982 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
983 possible to write unit files which when activated automatically gain
984 properly owned service specific directories in these locations, thus
985 making unit files self-contained and increasing compatibility with
986 stateless systems and factory reset where /etc or /var are
987 unpopulated at boot. Matching these new settings there's also
988 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
989 ConfigurationDirectoryMode= for configuring the access mode of these
990 directories. These settings are particularly useful in combination
991 with DynamicUser=yes as they provide secure, properly-owned,
992 writable, and stateful locations for storage, excluded from the
993 sandbox that such services live in otherwise.
995 * Automake support has been removed from this release. systemd is now
998 * systemd-journald will now aggressively cache client metadata during
999 runtime, speeding up log write performance under pressure. This comes
1000 at a small price though: as much of the metadata is read
1001 asynchronously from /proc/ (and isn't implicitly attached to log
1002 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
1003 metadata stored alongside a log entry might be slightly
1004 out-of-date. Previously it could only be slightly newer than the log
1005 message. The time window is small however, and given that the kernel
1006 is unlikely to be improved anytime soon in this regard, this appears
1009 * nss-myhostname/systemd-resolved will now by default synthesize an
1010 A/AAAA resource record for the "_gateway" hostname, pointing to the
1011 current default IP gateway. Previously it did that for the "gateway"
1012 name, hampering adoption, as some distributions wanted to leave that
1013 host name open for local use. The old behaviour may still be
1014 requested at build time.
1016 * systemd-networkd's [Address] section in .network files gained a new
1017 Scope= setting for configuring the IP address scope. The [Network]
1018 section gained a new boolean setting ConfigureWithoutCarrier= that
1019 tells systemd-networkd to ignore link sensing when configuring the
1020 device. The [DHCP] section gained a new Anonymize= boolean option for
1021 turning on a number of options suggested in RFC 7844. A new
1022 [RoutingPolicyRule] section has been added for configuring the IP
1023 routing policy. The [Route] section has gained support for a new
1024 Type= setting which permits configuring
1025 blackhole/unreachable/prohibit routes.
1027 * The [VRF] section in .netdev files gained a new Table= setting for
1028 configuring the routing table to use. The [Tunnel] section gained a
1029 new Independent= boolean field for configuring tunnels independent of
1030 an underlying network interface. The [Bridge] section gained a new
1031 GroupForwardMask= option for configuration of propagation of link
1032 local frames between bridge ports.
1034 * The WakeOnLan= setting in .link files gained support for a number of
1035 new modes. A new TCP6SegmentationOffload= setting has been added for
1036 configuring TCP/IPv6 hardware segmentation offload.
1038 * The IPv6 RA sender implementation may now optionally send out RDNSS
1039 and RDNSSL records to supply DNS configuration to peers.
1041 * systemd-nspawn gained support for a new --system-call-filter= command
1042 line option for adding and removing entries in the default system
1043 call filter it applies. Moreover systemd-nspawn has been changed to
1044 implement a system call whitelist instead of a blacklist.
1046 * systemd-run gained support for a new --pipe command line option. If
1047 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
1048 are directly passed on to the activated transient service
1049 executable. This allows invoking arbitrary processes as systemd
1050 services (for example to take benefit of dependency management,
1051 accounting management, resource management or log management that is
1052 done automatically for services) — while still allowing them to be
1053 integrated in a classic UNIX shell pipeline.
1055 * When a service sends RELOAD=1 via sd_notify() and reload propagation
1056 using ReloadPropagationTo= is configured, a reload is now propagated
1057 to configured units. (Previously this was only done on explicitly
1058 requested reloads, using "systemctl reload" or an equivalent
1061 * For each service unit a restart counter is now kept: it is increased
1062 each time the service is restarted due to Restart=, and may be
1063 queried using "systemctl show -p NRestarts …".
1065 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
1066 @signal and @timer have been added, for usage with SystemCallFilter=
1067 in unit files and the new --system-call-filter= command line option
1068 of systemd-nspawn (see above).
1070 * ExecStart= lines in unit files gained two new modifiers: when a
1071 command line is prefixed with "!" the command will be executed as
1072 configured, except for the credentials applied by
1073 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
1074 "+", but does still apply namespacing options unlike "+". There's
1075 also "!!" now, which is mostly identical, but becomes a NOP on
1076 systems that support ambient capabilities. This is useful to write
1077 unit files that work with ambient capabilities where possible but
1078 automatically fall back to traditional privilege dropping mechanisms
1079 on systems where this is not supported.
1081 * ListenNetlink= settings in socket units now support RDMA netlink
1084 * A new unit file setting LockPersonality= has been added which permits
1085 locking down the chosen execution domain ("personality") of a service
1088 * A new special target "getty-pre.target" has been added, which is
1089 ordered before all text logins, and may be used to order services
1090 before textual logins acquire access to the console.
1092 * systemd will now attempt to load the virtio-rng.ko kernel module very
1093 early on if a VM environment supporting this is detected. This should
1094 improve entropy during early boot in virtualized environments.
1096 * A _netdev option is now supported in /etc/crypttab that operates in a
1097 similar way as the same option in /etc/fstab: it permits configuring
1098 encrypted devices that need to be ordered after the network is up.
1099 Following this logic, two new special targets
1100 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
1101 added that are to cryptsetup.target what remote-fs.target and
1102 remote-fs-pre.target are to local-fs.target.
1104 * Service units gained a new UnsetEnvironment= setting which permits
1105 unsetting specific environment variables for services that are
1106 normally passed to it (for example in order to mask out locale
1107 settings for specific services that can't deal with it).
1109 * Units acquired a new boolean option IPAccounting=. When turned on, IP
1110 traffic accounting (packet count as well as byte count) is done for
1111 the service, and shown as part of "systemctl status" or "systemd-run
1114 * Service units acquired two new options IPAddressAllow= and
1115 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
1116 for configuring a simple IP access control list for all sockets of
1117 the unit. These options are available also on .slice and .socket
1118 units, permitting flexible access list configuration for individual
1119 services as well as groups of services (as defined by a slice unit),
1120 including system-wide. Note that IP ACLs configured this way are
1121 enforced on every single IPv4 and IPv6 socket created by any process
1122 of the service unit, and apply to ingress as well as egress traffic.
1124 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
1125 structured log message is generated each time the unit is stopped,
1126 containing information about the consumed resources of this
1129 * A new setting KeyringMode= has been added to unit files, which may be
1130 used to control how the kernel keyring is set up for executed
1133 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
1134 "systemctl kexec" and "systemctl exit" are now always asynchronous in
1135 behaviour (that is: these commands return immediately after the
1136 operation was enqueued instead of waiting for the operation to
1137 complete). Previously, "systemctl poweroff" and "systemctl reboot"
1138 were asynchronous on systems using systemd-logind (i.e. almost
1139 always, and like they were on sysvinit), and the other three commands
1140 were unconditionally synchronous. With this release this is cleaned
1141 up, and callers will see the same asynchronous behaviour on all
1142 systems for all five operations.
1144 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
1147 * .timer units now accept calendar specifications in other timezones
1148 than UTC or the local timezone.
1150 * The tmpfiles snippet var.conf has been changed to create
1151 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
1152 the "utmp" group already, and it appears to be generally understood
1153 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
1154 databases. Previously this was implemented correctly for all these
1155 databases excepts btmp, which has been opened up like this now
1156 too. Note that while the other databases are world-readable
1157 (i.e. 0644), btmp is not and remains more restrictive.
1159 * The systemd-resolve tool gained a new --reset-server-features
1160 switch. When invoked like this systemd-resolved will forget
1161 everything it learnt about the features supported by the configured
1162 upstream DNS servers, and restarts the feature probing logic on the
1163 next resolver look-up for them at the highest feature level
1166 * The status dump systemd-resolved sends to the logs upon receiving
1167 SIGUSR1 now also includes information about all DNS servers it is
1168 configured to use, and the features levels it probed for them.
1170 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
1171 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
1172 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
1173 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
1174 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
1175 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
1176 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
1177 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
1178 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
1179 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
1180 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
1181 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
1182 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
1183 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
1184 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
1185 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
1186 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
1187 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
1188 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
1189 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1191 — Berlin, 2017-10-06
1195 * Meson is now supported as build system in addition to Automake. It is
1196 our plan to remove Automake in one of our next releases, so that
1197 Meson becomes our exclusive build system. Hence, please start using
1198 the Meson build system in your downstream packaging. There's plenty
1199 of documentation around how to use Meson, the extremely brief
1202 ./autogen.sh && ./configure && make && sudo make install
1206 meson build && ninja -C build && sudo ninja -C build install
1208 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
1209 which permits configuring a timeout on the time a job is
1210 running. This is particularly useful for setting timeouts on jobs for
1213 * Unit files gained two new options ConditionUser= and ConditionGroup=
1214 for conditionalizing units based on the identity of the user/group
1215 running a systemd user instance.
1217 * systemd-networkd now understands a new FlowLabel= setting in the
1218 [VXLAN] section of .network files, as well as a Priority= in
1219 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
1220 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
1221 gained support for configuration of GENEVE links, and IPv6 address
1222 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
1224 * .link files now understand a new Port= setting.
1226 * systemd-networkd's DHCP support gained support for DHCP option 119
1227 (domain search list).
1229 * systemd-networkd gained support for serving IPv6 address ranges using
1230 the Router Advertisement protocol. The new .network configuration
1231 section [IPv6Prefix] may be used to configure the ranges to
1232 serve. This is implemented based on a new, minimal, native server
1233 implementation of RA.
1235 * journalctl's --output= switch gained support for a new parameter
1236 "short-iso-precise" for a mode where timestamps are shown as precise
1239 * systemd-udevd's "net_id" builtin may now generate stable network
1240 interface names from IBM PowerVM VIO devices as well as ACPI platform
1243 * MulticastDNS support in systemd-resolved may now be explicitly
1244 enabled/disabled using the new MulticastDNS= configuration file
1247 * systemd-resolved may now optionally use libidn2 instead of the libidn
1248 for processing internationalized domain names. Support for libidn2
1249 should be considered experimental and should not be enabled by
1252 * "machinectl pull-tar" and related call may now do verification of
1253 downloaded images using SUSE-style .sha256 checksum files in addition
1254 to the already existing support for validating using Ubuntu-style
1257 * sd-bus gained support for a new sd_bus_message_appendv() call which
1258 is va_list equivalent of sd_bus_message_append().
1260 * sd-boot gained support for validating images using SHIM/MOK.
1262 * The SMACK code learnt support for "onlycap".
1264 * systemd-mount --umount is now much smarter in figuring out how to
1265 properly unmount a device given its mount or device path.
1267 * The code to call libnss_dns as a fallback from libnss_resolve when
1268 the communication with systemd-resolved fails was removed. This
1269 fallback was redundant and interfered with the [!UNAVAIL=return]
1270 suffix. See nss-resolve(8) for the recommended configuration.
1272 * systemd-logind may now be restarted without losing state. It stores
1273 the file descriptors for devices it manages in the system manager
1274 using the FDSTORE= mechanism. Please note that further changes in
1275 other components may be required to make use of this (for example
1276 Xorg has code to listen for stops of systemd-logind and terminate
1277 itself when logind is stopped or restarted, in order to avoid using
1278 stale file descriptors for graphical devices, which is now
1279 counterproductive and must be reverted in order for restarts of
1280 systemd-logind to be safe. See
1281 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
1283 * All kernel install plugins are called with the environment variable
1284 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
1285 /etc/machine-id. If the file is missing or empty, the variable is
1286 empty and BOOT_DIR_ABS is the path of a temporary directory which is
1287 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
1288 is empty, all plugins should not put anything in BOOT_DIR_ABS.
1290 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
1291 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
1292 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
1293 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
1294 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
1295 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
1296 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
1297 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
1298 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
1299 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
1300 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
1301 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
1302 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
1303 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
1304 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
1305 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
1306 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
1307 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
1308 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
1309 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
1310 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
1311 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
1312 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
1313 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
1314 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
1315 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
1316 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
1317 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
1318 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
1321 — Berlin, 2017-07-12
1325 * The "hybrid" control group mode has been modified to improve
1326 compatibility with "legacy" cgroups-v1 setups. Specifically, the
1327 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
1328 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
1329 cgroups-v1 hierarchy), the only externally visible change being that
1330 the cgroups-v2 hierarchy is also mounted, to
1331 /sys/fs/cgroup/unified. This should provide a large degree of
1332 compatibility with "legacy" cgroups-v1, while taking benefit of the
1333 better management capabilities of cgroups-v2.
1335 * The default control group setup mode may be selected both a boot-time
1336 via a set of kernel command line parameters (specifically:
1337 systemd.unified_cgroup_hierarchy= and
1338 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
1339 default selected on the configure command line
1340 (--with-default-hierarchy=). The upstream default is "hybrid"
1341 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
1342 this will change in a future systemd version to be "unified" (pure
1343 cgroups-v2 mode). The third option for the compile time option is
1344 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
1345 distributions to default to "hybrid" mode for release distributions,
1346 starting with v233. We recommend "unified" for development
1347 distributions (specifically: distributions such as Fedora's rawhide)
1348 as that's where things are headed in the long run. Use "legacy" for
1349 greatest stability and compatibility only.
1351 * Note one current limitation of "unified" and "hybrid" control group
1352 setup modes: the kernel currently does not permit the systemd --user
1353 instance (i.e. unprivileged code) to migrate processes between two
1354 disconnected cgroup subtrees, even if both are managed and owned by
1355 the user. This effectively means "systemd-run --user --scope" doesn't
1356 work when invoked from outside of any "systemd --user" service or
1357 scope. Specifically, it is not supported from session scopes. We are
1358 working on fixing this in a future systemd version. (See #3388 for
1359 further details about this.)
1361 * DBus policy files are now installed into /usr rather than /etc. Make
1362 sure your system has dbus >= 1.9.18 running before upgrading to this
1363 version, or override the install path with --with-dbuspolicydir= .
1365 * All python scripts shipped with systemd (specifically: the various
1366 tests written in Python) now require Python 3.
1368 * systemd unit tests can now run standalone (without the source or
1369 build directories), and can be installed into /usr/lib/systemd/tests/
1370 with 'make install-tests'.
1372 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
1373 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
1376 * Support for the %c, %r, %R specifiers in unit files has been
1377 removed. Specifiers are not supposed to be dependent on configuration
1378 in the unit file itself (so that they resolve the same regardless
1379 where used in the unit files), but these specifiers were influenced
1380 by the Slice= option.
1382 * The shell invoked by debug-shell.service now defaults to /bin/sh in
1383 all cases. If distributions want to use a different shell for this
1384 purpose (for example Fedora's /sbin/sushell) they need to specify
1385 this explicitly at configure time using --with-debug-shell=.
1387 * The confirmation spawn prompt has been reworked to offer the
1390 (c)ontinue, proceed without asking anymore
1391 (D)ump, show the state of the unit
1392 (f)ail, don't execute the command and pretend it failed
1394 (i)nfo, show a short summary of the unit
1395 (j)obs, show jobs that are in progress
1396 (s)kip, don't execute the command and pretend it succeeded
1397 (y)es, execute the command
1399 The 'n' choice for the confirmation spawn prompt has been removed,
1400 because its meaning was confusing.
1402 The prompt may now also be redirected to an alternative console by
1403 specifying the console as parameter to systemd.confirm_spawn=.
1405 * Services of Type=notify require a READY=1 notification to be sent
1406 during startup. If no such message is sent, the service now fails,
1407 even if the main process exited with a successful exit code.
1409 * Services that fail to start up correctly now always have their
1410 ExecStopPost= commands executed. Previously, they'd enter "failed"
1411 state directly, without executing these commands.
1413 * The option MulticastDNS= of network configuration files has acquired
1414 an actual implementation. With MulticastDNS=yes a host can resolve
1415 names of remote hosts and reply to mDNS A and AAAA requests.
1417 * When units are about to be started an additional check is now done to
1418 ensure that all dependencies of type BindsTo= (when used in
1419 combination with After=) have been started.
1421 * systemd-analyze gained a new verb "syscall-filter" which shows which
1422 system call groups are defined for the SystemCallFilter= unit file
1423 setting, and which system calls they contain.
1425 * A new system call filter group "@filesystem" has been added,
1426 consisting of various file system related system calls. Group
1427 "@reboot" has been added, covering reboot, kexec and shutdown related
1428 calls. Finally, group "@swap" has been added covering swap
1429 configuration related calls.
1431 * A new unit file option RestrictNamespaces= has been added that may be
1432 used to restrict access to the various process namespace types the
1433 Linux kernel provides. Specifically, it may be used to take away the
1434 right for a service unit to create additional file system, network,
1435 user, and other namespaces. This sandboxing option is particularly
1436 relevant due to the high amount of recently discovered namespacing
1437 related vulnerabilities in the kernel.
1439 * systemd-udev's .link files gained support for a new AutoNegotiation=
1440 setting for configuring Ethernet auto-negotiation.
1442 * systemd-networkd's .network files gained support for a new
1443 ListenPort= setting in the [DHCP] section to explicitly configure the
1444 UDP client port the DHCP client shall listen on.
1446 * .network files gained a new Unmanaged= boolean setting for explicitly
1447 excluding one or more interfaces from management by systemd-networkd.
1449 * The systemd-networkd ProxyARP= option has been renamed to
1450 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
1451 renamed to ReduceARPProxy=. The old names continue to be available
1454 * systemd-networkd gained support for configuring IPv6 Proxy NDP
1455 addresses via the new IPv6ProxyNDPAddress= .network file setting.
1457 * systemd-networkd's bonding device support gained support for two new
1458 configuration options ActiveSlave= and PrimarySlave=.
1460 * The various options in the [Match] section of .network files gained
1461 support for negative matching.
1463 * New systemd-specific mount options are now understood in /etc/fstab:
1465 x-systemd.mount-timeout= may be used to configure the maximum
1466 permitted runtime of the mount command.
1468 x-systemd.device-bound may be set to bind a mount point to its
1469 backing device unit, in order to automatically remove a mount point
1470 if its backing device is unplugged. This option may also be
1471 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
1472 on the block device, which is now automatically set for all CDROM
1473 drives, so that mounted CDs are automatically unmounted when they are
1474 removed from the drive.
1476 x-systemd.after= and x-systemd.before= may be used to explicitly
1477 order a mount after or before another unit or mount point.
1479 * Enqueued start jobs for device units are now automatically garbage
1480 collected if there are no jobs waiting for them anymore.
1482 * systemctl list-jobs gained two new switches: with --after, for every
1483 queued job the jobs it's waiting for are shown; with --before the
1484 jobs which it's blocking are shown.
1486 * systemd-nspawn gained support for ephemeral boots from disk images
1487 (or in other words: --ephemeral and --image= may now be
1488 combined). Moreover, ephemeral boots are now supported for normal
1489 directories, even if the backing file system is not btrfs. Of course,
1490 if the file system does not support file system snapshots or
1491 reflinks, the initial copy operation will be relatively expensive, but
1492 this should still be suitable for many use cases.
1494 * Calendar time specifications in .timer units now support
1495 specifications relative to the end of a month by using "~" instead of
1496 "-" as separator between month and day. For example, "*-02~03" means
1497 "the third last day in February". In addition a new syntax for
1498 repeated events has been added using the "/" character. For example,
1499 "9..17/2:00" means "every two hours from 9am to 5pm".
1501 * systemd-socket-proxyd gained a new parameter --connections-max= for
1502 configuring the maximum number of concurrent connections.
1504 * sd-id128 gained a new API for generating unique IDs for the host in a
1505 way that does not leak the machine ID. Specifically,
1506 sd_id128_get_machine_app_specific() derives an ID based on the
1507 machine ID a in well-defined, non-reversible, stable way. This is
1508 useful whenever an identifier for the host is needed but where the
1509 identifier shall not be useful to identify the system beyond the
1510 scope of the application itself. (Internally this uses HMAC-SHA256 as
1511 keyed hash function using the machine ID as input.)
1513 * NotifyAccess= gained a new supported value "exec". When set
1514 notifications are accepted from all processes systemd itself invoked,
1515 including all control processes.
1517 * .nspawn files gained support for defining overlay mounts using the
1518 Overlay= and OverlayReadOnly= options. Previously this functionality
1519 was only available on the systemd-nspawn command line.
1521 * systemd-nspawn's --bind= and --overlay= options gained support for
1522 bind/overlay mounts whose source lies within the container tree by
1523 prefixing the source path with "+".
1525 * systemd-nspawn's --bind= and --overlay= options gained support for
1526 automatically allocating a temporary source directory in /var/tmp
1527 that is removed when the container dies. Specifically, if the source
1528 directory is specified as empty string this mechanism is selected. An
1529 example usage is --overlay=+/var::/var, which creates an overlay
1530 mount based on the original /var contained in the image, overlayed
1531 with a temporary directory in the host's /var/tmp. This way changes
1532 to /var are automatically flushed when the container shuts down.
1534 * systemd-nspawn --image= option does now permit raw file system block
1535 devices (in addition to images containing partition tables, as
1538 * The disk image dissection logic in systemd-nspawn gained support for
1539 automatically setting up LUKS encrypted as well as Verity protected
1540 partitions. When a container is booted from an encrypted image the
1541 passphrase is queried at start-up time. When a container with Verity
1542 data is started, the root hash is search in a ".roothash" file
1543 accompanying the disk image (alternatively, pass the root hash via
1544 the new --root-hash= command line option).
1546 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
1547 be used to dissect disk images the same way as systemd-nspawn does
1548 it, following the Bootable Partition Specification. It may even be
1549 used to mount disk images with complex partition setups (including
1550 LUKS and Verity partitions) to a local host directory, in order to
1551 inspect them. This tool is not considered public API (yet), and is
1552 thus not installed into /usr/bin. Please do not rely on its
1553 existence, since it might go away or be changed in later systemd
1556 * A new generator "systemd-verity-generator" has been added, similar in
1557 style to "systemd-cryptsetup-generator", permitting automatic setup of
1558 Verity root partitions when systemd boots up. In order to make use of
1559 this your partition setup should follow the Discoverable Partitions
1560 Specification, and the GPT partition ID of the root file system
1561 partition should be identical to the upper 128bit of the Verity root
1562 hash. The GPT partition ID of the Verity partition protecting it
1563 should be the lower 128bit of the Verity root hash. If the partition
1564 image follows this model it is sufficient to specify a single
1565 "roothash=" kernel command line argument to both configure which root
1566 image and verity partition to use as well as the root hash for
1567 it. Note that systemd-nspawn's Verity support follows the same
1568 semantics, meaning that disk images with proper Verity data in place
1569 may be booted in containers with systemd-nspawn as well as on
1570 physical systems via the verity generator. Also note that the "mkosi"
1571 tool available at https://github.com/systemd/mkosi has been updated
1572 to generate Verity protected disk images following this scheme. In
1573 fact, it has been updated to generate disk images that optionally
1574 implement a complete UEFI SecureBoot trust chain, involving a signed
1575 kernel and initrd image that incorporates such a root hash as well as
1576 a Verity-enabled root partition.
1578 * The hardware database (hwdb) udev supports has been updated to carry
1579 accelerometer quirks.
1581 * All system services are now run with a fresh kernel keyring set up
1582 for them. The invocation ID is stored by default in it, thus
1583 providing a safe, non-overridable way to determine the invocation
1586 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
1587 options for bind mounting arbitrary paths in a service-specific
1588 way. When these options are used, arbitrary host or service files and
1589 directories may be mounted to arbitrary locations in the service's
1592 * Documentation has been added that lists all of systemd's low-level
1593 environment variables:
1595 https://github.com/systemd/systemd/blob/master/doc/ENVIRONMENT.md
1597 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
1598 whether a specific socket file descriptor matches a specified socket
1601 * systemd-firstboot has been updated to check for the
1602 systemd.firstboot= kernel command line option. It accepts a boolean
1603 and when set to false the first boot questions are skipped.
1605 * systemd-fstab-generator has been updated to check for the
1606 systemd.volatile= kernel command line option, which either takes an
1607 optional boolean parameter or the special value "state". If used the
1608 system may be booted in a "volatile" boot mode. Specifically,
1609 "systemd.volatile" is used, the root directory will be mounted as
1610 tmpfs, and only /usr is mounted from the actual root file system. If
1611 "systemd.volatile=state" is used, the root directory will be mounted
1612 as usual, but /var is mounted as tmpfs. This concept provides similar
1613 functionality as systemd-nspawn's --volatile= option, but provides it
1614 on physical boots. Use this option for implementing stateless
1615 systems, or testing systems with all state and/or configuration reset
1616 to the defaults. (Note though that many distributions are not
1617 prepared to boot up without a populated /etc or /var, though.)
1619 * systemd-gpt-auto-generator gained support for LUKS encrypted root
1620 partitions. Previously it only supported LUKS encrypted partitions
1621 for all other uses, except for the root partition itself.
1623 * Socket units gained support for listening on AF_VSOCK sockets for
1624 communication in virtualized QEMU environments.
1626 * The "configure" script gained a new option --with-fallback-hostname=
1627 for specifying the fallback hostname to use if none is configured in
1628 /etc/hostname. For example, by specifying
1629 --with-fallback-hostname=fedora it is possible to default to a
1630 hostname of "fedora" on pristine installations.
1632 * systemd-cgls gained support for a new --unit= switch for listing only
1633 the control groups of a specific unit. Similar --user-unit= has been
1634 added for listing only the control groups of a specific user unit.
1636 * systemd-mount gained a new --umount switch for unmounting a mount or
1637 automount point (and all mount/automount points below it).
1639 * systemd will now refuse full configuration reloads (via systemctl
1640 daemon-reload and related calls) unless at least 16MiB of free space
1641 are available in /run. This is a safety precaution in order to ensure
1642 that generators can safely operate after the reload completed.
1644 * A new unit file option RootImage= has been added, which has a similar
1645 effect as RootDirectory= but mounts the service's root directory from
1646 a disk image instead of plain directory. This logic reuses the same
1647 image dissection and mount logic that systemd-nspawn already uses,
1648 and hence supports any disk images systemd-nspawn supports, including
1649 those following the Discoverable Partition Specification, as well as
1650 Verity enabled images. This option enables systemd to run system
1651 services directly off disk images acting as resource bundles,
1652 possibly even including full integrity data.
1654 * A new MountAPIVFS= unit file option has been added, taking a boolean
1655 argument. If enabled /proc, /sys and /dev (collectively called the
1656 "API VFS") will be mounted for the service. This is only relevant if
1657 RootDirectory= or RootImage= is used for the service, as these mounts
1658 are of course in place in the host mount namespace anyway.
1660 * systemd-nspawn gained support for a new --pivot-root= switch. If
1661 specified the root directory within the container image is pivoted to
1662 the specified mount point, while the original root disk is moved to a
1663 different place. This option enables booting of ostree images
1664 directly with systemd-nspawn.
1666 * The systemd build scripts will no longer complain if the NTP server
1667 addresses are not changed from the defaults. Google now supports
1668 these NTP servers officially. We still recommend downstreams to
1669 properly register an NTP pool with the NTP pool project though.
1671 * coredumpctl gained a new "--reverse" option for printing the list
1672 of coredumps in reverse order.
1674 * coredumpctl will now show additional information about truncated and
1675 inaccessible coredumps, as well as coredumps that are still being
1676 processed. It also gained a new --quiet switch for suppressing
1677 additional informational message in its output.
1679 * coredumpctl gained support for only showing coredumps newer and/or
1680 older than specific timestamps, using the new --since= and --until=
1681 options, reminiscent of journalctl's options by the same name.
1683 * The systemd-coredump logic has been improved so that it may be reused
1684 to collect backtraces in non-compiled languages, for example in
1685 scripting languages such as Python.
1687 * machinectl will now show the UID shift of local containers, if user
1688 namespacing is enabled for them.
1690 * systemd will now optionally run "environment generator" binaries at
1691 configuration load time. They may be used to add environment
1692 variables to the environment block passed to services invoked. One
1693 user environment generator is shipped by default that sets up
1694 environment variables based on files dropped into /etc/environment.d
1695 and ~/.config/environment.d/.
1697 * systemd-resolved now includes the new, recently published 2017 DNSSEC
1700 * hostnamed has been updated to report a new chassis type of
1701 "convertible" to cover "foldable" laptops that can both act as a
1702 tablet and as a laptop, such as various Lenovo Yoga devices.
1704 Contributions from: Adrián López, Alexander Galanin, Alexander
1705 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
1706 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
1707 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
1708 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
1709 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
1710 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
1711 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
1712 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
1713 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
1714 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
1715 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
1716 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
1717 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
1718 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
1719 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
1720 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
1721 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
1722 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
1723 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
1724 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
1725 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
1726 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
1727 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
1728 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
1729 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
1730 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
1731 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
1734 — Berlin, 2017-03-01
1738 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
1739 RestrictAddressFamilies= enabled. These sandboxing options should
1740 generally be compatible with the various external udev call-out
1741 binaries we are aware of, however there may be exceptions, in
1742 particular when exotic languages for these call-outs are used. In
1743 this case, consider turning off these settings locally.
1745 * The new RemoveIPC= option can be used to remove IPC objects owned by
1746 the user or group of a service when that service exits.
1748 * The new ProtectKernelModules= option can be used to disable explicit
1749 load and unload operations of kernel modules by a service. In
1750 addition access to /usr/lib/modules is removed if this option is set.
1752 * ProtectSystem= option gained a new value "strict", which causes the
1753 whole file system tree with the exception of /dev, /proc, and /sys,
1754 to be remounted read-only for a service.
1756 * The new ProtectKernelTunables= option can be used to disable
1757 modification of configuration files in /sys and /proc by a service.
1758 Various directories and files are remounted read-only, so access is
1759 restricted even if the file permissions would allow it.
1761 * The new ProtectControlGroups= option can be used to disable write
1762 access by a service to /sys/fs/cgroup.
1764 * Various systemd services have been hardened with
1765 ProtectKernelTunables=yes, ProtectControlGroups=yes,
1766 RestrictAddressFamilies=.
1768 * Support for dynamically creating users for the lifetime of a service
1769 has been added. If DynamicUser=yes is specified, user and group IDs
1770 will be allocated from the range 61184..65519 for the lifetime of the
1771 service. They can be resolved using the new nss-systemd.so NSS
1772 module. The module must be enabled in /etc/nsswitch.conf. Services
1773 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
1774 any resources allocated by the service will be cleaned up when the
1775 service exits. They also have ProtectHome=read-only and
1776 ProtectSystem=strict enabled, so they are not able to make any
1777 permanent modifications to the system.
1779 * The nss-systemd module also always resolves root and nobody, making
1780 it possible to have no /etc/passwd or /etc/group files in minimal
1781 container or chroot environments.
1783 * Services may be started with their own user namespace using the new
1784 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
1785 under which the service is running are mapped. All other users are
1788 * Support for the cgroup namespace has been added to systemd-nspawn. If
1789 supported by kernel, the container system started by systemd-nspawn
1790 will have its own view of the cgroup hierarchy. This new behaviour
1791 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
1793 * The new MemorySwapMax= option can be used to limit the maximum swap
1794 usage under the unified cgroup hierarchy.
1796 * Support for the CPU controller in the unified cgroup hierarchy has
1797 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
1798 options. This controller requires out-of-tree patches for the kernel
1799 and the support is provisional.
1801 * Mount and automount units may now be created transiently
1802 (i.e. dynamically at runtime via the bus API, instead of requiring
1803 unit files in the file system).
1805 * systemd-mount is a new tool which may mount file systems – much like
1806 mount(8), optionally pulling in additional dependencies through
1807 transient .mount and .automount units. For example, this tool
1808 automatically runs fsck on a backing block device before mounting,
1809 and allows the automount logic to be used dynamically from the
1810 command line for establishing mount points. This tool is particularly
1811 useful when dealing with removable media, as it will ensure fsck is
1812 run – if necessary – before the first access and that the file system
1813 is quickly unmounted after each access by utilizing the automount
1814 logic. This maximizes the chance that the file system on the
1815 removable media stays in a clean state, and if it isn't in a clean
1816 state is fixed automatically.
1818 * LazyUnmount=yes option for mount units has been added to expose the
1819 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
1822 * /efi will be used as the mount point of the EFI boot partition, if
1823 the directory is present, and the mount point was not configured
1824 through other means (e.g. fstab). If /efi directory does not exist,
1825 /boot will be used as before. This makes it easier to automatically
1826 mount the EFI partition on systems where /boot is used for something
1829 * When operating on GPT disk images for containers, systemd-nspawn will
1830 now mount the ESP to /boot or /efi according to the same rules as PID
1831 1 running on a host. This allows tools like "bootctl" to operate
1832 correctly within such containers, in order to make container images
1833 bootable on physical systems.
1835 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
1837 * Two new user session targets have been added to support running
1838 graphical sessions under the systemd --user instance:
1839 graphical-session.target and graphical-session-pre.target. See
1840 systemd.special(7) for a description of how those targets should be
1843 * The vconsole initialization code has been significantly reworked to
1844 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
1845 support unicode keymaps. Font and keymap configuration will now be
1846 copied to all allocated virtual consoles.
1848 * FreeBSD's bhyve virtualization is now detected.
1850 * Information recorded in the journal for core dumps now includes the
1851 contents of /proc/mountinfo and the command line of the process at
1852 the top of the process hierarchy (which is usually the init process
1855 * systemd-journal-gatewayd learned the --directory= option to serve
1856 files from the specified location.
1858 * journalctl --root=… can be used to peruse the journal in the
1859 /var/log/ directories inside of a container tree. This is similar to
1860 the existing --machine= option, but does not require the container to
1863 * The hardware database has been extended to support
1864 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
1867 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
1868 specify the click rate for mice which include a horizontal wheel with
1869 a click rate that is different than the one for the vertical wheel.
1871 * systemd-run gained a new --wait option that makes service execution
1872 synchronous. (Specifically, the command will not return until the
1873 specified service binary exited.)
1875 * systemctl gained a new --wait option that causes the start command to
1876 wait until the units being started have terminated again.
1878 * A new journal output mode "short-full" has been added which displays
1879 timestamps with abbreviated English day names and adds a timezone
1880 suffix. Those timestamps include more information than the default
1881 "short" output mode, and can be passed directly to journalctl's
1882 --since= and --until= options.
1884 * /etc/resolv.conf will be bind-mounted into containers started by
1885 systemd-nspawn, if possible, so any changes to resolv.conf contents
1886 are automatically propagated to the container.
1888 * The number of instances for socket-activated services originating
1889 from a single IP address can be limited with
1890 MaxConnectionsPerSource=, extending the existing setting of
1893 * systemd-networkd gained support for vcan ("Virtual CAN") interface
1896 * .netdev and .network configuration can now be extended through
1899 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
1900 Segmentation Offload, Generic Receive Offload, Large Receive Offload
1901 can be enabled and disabled using the new UDPSegmentationOffload=,
1902 TCPSegmentationOffload=, GenericSegmentationOffload=,
1903 GenericReceiveOffload=, LargeReceiveOffload= options in the
1904 [Link] section of .link files.
1906 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
1907 Port VLAN ID can be configured for bridge devices using the new STP=,
1908 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
1909 section of .netdev files.
1911 * The route table to which routes received over DHCP or RA should be
1912 added can be configured with the new RouteTable= option in the [DHCP]
1913 and [IPv6AcceptRA] sections of .network files.
1915 * The Address Resolution Protocol can be disabled on links managed by
1916 systemd-networkd using the ARP=no setting in the [Link] section of
1919 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
1920 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
1921 encode information about the result and exit codes of the current
1922 service runtime cycle.
1924 * systemd-sysctl will now configure kernel parameters in the order
1925 they occur in the configuration files. This matches what sysctl
1926 has been traditionally doing.
1928 * kernel-install "plugins" that are executed to perform various
1929 tasks after a new kernel is added and before an old one is removed
1930 can now return a special value to terminate the procedure and
1931 prevent any later plugins from running.
1933 * Journald's SplitMode=login setting has been deprecated. It has been
1934 removed from documentation, and its use is discouraged. In a future
1935 release it will be completely removed, and made equivalent to current
1936 default of SplitMode=uid.
1938 * Storage=both option setting in /etc/systemd/coredump.conf has been
1939 removed. With fast LZ4 compression storing the core dump twice is not
1942 * The --share-system systemd-nspawn option has been replaced with an
1943 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
1944 this functionality is discouraged. In addition the variables
1945 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
1946 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
1947 individual namespaces.
1949 * "machinectl list" now shows the IP address of running containers in
1950 the output, as well as OS release information.
1952 * "loginctl list" now shows the TTY of each session in the output.
1954 * sd-bus gained new API calls sd_bus_track_set_recursive(),
1955 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
1956 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
1957 tracking objects in a "recursive" mode, where a single client can be
1958 counted multiple times, if it takes multiple references.
1960 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
1961 sd_bus_get_exit_on_disconnect(). They may be used to make a
1962 process using sd-bus automatically exit if the bus connection is
1965 * Bus clients of the service manager may now "pin" loaded units into
1966 memory, by taking an explicit reference on them. This is useful to
1967 ensure the client can retrieve runtime data about the service even
1968 after the service completed execution. Taking such a reference is
1969 available only for privileged clients and should be helpful to watch
1970 running services in a race-free manner, and in particular collect
1971 information about exit statuses and results.
1973 * The nss-resolve module has been changed to strictly return UNAVAIL
1974 when communication via D-Bus with resolved failed, and NOTFOUND when
1975 a lookup completed but was negative. This means it is now possible to
1976 neatly configure fallbacks using nsswitch.conf result checking
1977 expressions. Taking benefit of this, the new recommended
1978 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
1980 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
1982 * A new setting CtrlAltDelBurstAction= has been added to
1983 /etc/systemd/system.conf which may be used to configure the precise
1984 behaviour if the user on the console presses Ctrl-Alt-Del more often
1985 than 7 times in 2s. Previously this would unconditionally result in
1986 an expedited, immediate reboot. With this new setting the precise
1987 operation may be configured in more detail, and also turned off
1990 * In .netdev files two new settings RemoteChecksumTx= and
1991 RemoteChecksumRx= are now understood that permit configuring the
1992 remote checksumming logic for VXLAN networks.
1994 * The service manager learnt a new "invocation ID" concept for invoked
1995 services. Each runtime cycle of a service will get a new invocation
1996 ID (a 128bit random UUID) assigned that identifies the current
1997 run of the service uniquely and globally. A new invocation ID
1998 is generated each time a service starts up. The journal will store
1999 the invocation ID of a service along with any logged messages, thus
2000 making the invocation ID useful for matching the online runtime of a
2001 service with the offline log data it generated in a safe way without
2002 relying on synchronized timestamps. In many ways this new service
2003 invocation ID concept is similar to the kernel's boot ID concept that
2004 uniquely and globally identifies the runtime of each boot. The
2005 invocation ID of a service is passed to the service itself via an
2006 environment variable ($INVOCATION_ID). A new bus call
2007 GetUnitByInvocationID() has been added that is similar to GetUnit()
2008 but instead of retrieving the bus path for a unit by its name
2009 retrieves it by its invocation ID. The returned path is valid only as
2010 long as the passed invocation ID is current.
2012 * systemd-resolved gained a new "DNSStubListener" setting in
2013 resolved.conf. It either takes a boolean value or the special values
2014 "udp" and "tcp", and configures whether to enable the stub DNS
2015 listener on 127.0.0.53:53.
2017 * IP addresses configured via networkd may now carry additional
2018 configuration settings supported by the kernel. New options include:
2019 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
2020 PrefixRoute=, AutoJoin=.
2022 * The PAM configuration fragment file for "user@.service" shipped with
2023 systemd (i.e. the --user instance of systemd) has been stripped to
2024 the minimum necessary to make the system boot. Previously, it
2025 contained Fedora-specific stanzas that did not apply to other
2026 distributions. It is expected that downstream distributions add
2027 additional configuration lines, matching their needs to this file,
2028 using it only as rough template of what systemd itself needs. Note
2029 that this reduced fragment does not even include an invocation of
2030 pam_limits which most distributions probably want to add, even though
2031 systemd itself does not need it. (There's also the new build time
2032 option --with-pamconfdir=no to disable installation of the PAM
2035 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
2036 capability is now also dropped from its set (in addition to
2037 CAP_SYS_MKNOD as before).
2039 * In service unit files it is now possible to connect a specific named
2040 file descriptor with stdin/stdout/stdout of an executed service. The
2041 name may be specified in matching .socket units using the
2042 FileDescriptorName= setting.
2044 * A number of journal settings may now be configured on the kernel
2045 command line. Specifically, the following options are now understood:
2046 systemd.journald.max_level_console=,
2047 systemd.journald.max_level_store=,
2048 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
2049 systemd.journald.max_level_wall=.
2051 * "systemctl is-enabled --full" will now show by which symlinks a unit
2052 file is enabled in the unit dependency tree.
2054 * Support for VeraCrypt encrypted partitions has been added to the
2055 "cryptsetup" logic and /etc/crypttab.
2057 * systemd-detect-virt gained support for a new --private-users switch
2058 that checks whether the invoking processes are running inside a user
2059 namespace. Similar, a new special value "private-users" for the
2060 existing ConditionVirtualization= setting has been added, permitting
2061 skipping of specific units in user namespace environments.
2063 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
2064 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
2065 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
2066 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
2067 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
2068 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
2069 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
2070 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
2071 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
2072 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
2073 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
2074 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
2075 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
2076 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
2077 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
2078 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
2079 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
2080 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
2081 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
2082 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
2083 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
2084 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
2085 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
2086 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
2087 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
2088 Jędrzejewski-Szmek, Zeal Jagannatha
2090 — Santa Fe, 2016-11-03
2094 * In service units the various ExecXYZ= settings have been extended
2095 with an additional special character as first argument of the
2096 assigned value: if the character '+' is used the specified command
2097 line it will be run with full privileges, regardless of User=,
2098 Group=, CapabilityBoundingSet= and similar options. The effect is
2099 similar to the existing PermissionsStartOnly= option, but allows
2100 configuration of this concept for each executed command line
2103 * Services may now alter the service watchdog timeout at runtime by
2104 sending a WATCHDOG_USEC= message via sd_notify().
2106 * MemoryLimit= and related unit settings now optionally take percentage
2107 specifications. The percentage is taken relative to the amount of
2108 physical memory in the system (or in case of containers, the assigned
2109 amount of memory). This allows scaling service resources neatly with
2110 the amount of RAM available on the system. Similarly, systemd-logind's
2111 RuntimeDirectorySize= option now also optionally takes percentage
2114 * In similar fashion TasksMax= takes percentage values now, too. The
2115 value is taken relative to the configured maximum number of processes
2116 on the system. The per-service task maximum has been changed to 15%
2117 using this functionality. (Effectively this is an increase of 512 →
2118 4915 for service units, given the kernel's default pid_max setting.)
2120 * Calendar time specifications in .timer units now understand a ".."
2121 syntax for time ranges. Example: "4..7:10" may now be used for
2122 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
2125 * The InaccessableDirectories=, ReadOnlyDirectories= and
2126 ReadWriteDirectories= unit file settings have been renamed to
2127 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
2128 applied to all kinds of file nodes, and not just directories, with
2129 the exception of symlinks. Specifically these settings may now be
2130 used on block and character device nodes, UNIX sockets and FIFOS as
2131 well as regular files. The old names of these settings remain
2132 available for compatibility.
2134 * systemd will now log about all service processes it kills forcibly
2135 (using SIGKILL) because they remained after the clean shutdown phase
2136 of the service completed. This should help identifying services that
2137 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
2138 systemd-logind's configuration a similar log message is generated for
2139 processes killed at the end of each session due to this setting.
2141 * systemd will now set the $JOURNAL_STREAM environment variable for all
2142 services whose stdout/stderr are connected to the Journal (which
2143 effectively means by default: all services). The variable contains
2144 the device and inode number of the file descriptor used for
2145 stdout/stderr. This may be used by invoked programs to detect whether
2146 their stdout/stderr is connected to the Journal, in which case they
2147 can switch over to direct Journal communication, thus being able to
2148 pass extended, structured metadata along with their log messages. As
2149 one example, this is now used by glib's logging primitives.
2151 * When using systemd's default tmp.mount unit for /tmp, the mount point
2152 will now be established with the "nosuid" and "nodev" options. This
2153 avoids privilege escalation attacks that put traps and exploits into
2154 /tmp. However, this might cause problems if you e. g. put container
2155 images or overlays into /tmp; if you need this, override tmp.mount's
2156 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
2159 * systemd now supports the "memory" cgroup controller also on
2162 * The systemd-cgtop tool now optionally takes a control group path as
2163 command line argument. If specified, the control group list shown is
2164 limited to subgroups of that group.
2166 * The SystemCallFilter= unit file setting gained support for
2167 pre-defined, named system call filter sets. For example
2168 SystemCallFilter=@clock is now an effective way to make all clock
2169 changing-related system calls unavailable to a service. A number of
2170 similar pre-defined groups are defined. Writing system call filters
2171 for system services is simplified substantially with this new
2172 concept. Accordingly, all of systemd's own, long-running services now
2173 enable system call filtering based on this, by default.
2175 * A new service setting MemoryDenyWriteExecute= has been added, taking
2176 a boolean value. If turned on, a service may no longer create memory
2177 mappings that are writable and executable at the same time. This
2178 enhances security for services where this is enabled as it becomes
2179 harder to dynamically write and then execute memory in exploited
2180 service processes. This option has been enabled for all of systemd's
2181 own long-running services.
2183 * A new RestrictRealtime= service setting has been added, taking a
2184 boolean argument. If set the service's processes may no longer
2185 acquire realtime scheduling. This improves security as realtime
2186 scheduling may otherwise be used to easily freeze the system.
2188 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
2189 value. This may be used for requesting that the system manager inside
2190 of the container reports start-up completion to nspawn which then
2191 propagates this notification further to the service manager
2192 supervising nspawn itself. A related option NotifyReady= in .nspawn
2193 files has been added too. This functionality allows ordering of the
2194 start-up of multiple containers using the usual systemd ordering
2197 * machinectl gained a new command "stop" that is an alias for
2200 * systemd-resolved gained support for contacting DNS servers on
2201 link-local IPv6 addresses.
2203 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
2204 its caches. A method call for requesting the same operation has been
2205 added to the bus API too, and is made available via "systemd-resolve
2208 * systemd-resolve gained a new --status switch. If passed a brief
2209 summary of the used DNS configuration with per-interface information
2212 * resolved.conf gained a new Cache= boolean option, defaulting to
2213 on. If turned off local DNS caching is disabled. This comes with a
2214 performance penalty in particular when DNSSEC is enabled. Note that
2215 resolved disables its internal caching implicitly anyway, when the
2216 configured DNS server is on a host-local IP address such as ::1 or
2217 127.0.0.1, thus automatically avoiding double local caching.
2219 * systemd-resolved now listens on the local IP address 127.0.0.53:53
2220 for DNS requests. This improves compatibility with local programs
2221 that do not use the libc NSS or systemd-resolved's bus APIs for name
2222 resolution. This minimal DNS service is only available to local
2223 programs and does not implement the full DNS protocol, but enough to
2224 cover local DNS clients. A new, static resolv.conf file, listing just
2225 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
2226 now recommended to make /etc/resolv.conf a symlink to this file in
2227 order to route all DNS lookups to systemd-resolved, regardless if
2228 done via NSS, the bus API or raw DNS packets. Note that this local
2229 DNS service is not as fully featured as the libc NSS or
2230 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
2231 used to deliver link-local address information (as this implies
2232 sending a local interface index along), LLMNR/mDNS support via this
2233 interface is severely restricted. It is thus strongly recommended for
2234 all applications to use the libc NSS API or native systemd-resolved
2237 * systemd-networkd's bridge support learned a new setting
2238 VLANFiltering= for controlling VLAN filtering. Moreover a new section
2239 in .network files has been added for configuring VLAN bridging in
2240 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
2242 * systemd-networkd's IPv6 Router Advertisement code now makes use of
2243 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
2244 now be acquired without relying on DHCPv6. Two new options
2245 UseDomains= and UseDNS= have been added to configure this behaviour.
2247 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
2248 renamed IPv6AcceptRA=, without altering its behaviour. The old
2249 setting name remains available for compatibility reasons.
2251 * The systemd-networkd VTI/VTI6 tunneling support gained new options
2252 Key=, InputKey= and OutputKey=.
2254 * systemd-networkd gained support for VRF ("Virtual Routing Function")
2255 interface configuration.
2257 * "systemctl edit" may now be used to create new unit files by
2258 specifying the --force switch.
2260 * sd-event gained a new function sd_event_get_iteration() for
2261 requesting the current iteration counter of the event loop. It starts
2262 at zero and is increased by one with each event loop iteration.
2264 * A new rpm macro %systemd_ordering is provided by the macros.systemd
2265 file. It can be used in lieu of %systemd_requires in packages which
2266 don't use any systemd functionality and are intended to be installed
2267 in minimal containers without systemd present. This macro provides
2268 ordering dependencies to ensure that if the package is installed in
2269 the same rpm transaction as systemd, systemd will be installed before
2270 the scriptlets for the package are executed, allowing unit presets
2273 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
2274 been added to simplify packaging of generators.
2276 * The os-release file gained VERSION_CODENAME field for the
2277 distribution nickname (e.g. VERSION_CODENAME=woody).
2279 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
2280 can be set to disable parsing of metadata and the creation
2281 of persistent symlinks for that device.
2283 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
2284 to make them available to logged-in users has been reverted.
2286 * Much of the common code of the various systemd components is now
2287 built into an internal shared library libsystemd-shared-231.so
2288 (incorporating the systemd version number in the name, to be updated
2289 with future releases) that the components link to. This should
2290 decrease systemd footprint both in memory during runtime and on
2291 disk. Note that the shared library is not for public use, and is
2292 neither API nor ABI stable, but is likely to change with every new
2293 released update. Packagers need to make sure that binaries
2294 linking to libsystemd-shared.so are updated in step with the
2297 * Configuration for "mkosi" is now part of the systemd
2298 repository. mkosi is a tool to easily build legacy-free OS images,
2299 and is available on github: https://github.com/systemd/mkosi. If
2300 "mkosi" is invoked in the build tree a new raw OS image is generated
2301 incorporating the systemd sources currently being worked on and a
2302 clean, fresh distribution installation. The generated OS image may be
2303 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
2304 UEFI PC. This functionality is particularly useful to easily test
2305 local changes made to systemd in a pristine, defined environment. See
2306 doc/HACKING for details.
2308 * configure learned the --with-support-url= option to specify the
2309 distribution's bugtracker.
2311 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
2312 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
2313 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
2314 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
2315 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
2316 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
2317 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
2318 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
2319 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
2320 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
2321 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
2322 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
2323 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
2324 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
2325 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
2326 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
2327 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
2328 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
2329 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
2331 — Berlin, 2016-07-25
2335 * DNSSEC is now turned on by default in systemd-resolved (in
2336 "allow-downgrade" mode), but may be turned off during compile time by
2337 passing "--with-default-dnssec=no" to "configure" (and of course,
2338 during runtime with DNSSEC= in resolved.conf). We recommend
2339 downstreams to leave this on at least during development cycles and
2340 report any issues with the DNSSEC logic upstream. We are very
2341 interested in collecting feedback about the DNSSEC validator and its
2342 limitations in the wild. Note however, that DNSSEC support is
2343 probably nothing downstreams should turn on in stable distros just
2344 yet, as it might create incompatibilities with a few DNS servers and
2345 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
2346 automatically whenever we detect such incompatible setups, but there
2347 might be systems we do not cover yet. Hence: please help us testing
2348 the DNSSEC code, leave this on where you can, report back, but then
2349 again don't consider turning this on in your stable, LTS or
2350 production release just yet. (Note that you have to enable
2351 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
2352 and its DNSSEC mode for host name resolution from local
2355 * systemd-resolve conveniently resolves DANE records with the --tlsa
2356 option and OPENPGPKEY records with the --openpgp option. It also
2357 supports dumping raw DNS record data via the new --raw= switch.
2359 * systemd-logind will now by default terminate user processes that are
2360 part of the user session scope unit (session-XX.scope) when the user
2361 logs out. This behavior is controlled by the KillUserProcesses=
2362 setting in logind.conf, and the previous default of "no" is now
2363 changed to "yes". This means that user sessions will be properly
2364 cleaned up after, but additional steps are necessary to allow
2365 intentionally long-running processes to survive logout.
2367 While the user is logged in at least once, user@.service is running,
2368 and any service that should survive the end of any individual login
2369 session can be started at a user service or scope using systemd-run.
2370 systemd-run(1) man page has been extended with an example which shows
2371 how to run screen in a scope unit underneath user@.service. The same
2372 command works for tmux.
2374 After the user logs out of all sessions, user@.service will be
2375 terminated too, by default, unless the user has "lingering" enabled.
2376 To effectively allow users to run long-term tasks even if they are
2377 logged out, lingering must be enabled for them. See loginctl(1) for
2378 details. The default polkit policy was modified to allow users to
2379 set lingering for themselves without authentication.
2381 Previous defaults can be restored at compile time by the
2382 --without-kill-user-processes option to "configure".
2384 * systemd-logind gained new configuration settings SessionsMax= and
2385 InhibitorsMax=, both with a default of 8192. It will not register new
2386 user sessions or inhibitors above this limit.
2388 * systemd-logind will now reload configuration on SIGHUP.
2390 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
2391 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
2392 enable. Also, support for the "io" cgroup controller in the unified
2393 hierarchy has been added, so that the "memory", "pids" and "io" are
2394 now the controllers that are supported on the unified hierarchy.
2396 WARNING: it is not possible to use previous systemd versions with
2397 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
2398 is necessary to also update systemd in the initramfs if using the
2399 unified hierarchy. An updated SELinux policy is also required.
2401 * LLDP support has been extended, and both passive (receive-only) and
2402 active (sender) modes are supported. Passive mode ("routers-only") is
2403 enabled by default in systemd-networkd. Active LLDP mode is enabled
2404 by default for containers on the internal network. The "networkctl
2405 lldp" command may be used to list information gathered. "networkctl
2406 status" will also show basic LLDP information on connected peers now.
2408 * The IAID and DUID unique identifier sent in DHCP requests may now be
2409 configured for the system and each .network file managed by
2410 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
2412 * systemd-networkd gained support for configuring proxy ARP support for
2413 each interface, via the ProxyArp= setting in .network files. It also
2414 gained support for configuring the multicast querier feature of
2415 bridge devices, via the new MulticastQuerier= setting in .netdev
2416 files. Similarly, snooping on the IGMP traffic can be controlled
2417 via the new setting MulticastSnooping=.
2419 A new setting PreferredLifetime= has been added for addresses
2420 configured in .network file to configure the lifetime intended for an
2423 The systemd-networkd DHCP server gained the option EmitRouter=, which
2424 defaults to yes, to configure whether the DHCP Option 3 (Router)
2427 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
2428 systemd-socket-activate and installed into /usr/bin. It is now fully
2431 * systemd-journald now uses separate threads to flush changes to disk
2432 when closing journal files, thus reducing impact of slow disk I/O on
2433 logging performance.
2435 * The sd-journal API gained two new calls
2436 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
2437 can be used to open journal files using file descriptors instead of
2438 file or directory paths. sd_journal_open_container() has been
2439 deprecated, sd_journal_open_directory_fd() should be used instead
2440 with the flag SD_JOURNAL_OS_ROOT.
2442 * journalctl learned a new output mode "-o short-unix" that outputs log
2443 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
2444 UTC). It also gained support for a new --no-hostname setting to
2445 suppress the hostname column in the family of "short" output modes.
2447 * systemd-ask-password now optionally skips printing of the password to
2448 stdout with --no-output which can be useful in scripts.
2450 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
2451 (devices tagged with ID_MAKER_TOOL) are now tagged with
2452 "uaccess" and are available to logged in users.
2454 * The DeviceAllow= unit setting now supports specifiers (with "%").
2456 * "systemctl show" gained a new --value switch, which allows print a
2457 only the contents of a specific unit property, without also printing
2458 the property's name. Similar support was added to "show*" verbs
2459 of loginctl and machinectl that output "key=value" lists.
2461 * A new unit type "generated" was added for files dynamically generated
2462 by generator tools. Similarly, a new unit type "transient" is used
2463 for unit files created using the runtime API. "systemctl enable" will
2464 refuse to operate on such files.
2466 * A new command "systemctl revert" has been added that may be used to
2467 revert to the vendor version of a unit file, in case local changes
2468 have been made by adding drop-ins or overriding the unit file.
2470 * "machinectl clean" gained a new verb to automatically remove all or
2471 just hidden container images.
2473 * systemd-tmpfiles gained support for a new line type "e" for emptying
2474 directories, if they exist, without creating them if they don't.
2476 * systemd-nspawn gained support for automatically patching the UID/GIDs
2477 of the owners and the ACLs of all files and directories in a
2478 container tree to match the UID/GID user namespacing range selected
2479 for the container invocation. This mode is enabled via the new
2480 --private-users-chown switch. It also gained support for
2481 automatically choosing a free, previously unused UID/GID range when
2482 starting a container, via the new --private-users=pick setting (which
2483 implies --private-users-chown). Together, these options for the first
2484 time make user namespacing for nspawn containers fully automatic and
2485 thus deployable. The systemd-nspawn@.service template unit file has
2486 been changed to use this functionality by default.
2488 * systemd-nspawn gained a new --network-zone= switch, that allows
2489 creating ad-hoc virtual Ethernet links between multiple containers,
2490 that only exist as long as at least one container referencing them is
2491 running. This allows easy connecting of multiple containers with a
2492 common link that implements an Ethernet broadcast domain. Each of
2493 these network "zones" may be named relatively freely by the user, and
2494 may be referenced by any number of containers, but each container may
2495 only reference one of these "zones". On the lower level, this is
2496 implemented by an automatically managed bridge network interface for
2497 each zone, that is created when the first container referencing its
2498 zone is created and removed when the last one referencing its zone
2501 * The default start timeout may now be configured on the kernel command
2502 line via systemd.default_timeout_start_sec=. It was already
2503 configurable via the DefaultTimeoutStartSec= option in
2504 /etc/systemd/system.conf.
2506 * Socket units gained a new TriggerLimitIntervalSec= and
2507 TriggerLimitBurst= setting to configure a limit on the activation
2508 rate of the socket unit.
2510 * The LimitNICE= setting now optionally takes normal UNIX nice values
2511 in addition to the raw integer limit value. If the specified
2512 parameter is prefixed with "+" or "-" and is in the range -20..19 the
2513 value is understood as UNIX nice value. If not prefixed like this it
2514 is understood as raw RLIMIT_NICE limit.
2516 * Note that the effect of the PrivateDevices= unit file setting changed
2517 slightly with this release: the per-device /dev file system will be
2518 mounted read-only from this version on, and will have "noexec"
2519 set. This (minor) change of behavior might cause some (exceptional)
2520 legacy software to break, when PrivateDevices=yes is set for its
2521 service. Please leave PrivateDevices= off if you run into problems
2524 * systemd-bootchart has been split out to a separate repository:
2525 https://github.com/systemd/systemd-bootchart
2527 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
2528 merged into the kernel in its current form.
2530 * The compatibility libraries libsystemd-daemon.so,
2531 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
2532 which have been deprecated since systemd-209 have been removed along
2533 with the corresponding pkg-config files. All symbols provided by
2534 those libraries are provided by libsystemd.so.
2536 * The Capabilities= unit file setting has been removed (it is ignored
2537 for backwards compatibility). AmbientCapabilities= and
2538 CapabilityBoundingSet= should be used instead.
2540 * A new special target has been added, initrd-root-device.target,
2541 which creates a synchronization point for dependencies of the root
2542 device in early userspace. Initramfs builders must ensure that this
2543 target is now included in early userspace.
2545 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
2546 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
2547 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
2548 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
2549 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
2550 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
2551 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
2552 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
2553 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
2554 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
2555 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
2556 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
2557 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
2558 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
2559 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
2560 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
2561 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
2562 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
2563 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
2564 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
2565 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
2566 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
2567 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
2568 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
2569 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
2572 — Fairfax, 2016-05-21
2576 * The systemd-resolved DNS resolver service has gained a substantial
2577 set of new features, most prominently it may now act as a DNSSEC
2578 validating stub resolver. DNSSEC mode is currently turned off by
2579 default, but is expected to be turned on by default in one of the
2580 next releases. For now, we invite everybody to test the DNSSEC logic
2581 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
2582 service also gained a full set of D-Bus interfaces, including calls
2583 to configure DNS and DNSSEC settings per link (for use by external
2584 network management software). systemd-resolved and systemd-networkd
2585 now distinguish between "search" and "routing" domains. The former
2586 are used to qualify single-label names, the latter are used purely
2587 for routing lookups within certain domains to specific links.
2588 resolved now also synthesizes RRs for all entries from /etc/hosts.
2590 * The systemd-resolve tool (which is a client utility for
2591 systemd-resolved) has been improved considerably and is now fully
2592 supported and documented. Hence it has moved from /usr/lib/systemd to
2595 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
2598 * The coredump collection logic has been reworked: when a coredump is
2599 collected it is now written to disk, compressed and processed
2600 (including stacktrace extraction) from a new instantiated service
2601 systemd-coredump@.service, instead of directly from the
2602 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
2603 processing large coredumps can take up a substantial amount of
2604 resources and time, and this previously happened entirely outside of
2605 systemd's service supervision. With the new logic the core_pattern
2606 hook only does minimal metadata collection before passing off control
2607 to the new instantiated service, which is configured with a time
2608 limit, a nice level and other settings to minimize negative impact on
2609 the rest of the system. Also note that the new logic will honour the
2610 RLIMIT_CORE setting of the crashed process, which now allows users
2611 and processes to turn off coredumping for their processes by setting
2614 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
2615 and all forked processes by default. Previously, PID 1 would leave
2616 the setting at "0" for all processes, as set by the kernel. Note that
2617 the resource limit traditionally has no effect on the generated
2618 coredumps on the system if the /proc/sys/kernel/core_pattern hook
2619 logic is used. Since the limit is now honoured (see above) its
2620 default has been changed so that the coredumping logic is enabled by
2621 default for all processes, while allowing specific opt-out.
2623 * When the stacktrace is extracted from processes of system users, this
2624 is now done as "systemd-coredump" user, in order to sandbox this
2625 potentially security sensitive parsing operation. (Note that when
2626 processing coredumps of normal users this is done under the user ID
2627 of process that crashed, as before.) Packagers should take notice
2628 that it is now necessary to create the "systemd-coredump" system user
2629 and group at package installation time.
2631 * The systemd-activate socket activation testing tool gained support
2632 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
2633 and --seqpacket switches. It also has been extended to support both
2634 new-style and inetd-style file descriptor passing. Use the new
2635 --inetd switch to request inetd-style file descriptor passing.
2637 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
2638 variable, which takes a boolean value. If set to false, ANSI color
2639 output is disabled in the tools even when run on a terminal that
2642 * The VXLAN support in networkd now supports two new settings
2643 DestinationPort= and PortRange=.
2645 * A new systemd.machine_id= kernel command line switch has been added,
2646 that may be used to set the machine ID in /etc/machine-id if it is
2647 not initialized yet. This command line option has no effect if the
2648 file is already initialized.
2650 * systemd-nspawn gained a new --as-pid2 switch that invokes any
2651 specified command line as PID 2 rather than PID 1 in the
2652 container. In this mode PID 1 is a minimal stub init process that
2653 implements the special POSIX and Linux semantics of PID 1 regarding
2654 signal and child process management. Note that this stub init process
2655 is implemented in nspawn itself and requires no support from the
2656 container image. This new logic is useful to support running
2657 arbitrary commands in the container, as normal processes are
2658 generally not prepared to run as PID 1.
2660 * systemd-nspawn gained a new --chdir= switch for setting the current
2661 working directory for the process started in the container.
2663 * "journalctl /dev/sda" will now output all kernel log messages for
2664 specified device from the current boot, in addition to all devices
2665 that are parents of it. This should make log output about devices
2666 pretty useful, as long as kernel drivers attach enough metadata to
2667 the log messages. (The usual SATA drivers do.)
2669 * The sd-journal API gained two new calls
2670 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
2671 that report whether log data from /run or /var has been found.
2673 * journalctl gained a new switch "--fields" that prints all journal
2674 record field names currently in use in the journal. This is backed
2675 by two new sd-journal API calls sd_journal_enumerate_fields() and
2676 sd_journal_restart_fields().
2678 * Most configurable timeouts in systemd now expect an argument of
2679 "infinity" to turn them off, instead of "0" as before. The semantics
2680 from now on is that a timeout of "0" means "now", and "infinity"
2681 means "never". To maintain backwards compatibility, "0" continues to
2682 turn off previously existing timeout settings.
2684 * "systemctl reload-or-try-restart" has been renamed to "systemctl
2685 try-reload-or-restart" to clarify what it actually does: the "try"
2686 logic applies to both reloading and restarting, not just restarting.
2687 The old name continues to be accepted for compatibility.
2689 * On boot-up, when PID 1 detects that the system clock is behind the
2690 release date of the systemd version in use, the clock is now set
2691 to the latter. Previously, this was already done in timesyncd, in order
2692 to avoid running with clocks set to the various clock epochs such as
2693 1902, 1938 or 1970. With this change the logic is now done in PID 1
2694 in addition to timesyncd during early boot-up, so that it is enforced
2695 before the first process is spawned by systemd. Note that the logic
2696 in timesyncd remains, as it is more comprehensive and ensures
2697 clock monotonicity by maintaining a persistent timestamp file in
2698 /var. Since /var is generally not available in earliest boot or the
2699 initrd, this part of the logic remains in timesyncd, and is not done
2702 * Support for tweaking details in net_cls.class_id through the
2703 NetClass= configuration directive has been removed, as the kernel
2704 people have decided to deprecate that controller in cgroup v2.
2705 Userspace tools such as nftables are moving over to setting rules
2706 that are specific to the full cgroup path of a task, which obsoletes
2707 these controllers anyway. The NetClass= directive is kept around for
2708 legacy compatibility reasons. For a more in-depth description of the
2709 kernel change, please refer to the respective upstream commit:
2711 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
2713 * A new service setting RuntimeMaxSec= has been added that may be used
2714 to specify a maximum runtime for a service. If the timeout is hit, the
2715 service is terminated and put into a failure state.
2717 * A new service setting AmbientCapabilities= has been added. It allows
2718 configuration of additional Linux process capabilities that are
2719 passed to the activated processes. This is only available on very
2722 * The process resource limit settings in service units may now be used
2723 to configure hard and soft limits individually.
2725 * The various libsystemd APIs such as sd-bus or sd-event now publicly
2726 expose support for gcc's __attribute__((cleanup())) C extension.
2727 Specifically, for many object destructor functions alternative
2728 versions have been added that have names suffixed with "p" and take a
2729 pointer to a pointer to the object to destroy, instead of just a
2730 pointer to the object itself. This is useful because these destructor
2731 functions may be used directly as parameters to the cleanup
2732 construct. Internally, systemd has been a heavy user of this GCC
2733 extension for a long time, and with this change similar support is
2734 now available to consumers of the library outside of systemd. Note
2735 that by using this extension in your sources compatibility with old
2736 and strictly ANSI compatible C compilers is lost. However, all gcc or
2737 LLVM versions of recent years support this extension.
2739 * Timer units gained support for a new setting RandomizedDelaySec= that
2740 allows configuring some additional randomized delay to the configured
2741 time. This is useful to spread out timer events to avoid load peaks in
2742 clusters or larger setups.
2744 * Calendar time specifications now support sub-second accuracy.
2746 * Socket units now support listening on SCTP and UDP-lite protocol
2749 * The sd-event API now comes with a full set of man pages.
2751 * Older versions of systemd contained experimental support for
2752 compressing journal files and coredumps with the LZ4 compressor that
2753 was not compatible with the lz4 binary (due to API limitations of the
2754 lz4 library). This support has been removed; only support for files
2755 compatible with the lz4 binary remains. This LZ4 logic is now
2756 officially supported and no longer considered experimental.
2758 * The dkr image import logic has been removed again from importd. dkr's
2759 micro-services focus doesn't fit into the machine image focus of
2760 importd, and quickly got out of date with the upstream dkr API.
2762 * Creation of the /run/lock/lockdev/ directory was dropped from
2763 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
2764 been available for many years. If you still need this, you need to
2765 create your own tmpfiles.d config file with:
2767 d /run/lock/lockdev 0775 root lock -
2769 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
2770 and RebootArgument= have been moved from the [Service] section of
2771 unit files to [Unit], and they are now supported on all unit types,
2772 not just service units. Of course, elogind will continue to
2773 understand these settings also at the old location, in order to
2774 maintain compatibility.
2776 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
2777 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
2778 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
2779 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
2780 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
2781 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
2782 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
2783 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
2784 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
2785 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
2786 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
2787 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
2788 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
2789 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
2790 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
2791 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
2792 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
2793 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
2794 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2796 — Berlin, 2016-02-11
2800 * A number of properties previously only settable in unit
2801 files are now also available as properties to set when
2802 creating transient units programmatically via the bus, as it
2803 is exposed with systemd-run's --property=
2804 setting. Specifically, these are: SyslogIdentifier=,
2805 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
2806 EnvironmentFile=, ReadWriteDirectories=,
2807 ReadOnlyDirectories=, InaccessibleDirectories=,
2808 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
2810 * When creating transient services via the bus API it is now
2811 possible to pass in a set of file descriptors to use as
2812 STDIN/STDOUT/STDERR for the invoked process.
2814 * Slice units may now be created transiently via the bus APIs,
2815 similar to the way service and scope units may already be
2816 created transiently.
2818 * Wherever systemd expects a calendar timestamp specification
2819 (like in journalctl's --since= and --until= switches) UTC
2820 timestamps are now supported. Timestamps suffixed with "UTC"
2821 are now considered to be in Universal Time Coordinated
2822 instead of the local timezone. Also, timestamps may now
2823 optionally be specified with sub-second accuracy. Both of
2824 these additions also apply to recurring calendar event
2825 specification, such as OnCalendar= in timer units.
2827 * journalctl gained a new "--sync" switch that asks the
2828 journal daemon to write all so far unwritten log messages to
2829 disk and sync the files, before returning.
2831 * systemd-tmpfiles learned two new line types "q" and "Q" that
2832 operate like "v", but also set up a basic btrfs quota
2833 hierarchy when used on a btrfs file system with quota
2836 * tmpfiles' "v", "q" and "Q" will now create a plain directory
2837 instead of a subvolume (even on a btrfs file system) if the
2838 root directory is a plain directory, and not a
2839 subvolume. This should simplify things with certain chroot()
2840 environments which are not aware of the concept of btrfs
2843 * systemd-detect-virt gained a new --chroot switch to detect
2844 whether execution takes place in a chroot() environment.
2846 * CPUAffinity= now takes CPU index ranges in addition to
2849 * The various memory-related resource limit settings (such as
2850 LimitAS=) now understand the usual K, M, G, ... suffixes to
2851 the base of 1024 (IEC). Similar, the time-related resource
2852 limit settings understand the usual min, h, day, ...
2855 * There's a new system.conf setting DefaultTasksMax= to
2856 control the default TasksMax= setting for services and
2857 scopes running on the system. (TasksMax= is the primary
2858 setting that exposes the "pids" cgroup controller on systemd
2859 and was introduced in the previous systemd release.) The
2860 setting now defaults to 512, which means services that are
2861 not explicitly configured otherwise will only be able to
2862 create 512 processes or threads at maximum, from this
2863 version on. Note that this means that thread- or
2864 process-heavy services might need to be reconfigured to set
2865 TasksMax= to a higher value. It is sufficient to set
2866 TasksMax= in these specific unit files to a higher value, or
2867 even "infinity". Similar, there's now a logind.conf setting
2868 UserTasksMax= that defaults to 4096 and limits the total
2869 number of processes or tasks each user may own
2870 concurrently. nspawn containers also have the TasksMax=
2871 value set by default now, to 8192. Note that all of this
2872 only has an effect if the "pids" cgroup controller is
2873 enabled in the kernel. The general benefit of these changes
2874 should be a more robust and safer system, that provides a
2875 certain amount of per-service fork() bomb protection.
2877 * systemd-nspawn gained the new --network-veth-extra= switch
2878 to define additional and arbitrarily-named virtual Ethernet
2879 links between the host and the container.
2881 * A new service execution setting PassEnvironment= has been
2882 added that allows importing select environment variables
2883 from PID1's environment block into the environment block of
2886 * Timer units gained support for a new RemainAfterElapse=
2887 setting which takes a boolean argument. It defaults to on,
2888 exposing behaviour unchanged to previous releases. If set to
2889 off, timer units are unloaded after they elapsed if they
2890 cannot elapse again. This is particularly useful for
2891 transient timer units, which shall not stay around longer
2892 than until they first elapse.
2894 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
2895 default now (the kernel default is 16). This is beneficial
2896 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
2897 allows substantially larger numbers of queued
2898 datagrams. This should increase the capability of systemd to
2899 parallelize boot-up, as logging and sd_notify() are unlikely
2900 to stall execution anymore. If you need to change the value
2901 from the new defaults, use the usual sysctl.d/ snippets.
2903 * The compression framing format used by the journal or
2904 coredump processing has changed to be in line with what the
2905 official LZ4 tools generate. LZ4 compression support in
2906 systemd was considered unsupported previously, as the format
2907 was not compatible with the normal tools. With this release
2908 this has changed now, and it is hence safe for downstream
2909 distributions to turn it on. While not compressing as well
2910 as the XZ, LZ4 is substantially faster, which makes
2911 it a good default choice for the compression logic in the
2912 journal and in coredump handling.
2914 * Any reference to /etc/mtab has been dropped from
2915 systemd. The file has been obsolete since a while, but
2916 systemd refused to work on systems where it was incorrectly
2917 set up (it should be a symlink or non-existent). Please make
2918 sure to update to util-linux 2.27.1 or newer in conjunction
2919 with this systemd release, which also drops any reference to
2920 /etc/mtab. If you maintain a distribution make sure that no
2921 software you package still references it, as this is a
2922 likely source of bugs. There's also a glibc bug pending,
2923 asking for removal of any reference to this obsolete file:
2925 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
2927 Note that only util-linux versions built with
2928 --enable-libmount-force-mountinfo are supported.
2930 * Support for the ".snapshot" unit type has been removed. This
2931 feature turned out to be little useful and little used, and
2932 has now been removed from the core and from systemctl.
2934 * The dependency types RequiresOverridable= and
2935 RequisiteOverridable= have been removed from systemd. They
2936 have been used only very sparingly to our knowledge and
2937 other options that provide a similar effect (such as
2938 systemctl --mode=ignore-dependencies) are much more useful
2939 and commonly used. Moreover, they were only half-way
2940 implemented as the option to control behaviour regarding
2941 these dependencies was never added to systemctl. By removing
2942 these dependency types the execution engine becomes a bit
2943 simpler. Unit files that use these dependencies should be
2944 changed to use the non-Overridable dependency types
2945 instead. In fact, when parsing unit files with these
2946 options, that's what systemd will automatically convert them
2947 too, but it will also warn, asking users to fix the unit
2948 files accordingly. Removal of these dependency types should
2949 only affect a negligible number of unit files in the wild.
2951 * Behaviour of networkd's IPForward= option changed
2952 (again). It will no longer maintain a per-interface setting,
2953 but propagate one way from interfaces where this is enabled
2954 to the global kernel setting. The global setting will be
2955 enabled when requested by a network that is set up, but
2956 never be disabled again. This change was made to make sure
2957 IPv4 and IPv6 behaviour regarding packet forwarding is
2958 similar (as the Linux IPv6 stack does not support
2959 per-interface control of this setting) and to minimize
2962 * In unit files the behaviour of %u, %U, %h, %s has
2963 changed. These specifiers will now unconditionally resolve
2964 to the various user database fields of the user that the
2965 systemd instance is running as, instead of the user
2966 configured in the specific unit via User=. Note that this
2967 effectively doesn't change much, as resolving of these
2968 specifiers was already turned off in the --system instance
2969 of systemd, as we cannot do NSS lookups from PID 1. In the
2970 --user instance of systemd these specifiers where correctly
2971 resolved, but hardly made any sense, since the user instance
2972 lacks privileges to do user switches anyway, and User= is
2973 hence useless. Moreover, even in the --user instance of
2974 systemd behaviour was awkward as it would only take settings
2975 from User= assignment placed before the specifier into
2976 account. In order to unify and simplify the logic around
2977 this the specifiers will now always resolve to the
2978 credentials of the user invoking the manager (which in case
2979 of PID 1 is the root user).
2981 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
2982 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
2983 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
2984 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
2985 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
2986 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
2987 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
2988 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
2989 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
2990 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
2991 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
2992 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
2993 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
2994 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
2997 — Berlin, 2015-11-18
3001 * systemd now depends on util-linux v2.27. More specifically,
3002 the newly added mount monitor feature in libmount now
3003 replaces systemd's former own implementation.
3005 * libmount mandates /etc/mtab not to be regular file, and
3006 systemd now enforces this condition at early boot.
3007 /etc/mtab has been deprecated and warned about for a very
3008 long time, so systems running systemd should already have
3009 stopped having this file around as anything else than a
3010 symlink to /proc/self/mounts.
3012 * Support for the "pids" cgroup controller has been added. It
3013 allows accounting the number of tasks in a cgroup and
3014 enforcing limits on it. This adds two new setting
3015 TasksAccounting= and TasksMax= to each unit, as well as a
3016 global option DefaultTasksAccounting=.
3018 * Support for the "net_cls" cgroup controller has been added.
3019 It allows assigning a net class ID to each task in the
3020 cgroup, which can then be used in firewall rules and traffic
3021 shaping configurations. Note that the kernel netfilter net
3022 class code does not currently work reliably for ingress
3023 packets on unestablished sockets.
3025 This adds a new config directive called NetClass= to CGroup
3026 enabled units. Allowed values are positive numbers for fixed
3027 assignments and "auto" for picking a free value
3030 * 'systemctl is-system-running' now returns 'offline' if the
3031 system is not booted with systemd. This command can now be
3032 used as a substitute for 'systemd-notify --booted'.
3034 * Watchdog timeouts have been increased to 3 minutes for all
3035 in-tree service files. Apparently, disk IO issues are more
3036 frequent than we hoped, and user reported >1 minute waiting
3039 * 'machine-id-commit' functionality has been merged into
3040 'machine-id-setup --commit'. The separate binary has been
3043 * The WorkingDirectory= directive in unit files may now be set
3044 to the special value '~'. In this case, the working
3045 directory is set to the home directory of the user
3046 configured in User=.
3048 * "machinectl shell" will now open the shell in the home
3049 directory of the selected user by default.
3051 * The CrashChVT= configuration file setting is renamed to
3052 CrashChangeVT=, following our usual logic of not
3053 abbreviating unnecessarily. The old directive is still
3054 supported for compat reasons. Also, this directive now takes
3055 an integer value between 1 and 63, or a boolean value. The
3056 formerly supported '-1' value for disabling stays around for
3059 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
3060 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
3061 RootDirectory= properties can now be set for transient
3064 * The systemd-analyze tool gained a new "set-log-target" verb
3065 to change the logging target the system manager logs to
3066 dynamically during runtime. This is similar to how
3067 "systemd-analyze set-log-level" already changes the log
3070 * In nspawn /sys is now mounted as tmpfs, with only a selected
3071 set of subdirectories mounted in from the real sysfs. This
3072 enhances security slightly, and is useful for ensuring user
3073 namespaces work correctly.
3075 * Support for USB FunctionFS activation has been added. This
3076 allows implementation of USB gadget services that are
3077 activated as soon as they are requested, so that they don't
3078 have to run continuously, similar to classic socket
3081 * The "systemctl exit" command now optionally takes an
3082 additional parameter that sets the exit code to return from
3083 the systemd manager when exiting. This is only relevant when
3084 running the systemd user instance, or when running the
3085 system instance in a container.
3087 * sd-bus gained the new API calls sd_bus_path_encode_many()
3088 and sd_bus_path_decode_many() that allow easy encoding and
3089 decoding of multiple identifier strings inside a D-Bus
3090 object path. Another new call sd_bus_default_flush_close()
3091 has been added to flush and close per-thread default
3094 * systemd-cgtop gained support for a -M/--machine= switch to
3095 show the control groups within a certain container only.
3097 * "systemctl kill" gained support for an optional --fail
3098 switch. If specified the requested operation will fail of no
3099 processes have been killed, because the unit had no
3100 processes attached, or similar.
3102 * A new systemd.crash_reboot=1 kernel command line option has
3103 been added that triggers a reboot after crashing. This can
3104 also be set through CrashReboot= in systemd.conf.
3106 * The RuntimeDirectory= setting now understands unit
3107 specifiers like %i or %f.
3109 * A new (still internal) library API sd-ipv4acd has been added,
3110 that implements address conflict detection for IPv4. It's
3111 based on code from sd-ipv4ll, and will be useful for
3112 detecting DHCP address conflicts.
3114 * File descriptors passed during socket activation may now be
3115 named. A new API sd_listen_fds_with_names() is added to
3116 access the names. The default names may be overridden,
3117 either in the .socket file using the FileDescriptorName=
3118 parameter, or by passing FDNAME= when storing the file
3119 descriptors using sd_notify().
3121 * systemd-networkd gained support for:
3123 - Setting the IPv6 Router Advertisement settings via
3124 IPv6AcceptRouterAdvertisements= in .network files.
3126 - Configuring the HelloTimeSec=, MaxAgeSec= and
3127 ForwardDelaySec= bridge parameters in .netdev files.
3129 - Configuring PreferredSource= for static routes in
3132 * The "ask-password" framework used to query for LUKS harddisk
3133 passwords or SSL passwords during boot gained support for
3134 caching passwords in the kernel keyring, if it is
3135 available. This makes sure that the user only has to type in
3136 a passphrase once if there are multiple objects to unlock
3137 with the same one. Previously, such password caching was
3138 available only when Plymouth was used; this moves the
3139 caching logic into the systemd codebase itself. The
3140 "systemd-ask-password" utility gained a new --keyname=
3141 switch to control which kernel keyring key to use for
3142 caching a password in. This functionality is also useful for
3143 enabling display managers such as gdm to automatically
3144 unlock the user's GNOME keyring if its passphrase, the
3145 user's password and the harddisk password are the same, if
3146 gdm-autologin is used.
3148 * When downloading tar or raw images using "machinectl
3149 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
3150 file is now also downloaded, if it is available and stored
3151 next to the image file.
3153 * Units of type ".socket" gained a new boolean setting
3154 Writable= which is only useful in conjunction with
3155 ListenSpecial=. If true, enables opening the specified
3156 special file in O_RDWR mode rather than O_RDONLY mode.
3158 * systemd-rfkill has been reworked to become a singleton
3159 service that is activated through /dev/rfkill on each rfkill
3160 state change and saves the settings to disk. This way,
3161 systemd-rfkill is now compatible with devices that exist
3162 only intermittendly, and even restores state if the previous
3163 system shutdown was abrupt rather than clean.
3165 * The journal daemon gained support for vacuuming old journal
3166 files controlled by the number of files that shall remain,
3167 in addition to the already existing control by size and by
3168 date. This is useful as journal interleaving performance
3169 degrades with too many separate journal files, and allows
3170 putting an effective limit on them. The new setting defaults
3171 to 100, but this may be changed by setting SystemMaxFiles=
3172 and RuntimeMaxFiles= in journald.conf. Also, the
3173 "journalctl" tool gained the new --vacuum-files= switch to
3174 manually vacuum journal files to leave only the specified
3175 number of files in place.
3177 * udev will now create /dev/disk/by-path links for ATA devices
3178 on kernels where that is supported.
3180 * Galician, Serbian, Turkish and Korean translations were added.
3182 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
3183 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
3184 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
3185 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
3186 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
3187 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
3188 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
3189 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
3190 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
3191 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
3192 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
3193 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
3194 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
3195 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
3196 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
3197 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
3198 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
3199 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
3201 — Berlin, 2015-10-07
3205 * The DHCP implementation of systemd-networkd gained a set of
3208 - The DHCP server now supports emitting DNS and NTP
3209 information. It may be enabled and configured via
3210 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
3211 and NTP information is enabled, but no servers are
3212 configured, the corresponding uplink information (if there
3213 is any) is propagated.
3215 - Server and client now support transmission and reception
3216 of timezone information. It can be configured via the
3217 newly introduced network options UseTimezone=,
3218 EmitTimezone=, and Timezone=. Transmission of timezone
3219 information is enabled between host and containers by
3220 default now: the container will change its local timezone
3221 to what the host has set.
3223 - Lease timeouts can now be configured via
3224 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
3226 - The DHCP server improved on the stability of
3227 leases. Clients are more likely to get the same lease
3228 information back, even if the server loses state.
3230 - The DHCP server supports two new configuration options to
3231 control the lease address pool metrics, PoolOffset= and
3234 * The encapsulation limit of tunnels in systemd-networkd may
3235 now be configured via 'EncapsulationLimit='. It allows
3236 modifying the maximum additional levels of encapsulation
3237 that are permitted to be prepended to a packet.
3239 * systemd now supports the concept of user buses replacing
3240 session buses, if used with dbus-1.10 (and enabled via dbus
3241 --enable-user-session). It previously only supported this on
3242 kdbus-enabled systems, and this release expands this to
3243 'dbus-daemon' systems.
3245 * systemd-networkd now supports predictable interface names
3248 * systemd now optionally supports the new Linux kernel
3249 "unified" control group hierarchy. If enabled via the kernel
3250 command-line option 'systemd.unified_cgroup_hierarchy=1',
3251 systemd will try to mount the unified cgroup hierarchy
3252 directly on /sys/fs/cgroup. If not enabled, or not
3253 available, systemd will fall back to the legacy cgroup
3254 hierarchy setup, as before. Host system and containers can
3255 mix and match legacy and unified hierarchies as they
3256 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
3257 environment variable to individually select the hierarchy to
3258 use for executed containers. By default, nspawn will use the
3259 unified hierarchy for the containers if the host uses the
3260 unified hierarchy, and the legacy hierarchy otherwise.
3261 Please note that at this point the unified hierarchy is an
3262 experimental kernel feature and is likely to change in one
3263 of the next kernel releases. Therefore, it should not be
3264 enabled by default in downstream distributions yet. The
3265 minimum required kernel version for the unified hierarchy to
3266 work is 4.2. Note that when the unified hierarchy is used
3267 for the first time delegated access to controllers is
3268 safe. Because of this systemd-nspawn containers will get
3269 access to controllers now, as will systemd user
3270 sessions. This means containers and user sessions may now
3271 manage their own resources, partitioning up what the system
3274 * A new special scope unit "init.scope" has been introduced
3275 that encapsulates PID 1 of the system. It may be used to
3276 determine resource usage and enforce resource limits on PID
3277 1 itself. PID 1 hence moved out of the root of the control
3280 * The cgtop tool gained support for filtering out kernel
3281 threads when counting tasks in a control group. Also, the
3282 count of processes is now recursively summed up by
3283 default. Two options -k and --recursive= have been added to
3284 revert to old behaviour. The tool has also been updated to
3285 work correctly in containers now.
3287 * systemd-nspawn's --bind= and --bind-ro= options have been
3288 extended to allow creation of non-recursive bind mounts.
3290 * libsystemd gained two new calls sd_pid_get_cgroup() and
3291 sd_peer_get_cgroup() which return the control group path of
3292 a process or peer of a connected AF_UNIX socket. This
3293 function call is particularly useful when implementing
3294 delegated subtrees support in the control group hierarchy.
3296 * The "sd-event" event loop API of libsystemd now supports
3297 correct dequeuing of real-time signals, without losing
3300 * When systemd requests a PolicyKit decision when managing
3301 units it will now add additional fields to the request,
3302 including unit name and desired operation. This enables more
3303 powerful PolicyKit policies, that make decisions depending
3304 on these parameters.
3306 * nspawn learnt support for .nspawn settings files, that may
3307 accompany the image files or directories of containers, and
3308 may contain additional settings for the container. This is
3309 an alternative to configuring container parameters via the
3310 nspawn command line.
3312 Contributions from: Cristian Rodríguez, Daniel Mack, David
3313 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
3314 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
3315 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
3316 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
3317 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
3318 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
3319 Andersen, Tom Gundersen, Torstein Husebø
3321 — Berlin, 2015-09-08
3325 * machinectl gained a new verb 'shell' which opens a fresh
3326 shell on the target container or the host. It is similar to
3327 the existing 'login' command of machinectl, but spawns the
3328 shell directly without prompting for username or
3329 password. The pseudo machine '.host' now refers to the local
3330 host and is used by default. Hence, 'machinectl shell' can
3331 be used as replacement for 'su -' which spawns a session as
3332 a fresh systemd unit in a way that is fully isolated from
3333 the originating session.
3335 * systemd-networkd learned to cope with private-zone DHCP
3336 options and allows other programs to query the values.
3338 * SELinux access control when enabling/disabling units is no
3339 longer enforced with this release. The previous
3340 implementation was incorrect, and a new corrected
3341 implementation is not yet available. As unit file operations
3342 are still protected via PolicyKit and D-Bus policy this is
3343 not a security problem. Yet, distributions which care about
3344 optimal SELinux support should probably not stabilize on
3347 * sd-bus gained support for matches of type "arg0has=", that
3348 test for membership of strings in string arrays sent in bus
3351 * systemd-resolved now dumps the contents of its DNS and LLMNR
3352 caches to the logs on reception of the SIGUSR1 signal. This
3353 is useful to debug DNS behaviour.
3355 * The coredumpctl tool gained a new --directory= option to
3356 operate on journal files in a specific directory.
3358 * "systemctl reboot" and related commands gained a new
3359 "--message=" option which may be used to set a free-text
3360 wall message when shutting down or rebooting the
3361 system. This message is also logged, which is useful for
3362 figuring out the reason for a reboot or shutdown a
3365 * The "systemd-resolve-host" tool's -i switch now takes
3366 network interface numbers as alternative to interface names.
3368 * A new unit file setting for services has been introduced:
3369 UtmpMode= allows configuration of how precisely systemd
3370 handles utmp and wtmp entries for the service if this is
3371 enabled. This allows writing services that appear similar to
3372 user sessions in the output of the "w", "who", "last" and
3375 * systemd-resolved will now locally synthesize DNS resource
3376 records for the "localhost" and "gateway" domains as well as
3377 the local hostname. This should ensure that clients querying
3378 RRs via resolved will get similar results as those going via
3379 NSS, if nss-myhostname is enabled.
3381 Contributions from: Alastair Hughes, Alex Crawford, Daniel
3382 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
3383 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
3384 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
3385 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
3386 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
3387 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
3388 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
3389 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
3390 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
3391 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
3392 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3394 — Berlin, 2015-08-27
3398 * The systemd-efi-boot-generator functionality was merged into
3399 systemd-gpt-auto-generator.
3401 * systemd-networkd now supports Group Policy for vxlan
3402 devices. It can be enabled via the new boolean configuration
3403 option called 'GroupPolicyExtension='.
3405 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
3406 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
3407 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
3409 — Berlin, 2015-07-31
3413 * The python-systemd code has been removed from the systemd repository.
3414 A new repository has been created which accommodates the code from
3415 now on, and we kindly ask distributions to create a separate package
3416 for this: https://github.com/systemd/python-systemd
3418 * The systemd daemon will now reload its main configuration
3419 (/etc/systemd/system.conf) on daemon-reload.
3421 * sd-dhcp now exposes vendor specific extensions via
3422 sd_dhcp_lease_get_vendor_specific().
3424 * systemd-networkd gained a number of new configuration options.
3426 - A new boolean configuration option for TAP devices called
3427 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
3428 device, thus allowing to send and receive GSO packets.
3430 - A new tunnel configuration option called 'CopyDSCP='.
3431 If enabled, the DSCP field of ip6 tunnels is copied into the
3432 decapsulated packet.
3434 - A set of boolean bridge configuration options were added.
3435 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
3436 and 'UnicastFlood=' are now parsed by networkd and applied to the
3437 respective bridge link device via the respective IFLA_BRPORT_*
3440 - A new string configuration option to override the hostname sent
3441 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
3442 is true, networkd will use the configured hostname instead of the
3443 system hostname when sending DHCP requests.
3445 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
3446 networkd will configure the IPv6 flow-label of the tunnel device
3447 according to RFC2460.
3449 - The 'macvtap' virtual network devices are now supported, similar to
3450 the already supported 'macvlan' devices.
3452 * systemd-resolved now implements RFC5452 to improve resilience against
3453 cache poisoning. Additionally, source port randomization is enabled
3454 by default to further protect against DNS spoofing attacks.
3456 * nss-mymachines now supports translating UIDs and GIDs of running
3457 containers with user-namespaces enabled. If a container 'foo'
3458 translates a host uid 'UID' to the container uid 'TUID', then
3459 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
3460 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
3461 mapped as 'vg-foo-TGID'.
3463 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
3464 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
3465 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
3466 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
3467 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
3468 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
3469 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
3470 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
3471 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
3472 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
3474 — Berlin, 2015-07-29
3478 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
3479 There are no known issues with current sysfs, and udev does not need
3480 or should be used to work around such bugs.
3482 * udev does no longer enable USB HID power management. Several reports
3483 indicate, that some devices cannot handle that setting.
3485 * The udev accelerometer helper was removed. The functionality
3486 is now fully included in iio-sensor-proxy. But this means,
3487 older iio-sensor-proxy versions will no longer provide
3488 accelerometer/orientation data with this systemd version.
3489 Please upgrade iio-sensor-proxy to version 1.0.
3491 * networkd gained a new configuration option IPv6PrivacyExtensions=
3492 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
3493 for Stateless Address") on selected networks.
3495 * For the sake of fewer build-time dependencies and less code in the
3496 main repository, the python bindings are about to be removed in the
3497 next release. A new repository has been created which accommodates
3498 the code from now on, and we kindly ask distributions to create a
3499 separate package for this. The removal will take place in v223.
3501 https://github.com/systemd/python-systemd
3503 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
3504 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
3505 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
3506 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
3507 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
3508 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
3509 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
3510 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
3511 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3512 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3514 — Berlin, 2015-07-07
3518 * The sd-bus.h and sd-event.h APIs have now been declared
3519 stable and have been added to the official interface of
3520 libsystemd.so. sd-bus implements an alternative D-Bus client
3521 library, that is relatively easy to use, very efficient and
3522 supports both classic D-Bus as well as kdbus as transport
3523 backend. sd-event is a generic event loop abstraction that
3524 is built around Linux epoll, but adds features such as event
3525 prioritization or efficient timer handling. Both APIs are good
3526 choices for C programs looking for a bus and/or event loop
3527 implementation that is minimal and does not have to be
3528 portable to other kernels.
3530 * kdbus support is no longer compile-time optional. It is now
3531 always built-in. However, it can still be disabled at
3532 runtime using the kdbus=0 kernel command line setting, and
3533 that setting may be changed to default to off, by specifying
3534 --disable-kdbus at build-time. Note though that the kernel
3535 command line setting has no effect if the kdbus.ko kernel
3536 module is not installed, in which case kdbus is (obviously)
3537 also disabled. We encourage all downstream distributions to
3538 begin testing kdbus by adding it to the kernel images in the
3539 development distributions, and leaving kdbus support in
3542 * The minimal required util-linux version has been bumped to
3545 * Support for chkconfig (--enable-chkconfig) was removed in
3546 favor of calling an abstraction tool
3547 /lib/systemd/systemd-sysv-install. This needs to be
3548 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
3549 in README for details.
3551 * If there's a systemd unit and a SysV init script for the
3552 same service name, and the user executes "systemctl enable"
3553 for it (or a related call), then this will now enable both
3554 (or execute the related operation on both), not just the
3557 * The libudev API documentation has been converted from gtkdoc
3560 * gudev has been removed from the systemd tree, it is now an
3563 * The systemd-cgtop tool learnt a new --raw switch to generate
3564 "raw" (machine parsable) output.
3566 * networkd's IPForwarding= .network file setting learnt the
3567 new setting "kernel", which ensures that networkd does not
3568 change the IP forwarding sysctl from the default kernel
3571 * The systemd-logind bus API now exposes a new boolean
3572 property "Docked" that reports whether logind considers the
3573 system "docked", i.e. connected to a docking station or not.
3575 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
3576 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
3577 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
3578 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
3579 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
3580 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
3581 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
3582 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
3583 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
3584 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
3585 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
3586 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
3587 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
3588 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3589 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
3590 Fink, Zbigniew Jędrzejewski-Szmek
3592 — Berlin, 2015-06-19
3596 * The gudev library has been extracted into a separate repository
3597 available at: https://git.gnome.org/browse/libgudev/
3598 It is now managed as part of the Gnome project. Distributions
3599 are recommended to pass --disable-gudev to systemd and use
3600 gudev from the Gnome project instead. gudev is still included
3601 in systemd, for now. It will be removed soon, though. Please
3602 also see the announcement-thread on systemd-devel:
3603 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
3605 * systemd now exposes a CPUUsageNSec= property for each
3606 service unit on the bus, that contains the overall consumed
3607 CPU time of a service (the sum of what each process of the
3608 service consumed). This value is only available if
3609 CPUAccounting= is turned on for a service, and is then shown
3610 in the "systemctl status" output.
3612 * Support for configuring alternative mappings of the old SysV
3613 runlevels to systemd targets has been removed. They are now
3614 hardcoded in a way that runlevels 2, 3, 4 all map to
3615 multi-user.target and 5 to graphical.target (which
3616 previously was already the default behaviour).
3618 * The auto-mounter logic gained support for mount point
3619 expiry, using a new TimeoutIdleSec= setting in .automount
3620 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
3622 * The EFI System Partition (ESP) as mounted to /boot by
3623 systemd-efi-boot-generator will now be unmounted
3624 automatically after 2 minutes of not being used. This should
3625 minimize the risk of ESP corruptions.
3627 * New /etc/fstab options x-systemd.requires= and
3628 x-systemd.requires-mounts-for= are now supported to express
3629 additional dependencies for mounts. This is useful for
3630 journalling file systems that support external journal
3631 devices or overlay file systems that require underlying file
3632 systems to be mounted.
3634 * systemd does not support direct live-upgrades (via systemctl
3635 daemon-reexec) from versions older than v44 anymore. As no
3636 distribution we are aware of shipped such old versions in a
3637 stable release this should not be problematic.
3639 * When systemd forks off a new per-connection service instance
3640 it will now set the $REMOTE_ADDR environment variable to the
3641 remote IP address, and $REMOTE_PORT environment variable to
3642 the remote IP port. This behaviour is similar to the
3643 corresponding environment variables defined by CGI.
3645 * systemd-networkd gained support for uplink failure
3646 detection. The BindCarrier= option allows binding interface
3647 configuration dynamically to the link sense of other
3648 interfaces. This is useful to achieve behaviour like in
3651 * systemd-networkd gained support for configuring the DHCP
3652 client identifier to use when requesting leases.
3654 * systemd-networkd now has a per-network UseNTP= option to
3655 configure whether NTP server information acquired via DHCP
3656 is passed on to services like systemd-timesyncd.
3658 * systemd-networkd gained support for vti6 tunnels.
3660 * Note that systemd-networkd manages the sysctl variable
3661 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
3662 it is configured for since v219. The variable controls IP
3663 forwarding, and is a per-interface alternative to the global
3664 /proc/sys/net/ipv[46]/ip_forward. This setting is
3665 configurable in the IPForward= option, which defaults to
3666 "no". This means if networkd is used for an interface it is
3667 no longer sufficient to set the global sysctl option to turn
3668 on IP forwarding! Instead, the .network file option
3669 IPForward= needs to be turned on! Note that the
3670 implementation of this behaviour was broken in v219 and has
3673 * Many bonding and vxlan options are now configurable in
3676 * systemd-nspawn gained a new --property= setting to set unit
3677 properties for the container scope. This is useful for
3678 setting resource parameters (e.g. "CPUShares=500") on
3679 containers started from the command line.
3681 * systemd-nspawn gained a new --private-users= switch to make
3682 use of user namespacing available on recent Linux kernels.
3684 * systemd-nspawn may now be called as part of a shell pipeline
3685 in which case the pipes used for stdin and stdout are passed
3686 directly to the process invoked in the container, without
3687 indirection via a pseudo tty.
3689 * systemd-nspawn gained a new switch to control the UNIX
3690 signal to use when killing the init process of the container
3693 * systemd-nspawn gained a new --overlay= switch for mounting
3694 overlay file systems into the container using the new kernel
3697 * When a container image is imported via systemd-importd and
3698 the host file system is not btrfs, a loopback block device
3699 file is created in /var/lib/machines.raw with a btrfs file
3700 system inside. It is then mounted to /var/lib/machines to
3701 enable btrfs features for container management. The loopback
3702 file and btrfs file system is grown as needed when container
3703 images are imported via systemd-importd.
3705 * systemd-machined/systemd-importd gained support for btrfs
3706 quota, to enforce container disk space limits on disk. This
3707 is exposed in "machinectl set-limit".
3709 * systemd-importd now can import containers from local .tar,
3710 .raw and .qcow2 images, and export them to .tar and .raw. It
3711 can also import dkr v2 images now from the network (on top
3714 * systemd-importd gained support for verifying downloaded
3715 images with gpg2 (previously only gpg1 was supported).
3717 * systemd-machined, systemd-logind, systemd: most bus calls
3718 are now accessible to unprivileged processes via
3719 PolicyKit. Also, systemd-logind will now allow users to kill
3720 their own sessions without further privileges or
3723 * systemd-shutdownd has been removed. This service was
3724 previously responsible for implementing scheduled shutdowns
3725 as exposed in /usr/bin/shutdown's time parameter. This
3726 functionality has now been moved into systemd-logind and is
3727 accessible via a bus interface.
3729 * "systemctl reboot" gained a new switch --firmware-setup that
3730 can be used to reboot into the EFI firmware setup, if that
3731 is available. systemd-logind now exposes an API on the bus
3732 to trigger such reboots, in case graphical desktop UIs want
3733 to cover this functionality.
3735 * "systemctl enable", "systemctl disable" and "systemctl mask"
3736 now support a new "--now" switch. If specified the units
3737 that are enabled will also be started, and the ones
3738 disabled/masked also stopped.
3740 * The Gummiboot EFI boot loader tool has been merged into
3741 systemd, and renamed to "systemd-boot". The bootctl tool has been
3742 updated to support systemd-boot.
3744 * An EFI kernel stub has been added that may be used to create
3745 kernel EFI binaries that contain not only the actual kernel,
3746 but also an initrd, boot splash, command line and OS release
3747 information. This combined binary can then be signed as a
3748 single image, so that the firmware can verify it all in one
3749 step. systemd-boot has special support for EFI binaries created
3750 like this and can extract OS release information from them
3751 and show them in the boot menu. This functionality is useful
3752 to implement cryptographically verified boot schemes.
3754 * Optional support has been added to systemd-fsck to pass
3755 fsck's progress report to an AF_UNIX socket in the file
3758 * udev will no longer create device symlinks for all block
3759 devices by default. A blacklist for excluding special block
3760 devices from this logic has been turned into a whitelist
3761 that requires picking block devices explicitly that require
3764 * A new (currently still internal) API sd-device.h has been
3765 added to libsystemd. This modernized API is supposed to
3766 replace libudev eventually. In fact, already much of libudev
3767 is now just a wrapper around sd-device.h.
3769 * A new hwdb database for storing metadata about pointing
3770 stick devices has been added.
3772 * systemd-tmpfiles gained support for setting file attributes
3773 similar to the "chattr" tool with new 'h' and 'H' lines.
3775 * systemd-journald will no longer unconditionally set the
3776 btrfs NOCOW flag on new journal files. This is instead done
3777 with tmpfiles snippet using the new 'h' line type. This
3778 allows easy disabling of this logic, by masking the
3779 journal-nocow.conf tmpfiles file.
3781 * systemd-journald will now translate audit message types to
3782 human readable identifiers when writing them to the
3783 journal. This should improve readability of audit messages.
3785 * The LUKS logic gained support for the offset= and skip=
3786 options in /etc/crypttab, as previously implemented by
3789 * /usr/lib/os-release gained a new optional field VARIANT= for
3790 distributions that support multiple variants (such as a
3791 desktop edition, a server edition, ...)
3793 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
3794 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
3795 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
3796 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
3797 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
3798 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
3799 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
3800 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
3801 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
3802 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
3803 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
3804 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
3805 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
3806 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
3807 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
3808 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
3809 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
3810 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
3811 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
3812 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
3813 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
3814 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
3815 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
3816 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
3817 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
3818 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
3819 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
3821 — Berlin, 2015-05-22
3825 * Introduce a new API "sd-hwdb.h" for querying the hardware
3826 metadata database. With this minimal interface one can query
3827 and enumerate the udev hwdb, decoupled from the old libudev
3828 library. libudev's interface for this is now only a wrapper
3829 around sd-hwdb. A new tool systemd-hwdb has been added to
3830 interface with and update the database.
3832 * When any of systemd's tools copies files (for example due to
3833 tmpfiles' C lines) a btrfs reflink will attempted first,
3834 before bytewise copying is done.
3836 * systemd-nspawn gained a new --ephemeral switch. When
3837 specified a btrfs snapshot is taken of the container's root
3838 directory, and immediately removed when the container
3839 terminates again. Thus, a container can be started whose
3840 changes never alter the container's root directory, and are
3841 lost on container termination. This switch can also be used
3842 for starting a container off the root file system of the
3843 host without affecting the host OS. This switch is only
3844 available on btrfs file systems.
3846 * systemd-nspawn gained a new --template= switch. It takes the
3847 path to a container tree to use as template for the tree
3848 specified via --directory=, should that directory be
3849 missing. This allows instantiating containers dynamically,
3850 on first run. This switch is only available on btrfs file
3853 * When a .mount unit refers to a mount point on which multiple
3854 mounts are stacked, and the .mount unit is stopped all of
3855 the stacked mount points will now be unmounted until no
3856 mount point remains.
3858 * systemd now has an explicit notion of supported and
3859 unsupported unit types. Jobs enqueued for unsupported unit
3860 types will now fail with an "unsupported" error code. More
3861 specifically .swap, .automount and .device units are not
3862 supported in containers, .busname units are not supported on
3863 non-kdbus systems. .swap and .automount are also not
3864 supported if their respective kernel compile time options
3867 * machinectl gained support for two new "copy-from" and
3868 "copy-to" commands for copying files from a running
3869 container to the host or vice versa.
3871 * machinectl gained support for a new "bind" command to bind
3872 mount host directories into local containers. This is
3873 currently only supported for nspawn containers.
3875 * networkd gained support for configuring bridge forwarding
3876 database entries (fdb) from .network files.
3878 * A new tiny daemon "systemd-importd" has been added that can
3879 download container images in tar, raw, qcow2 or dkr formats,
3880 and make them available locally in /var/lib/machines, so
3881 that they can run as nspawn containers. The daemon can GPG
3882 verify the downloads (not supported for dkr, since it has no
3883 provisions for verifying downloads). It will transparently
3884 decompress bz2, xz, gzip compressed downloads if necessary,
3885 and restore sparse files on disk. The daemon uses privilege
3886 separation to ensure the actual download logic runs with
3887 fewer privileges than the daemon itself. machinectl has
3888 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
3889 make the functionality of importd available to the
3890 user. With this in place the Fedora and Ubuntu "Cloud"
3891 images can be downloaded and booted as containers unmodified
3892 (the Fedora images lack the appropriate GPG signature files
3893 currently, so they cannot be verified, but this will change
3894 soon, hopefully). Note that downloading images is currently
3895 only fully supported on btrfs.
3897 * machinectl is now able to list container images found in
3898 /var/lib/machines, along with some metadata about sizes of
3899 disk and similar. If the directory is located on btrfs and
3900 quota is enabled, this includes quota display. A new command
3901 "image-status" has been added that shows additional
3902 information about images.
3904 * machinectl is now able to clone container images
3905 efficiently, if the underlying file system (btrfs) supports
3906 it, with the new "machinectl clone" command. It also
3907 gained commands for renaming and removing images, as well as
3908 marking them read-only or read-write (supported also on
3909 legacy file systems).
3911 * networkd gained support for collecting LLDP network
3912 announcements, from hardware that supports this. This is
3913 shown in networkctl output.
3915 * systemd-run gained support for a new -t (--pty) switch for
3916 invoking a binary on a pty whose input and output is
3917 connected to the invoking terminal. This allows executing
3918 processes as system services while interactively
3919 communicating with them via the terminal. Most interestingly
3920 this is supported across container boundaries. Invoking
3921 "systemd-run -t /bin/bash" is an alternative to running a
3922 full login session, the difference being that the former
3923 will not register a session, nor go through the PAM session
3926 * tmpfiles gained support for a new "v" line type for creating
3927 btrfs subvolumes. If the underlying file system is a legacy
3928 file system, this automatically degrades to creating a
3929 normal directory. Among others /var/lib/machines is now
3930 created like this at boot, should it be missing.
3932 * The directory /var/lib/containers/ has been deprecated and
3933 been replaced by /var/lib/machines. The term "machines" has
3934 been used in the systemd context as generic term for both
3935 VMs and containers, and hence appears more appropriate for
3936 this, as the directory can also contain raw images bootable
3939 * systemd-nspawn when invoked with -M but without --directory=
3940 or --image= is now capable of searching for the container
3941 root directory, subvolume or disk image automatically, in
3942 /var/lib/machines. systemd-nspawn@.service has been updated
3943 to make use of this, thus allowing it to be used for raw
3946 * A new machines.target unit has been introduced that is
3947 supposed to group all containers/VMs invoked as services on
3948 the system. systemd-nspawn@.service has been updated to
3949 integrate with that.
3951 * machinectl gained a new "start" command, for invoking a
3952 container as a service. "machinectl start foo" is mostly
3953 equivalent to "systemctl start systemd-nspawn@foo.service",
3954 but handles escaping in a nicer way.
3956 * systemd-nspawn will now mount most of the cgroupfs tree
3957 read-only into each container, with the exception of the
3958 container's own subtree in the name=systemd hierarchy.
3960 * journald now sets the special FS_NOCOW file flag for its
3961 journal files. This should improve performance on btrfs, by
3962 avoiding heavy fragmentation when journald's write-pattern
3963 is used on COW file systems. It degrades btrfs' data
3964 integrity guarantees for the files to the same levels as for
3965 ext3/ext4 however. This should be OK though as journald does
3966 its own data integrity checks and all its objects are
3967 checksummed on disk. Also, journald should handle btrfs disk
3968 full events a lot more gracefully now, by processing SIGBUS
3969 errors, and not relying on fallocate() anymore.
3971 * When journald detects that journal files it is writing to
3972 have been deleted it will immediately start new journal
3975 * systemd now provides a way to store file descriptors
3976 per-service in PID 1. This is useful for daemons to ensure
3977 that fds they require are not lost during a daemon
3978 restart. The fds are passed to the daemon on the next
3979 invocation in the same way socket activation fds are
3980 passed. This is now used by journald to ensure that the
3981 various sockets connected to all the system's stdout/stderr
3982 are not lost when journald is restarted. File descriptors
3983 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
3984 an extension to sd_notify(). Note that a limit is enforced
3985 on the number of fds a service can store in PID 1, and it
3986 defaults to 0, so that no fds may be stored, unless this is
3987 explicitly turned on.
3989 * The default TERM variable to use for units connected to a
3990 terminal, when no other value is explicitly is set is now
3991 vt220 rather than vt102. This should be fairly safe still,
3992 but allows PgUp/PgDn work.
3994 * The /etc/crypttab option header= as known from Debian is now
3997 * "loginctl user-status" and "loginctl session-status" will
3998 now show the last 10 lines of log messages of the
3999 user/session following the status output. Similar,
4000 "machinectl status" will show the last 10 log lines
4001 associated with a virtual machine or container
4002 service. (Note that this is usually not the log messages
4003 done in the VM/container itself, but simply what the
4004 container manager logs. For nspawn this includes all console
4007 * "loginctl session-status" without further argument will now
4008 show the status of the session of the caller. Similar,
4009 "lock-session", "unlock-session", "activate",
4010 "enable-linger", "disable-linger" may now be called without
4011 session/user parameter in which case they apply to the
4012 caller's session/user.
4014 * An X11 session scriptlet is now shipped that uploads
4015 $DISPLAY and $XAUTHORITY into the environment of the systemd
4016 --user daemon if a session begins. This should improve
4017 compatibility with X11 enabled applications run as systemd
4020 * Generators are now subject to masking via /etc and /run, the
4021 same way as unit files.
4023 * networkd .network files gained support for configuring
4024 per-link IPv4/IPv6 packet forwarding as well as IPv4
4025 masquerading. This is by default turned on for veth links to
4026 containers, as registered by systemd-nspawn. This means that
4027 nspawn containers run with --network-veth will now get
4028 automatic routed access to the host's networks without any
4029 further configuration or setup, as long as networkd runs on
4032 * systemd-nspawn gained the --port= (-p) switch to expose TCP
4033 or UDP posts of a container on the host. With this in place
4034 it is possible to run containers with private veth links
4035 (--network-veth), and have their functionality exposed on
4036 the host as if their services were running directly on the
4039 * systemd-nspawn's --network-veth switch now gained a short
4040 version "-n", since with the changes above it is now truly
4041 useful out-of-the-box. The systemd-nspawn@.service has been
4042 updated to make use of it too by default.
4044 * systemd-nspawn will now maintain a per-image R/W lock, to
4045 ensure that the same image is not started more than once
4046 writable. (It's OK to run an image multiple times
4047 simultaneously in read-only mode.)
4049 * systemd-nspawn's --image= option is now capable of
4050 dissecting and booting MBR and GPT disk images that contain
4051 only a single active Linux partition. Previously it
4052 supported only GPT disk images with proper GPT type
4053 IDs. This allows running cloud images from major
4054 distributions directly with systemd-nspawn, without
4057 * In addition to collecting mouse dpi data in the udev
4058 hardware database, there's now support for collecting angle
4059 information for mouse scroll wheels. The database is
4060 supposed to guarantee similar scrolling behavior on mice
4061 that it knows about. There's also support for collecting
4062 information about Touchpad types.
4064 * udev's input_id built-in will now also collect touch screen
4065 dimension data and attach it to probed devices.
4067 * /etc/os-release gained support for a Distribution Privacy
4070 * networkd gained support for creating "ipvlan", "gretap",
4071 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
4073 * systemd-tmpfiles gained support for "a" lines for setting
4076 * systemd-nspawn will now mount /tmp in the container to
4077 tmpfs, automatically.
4079 * systemd now exposes the memory.usage_in_bytes cgroup
4080 attribute and shows it for each service in the "systemctl
4081 status" output, if available.
4083 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
4084 immediate reboot is triggered. This useful if shutdown is
4085 hung and is unable to complete, to expedite the
4086 operation. Note that this kind of reboot will still unmount
4087 all file systems, and hence should not result in fsck being
4090 * A .device unit for an optical block device will now be
4091 considered active only when a medium is in the drive. Also,
4092 mount units are now bound to their backing devices thus
4093 triggering automatic unmounting when devices become
4094 unavailable. With this in place systemd will now
4095 automatically unmount left-over mounts when a CD-ROM is
4096 ejected or an USB stick is yanked from the system.
4098 * networkd-wait-online now has support for waiting for
4099 specific interfaces only (with globbing), and for giving up
4100 after a configurable timeout.
4102 * networkd now exits when idle. It will be automatically
4103 restarted as soon as interfaces show up, are removed or
4104 change state. networkd will stay around as long as there is
4105 at least one DHCP state machine or similar around, that keep
4108 * networkd may now configure IPv6 link-local addressing in
4109 addition to IPv4 link-local addressing.
4111 * The IPv6 "token" for use in SLAAC may now be configured for
4112 each .network interface in networkd.
4114 * Routes configured with networkd may now be assigned a scope
4117 * networkd's [Match] sections now support globbing and lists
4118 of multiple space-separated matches per item.
4120 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
4121 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
4122 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
4123 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
4124 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
4125 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
4126 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
4127 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
4128 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
4129 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
4130 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
4131 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
4132 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
4133 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
4134 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
4135 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
4136 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
4137 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
4138 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
4139 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
4140 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
4141 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
4142 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
4143 Hoffmann, Zbigniew Jędrzejewski-Szmek
4145 — Berlin, 2015-02-16
4149 * When querying unit file enablement status (for example via
4150 "systemctl is-enabled"), a new state "indirect" is now known
4151 which indicates that a unit might not be enabled itself, but
4152 another unit listed in its Also= setting might be.
4154 * Similar to the various existing ConditionXYZ= settings for
4155 units, there are now matching AssertXYZ= settings. While
4156 failing conditions cause a unit to be skipped, but its job
4157 to succeed, failing assertions declared like this will cause
4158 a unit start operation and its job to fail.
4160 * hostnamed now knows a new chassis type "embedded".
4162 * systemctl gained a new "edit" command. When used on a unit
4163 file, this allows extending unit files with .d/ drop-in
4164 configuration snippets or editing the full file (after
4165 copying it from /usr/lib to /etc). This will invoke the
4166 user's editor (as configured with $EDITOR), and reload the
4167 modified configuration after editing.
4169 * "systemctl status" now shows the suggested enablement state
4170 for a unit, as declared in the (usually vendor-supplied)
4171 system preset files.
4173 * nss-myhostname will now resolve the single-label host name
4174 "gateway" to the locally configured default IP routing
4175 gateways, ordered by their metrics. This assigns a stable
4176 name to the used gateways, regardless which ones are
4177 currently configured. Note that the name will only be
4178 resolved after all other name sources (if nss-myhostname is
4179 configured properly) and should hence not negatively impact
4180 systems that use the single-label host name "gateway" in
4183 * systemd-inhibit now allows filtering by mode when listing
4186 * Scope and service units gained a new "Delegate" boolean
4187 property, which, when set, allows processes running inside the
4188 unit to further partition resources. This is primarily
4189 useful for systemd user instances as well as container
4192 * journald will now pick up audit messages directly from
4193 the kernel, and log them like any other log message. The
4194 audit fields are split up and fully indexed. This means that
4195 journalctl in many ways is now a (nicer!) alternative to
4196 ausearch, the traditional audit client. Note that this
4197 implements only a minimal audit client. If you want the
4198 special audit modes like reboot-on-log-overflow, please use
4199 the traditional auditd instead, which can be used in
4200 parallel to journald.
4202 * The ConditionSecurity= unit file option now understands the
4203 special string "audit" to check whether auditing is
4206 * journalctl gained two new commands --vacuum-size= and
4207 --vacuum-time= to delete old journal files until the
4208 remaining ones take up no more than the specified size on disk,
4209 or are not older than the specified time.
4211 * A new, native PPPoE library has been added to sd-network,
4212 systemd's library of light-weight networking protocols. This
4213 library will be used in a future version of networkd to
4214 enable PPPoE communication without an external pppd daemon.
4216 * The busctl tool now understands a new "capture" verb that
4217 works similar to "monitor", but writes a packet capture
4218 trace to STDOUT that can be redirected to a file which is
4219 compatible with libcap's capture file format. This can then
4220 be loaded in Wireshark and similar tools to inspect bus
4223 * The busctl tool now understands a new "tree" verb that shows
4224 the object trees of a specific service on the bus, or of all
4227 * The busctl tool now understands a new "introspect" verb that
4228 shows all interfaces and members of objects on the bus,
4229 including their signature and values. This is particularly
4230 useful to get more information about bus objects shown by
4231 the new "busctl tree" command.
4233 * The busctl tool now understands new verbs "call",
4234 "set-property" and "get-property" for invoking bus method
4235 calls, setting and getting bus object properties in a
4238 * busctl gained a new --augment-creds= argument that controls
4239 whether the tool shall augment credential information it
4240 gets from the bus with data from /proc, in a possibly
4243 * nspawn's --link-journal= switch gained two new values
4244 "try-guest" and "try-host" that work like "guest" and
4245 "host", but do not fail if the host has no persistent
4246 journalling enabled. -j is now equivalent to
4247 --link-journal=try-guest.
4249 * macvlan network devices created by nspawn will now have
4250 stable MAC addresses.
4252 * A new SmackProcessLabel= unit setting has been added, which
4253 controls the SMACK security label processes forked off by
4254 the respective unit shall use.
4256 * If compiled with --enable-xkbcommon, systemd-localed will
4257 verify x11 keymap settings by compiling the given keymap. It
4258 will spew out warnings if the compilation fails. This
4259 requires libxkbcommon to be installed.
4261 * When a coredump is collected, a larger number of metadata
4262 fields is now collected and included in the journal records
4263 created for it. More specifically, control group membership,
4264 environment variables, memory maps, working directory,
4265 chroot directory, /proc/$PID/status, and a list of open file
4266 descriptors is now stored in the log entry.
4268 * The udev hwdb now contains DPI information for mice. For
4271 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
4273 * All systemd programs that read standalone configuration
4274 files in /etc now also support a corresponding series of
4275 .conf.d configuration directories in /etc/, /run/,
4276 /usr/local/lib/, /usr/lib/, and (if configured with
4277 --enable-split-usr) /lib/. In particular, the following
4278 configuration files now have corresponding configuration
4279 directories: system.conf user.conf, logind.conf,
4280 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
4281 resolved.conf, timesyncd.conf, journal-remote.conf, and
4282 journal-upload.conf. Note that distributions should use the
4283 configuration directories in /usr/lib/; the directories in
4284 /etc/ are reserved for the system administrator.
4286 * systemd-rfkill will no longer take the rfkill device name
4287 into account when storing rfkill state on disk, as the name
4288 might be dynamically assigned and not stable. Instead, the
4289 ID_PATH udev variable combined with the rfkill type (wlan,
4290 bluetooth, ...) is used.
4292 * A new service systemd-machine-id-commit.service has been
4293 added. When used on systems where /etc is read-only during
4294 boot, and /etc/machine-id is not initialized (but an empty
4295 file), this service will copy the temporary machine ID
4296 created as replacement into /etc after the system is fully
4297 booted up. This is useful for systems that are freshly
4298 installed with a non-initialized machine ID, but should get
4299 a fixed machine ID for subsequent boots.
4301 * networkd's .netdev files now provide a large set of
4302 configuration parameters for VXLAN devices. Similarly, the
4303 bridge port cost parameter is now configurable in .network
4304 files. There's also new support for configuring IP source
4305 routing. networkd .link files gained support for a new
4306 OriginalName= match that is useful to match against the
4307 original interface name the kernel assigned. .network files
4308 may include MTU= and MACAddress= fields for altering the MTU
4309 and MAC address while being connected to a specific network
4312 * The LUKS logic gained supported for configuring
4313 UUID-specific key files. There's also new support for naming
4314 LUKS device from the kernel command line, using the new
4315 luks.name= argument.
4317 * Timer units may now be transiently created via the bus API
4318 (this was previously already available for scope and service
4319 units). In addition it is now possible to create multiple
4320 transient units at the same time with a single bus call. The
4321 "systemd-run" tool has been updated to make use of this for
4322 running commands on a specified time, in at(1)-style.
4324 * tmpfiles gained support for "t" lines, for assigning
4325 extended attributes to files. Among other uses this may be
4326 used to assign SMACK labels to files.
4328 Contributions from: Alin Rauta, Alison Chaiken, Andrej
4329 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
4330 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
4331 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
4332 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
4333 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
4334 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
4335 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
4336 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
4337 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
4338 Michael Biebl, Michael Chapman, Michael Marineau, Michal
4339 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
4340 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
4341 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
4342 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
4343 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4344 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
4345 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
4347 — Berlin, 2014-12-10
4351 * journalctl gained the new options -t/--identifier= to match
4352 on the syslog identifier (aka "tag"), as well as --utc to
4353 show log timestamps in the UTC timezone. journalctl now also
4354 accepts -n/--lines=all to disable line capping in a pager.
4356 * journalctl gained a new switch, --flush, that synchronously
4357 flushes logs from /run/log/journal to /var/log/journal if
4358 persistent storage is enabled. systemd-journal-flush.service
4359 now waits until the operation is complete.
4361 * Services can notify the manager before they start a reload
4362 (by sending RELOADING=1) or shutdown (by sending
4363 STOPPING=1). This allows the manager to track and show the
4364 internal state of daemons and closes a race condition when
4365 the process is still running but has closed its D-Bus
4368 * Services with Type=oneshot do not have to have any ExecStart
4371 * User units are now loaded also from
4372 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
4373 /run/systemd/user directory that was already previously
4374 supported, but is under the control of the user.
4376 * Job timeouts (i.e. time-outs on the time a job that is
4377 queued stays in the run queue) can now optionally result in
4378 immediate reboot or power-off actions (JobTimeoutAction= and
4379 JobTimeoutRebootArgument=). This is useful on ".target"
4380 units, to limit the maximum time a target remains
4381 undispatched in the run queue, and to trigger an emergency
4382 operation in such a case. This is now used by default to
4383 turn off the system if boot-up (as defined by everything in
4384 basic.target) hangs and does not complete for at least
4385 15min. Also, if power-off or reboot hang for at least 30min
4386 an immediate power-off/reboot operation is triggered. This
4387 functionality is particularly useful to increase reliability
4388 on embedded devices, but also on laptops which might
4389 accidentally get powered on when carried in a backpack and
4390 whose boot stays stuck in a hard disk encryption passphrase
4393 * systemd-logind can be configured to also handle lid switch
4394 events even when the machine is docked or multiple displays
4395 are attached (HandleLidSwitchDocked= option).
4397 * A helper binary and a service have been added which can be
4398 used to resume from hibernation in the initramfs. A
4399 generator will parse the resume= option on the kernel
4400 command line to trigger resume.
4402 * A user console daemon systemd-consoled has been
4403 added. Currently, it is a preview, and will so far open a
4404 single terminal on each session of the user marked as
4405 Desktop=systemd-console.
4407 * Route metrics can be specified for DHCP routes added by
4410 * The SELinux context of socket-activated services can be set
4411 from the information provided by the networking stack
4412 (SELinuxContextFromNet= option).
4414 * Userspace firmware loading support has been removed and
4415 the minimum supported kernel version is thus bumped to 3.7.
4417 * Timeout for udev workers has been increased from 1 to 3
4418 minutes, but a warning will be printed after 1 minute to
4419 help diagnose kernel modules that take a long time to load.
4421 * Udev rules can now remove tags on devices with TAG-="foobar".
4423 * systemd's readahead implementation has been removed. In many
4424 circumstances it didn't give expected benefits even for
4425 rotational disk drives and was becoming less relevant in the
4426 age of SSDs. As none of the developers has been using
4427 rotating media anymore, and nobody stepped up to actively
4428 maintain this component of systemd it has now been removed.
4430 * Swap units can use Options= to specify discard options.
4431 Discard options specified for swaps in /etc/fstab are now
4434 * Docker containers are now detected as a separate type of
4437 * The Password Agent protocol gained support for queries where
4438 the user input is shown, useful e.g. for user names.
4439 systemd-ask-password gained a new --echo option to turn that
4442 * The default sysctl.d/ snippets will now set:
4444 net.core.default_qdisc = fq_codel
4446 This selects Fair Queuing Controlled Delay as the default
4447 queuing discipline for network interfaces. fq_codel helps
4448 fight the network bufferbloat problem. It is believed to be
4449 a good default with no tuning required for most workloads.
4450 Downstream distributions may override this choice. On 10Gbit
4451 servers that do not do forwarding, "fq" may perform better.
4452 Systems without a good clocksource should use "pfifo_fast".
4454 * If kdbus is enabled during build a new option BusPolicy= is
4455 available for service units, that allows locking all service
4456 processes into a stricter bus policy, in order to limit
4457 access to various bus services, or even hide most of them
4458 from the service's view entirely.
4460 * networkctl will now show the .network and .link file
4461 networkd has applied to a specific interface.
4463 * sd-login gained a new API call sd_session_get_desktop() to
4464 query which desktop environment has been selected for a
4467 * UNIX utmp support is now compile-time optional to support
4468 legacy-free systems.
4470 * systemctl gained two new commands "add-wants" and
4471 "add-requires" for pulling in units from specific targets
4474 * If the word "rescue" is specified on the kernel command line
4475 the system will now boot into rescue mode (aka
4476 rescue.target), which was previously available only by
4477 specifying "1" or "systemd.unit=rescue.target" on the kernel
4478 command line. This new kernel command line option nicely
4479 mirrors the already existing "emergency" kernel command line
4482 * New kernel command line options mount.usr=, mount.usrflags=,
4483 mount.usrfstype= have been added that match root=, rootflags=,
4484 rootfstype= but allow mounting a specific file system to
4487 * The $NOTIFY_SOCKET is now also passed to control processes of
4488 services, not only the main process.
4490 * This version reenables support for fsck's -l switch. This
4491 means at least version v2.25 of util-linux is required for
4492 operation, otherwise dead-locks on device nodes may
4493 occur. Again: you need to update util-linux to at least
4494 v2.25 when updating systemd to v217.
4496 * The "multi-seat-x" tool has been removed from systemd, as
4497 its functionality has been integrated into X servers 1.16,
4498 and the tool is hence redundant. It is recommended to update
4499 display managers invoking this tool to simply invoke X
4500 directly from now on, again.
4502 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
4503 message flag has been added for all of systemd's PolicyKit
4504 authenticated method calls has been added. In particular
4505 this now allows optional interactive authorization via
4506 PolicyKit for many of PID1's privileged operations such as
4507 unit file enabling and disabling.
4509 * "udevadm hwdb --update" learnt a new switch "--usr" for
4510 placing the rebuilt hardware database in /usr instead of
4511 /etc. When used only hardware database entries stored in
4512 /usr will be used, and any user database entries in /etc are
4513 ignored. This functionality is useful for vendors to ship a
4514 pre-built database on systems where local configuration is
4515 unnecessary or unlikely.
4517 * Calendar time specifications in .timer units now also
4518 understand the strings "semi-annually", "quarterly" and
4519 "minutely" as shortcuts (in addition to the preexisting
4520 "anually", "hourly", ...).
4522 * systemd-tmpfiles will now correctly create files in /dev
4523 at boot which are marked for creation only at boot. It is
4524 recommended to always create static device nodes with 'c!'
4525 and 'b!', so that they are created only at boot and not
4526 overwritten at runtime.
4528 * When the watchdog logic is used for a service (WatchdogSec=)
4529 and the watchdog timeout is hit the service will now be
4530 terminated with SIGABRT (instead of just SIGTERM), in order
4531 to make sure a proper coredump and backtrace is
4532 generated. This ensures that hanging services will result in
4533 similar coredump/backtrace behaviour as services that hit a
4536 Contributions from: Andreas Henriksson, Andrei Borzenkov,
4537 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
4538 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
4539 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
4540 Herrmann, David Sommerseth, David Strauss, Emil Renner
4541 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
4542 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
4543 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
4544 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
4545 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
4546 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
4547 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
4548 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
4549 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
4550 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
4551 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
4552 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
4553 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
4554 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
4555 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
4556 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
4559 — Berlin, 2014-10-28
4563 * timedated no longer reads NTP implementation unit names from
4564 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
4565 implementations should add a
4567 Conflicts=systemd-timesyncd.service
4569 to their unit files to take over and replace systemd's NTP
4570 default functionality.
4572 * systemd-sysusers gained a new line type "r" for configuring
4573 which UID/GID ranges to allocate system users/groups
4574 from. Lines of type "u" may now add an additional column
4575 that specifies the home directory for the system user to be
4576 created. Also, systemd-sysusers may now optionally read user
4577 information from STDIN instead of a file. This is useful for
4578 invoking it from RPM preinst scriptlets that need to create
4579 users before the first RPM file is installed since these
4580 files might need to be owned by them. A new
4581 %sysusers_create_inline RPM macro has been introduced to do
4582 just that. systemd-sysusers now updates the shadow files as
4583 well as the user/group databases, which should enhance
4584 compatibility with certain tools like grpck.
4586 * A number of bus APIs of PID 1 now optionally consult
4587 PolicyKit to permit access for otherwise unprivileged
4588 clients under certain conditions. Note that this currently
4589 doesn't support interactive authentication yet, but this is
4590 expected to be added eventually, too.
4592 * /etc/machine-info now has new fields for configuring the
4593 deployment environment of the machine, as well as the
4594 location of the machine. hostnamectl has been updated with
4595 new command to update these fields.
4597 * systemd-timesyncd has been updated to automatically acquire
4598 NTP server information from systemd-networkd, which might
4599 have been discovered via DHCP.
4601 * systemd-resolved now includes a caching DNS stub resolver
4602 and a complete LLMNR name resolution implementation. A new
4603 NSS module "nss-resolve" has been added which can be used
4604 instead of glibc's own "nss-dns" to resolve hostnames via
4605 systemd-resolved. Hostnames, addresses and arbitrary RRs may
4606 be resolved via systemd-resolved D-Bus APIs. In contrast to
4607 the glibc internal resolver systemd-resolved is aware of
4608 multi-homed system, and keeps DNS server and caches separate
4609 and per-interface. Queries are sent simultaneously on all
4610 interfaces that have DNS servers configured, in order to
4611 properly handle VPNs and local LANs which might resolve
4612 separate sets of domain names. systemd-resolved may acquire
4613 DNS server information from systemd-networkd automatically,
4614 which in turn might have discovered them via DHCP. A tool
4615 "systemd-resolve-host" has been added that may be used to
4616 query the DNS logic in resolved. systemd-resolved implements
4617 IDNA and automatically uses IDNA or UTF-8 encoding depending
4618 on whether classic DNS or LLMNR is used as transport. In the
4619 next releases we intend to add a DNSSEC and mDNS/DNS-SD
4620 implementation to systemd-resolved.
4622 * A new NSS module nss-mymachines has been added, that
4623 automatically resolves the names of all local registered
4624 containers to their respective IP addresses.
4626 * A new client tool "networkctl" for systemd-networkd has been
4627 added. It currently is entirely passive and will query
4628 networking configuration from udev, rtnetlink and networkd,
4629 and present it to the user in a very friendly
4630 way. Eventually, we hope to extend it to become a full
4631 control utility for networkd.
4633 * .socket units gained a new DeferAcceptSec= setting that
4634 controls the kernels' TCP_DEFER_ACCEPT sockopt for
4635 TCP. Similarly, support for controlling TCP keep-alive
4636 settings has been added (KeepAliveTimeSec=,
4637 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
4638 turning off Nagle's algorithm on TCP has been added
4641 * logind learned a new session type "web", for use in projects
4642 like Cockpit which register web clients as PAM sessions.
4644 * timer units with at least one OnCalendar= setting will now
4645 be started only after time-sync.target has been
4646 reached. This way they will not elapse before the system
4647 clock has been corrected by a local NTP client or
4648 similar. This is particular useful on RTC-less embedded
4649 machines, that come up with an invalid system clock.
4651 * systemd-nspawn's --network-veth= switch should now result in
4652 stable MAC addresses for both the outer and the inner side
4655 * systemd-nspawn gained a new --volatile= switch for running
4656 container instances with /etc or /var unpopulated.
4658 * The kdbus client code has been updated to use the new Linux
4659 3.17 memfd subsystem instead of the old kdbus-specific one.
4661 * systemd-networkd's DHCP client and server now support
4662 FORCERENEW. There are also new configuration options to
4663 configure the vendor client identifier and broadcast mode
4666 * systemd will no longer inform the kernel about the current
4667 timezone, as this is necessarily incorrect and racy as the
4668 kernel has no understanding of DST and similar
4669 concepts. This hence means FAT timestamps will be always
4670 considered UTC, similar to what Android is already
4671 doing. Also, when the RTC is configured to the local time
4672 (rather than UTC) systemd will never synchronize back to it,
4673 as this might confuse Windows at a later boot.
4675 * systemd-analyze gained a new command "verify" for offline
4676 validation of unit files.
4678 * systemd-networkd gained support for a couple of additional
4679 settings for bonding networking setups. Also, the metric for
4680 statically configured routes may now be configured. For
4681 network interfaces where this is appropriate the peer IP
4682 address may now be configured.
4684 * systemd-networkd's DHCP client will no longer request
4685 broadcasting by default, as this tripped up some networks.
4686 For hardware where broadcast is required the feature should
4687 be switched back on using RequestBroadcast=yes.
4689 * systemd-networkd will now set up IPv4LL addresses (when
4690 enabled) even if DHCP is configured successfully.
4692 * udev will now default to respect network device names given
4693 by the kernel when the kernel indicates that these are
4694 predictable. This behavior can be tweaked by changing
4695 NamePolicy= in the relevant .link file.
4697 * A new library systemd-terminal has been added that
4698 implements full TTY stream parsing and rendering. This
4699 library is supposed to be used later on for implementing a
4700 full userspace VT subsystem, replacing the current kernel
4703 * A new tool systemd-journal-upload has been added to push
4704 journal data to a remote system running
4705 systemd-journal-remote.
4707 * journald will no longer forward all local data to another
4708 running syslog daemon. This change has been made because
4709 rsyslog (which appears to be the most commonly used syslog
4710 implementation these days) no longer makes use of this, and
4711 instead pulls the data out of the journal on its own. Since
4712 forwarding the messages to a non-existent syslog server is
4713 more expensive than we assumed we have now turned this
4714 off. If you run a syslog server that is not a recent rsyslog
4715 version, you have to turn this option on again
4716 (ForwardToSyslog= in journald.conf).
4718 * journald now optionally supports the LZ4 compressor for
4719 larger journal fields. This compressor should perform much
4720 better than XZ which was the previous default.
4722 * machinectl now shows the IP addresses of local containers,
4723 if it knows them, plus the interface name of the container.
4725 * A new tool "systemd-escape" has been added that makes it
4726 easy to escape strings to build unit names and similar.
4728 * sd_notify() messages may now include a new ERRNO= field
4729 which is parsed and collected by systemd and shown among the
4730 "systemctl status" output for a service.
4732 * A new component "systemd-firstboot" has been added that
4733 queries the most basic systemd information (timezone,
4734 hostname, root password) interactively on first
4735 boot. Alternatively it may also be used to provision these
4736 things offline on OS images installed into directories.
4738 * The default sysctl.d/ snippets will now set
4740 net.ipv4.conf.default.promote_secondaries=1
4742 This has the benefit of no flushing secondary IP addresses
4743 when primary addresses are removed.
4745 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
4746 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
4747 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
4748 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
4749 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
4750 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
4751 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
4752 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
4753 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
4754 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
4755 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
4756 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
4757 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
4758 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
4759 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
4761 — Berlin, 2014-08-19
4765 * A new tool systemd-sysusers has been added. This tool
4766 creates system users and groups in /etc/passwd and
4767 /etc/group, based on static declarative system user/group
4768 definitions in /usr/lib/sysusers.d/. This is useful to
4769 enable factory resets and volatile systems that boot up with
4770 an empty /etc directory, and thus need system users and
4771 groups created during early boot. systemd now also ships
4772 with two default sysusers.d/ files for the most basic
4773 users and groups systemd and the core operating system
4776 * A new tmpfiles snippet has been added that rebuilds the
4777 essential files in /etc on boot, should they be missing.
4779 * A directive for ensuring automatic clean-up of
4780 /var/cache/man/ has been removed from the default
4781 configuration. This line should now be shipped by the man
4782 implementation. The necessary change has been made to the
4783 man-db implementation. Note that you need to update your man
4784 implementation to one that ships this line, otherwise no
4785 automatic clean-up of /var/cache/man will take place.
4787 * A new condition ConditionNeedsUpdate= has been added that
4788 may conditionalize services to only run when /etc or /var
4789 are "older" than the vendor operating system resources in
4790 /usr. This is useful for reconstructing or updating /etc
4791 after an offline update of /usr or a factory reset, on the
4792 next reboot. Services that want to run once after such an
4793 update or reset should use this condition and order
4794 themselves before the new systemd-update-done.service, which
4795 will mark the two directories as fully updated. A number of
4796 service files have been added making use of this, to rebuild
4797 the udev hardware database, the journald message catalog and
4798 dynamic loader cache (ldconfig). The systemd-sysusers tool
4799 described above also makes use of this now. With this in
4800 place it is now possible to start up a minimal operating
4801 system with /etc empty cleanly. For more information on the
4802 concepts involved see this recent blog story:
4804 http://0pointer.de/blog/projects/stateless.html
4806 * A new system group "input" has been introduced, and all
4807 input device nodes get this group assigned. This is useful
4808 for system-level software to get access to input devices. It
4809 complements what is already done for "audio" and "video".
4811 * systemd-networkd learnt minimal DHCPv4 server support in
4812 addition to the existing DHCPv4 client support. It also
4813 learnt DHCPv6 client and IPv6 Router Solicitation client
4814 support. The DHCPv4 client gained support for static routes
4815 passed in from the server. Note that the [DHCPv4] section
4816 known in older systemd-networkd versions has been renamed to
4817 [DHCP] and is now also used by the DHCPv6 client. Existing
4818 .network files using settings of this section should be
4819 updated, though compatibility is maintained. Optionally, the
4820 client hostname may now be sent to the DHCP server.
4822 * networkd gained support for vxlan virtual networks as well
4823 as tun/tap and dummy devices.
4825 * networkd gained support for automatic allocation of address
4826 ranges for interfaces from a system-wide pool of
4827 addresses. This is useful for dynamically managing a large
4828 number of interfaces with a single network configuration
4829 file. In particular this is useful to easily assign
4830 appropriate IP addresses to the veth links of a large number
4831 of nspawn instances.
4833 * RPM macros for processing sysusers, sysctl and binfmt
4834 drop-in snippets at package installation time have been
4837 * The /etc/os-release file should now be placed in
4838 /usr/lib/os-release. The old location is automatically
4839 created as symlink. /usr/lib is the more appropriate
4840 location of this file, since it shall actually describe the
4841 vendor operating system shipped in /usr, and not the
4842 configuration stored in /etc.
4844 * .mount units gained a new boolean SloppyOptions= setting
4845 that maps to mount(8)'s -s option which enables permissive
4846 parsing of unknown mount options.
4848 * tmpfiles learnt a new "L+" directive which creates a symlink
4849 but (unlike "L") deletes a pre-existing file first, should
4850 it already exist and not already be the correct
4851 symlink. Similarly, "b+", "c+" and "p+" directives have been
4852 added as well, which create block and character devices, as
4853 well as fifos in the filesystem, possibly removing any
4854 pre-existing files of different types.
4856 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
4857 'argument' field (which so far specified the source to
4858 symlink/copy the files from) is now optional. If omitted the
4859 same file os copied from /usr/share/factory/ suffixed by the
4860 full destination path. This is useful for populating /etc
4861 with essential files, by copying them from vendor defaults
4862 shipped in /usr/share/factory/etc.
4864 * A new command "systemctl preset-all" has been added that
4865 applies the service preset settings to all installed unit
4866 files. A new switch --preset-mode= has been added that
4867 controls whether only enable or only disable operations
4870 * A new command "systemctl is-system-running" has been added
4871 that allows checking the overall state of the system, for
4872 example whether it is fully up and running.
4874 * When the system boots up with an empty /etc, the equivalent
4875 to "systemctl preset-all" is executed during early boot, to
4876 make sure all default services are enabled after a factory
4879 * systemd now contains a minimal preset file that enables the
4880 most basic services systemd ships by default.
4882 * Unit files' [Install] section gained a new DefaultInstance=
4883 field for defining the default instance to create if a
4884 template unit is enabled with no instance specified.
4886 * A new passive target cryptsetup-pre.target has been added
4887 that may be used by services that need to make they run and
4888 finish before the first LUKS cryptographic device is set up.
4890 * The /dev/loop-control and /dev/btrfs-control device nodes
4891 are now owned by the "disk" group by default, opening up
4892 access to this group.
4894 * systemd-coredump will now automatically generate a
4895 stack trace of all core dumps taking place on the system,
4896 based on elfutils' libdw library. This stack trace is logged
4899 * systemd-coredump may now optionally store coredumps directly
4900 on disk (in /var/lib/systemd/coredump, possibly compressed),
4901 instead of storing them unconditionally in the journal. This
4902 mode is the new default. A new configuration file
4903 /etc/systemd/coredump.conf has been added to configure this
4904 and other parameters of systemd-coredump.
4906 * coredumpctl gained a new "info" verb to show details about a
4907 specific coredump. A new switch "-1" has also been added
4908 that makes sure to only show information about the most
4909 recent entry instead of all entries. Also, as the tool is
4910 generally useful now the "systemd-" prefix of the binary
4911 name has been removed. Distributions that want to maintain
4912 compatibility with the old name should add a symlink from
4913 the old name to the new name.
4915 * journald's SplitMode= now defaults to "uid". This makes sure
4916 that unprivileged users can access their own coredumps with
4917 coredumpctl without restrictions.
4919 * New kernel command line options "systemd.wants=" (for
4920 pulling an additional unit during boot), "systemd.mask="
4921 (for masking a specific unit for the boot), and
4922 "systemd.debug-shell" (for enabling the debug shell on tty9)
4923 have been added. This is implemented in the new generator
4924 "systemd-debug-generator".
4926 * systemd-nspawn will now by default filter a couple of
4927 syscalls for containers, among them those required for
4928 kernel module loading, direct x86 IO port access, swap
4929 management, and kexec. Most importantly though
4930 open_by_handle_at() is now prohibited for containers,
4931 closing a hole similar to a recently discussed vulnerability
4932 in docker regarding access to files on file hierarchies the
4933 container should normally not have access to. Note that, for
4934 nspawn, we generally make no security claims anyway (and
4935 this is explicitly documented in the man page), so this is
4936 just a fix for one of the most obvious problems.
4938 * A new man page file-hierarchy(7) has been added that
4939 contains a minimized, modernized version of the file system
4940 layout systemd expects, similar in style to the FHS
4941 specification or hier(5). A new tool systemd-path(1) has
4942 been added to query many of these paths for the local
4945 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
4946 longer done. Since the directory now has a per-user size
4947 limit, and is cleaned on logout this appears unnecessary,
4948 in particular since this now brings the lifecycle of this
4949 directory closer in line with how IPC objects are handled.
4951 * systemd.pc now exports a number of additional directories,
4952 including $libdir (which is useful to identify the library
4953 path for the primary architecture of the system), and a
4954 couple of drop-in directories.
4956 * udev's predictable network interface names now use the dev_port
4957 sysfs attribute, introduced in linux 3.15 instead of dev_id to
4958 distinguish between ports of the same PCI function. dev_id should
4959 only be used for ports using the same HW address, hence the need
4962 * machined has been updated to export the OS version of a
4963 container (read from /etc/os-release and
4964 /usr/lib/os-release) on the bus. This is now shown in
4965 "machinectl status" for a machine.
4967 * A new service setting RestartForceExitStatus= has been
4968 added. If configured to a set of exit signals or process
4969 return values, the service will be restarted when the main
4970 daemon process exits with any of them, regardless of the
4973 * systemctl's -H switch for connecting to remote systemd
4974 machines has been extended so that it may be used to
4975 directly connect to a specific container on the
4976 host. "systemctl -H root@foobar:waldi" will now connect as
4977 user "root" to host "foobar", and then proceed directly to
4978 the container named "waldi". Note that currently you have to
4979 authenticate as user "root" for this to work, as entering
4980 containers is a privileged operation.
4982 Contributions from: Andreas Henriksson, Benjamin Steinwender,
4983 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
4984 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
4985 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
4986 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
4987 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
4988 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
4989 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
4990 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
4991 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
4992 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
4993 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
4995 — Berlin, 2014-07-03
4999 * As an experimental feature, udev now tries to lock the
5000 disk device node (flock(LOCK_SH|LOCK_NB)) while it
5001 executes events for the disk or any of its partitions.
5002 Applications like partitioning programs can lock the
5003 disk device node (flock(LOCK_EX)) and claim temporary
5004 device ownership that way; udev will entirely skip all event
5005 handling for this disk and its partitions. If the disk
5006 was opened for writing, the close will trigger a partition
5007 table rescan in udev's "watch" facility, and if needed
5008 synthesize "change" events for the disk and all its partitions.
5009 This is now unconditionally enabled, and if it turns out to
5010 cause major problems, we might turn it on only for specific
5011 devices, or might need to disable it entirely. Device Mapper
5012 devices are excluded from this logic.
5014 * We temporarily dropped the "-l" switch for fsck invocations,
5015 since they collide with the flock() logic above. util-linux
5016 upstream has been changed already to avoid this conflict,
5017 and we will readd "-l" as soon as util-linux with this
5018 change has been released.
5020 * The dependency on libattr has been removed. Since a long
5021 time, the extended attribute calls have moved to glibc, and
5022 libattr is thus unnecessary.
5024 * Virtualization detection works without privileges now. This
5025 means the systemd-detect-virt binary no longer requires
5026 CAP_SYS_PTRACE file capabilities, and our daemons can run
5027 with fewer privileges.
5029 * systemd-networkd now runs under its own "systemd-network"
5030 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
5031 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
5032 loses the ability to write to files owned by root this way.
5034 * Similarly, systemd-resolved now runs under its own
5035 "systemd-resolve" user with no capabilities remaining.
5037 * Similarly, systemd-bus-proxyd now runs under its own
5038 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
5040 * systemd-networkd gained support for setting up "veth"
5041 virtual Ethernet devices for container connectivity, as well
5042 as GRE and VTI tunnels.
5044 * systemd-networkd will no longer automatically attempt to
5045 manually load kernel modules necessary for certain tunnel
5046 transports. Instead, it is assumed the kernel loads them
5047 automatically when required. This only works correctly on
5048 very new kernels. On older kernels, please consider adding
5049 the kernel modules to /etc/modules-load.d/ as a work-around.
5051 * The resolv.conf file systemd-resolved generates has been
5052 moved to /run/systemd/resolve/. If you have a symlink from
5053 /etc/resolv.conf, it might be necessary to correct it.
5055 * Two new service settings, ProtectHome= and ProtectSystem=,
5056 have been added. When enabled, they will make the user data
5057 (such as /home) inaccessible or read-only and the system
5058 (such as /usr) read-only, for specific services. This allows
5059 very light-weight per-service sandboxing to avoid
5060 modifications of user data or system files from
5061 services. These two new switches have been enabled for all
5062 of systemd's long-running services, where appropriate.
5064 * Socket units gained new SocketUser= and SocketGroup=
5065 settings to set the owner user and group of AF_UNIX sockets
5066 and FIFOs in the file system.
5068 * Socket units gained a new RemoveOnStop= setting. If enabled,
5069 all FIFOS and sockets in the file system will be removed
5070 when the specific socket unit is stopped.
5072 * Socket units gained a new Symlinks= setting. It takes a list
5073 of symlinks to create to file system sockets or FIFOs
5074 created by the specific Unix sockets. This is useful to
5075 manage symlinks to socket nodes with the same life-cycle as
5078 * The /dev/log socket and /dev/initctl FIFO have been moved to
5079 /run, and have been replaced by symlinks. This allows
5080 connecting to these facilities even if PrivateDevices=yes is
5081 used for a service (which makes /dev/log itself unavailable,
5082 but /run is left). This also has the benefit of ensuring
5083 that /dev only contains device nodes, directories and
5084 symlinks, and nothing else.
5086 * sd-daemon gained two new calls sd_pid_notify() and
5087 sd_pid_notifyf(). They are similar to sd_notify() and
5088 sd_notifyf(), but allow overriding of the source PID of
5089 notification messages if permissions permit this. This is
5090 useful to send notify messages on behalf of a different
5091 process (for example, the parent process). The
5092 systemd-notify tool has been updated to make use of this
5093 when sending messages (so that notification messages now
5094 originate from the shell script invoking systemd-notify and
5095 not the systemd-notify process itself. This should minimize
5096 a race where systemd fails to associate notification
5097 messages to services when the originating process already
5100 * A new "on-abnormal" setting for Restart= has been added. If
5101 set, it will result in automatic restarts on all "abnormal"
5102 reasons for a process to exit, which includes unclean
5103 signals, core dumps, timeouts and watchdog timeouts, but
5104 does not include clean and unclean exit codes or clean
5105 signals. Restart=on-abnormal is an alternative for
5106 Restart=on-failure for services that shall be able to
5107 terminate and avoid restarts on certain errors, by
5108 indicating so with an unclean exit code. Restart=on-failure
5109 or Restart=on-abnormal is now the recommended setting for
5110 all long-running services.
5112 * If the InaccessibleDirectories= service setting points to a
5113 mount point (or if there are any submounts contained within
5114 it), it is now attempted to completely unmount it, to make
5115 the file systems truly unavailable for the respective
5118 * The ReadOnlyDirectories= service setting and
5119 systemd-nspawn's --read-only parameter are now recursively
5120 applied to all submounts, too.
5122 * Mount units may now be created transiently via the bus APIs.
5124 * The support for SysV and LSB init scripts has been removed
5125 from the systemd daemon itself. Instead, it is now
5126 implemented as a generator that creates native systemd units
5127 from these scripts when needed. This enables us to remove a
5128 substantial amount of legacy code from PID 1, following the
5129 fact that many distributions only ship a very small number
5130 of LSB/SysV init scripts nowadays.
5132 * Privileged Xen (dom0) domains are not considered
5133 virtualization anymore by the virtualization detection
5134 logic. After all, they generally have unrestricted access to
5135 the hardware and usually are used to manage the unprivileged
5138 * systemd-tmpfiles gained a new "C" line type, for copying
5139 files or entire directories.
5141 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
5142 lines. So far, they have been non-globbing versions of the
5143 latter, and have thus been redundant. In future, it is
5144 recommended to only use "z". "m" has hence been removed
5145 from the documentation, even though it stays supported.
5147 * A tmpfiles snippet to recreate the most basic structure in
5148 /var has been added. This is enough to create the /var/run →
5149 /run symlink and create a couple of structural
5150 directories. This allows systems to boot up with an empty or
5151 volatile /var. Of course, while with this change, the core OS
5152 now is capable with dealing with a volatile /var, not all
5153 user services are ready for it. However, we hope that sooner
5154 or later, many service daemons will be changed upstream so
5155 that they are able to automatically create their necessary
5156 directories in /var at boot, should they be missing. This is
5157 the first step to allow state-less systems that only require
5158 the vendor image for /usr to boot.
5160 * systemd-nspawn has gained a new --tmpfs= switch to mount an
5161 empty tmpfs instance to a specific directory. This is
5162 particularly useful for making use of the automatic
5163 reconstruction of /var (see above), by passing --tmpfs=/var.
5165 * Access modes specified in tmpfiles snippets may now be
5166 prefixed with "~", which indicates that they shall be masked
5167 by whether the existing file or directory is currently
5168 writable, readable or executable at all. Also, if specified,
5169 the sgid/suid/sticky bits will be masked for all
5172 * A new passive target unit "network-pre.target" has been
5173 added which is useful for services that shall run before any
5174 network is configured, for example firewall scripts.
5176 * The "floppy" group that previously owned the /dev/fd*
5177 devices is no longer used. The "disk" group is now used
5178 instead. Distributions should probably deprecate usage of
5181 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
5182 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
5183 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
5184 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
5185 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
5186 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
5189 — Berlin, 2014-06-11
5193 * A new "systemd-timesyncd" daemon has been added for
5194 synchronizing the system clock across the network. It
5195 implements an SNTP client. In contrast to NTP
5196 implementations such as chrony or the NTP reference server,
5197 this only implements a client side, and does not bother with
5198 the full NTP complexity, focusing only on querying time from
5199 one remote server and synchronizing the local clock to
5200 it. Unless you intend to serve NTP to networked clients or
5201 want to connect to local hardware clocks, this simple NTP
5202 client should be more than appropriate for most
5203 installations. The daemon runs with minimal privileges, and
5204 has been hooked up with networkd to only operate when
5205 network connectivity is available. The daemon saves the
5206 current clock to disk every time a new NTP sync has been
5207 acquired, and uses this to possibly correct the system clock
5208 early at bootup, in order to accommodate for systems that
5209 lack an RTC such as the Raspberry Pi and embedded devices,
5210 and to make sure that time monotonically progresses on these
5211 systems, even if it is not always correct. To make use of
5212 this daemon, a new system user and group "systemd-timesync"
5213 needs to be created on installation of systemd.
5215 * The queue "seqnum" interface of libudev has been disabled, as
5216 it was generally incompatible with device namespacing as
5217 sequence numbers of devices go "missing" if the devices are
5218 part of a different namespace.
5220 * "systemctl list-timers" and "systemctl list-sockets" gained
5221 a --recursive switch for showing units of these types also
5222 for all local containers, similar in style to the already
5223 supported --recursive switch for "systemctl list-units".
5225 * A new RebootArgument= setting has been added for service
5226 units, which may be used to specify a kernel reboot argument
5227 to use when triggering reboots with StartLimitAction=.
5229 * A new FailureAction= setting has been added for service
5230 units which may be used to specify an operation to trigger
5231 when a service fails. This works similarly to
5232 StartLimitAction=, but unlike it, controls what is done
5233 immediately rather than only after several attempts to
5234 restart the service in question.
5236 * hostnamed got updated to also expose the kernel name,
5237 release, and version on the bus. This is useful for
5238 executing commands like hostnamectl with the -H switch.
5239 systemd-analyze makes use of this to properly display
5240 details when running non-locally.
5242 * The bootchart tool can now show cgroup information in the
5243 graphs it generates.
5245 * The CFS CPU quota cgroup attribute is now exposed for
5246 services. The new CPUQuota= switch has been added for this
5247 which takes a percentage value. Setting this will have the
5248 result that a service may never get more CPU time than the
5249 specified percentage, even if the machine is otherwise idle.
5251 * systemd-networkd learned IPIP and SIT tunnel support.
5253 * LSB init scripts exposing a dependency on $network will now
5254 get a dependency on network-online.target rather than simply
5255 network.target. This should bring LSB handling closer to
5256 what it was on SysV systems.
5258 * A new fsck.repair= kernel option has been added to control
5259 how fsck shall deal with unclean file systems at boot.
5261 * The (.ini) configuration file parser will now silently
5262 ignore sections whose name begins with "X-". This may be
5263 used to maintain application-specific extension sections in unit
5266 * machined gained a new API to query the IP addresses of
5267 registered containers. "machinectl status" has been updated
5268 to show these addresses in its output.
5270 * A new call sd_uid_get_display() has been added to the
5271 sd-login APIs for querying the "primary" session of a
5272 user. The "primary" session of the user is elected from the
5273 user's sessions and generally a graphical session is
5274 preferred over a text one.
5276 * A minimal systemd-resolved daemon has been added. It
5277 currently simply acts as a companion to systemd-networkd and
5278 manages resolv.conf based on per-interface DNS
5279 configuration, possibly supplied via DHCP. In the long run
5280 we hope to extend this into a local DNSSEC enabled DNS and
5283 * The systemd-networkd-wait-online tool is now enabled by
5284 default. It will delay network-online.target until a network
5285 connection has been configured. The tool primarily integrates
5286 with networkd, but will also make a best effort to make sense
5287 of network configuration performed in some other way.
5289 * Two new service options StartupCPUShares= and
5290 StartupBlockIOWeight= have been added that work similarly to
5291 CPUShares= and BlockIOWeight= however only apply during
5292 system startup. This is useful to prioritize certain services
5293 differently during bootup than during normal runtime.
5295 * hostnamed has been changed to prefer the statically
5296 configured hostname in /etc/hostname (unless set to
5297 'localhost' or empty) over any dynamic one supplied by
5298 dhcp. With this change, the rules for picking the hostname
5299 match more closely the rules of other configuration settings
5300 where the local administrator's configuration in /etc always
5301 overrides any other settings.
5303 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
5304 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
5305 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
5306 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
5307 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
5308 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
5309 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
5310 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
5311 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
5312 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
5313 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
5314 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
5315 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
5316 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
5317 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
5318 Lindskog, WaLyong Cho, Will Woods, Zbigniew
5321 — Beijing, 2014-05-28
5325 * When restoring the screen brightness at boot, stay away from
5326 the darkest setting or from the lowest 5% of the available
5327 range, depending on which is the larger value of both. This
5328 should effectively protect the user from rebooting into a
5329 black screen, should the brightness have been set to minimum
5332 * sd-login gained a new sd_machine_get_class() call to
5333 determine the class ("vm" or "container") of a machine
5334 registered with machined.
5336 * sd-login gained new calls
5337 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
5338 to query the identity of the peer of a local AF_UNIX
5339 connection. They operate similarly to their sd_pid_get_xyz()
5342 * PID 1 will now maintain a system-wide system state engine
5343 with the states "starting", "running", "degraded",
5344 "maintenance", "stopping". These states are bound to system
5345 startup, normal runtime, runtime with at least one failed
5346 service, rescue/emergency mode and system shutdown. This
5347 state is shown in the "systemctl status" output when no unit
5348 name is passed. It is useful to determine system state, in
5349 particularly when doing so for many systems or containers at
5352 * A new command "list-machines" has been added to "systemctl"
5353 that lists all local OS containers and shows their system
5354 state (see above), if systemd runs inside of them.
5356 * systemctl gained a new "-r" switch to recursively enumerate
5357 units on all local containers, when used with the
5358 "list-unit" command (which is the default one that is
5359 executed when no parameters are specified).
5361 * The GPT automatic partition discovery logic will now honour
5362 two GPT partition flags: one may be set on a partition to
5363 cause it to be mounted read-only, and the other may be set
5364 on a partition to ignore it during automatic discovery.
5366 * Two new GPT type UUIDs have been added for automatic root
5367 partition discovery, for 32-bit and 64-bit ARM. This is not
5368 particularly useful for discovering the root directory on
5369 these architectures during bare-metal boots (since UEFI is
5370 not common there), but still very useful to allow booting of
5371 ARM disk images in nspawn with the -i option.
5373 * MAC addresses of interfaces created with nspawn's
5374 --network-interface= switch will now be generated from the
5375 machine name, and thus be stable between multiple invocations
5378 * logind will now automatically remove all IPC objects owned
5379 by a user if she or he fully logs out. This makes sure that
5380 users who are logged out cannot continue to consume IPC
5381 resources. This covers SysV memory, semaphores and message
5382 queues as well as POSIX shared memory and message
5383 queues. Traditionally, SysV and POSIX IPC had no life-cycle
5384 limits. With this functionality, that is corrected. This may
5385 be turned off by using the RemoveIPC= switch of logind.conf.
5387 * The systemd-machine-id-setup and tmpfiles tools gained a
5388 --root= switch to operate on a specific root directory,
5391 * journald can now forward logged messages to the TTYs of all
5392 logged in users ("wall"). This is the default for all
5393 emergency messages now.
5395 * A new tool systemd-journal-remote has been added to stream
5396 journal log messages across the network.
5398 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
5399 controller trees are mounted into it. Note that the
5400 directories mounted beneath it are not read-only. This is a
5401 security measure and is particularly useful because glibc
5402 actually includes a search logic to pick any tmpfs it can
5403 find to implement shm_open() if /dev/shm is not available
5404 (which it might very well be in namespaced setups).
5406 * machinectl gained a new "poweroff" command to cleanly power
5407 down a local OS container.
5409 * The PrivateDevices= unit file setting will now also drop the
5410 CAP_MKNOD capability from the capability bound set, and
5411 imply DevicePolicy=closed.
5413 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
5414 comprehensively on all long-running systemd services where
5415 this is appropriate.
5417 * systemd-udevd will now run in a disassociated mount
5418 namespace. To mount directories from udev rules, make sure to
5419 pull in mount units via SYSTEMD_WANTS properties.
5421 * The kdbus support gained support for uploading policy into
5422 the kernel. sd-bus gained support for creating "monitoring"
5423 connections that can eavesdrop into all bus communication
5424 for debugging purposes.
5426 * Timestamps may now be specified in seconds since the UNIX
5427 epoch Jan 1st, 1970 by specifying "@" followed by the value
5430 * Native tcpwrap support in systemd has been removed. tcpwrap
5431 is old code, not really maintained anymore and has serious
5432 shortcomings, and better options such as firewalls
5433 exist. For setups that require tcpwrap usage, please
5434 consider invoking your socket-activated service via tcpd,
5435 like on traditional inetd.
5437 * A new system.conf configuration option
5438 DefaultTimerAccuracySec= has been added that controls the
5439 default AccuracySec= setting of .timer units.
5441 * Timer units gained a new WakeSystem= switch. If enabled,
5442 timers configured this way will cause the system to resume
5443 from system suspend (if the system supports that, which most
5446 * Timer units gained a new Persistent= switch. If enabled,
5447 timers configured this way will save to disk when they have
5448 been last triggered. This information is then used on next
5449 reboot to possible execute overdue timer events, that
5450 could not take place because the system was powered off.
5451 This enables simple anacron-like behaviour for timer units.
5453 * systemctl's "list-timers" will now also list the time a
5454 timer unit was last triggered in addition to the next time
5455 it will be triggered.
5457 * systemd-networkd will now assign predictable IPv4LL
5458 addresses to its local interfaces.
5460 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
5461 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
5462 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
5463 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
5464 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
5465 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
5466 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
5467 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
5470 — Berlin, 2014-03-25
5474 * A new unit file setting RestrictAddressFamilies= has been
5475 added to restrict which socket address families unit
5476 processes gain access to. This takes address family names
5477 like "AF_INET" or "AF_UNIX", and is useful to minimize the
5478 attack surface of services via exotic protocol stacks. This
5479 is built on seccomp system call filters.
5481 * Two new unit file settings RuntimeDirectory= and
5482 RuntimeDirectoryMode= have been added that may be used to
5483 manage a per-daemon runtime directories below /run. This is
5484 an alternative for setting up directory permissions with
5485 tmpfiles snippets, and has the advantage that the runtime
5486 directory's lifetime is bound to the daemon runtime and that
5487 the daemon starts up with an empty directory each time. This
5488 is particularly useful when writing services that drop
5489 privileges using the User= or Group= setting.
5491 * The DeviceAllow= unit setting now supports globbing for
5492 matching against device group names.
5494 * The systemd configuration file system.conf gained new
5495 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
5496 DefaultMemoryAccounting= to globally turn on/off accounting
5497 for specific resources (cgroups) for all units. These
5498 settings may still be overridden individually in each unit
5501 * systemd-gpt-auto-generator is now able to discover /srv and
5502 root partitions in addition to /home and swap partitions. It
5503 also supports LUKS-encrypted partitions now. With this in
5504 place, automatic discovery of partitions to mount following
5505 the Discoverable Partitions Specification
5506 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
5507 is now a lot more complete. This allows booting without
5508 /etc/fstab and without root= on the kernel command line on
5509 systems prepared appropriately.
5511 * systemd-nspawn gained a new --image= switch which allows
5512 booting up disk images and Linux installations on any block
5513 device that follow the Discoverable Partitions Specification
5514 (see above). This means that installations made with
5515 appropriately updated installers may now be started and
5516 deployed using container managers, completely
5517 unmodified. (We hope that libvirt-lxc will add support for
5518 this feature soon, too.)
5520 * systemd-nspawn gained a new --network-macvlan= setting to
5521 set up a private macvlan interface for the
5522 container. Similarly, systemd-networkd gained a new
5523 Kind=macvlan setting in .netdev files.
5525 * systemd-networkd now supports configuring local addresses
5528 * A new tool systemd-network-wait-online has been added to
5529 synchronously wait for network connectivity using
5532 * The sd-bus.h bus API gained a new sd_bus_track object for
5533 tracking the life-cycle of bus peers. Note that sd-bus.h is
5534 still not a public API though (unless you specify
5535 --enable-kdbus on the configure command line, which however
5536 voids your warranty and you get no API stability guarantee).
5538 * The $XDG_RUNTIME_DIR runtime directories for each user are
5539 now individual tmpfs instances, which has the benefit of
5540 introducing separate pools for each user, with individual
5541 size limits, and thus making sure that unprivileged clients
5542 can no longer negatively impact the system or other users by
5543 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
5544 RuntimeDirectorySize= has been introduced that allows
5545 controlling the default size limit for all users. It
5546 defaults to 10% of the available physical memory. This is no
5547 replacement for quotas on tmpfs though (which the kernel
5548 still does not support), as /dev/shm and /tmp are still
5549 shared resources used by both the system and unprivileged
5552 * logind will now automatically turn off automatic suspending
5553 on laptop lid close when more than one display is
5554 connected. This was previously expected to be implemented
5555 individually in desktop environments (such as GNOME),
5556 however has been added to logind now, in order to fix a
5557 boot-time race where a desktop environment might not have
5558 been started yet and thus not been able to take an inhibitor
5559 lock at the time where logind already suspends the system
5560 due to a closed lid.
5562 * logind will now wait at least 30s after each system
5563 suspend/resume cycle, and 3min after system boot before
5564 suspending the system due to a closed laptop lid. This
5565 should give USB docking stations and similar enough time to
5566 be probed and configured after system resume and boot in
5567 order to then act as suspend blocker.
5569 * systemd-run gained a new --property= setting which allows
5570 initialization of resource control properties (and others)
5571 for the created scope or service unit. Example: "systemd-run
5572 --property=BlockIOWeight=10 updatedb" may be used to run
5573 updatedb at a low block IO scheduling weight.
5575 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
5576 now also work in --scope mode.
5578 * When systemd is compiled with kdbus support, basic support
5579 for enforced policies is now in place. (Note that enabling
5580 kdbus still voids your warranty and no API compatibility
5583 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
5584 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5585 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
5586 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
5587 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
5588 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
5589 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
5590 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
5591 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
5592 Zbigniew Jędrzejewski-Szmek
5594 — Berlin, 2014-03-12
5598 * systemd will now relabel /dev after loading the SMACK policy
5599 according to SMACK rules.
5601 * A new unit file option AppArmorProfile= has been added to
5602 set the AppArmor profile for the processes of a unit.
5604 * A new condition check ConditionArchitecture= has been added
5605 to conditionalize units based on the system architecture, as
5606 reported by uname()'s "machine" field.
5608 * systemd-networkd now supports matching on the system
5609 virtualization, architecture, kernel command line, host name
5612 * logind is now a lot more aggressive when suspending the
5613 machine due to a closed laptop lid. Instead of acting only
5614 on the lid close action, it will continuously watch the lid
5615 status and act on it. This is useful for laptops where the
5616 power button is on the outside of the chassis so that it can
5617 be reached without opening the lid (such as the Lenovo
5618 Yoga). On those machines, logind will now immediately
5619 re-suspend the machine if the power button has been
5620 accidentally pressed while the laptop was suspended and in a
5621 backpack or similar.
5623 * logind will now watch SW_DOCK switches and inhibit reaction
5624 to the lid switch if it is pressed. This means that logind
5625 will not suspend the machine anymore if the lid is closed
5626 and the system is docked, if the laptop supports SW_DOCK
5627 notifications via the input layer. Note that ACPI docking
5628 stations do not generate this currently. Also note that this
5629 logic is usually not fully sufficient and Desktop
5630 Environments should take a lid switch inhibitor lock when an
5631 external display is connected, as systemd will not watch
5634 * nspawn will now make use of the devices cgroup controller by
5635 default, and only permit creation of and access to the usual
5636 API device nodes like /dev/null or /dev/random, as well as
5637 access to (but not creation of) the pty devices.
5639 * We will now ship a default .network file for
5640 systemd-networkd that automatically configures DHCP for
5641 network interfaces created by nspawn's --network-veth or
5642 --network-bridge= switches.
5644 * systemd will now understand the usual M, K, G, T suffixes
5645 according to SI conventions (i.e. to the base 1000) when
5646 referring to throughput and hardware metrics. It will stay
5647 with IEC conventions (i.e. to the base 1024) for software
5648 metrics, according to what is customary according to
5649 Wikipedia. We explicitly document which base applies for
5650 each configuration option.
5652 * The DeviceAllow= setting in unit files now supports a syntax
5653 to whitelist an entire group of devices node majors at once,
5654 based on the /proc/devices listing. For example, with the
5655 string "char-pts", it is now possible to whitelist all
5656 current and future pseudo-TTYs at once.
5658 * sd-event learned a new "post" event source. Event sources of
5659 this type are triggered by the dispatching of any event
5660 source of a type that is not "post". This is useful for
5661 implementing clean-up and check event sources that are
5662 triggered by other work being done in the program.
5664 * systemd-networkd is no longer statically enabled, but uses
5665 the usual [Install] sections so that it can be
5666 enabled/disabled using systemctl. It still is enabled by
5669 * When creating a veth interface pair with systemd-nspawn, the
5670 host side will now be prefixed with "vb-" if
5671 --network-bridge= is used, and with "ve-" if --network-veth
5672 is used. This way, it is easy to distinguish these cases on
5673 the host, for example to apply different configuration to
5674 them with systemd-networkd.
5676 * The compatibility libraries for libsystemd-journal.so,
5677 libsystem-id128.so, libsystemd-login.so and
5678 libsystemd-daemon.so do not make use of IFUNC
5679 anymore. Instead, we now build libsystemd.so multiple times
5680 under these alternative names. This means that the footprint
5681 is drastically increased, but given that these are
5682 transitional compatibility libraries, this should not matter
5683 much. This change has been made necessary to support the ARM
5684 platform for these compatibility libraries, as the ARM
5685 toolchain is not really at the same level as the toolchain
5686 for other architectures like x86 and does not support
5687 IFUNC. Please make sure to use --enable-compat-libs only
5688 during a transitional period!
5690 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
5691 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5692 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
5693 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
5694 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
5695 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
5696 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
5697 Zbigniew Jędrzejewski-Szmek
5699 — Berlin, 2014-02-24
5703 * A new component "systemd-networkd" has been added that can
5704 be used to configure local network interfaces statically or
5705 via DHCP. It is capable of bringing up bridges, VLANs, and
5706 bonding. Currently, no hook-ups for interactive network
5707 configuration are provided. Use this for your initrd,
5708 container, embedded, or server setup if you need a simple,
5709 yet powerful, network configuration solution. This
5710 configuration subsystem is quite nifty, as it allows wildcard
5711 hotplug matching in interfaces. For example, with a single
5712 configuration snippet, you can configure that all Ethernet
5713 interfaces showing up are automatically added to a bridge,
5714 or similar. It supports link-sensing and more.
5716 * A new tool "systemd-socket-proxyd" has been added which can
5717 act as a bidirectional proxy for TCP sockets. This is
5718 useful for adding socket activation support to services that
5719 do not actually support socket activation, including virtual
5720 machines and the like.
5722 * Add a new tool to save/restore rfkill state on
5725 * Save/restore state of keyboard backlights in addition to
5726 display backlights on shutdown/boot.
5728 * udev learned a new SECLABEL{} construct to label device
5729 nodes with a specific security label when they appear. For
5730 now, only SECLABEL{selinux} is supported, but the syntax is
5731 prepared for additional security frameworks.
5733 * udev gained a new scheme to configure link-level attributes
5734 from files in /etc/systemd/network/*.link. These files can
5735 match against MAC address, device path, driver name and type,
5736 and will apply attributes like the naming policy, link speed,
5737 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
5738 address assignment policy (randomized, ...).
5740 * The configuration of network interface naming rules for
5741 "permanent interface names" has changed: a new NamePolicy=
5742 setting in the [Link] section of .link files determines the
5743 priority of possible naming schemes (onboard, slot, MAC,
5744 path). The default value of this setting is determined by
5745 /usr/lib/net/links/99-default.link. Old
5746 80-net-name-slot.rules udev configuration file has been
5747 removed, so local configuration overriding this file should
5748 be adapted to override 99-default.link instead.
5750 * When the User= switch is used in a unit file, also
5751 initialize $SHELL= based on the user database entry.
5753 * systemd no longer depends on libdbus. All communication is
5754 now done with sd-bus, systemd's low-level bus library
5757 * kdbus support has been added to PID 1 itself. When kdbus is
5758 enabled, this causes PID 1 to set up the system bus and
5759 enable support for a new ".busname" unit type that
5760 encapsulates bus name activation on kdbus. It works a little
5761 bit like ".socket" units, except for bus names. A new
5762 generator has been added that converts classic dbus1 service
5763 activation files automatically into native systemd .busname
5766 * sd-bus: add a light-weight vtable implementation that allows
5767 defining objects on the bus with a simple static const
5768 vtable array of its methods, signals and properties.
5770 * systemd will not generate or install static dbus
5771 introspection data anymore to /usr/share/dbus-1/interfaces,
5772 as the precise format of these files is unclear, and
5773 nothing makes use of it.
5775 * A proxy daemon is now provided to proxy clients connecting
5776 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
5777 compatibility with classic D-Bus.
5779 * A bus driver implementation has been added that supports the
5780 classic D-Bus bus driver calls on kdbus, also for
5781 compatibility purposes.
5783 * A new API "sd-event.h" has been added that implements a
5784 minimal event loop API built around epoll. It provides a
5785 couple of features that direct epoll usage is lacking:
5786 prioritization of events, scales to large numbers of timer
5787 events, per-event timer slack (accuracy), system-wide
5788 coalescing of timer events, exit handlers, watchdog
5789 supervision support using systemd's sd_notify() API, child
5792 * A new API "sd-rntl.h" has been added that provides an API
5793 around the route netlink interface of the kernel, similar in
5794 style to "sd-bus.h".
5796 * A new API "sd-dhcp-client.h" has been added that provides a
5797 small DHCPv4 client-side implementation. This is used by
5800 * There is a new kernel command line option
5801 "systemd.restore_state=0|1". When set to "0", none of the
5802 systemd tools will restore saved runtime state to hardware
5803 devices. More specifically, the rfkill and backlight states
5806 * The FsckPassNo= compatibility option in mount/service units
5807 has been removed. The fstab generator will now add the
5808 necessary dependencies automatically, and does not require
5809 PID1's support for that anymore.
5811 * journalctl gained a new switch, --list-boots, that lists
5812 recent boots with their times and boot IDs.
5814 * The various tools like systemctl, loginctl, timedatectl,
5815 busctl, systemd-run, ... have gained a new switch "-M" to
5816 connect to a specific, local OS container (as direct
5817 connection, without requiring SSH). This works on any
5818 container that is registered with machined, such as those
5819 created by libvirt-lxc or nspawn.
5821 * systemd-run and systemd-analyze also gained support for "-H"
5822 to connect to remote hosts via SSH. This is particularly
5823 useful for systemd-run because it enables queuing of jobs
5824 onto remote systems.
5826 * machinectl gained a new command "login" to open a getty
5827 login in any local container. This works with any container
5828 that is registered with machined (such as those created by
5829 libvirt-lxc or nspawn), and which runs systemd inside.
5831 * machinectl gained a new "reboot" command that may be used to
5832 trigger a reboot on a specific container that is registered
5833 with machined. This works on any container that runs an init
5834 system of some kind.
5836 * systemctl gained a new "list-timers" command to print a nice
5837 listing of installed timer units with the times they elapse
5840 * Alternative reboot() parameters may now be specified on the
5841 "systemctl reboot" command line and are passed to the
5842 reboot() system call.
5844 * systemctl gained a new --job-mode= switch to configure the
5845 mode to queue a job with. This is a more generic version of
5846 --fail, --irreversible, and --ignore-dependencies, which are
5847 still available but not advertised anymore.
5849 * /etc/systemd/system.conf gained new settings to configure
5850 various default timeouts of units, as well as the default
5851 start limit interval and burst. These may still be overridden
5854 * PID1 will now export on the bus profile data of the security
5855 policy upload process (such as the SELinux policy upload to
5858 * journald: when forwarding logs to the console, include
5859 timestamps (following the setting in
5860 /sys/module/printk/parameters/time).
5862 * OnCalendar= in timer units now understands the special
5863 strings "yearly" and "annually". (Both are equivalent)
5865 * The accuracy of timer units is now configurable with the new
5866 AccuracySec= setting. It defaults to 1min.
5868 * A new dependency type JoinsNamespaceOf= has been added that
5869 allows running two services within the same /tmp and network
5870 namespace, if PrivateNetwork= or PrivateTmp= are used.
5872 * A new command "cat" has been added to systemctl. It outputs
5873 the original unit file of a unit, and concatenates the
5874 contents of additional "drop-in" unit file snippets, so that
5875 the full configuration is shown.
5877 * systemctl now supports globbing on the various "list-xyz"
5878 commands, like "list-units" or "list-sockets", as well as on
5879 those commands which take multiple unit names.
5881 * journalctl's --unit= switch gained support for globbing.
5883 * All systemd daemons now make use of the watchdog logic so
5884 that systemd automatically notices when they hang.
5886 * If the $container_ttys environment variable is set,
5887 getty-generator will automatically spawn a getty for each
5888 listed tty. This is useful for container managers to request
5889 login gettys to be spawned on as many ttys as needed.
5891 * %h, %s, %U specifier support is not available anymore when
5892 used in unit files for PID 1. This is because NSS calls are
5893 not safe from PID 1. They stay available for --user
5894 instances of systemd, and as special case for the root user.
5896 * loginctl gained a new "--no-legend" switch to turn off output
5899 * The "sd-login.h" API gained three new calls:
5900 sd_session_is_remote(), sd_session_get_remote_user(),
5901 sd_session_get_remote_host() to query information about
5904 * The udev hardware database now also carries vendor/product
5905 information of SDIO devices.
5907 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
5908 determine whether watchdog notifications are requested by
5911 * Socket-activated per-connection services now include a
5912 short description of the connection parameters in the
5915 * tmpfiles gained a new "--boot" option. When this is not used,
5916 only lines where the command character is not suffixed with
5917 "!" are executed. When this option is specified, those
5918 options are executed too. This partitions tmpfiles
5919 directives into those that can be safely executed at any
5920 time, and those which should be run only at boot (for
5921 example, a line that creates /run/nologin).
5923 * A new API "sd-resolve.h" has been added which provides a simple
5924 asynchronous wrapper around glibc NSS host name resolution
5925 calls, such as getaddrinfo(). In contrast to glibc's
5926 getaddrinfo_a(), it does not use signals. In contrast to most
5927 other asynchronous name resolution libraries, this one does
5928 not reimplement DNS, but reuses NSS, so that alternate
5929 host name resolution systems continue to work, such as mDNS,
5930 LDAP, etc. This API is based on libasyncns, but it has been
5931 cleaned up for inclusion in systemd.
5933 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
5934 "sd-daemon.h" are no longer found in individual libraries
5935 libsystemd-journal.so, libsystemd-login.so,
5936 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
5937 merged them into a single library, libsystemd.so, which
5938 provides all symbols. The reason for this is cyclic
5939 dependencies, as these libraries tend to use each other's
5940 symbols. So far, we have managed to workaround that by linking
5941 a copy of a good part of our code into each of these
5942 libraries again and again, which, however, makes certain
5943 things hard to do, like sharing static variables. Also, it
5944 substantially increases footprint. With this change, there
5945 is only one library for the basic APIs systemd
5946 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
5947 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
5948 library as well, however are subject to the --enable-kdbus
5949 switch (see below). Note that "sd-dhcp-client.h" is not part
5950 of this library (this is because it only consumes, never
5951 provides, services of/to other APIs). To make the transition
5952 easy from the separate libraries to the unified one, we
5953 provide the --enable-compat-libs compile-time switch which
5954 will generate stub libraries that are compatible with the
5955 old ones but redirect all calls to the new one.
5957 * All of the kdbus logic and the new APIs "sd-bus.h",
5958 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
5959 and "sd-utf8.h" are compile-time optional via the
5960 "--enable-kdbus" switch, and they are not compiled in by
5961 default. To make use of kdbus, you have to explicitly enable
5962 the switch. Note however, that neither the kernel nor the
5963 userspace API for all of this is considered stable yet. We
5964 want to maintain the freedom to still change the APIs for
5965 now. By specifying this build-time switch, you acknowledge
5966 that you are aware of the instability of the current
5969 * Also, note that while kdbus is pretty much complete,
5970 it lacks one thing: proper policy support. This means you
5971 can build a fully working system with all features; however,
5972 it will be highly insecure. Policy support will be added in
5973 one of the next releases, at the same time that we will
5974 declare the APIs stable.
5976 * When the kernel command line argument "kdbus" is specified,
5977 systemd will automatically load the kdbus.ko kernel module. At
5978 this stage of development, it is only useful for testing kdbus
5979 and should not be used in production. Note: if "--enable-kdbus"
5980 is specified, and the kdbus.ko kernel module is available, and
5981 "kdbus" is added to the kernel command line, the entire system
5982 runs with kdbus instead of dbus-daemon, with the above mentioned
5983 problem of missing the system policy enforcement. Also a future
5984 version of kdbus.ko or a newer systemd will not be compatible with
5985 each other, and will unlikely be able to boot the machine if only
5986 one of them is updated.
5988 * systemctl gained a new "import-environment" command which
5989 uploads the caller's environment (or parts thereof) into the
5990 service manager so that it is inherited by services started
5991 by the manager. This is useful to upload variables like
5992 $DISPLAY into the user service manager.
5994 * A new PrivateDevices= switch has been added to service units
5995 which allows running a service with a namespaced /dev
5996 directory that does not contain any device nodes for
5997 physical devices. More specifically, it only includes devices
5998 such as /dev/null, /dev/urandom, and /dev/zero which are API
6001 * logind has been extended to support behaviour like VT
6002 switching on seats that do not support a VT. This makes
6003 multi-session available on seats that are not the first seat
6004 (seat0), and on systems where kernel support for VTs has
6005 been disabled at compile-time.
6007 * If a process holds a delay lock for system sleep or shutdown
6008 and fails to release it in time, we will now log its
6009 identity. This makes it easier to identify processes that
6010 cause slow suspends or power-offs.
6012 * When parsing /etc/crypttab, support for a new key-slot=
6013 option as supported by Debian is added. It allows indicating
6014 which LUKS slot to use on disk, speeding up key loading.
6016 * The sd_journal_sendv() API call has been checked and
6017 officially declared to be async-signal-safe so that it may
6018 be invoked from signal handlers for logging purposes.
6020 * Boot-time status output is now enabled automatically after a
6021 short timeout if boot does not progress, in order to give
6022 the user an indication what she or he is waiting for.
6024 * The boot-time output has been improved to show how much time
6025 remains until jobs expire.
6027 * The KillMode= switch in service units gained a new possible
6028 value "mixed". If set, and the unit is shut down, then the
6029 initial SIGTERM signal is sent only to the main daemon
6030 process, while the following SIGKILL signal is sent to
6031 all remaining processes of the service.
6033 * When a scope unit is registered, a new property "Controller"
6034 may be set. If set to a valid bus name, systemd will send a
6035 RequestStop() signal to this name when it would like to shut
6036 down the scope. This may be used to hook manager logic into
6037 the shutdown logic of scope units. Also, scope units may now
6038 be put in a special "abandoned" state, in which case the
6039 manager process which created them takes no further
6040 responsibilities for it.
6042 * When reading unit files, systemd will now verify
6043 the access mode of these files, and warn about certain
6044 suspicious combinations. This has been added to make it
6045 easier to track down packaging bugs where unit files are
6046 marked executable or world-writable.
6048 * systemd-nspawn gained a new "--setenv=" switch to set
6049 container-wide environment variables. The similar option in
6050 systemd-activate was renamed from "--environment=" to
6051 "--setenv=" for consistency.
6053 * systemd-nspawn has been updated to create a new kdbus domain
6054 for each container that is invoked, thus allowing each
6055 container to have its own set of system and user buses,
6056 independent of the host.
6058 * systemd-nspawn gained a new --drop-capability= switch to run
6059 the container with less capabilities than the default. Both
6060 --drop-capability= and --capability= now take the special
6061 string "all" for dropping or keeping all capabilities.
6063 * systemd-nspawn gained new switches for executing containers
6064 with specific SELinux labels set.
6066 * systemd-nspawn gained a new --quiet switch to not generate
6067 any additional output but the container's own console
6070 * systemd-nspawn gained a new --share-system switch to run a
6071 container without PID namespacing enabled.
6073 * systemd-nspawn gained a new --register= switch to control
6074 whether the container is registered with systemd-machined or
6075 not. This is useful for containers that do not run full
6076 OS images, but only specific apps.
6078 * systemd-nspawn gained a new --keep-unit which may be used
6079 when invoked as the only program from a service unit, and
6080 results in registration of the unit service itself in
6081 systemd-machined, instead of a newly opened scope unit.
6083 * systemd-nspawn gained a new --network-interface= switch for
6084 moving arbitrary interfaces to the container. The new
6085 --network-veth switch creates a virtual Ethernet connection
6086 between host and container. The new --network-bridge=
6087 switch then allows assigning the host side of this virtual
6088 Ethernet connection to a bridge device.
6090 * systemd-nspawn gained a new --personality= switch for
6091 setting the kernel personality for the container. This is
6092 useful when running a 32-bit container on a 64-bit host. A
6093 similar option Personality= is now also available for service
6096 * logind will now also track a "Desktop" identifier for each
6097 session which encodes the desktop environment of it. This is
6098 useful for desktop environments that want to identify
6099 multiple running sessions of itself easily.
6101 * A new SELinuxContext= setting for service units has been
6102 added that allows setting a specific SELinux execution
6103 context for a service.
6105 * Most systemd client tools will now honour $SYSTEMD_LESS for
6106 settings of the "less" pager. By default, these tools will
6107 override $LESS to allow certain operations to work, such as
6108 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
6109 influence this logic.
6111 * systemd's "seccomp" hook-up has been changed to make use of
6112 the libseccomp library instead of using its own
6113 implementation. This has benefits for portability among
6116 * For usage together with SystemCallFilter=, a new
6117 SystemCallErrorNumber= setting has been introduced that
6118 allows configuration of a system error number to be returned
6119 on filtered system calls, instead of immediately killing the
6120 process. Also, SystemCallArchitectures= has been added to
6121 limit access to system calls of a particular architecture
6122 (in order to turn off support for unused secondary
6123 architectures). There is also a global
6124 SystemCallArchitectures= setting in system.conf now to turn
6125 off support for non-native system calls system-wide.
6127 * systemd requires a kernel with a working name_to_handle_at(),
6128 please see the kernel config requirements in the README file.
6130 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
6131 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
6132 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
6133 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
6134 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
6135 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
6136 Elia Pinto, Florian Weimer, George McCollister, Goffredo
6137 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
6138 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
6139 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
6140 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
6141 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
6142 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
6143 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
6144 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
6145 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
6146 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
6147 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
6148 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
6149 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
6150 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
6151 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
6152 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
6153 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
6155 — Berlin, 2014-02-20
6159 * logind has gained support for facilitating privileged input
6160 and drm device access for unprivileged clients. This work is
6161 useful to allow Wayland display servers (and similar
6162 programs, such as kmscon) to run under the user's ID and
6163 access input and drm devices which are normally
6164 protected. When this is used (and the kernel is new enough)
6165 logind will "mute" IO on the file descriptors passed to
6166 Wayland as long as it is in the background and "unmute" it
6167 if it returns into the foreground. This allows secure
6168 session switching without allowing background sessions to
6169 eavesdrop on input and display data. This also introduces
6170 session switching support if VT support is turned off in the
6171 kernel, and on seats that are not seat0.
6173 * A new kernel command line option luks.options= is understood
6174 now which allows specifying LUKS options for usage for LUKS
6175 encrypted partitions specified with luks.uuid=.
6177 * tmpfiles.d(5) snippets may now use specifier expansion in
6178 path names. More specifically %m, %b, %H, %v, are now
6179 replaced by the local machine id, boot id, hostname, and
6180 kernel version number.
6182 * A new tmpfiles.d(5) command "m" has been introduced which
6183 may be used to change the owner/group/access mode of a file
6184 or directory if it exists, but do nothing if it does not.
6186 * This release removes high-level support for the
6187 MemorySoftLimit= cgroup setting. The underlying kernel
6188 cgroup attribute memory.soft_limit= is currently badly
6189 designed and likely to be removed from the kernel API in its
6190 current form, hence we should not expose it for now.
6192 * The memory.use_hierarchy cgroup attribute is now enabled for
6193 all cgroups systemd creates in the memory cgroup
6194 hierarchy. This option is likely to be come the built-in
6195 default in the kernel anyway, and the non-hierarchical mode
6196 never made much sense in the intrinsically hierarchical
6199 * A new field _SYSTEMD_SLICE= is logged along with all journal
6200 messages containing the slice a message was generated
6201 from. This is useful to allow easy per-customer filtering of
6202 logs among other things.
6204 * systemd-journald will no longer adjust the group of journal
6205 files it creates to the "systemd-journal" group. Instead we
6206 rely on the journal directory to be owned by the
6207 "systemd-journal" group, and its setgid bit set, so that the
6208 kernel file system layer will automatically enforce that
6209 journal files inherit this group assignment. The reason for
6210 this change is that we cannot allow NSS look-ups from
6211 journald which would be necessary to resolve
6212 "systemd-journal" to a numeric GID, because this might
6213 create deadlocks if NSS involves synchronous queries to
6214 other daemons (such as nscd, or sssd) which in turn are
6215 logging clients of journald and might block on it, which
6216 would then dead lock. A tmpfiles.d(5) snippet included in
6217 systemd will make sure the setgid bit and group are
6218 properly set on the journal directory if it exists on every
6219 boot. However, we recommend adjusting it manually after
6220 upgrades too (or from RPM scriptlets), so that the change is
6221 not delayed until next reboot.
6223 * Backlight and random seed files in /var/lib/ have moved into
6224 the /var/lib/systemd/ directory, in order to centralize all
6225 systemd generated files in one directory.
6227 * Boot time performance measurements (as displayed by
6228 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
6229 performance information if that's available to determine how
6230 much time BIOS and boot loader initialization required. With
6231 a sufficiently new BIOS you hence no longer need to boot
6232 with Gummiboot to get access to such information.
6234 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
6235 Cristian Rodríguez, Dave Reisner, David Herrmann, David
6236 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
6237 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
6238 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
6239 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
6240 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6242 — Berlin, 2013-10-02
6246 * The Restart= option for services now understands a new
6247 on-watchdog setting, which will restart the service
6248 automatically if the service stops sending out watchdog keep
6249 alive messages (as configured with WatchdogSec=).
6251 * The getty generator (which is responsible for bringing up a
6252 getty on configured serial consoles) will no longer only
6253 start a getty on the primary kernel console but on all
6254 others, too. This makes the order in which console= is
6255 specified on the kernel command line less important.
6257 * libsystemd-logind gained a new sd_session_get_vt() call to
6258 retrieve the VT number of a session.
6260 * If the option "tries=0" is set for an entry of /etc/crypttab
6261 its passphrase is queried indefinitely instead of any
6262 maximum number of tries.
6264 * If a service with a configure PID file terminates its PID
6265 file will now be removed automatically if it still exists
6266 afterwards. This should put an end to stale PID files.
6268 * systemd-run will now also take relative binary path names
6269 for execution and no longer insists on absolute paths.
6271 * InaccessibleDirectories= and ReadOnlyDirectories= now take
6272 paths that are optionally prefixed with "-" to indicate that
6273 it should not be considered a failure if they do not exist.
6275 * journalctl -o (and similar commands) now understands a new
6276 output mode "short-precise", it is similar to "short" but
6277 shows timestamps with usec accuracy.
6279 * The option "discard" (as known from Debian) is now
6280 synonymous to "allow-discards" in /etc/crypttab. In fact,
6281 "discard" is preferred now (since it is easier to remember
6284 * Some licensing clean-ups were made, so that more code is now
6285 LGPL-2.1 licensed than before.
6287 * A minimal tool to save/restore the display backlight
6288 brightness across reboots has been added. It will store the
6289 backlight setting as late as possible at shutdown, and
6290 restore it as early as possible during reboot.
6292 * A logic to automatically discover and enable home and swap
6293 partitions on GPT disks has been added. With this in place
6294 /etc/fstab becomes optional for many setups as systemd can
6295 discover certain partitions located on the root disk
6296 automatically. Home partitions are recognized under their
6297 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
6298 partitions are recognized under their GPT type ID
6299 0657fd6da4ab43c484e50933c84b4f4f.
6301 * systemd will no longer pass any environment from the kernel
6302 or initrd to system services. If you want to set an
6303 environment for all services, do so via the kernel command
6304 line systemd.setenv= assignment.
6306 * The systemd-sysctl tool no longer natively reads the file
6307 /etc/sysctl.conf. If desired, the file should be symlinked
6308 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
6309 legacy support by a symlink rather than built-in code, it
6310 also makes the otherwise hidden order of application of the
6311 different files visible. (Note that this partly reverts to a
6312 pre-198 application order of sysctl knobs!)
6314 * The "systemctl set-log-level" and "systemctl dump" commands
6315 have been moved to systemd-analyze.
6317 * systemd-run learned the new --remain-after-exit switch,
6318 which causes the scope unit not to be cleaned up
6319 automatically after the process terminated.
6321 * tmpfiles learned a new --exclude-prefix= switch to exclude
6322 certain paths from operation.
6324 * journald will now automatically flush all messages to disk
6325 as soon as a message at the log level CRIT, ALERT or EMERG
6328 Contributions from: Andrew Cook, Brandon Philips, Christian
6329 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
6330 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
6331 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
6332 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
6333 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
6334 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
6335 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
6336 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
6337 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
6338 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
6339 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
6340 William Giokas, Zbigniew Jędrzejewski-Szmek
6342 — Berlin, 2013-09-13
6346 * The documentation has been updated to cover the various new
6347 concepts introduced with 205.
6349 * Unit files now understand the new %v specifier which
6350 resolves to the kernel version string as returned by "uname
6353 * systemctl now supports filtering the unit list output by
6354 load state, active state and sub state, using the new
6357 * "systemctl status" will now show the results of the
6358 condition checks (like ConditionPathExists= and similar) of
6359 the last start attempts of the unit. They are also logged to
6362 * "journalctl -b" may now be used to look for boot output of a
6363 specific boot. Try "journalctl -b -1" for the previous boot,
6364 but the syntax is substantially more powerful.
6366 * "journalctl --show-cursor" has been added which prints the
6367 cursor string the last shown log line. This may then be used
6368 with the new "journalctl --after-cursor=" switch to continue
6369 browsing logs from that point on.
6371 * "journalctl --force" may now be used to force regeneration
6374 * Creation of "dead" device nodes has been moved from udev
6375 into kmod and tmpfiles. Previously, udev would read the kmod
6376 databases to pre-generate dead device nodes based on meta
6377 information contained in kernel modules, so that these would
6378 be auto-loaded on access rather then at boot. As this
6379 does not really have much to do with the exposing actual
6380 kernel devices to userspace this has always been slightly
6381 alien in the udev codebase. Following the new scheme kmod
6382 will now generate a runtime snippet for tmpfiles from the
6383 module meta information and it now is tmpfiles' job to the
6384 create the nodes. This also allows overriding access and
6385 other parameters for the nodes using the usual tmpfiles
6386 facilities. As side effect this allows us to remove the
6387 CAP_SYS_MKNOD capability bit from udevd entirely.
6389 * logind's device ACLs may now be applied to these "dead"
6390 devices nodes too, thus finally allowing managed access to
6391 devices such as /dev/snd/sequencer without loading the
6392 backing module right-away.
6394 * A new RPM macro has been added that may be used to apply
6395 tmpfiles configuration during package installation.
6397 * systemd-detect-virt and ConditionVirtualization= now can
6398 detect User-Mode-Linux machines (UML).
6400 * journald will now implicitly log the effective capabilities
6401 set of processes in the message metadata.
6403 * systemd-cryptsetup has gained support for TrueCrypt volumes.
6405 * The initrd interface has been simplified (more specifically,
6406 support for passing performance data via environment
6407 variables and fsck results via files in /run has been
6408 removed). These features were non-essential, and are
6409 nowadays available in a much nicer way by having systemd in
6410 the initrd serialize its state and have the hosts systemd
6411 deserialize it again.
6413 * The udev "keymap" data files and tools to apply keyboard
6414 specific mappings of scan to key codes, and force-release
6415 scan code lists have been entirely replaced by a udev
6416 "keyboard" builtin and a hwdb data file.
6418 * systemd will now honour the kernel's "quiet" command line
6419 argument also during late shutdown, resulting in a
6420 completely silent shutdown when used.
6422 * There's now an option to control the SO_REUSEPORT socket
6423 option in .socket units.
6425 * Instance units will now automatically get a per-template
6426 subslice of system.slice unless something else is explicitly
6427 configured. For example, instances of sshd@.service will now
6428 implicitly be placed in system-sshd.slice rather than
6429 system.slice as before.
6431 * Test coverage support may now be enabled at build time.
6433 Contributions from: Dave Reisner, Frederic Crozat, Harald
6434 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
6435 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
6436 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
6437 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
6438 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
6439 Giokas, Zbigniew Jędrzejewski-Szmek
6441 — Berlin, 2013-07-23
6445 * Two new unit types have been introduced:
6447 Scope units are very similar to service units, however, are
6448 created out of pre-existing processes — instead of PID 1
6449 forking off the processes. By using scope units it is
6450 possible for system services and applications to group their
6451 own child processes (worker processes) in a powerful way
6452 which then maybe used to organize them, or kill them
6453 together, or apply resource limits on them.
6455 Slice units may be used to partition system resources in an
6456 hierarchical fashion and then assign other units to them. By
6457 default there are now three slices: system.slice (for all
6458 system services), user.slice (for all user sessions),
6459 machine.slice (for VMs and containers).
6461 Slices and scopes have been introduced primarily in
6462 context of the work to move cgroup handling to a
6463 single-writer scheme, where only PID 1
6464 creates/removes/manages cgroups.
6466 * There's a new concept of "transient" units. In contrast to
6467 normal units these units are created via an API at runtime,
6468 not from configuration from disk. More specifically this
6469 means it is now possible to run arbitrary programs as
6470 independent services, with all execution parameters passed
6471 in via bus APIs rather than read from disk. Transient units
6472 make systemd substantially more dynamic then it ever was,
6473 and useful as a general batch manager.
6475 * logind has been updated to make use of scope and slice units
6476 for managing user sessions. As a user logs in he will get
6477 his own private slice unit, to which all sessions are added
6478 as scope units. We also added support for automatically
6479 adding an instance of user@.service for the user into the
6480 slice. Effectively logind will no longer create cgroup
6481 hierarchies on its own now, it will defer entirely to PID 1
6482 for this by means of scope, service and slice units. Since
6483 user sessions this way become entities managed by PID 1
6484 the output of "systemctl" is now a lot more comprehensive.
6486 * A new mini-daemon "systemd-machined" has been added which
6487 may be used by virtualization managers to register local
6488 VMs/containers. nspawn has been updated accordingly, and
6489 libvirt will be updated shortly. machined will collect a bit
6490 of meta information about the VMs/containers, and assign
6491 them their own scope unit (see above). The collected
6492 meta-data is then made available via the "machinectl" tool,
6493 and exposed in "ps" and similar tools. machined/machinectl
6494 is compile-time optional.
6496 * As discussed earlier, the low-level cgroup configuration
6497 options ControlGroup=, ControlGroupModify=,
6498 ControlGroupPersistent=, ControlGroupAttribute= have been
6499 removed. Please use high-level attribute settings instead as
6500 well as slice units.
6502 * A new bus call SetUnitProperties() has been added to alter
6503 various runtime parameters of a unit. This is primarily
6504 useful to alter cgroup parameters dynamically in a nice way,
6505 but will be extended later on to make more properties
6506 modifiable at runtime. systemctl gained a new set-properties
6507 command that wraps this call.
6509 * A new tool "systemd-run" has been added which can be used to
6510 run arbitrary command lines as transient services or scopes,
6511 while configuring a number of settings via the command
6512 line. This tool is currently very basic, however already
6513 very useful. We plan to extend this tool to even allow
6514 queuing of execution jobs with time triggers from the
6515 command line, similar in fashion to "at".
6517 * nspawn will now inform the user explicitly that kernels with
6518 audit enabled break containers, and suggest the user to turn
6521 * Support for detecting the IMA and AppArmor security
6522 frameworks with ConditionSecurity= has been added.
6524 * journalctl gained a new "-k" switch for showing only kernel
6525 messages, mimicking dmesg output; in addition to "--user"
6526 and "--system" switches for showing only user's own logs
6529 * systemd-delta can now show information about drop-in
6530 snippets extending unit files.
6532 * libsystemd-bus has been substantially updated but is still
6533 not available as public API.
6535 * systemd will now look for the "debug" argument on the kernel
6536 command line and enable debug logging, similar to what
6537 "systemd.log_level=debug" already did before.
6539 * "systemctl set-default", "systemctl get-default" has been
6540 added to configure the default.target symlink, which
6541 controls what to boot into by default.
6543 * "systemctl set-log-level" has been added as a convenient
6544 way to raise and lower systemd logging threshold.
6546 * "systemd-analyze plot" will now show the time the various
6547 generators needed for execution, as well as information
6548 about the unit file loading.
6550 * libsystemd-journal gained a new sd_journal_open_files() call
6551 for opening specific journal files. journactl also gained a
6552 new switch to expose this new functionality. Previously we
6553 only supported opening all files from a directory, or all
6554 files from the system, as opening individual files only is
6555 racy due to journal file rotation.
6557 * systemd gained the new DefaultEnvironment= setting in
6558 /etc/systemd/system.conf to set environment variables for
6561 * If a privileged process logs a journal message with the
6562 OBJECT_PID= field set, then journald will automatically
6563 augment this with additional OBJECT_UID=, OBJECT_GID=,
6564 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
6565 system services want to log events about specific client
6566 processes. journactl/systemctl has been updated to make use
6567 of this information if all log messages regarding a specific
6570 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
6571 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
6572 Reisner, David Coppa, David King, David Strauss, Eelco
6573 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
6574 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
6575 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
6576 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
6577 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
6578 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
6579 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
6580 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
6581 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
6582 Łukasz Stelmach, 장동준
6586 * The Python bindings gained some minimal support for the APIs
6587 exposed by libsystemd-logind.
6589 * ConditionSecurity= gained support for detecting SMACK. Since
6590 this condition already supports SELinux and AppArmor we only
6591 miss IMA for this. Patches welcome!
6593 Contributions from: Karol Lewandowski, Lennart Poettering,
6594 Zbigniew Jędrzejewski-Szmek
6598 * systemd-nspawn will now create /etc/resolv.conf if
6599 necessary, before bind-mounting the host's file onto it.
6601 * systemd-nspawn will now store meta information about a
6602 container on the container's cgroup as extended attribute
6603 fields, including the root directory.
6605 * The cgroup hierarchy has been reworked in many ways. All
6606 objects any of the components systemd creates in the cgroup
6607 tree are now suffixed. More specifically, user sessions are
6608 now placed in cgroups suffixed with ".session", users in
6609 cgroups suffixed with ".user", and nspawn containers in
6610 cgroups suffixed with ".nspawn". Furthermore, all cgroup
6611 names are now escaped in a simple scheme to avoid collision
6612 of userspace object names with kernel filenames. This work
6613 is preparation for making these objects relocatable in the
6614 cgroup tree, in order to allow easy resource partitioning of
6615 these objects without causing naming conflicts.
6617 * systemctl list-dependencies gained the new switches
6618 --plain, --reverse, --after and --before.
6620 * systemd-inhibit now shows the process name of processes that
6621 have taken an inhibitor lock.
6623 * nss-myhostname will now also resolve "localhost"
6624 implicitly. This makes /etc/hosts an optional file and
6625 nicely handles that on IPv6 ::1 maps to both "localhost" and
6628 * libsystemd-logind.so gained a new call
6629 sd_get_machine_names() to enumerate running containers and
6630 VMs (currently only supported by very new libvirt and
6631 nspawn). sd_login_monitor can now be used to watch
6632 VMs/containers coming and going.
6634 * .include is not allowed recursively anymore, and only in
6635 unit files. Usually it is better to use drop-in snippets in
6636 .d/*.conf anyway, as introduced with systemd 198.
6638 * systemd-analyze gained a new "critical-chain" command that
6639 determines the slowest chain of units run during system
6640 boot-up. It is very useful for tracking down where
6641 optimizing boot time is the most beneficial.
6643 * systemd will no longer allow manipulating service paths in
6644 the name=systemd:/system cgroup tree using ControlGroup= in
6645 units. (But is still fine with it in all other dirs.)
6647 * There's a new systemd-nspawn@.service service file that may
6648 be used to easily run nspawn containers as system
6649 services. With the container's root directory in
6650 /var/lib/container/foobar it is now sufficient to run
6651 "systemctl start systemd-nspawn@foobar.service" to boot it.
6653 * systemd-cgls gained a new parameter "--machine" to list only
6654 the processes within a certain container.
6656 * ConditionSecurity= now can check for "apparmor". We still
6657 are lacking checks for SMACK and IMA for this condition
6658 check though. Patches welcome!
6660 * A new configuration file /etc/systemd/sleep.conf has been
6661 added that may be used to configure which kernel operation
6662 systemd is supposed to execute when "suspend", "hibernate"
6663 or "hybrid-sleep" is requested. This makes the new kernel
6664 "freeze" state accessible to the user.
6666 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
6667 the passed argument if applicable.
6669 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
6670 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
6671 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
6672 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
6673 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
6674 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
6675 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
6680 * The output of 'systemctl list-jobs' got some polishing. The
6681 '--type=' argument may now be passed more than once. A new
6682 command 'systemctl list-sockets' has been added which shows
6683 a list of kernel sockets systemd is listening on with the
6684 socket units they belong to, plus the units these socket
6687 * The experimental libsystemd-bus library got substantial
6688 updates to work in conjunction with the (also experimental)
6689 kdbus kernel project. It works well enough to exchange
6690 messages with some sophistication. Note that kdbus is not
6691 ready yet, and the library is mostly an elaborate test case
6692 for now, and not installable.
6694 * systemd gained a new unit 'systemd-static-nodes.service'
6695 that generates static device nodes earlier during boot, and
6696 can run in conjunction with udev.
6698 * libsystemd-login gained a new call sd_pid_get_user_unit()
6699 to retrieve the user systemd unit a process is running
6700 in. This is useful for systems where systemd is used as
6703 * systemd-nspawn now places all containers in the new /machine
6704 top-level cgroup directory in the name=systemd
6705 hierarchy. libvirt will soon do the same, so that we get a
6706 uniform separation of /system, /user and /machine for system
6707 services, user processes and containers/virtual
6708 machines. This new cgroup hierarchy is also useful to stick
6709 stable names to specific container instances, which can be
6710 recognized later this way (this name may be controlled
6711 via systemd-nspawn's new -M switch). libsystemd-login also
6712 gained a new call sd_pid_get_machine_name() to retrieve the
6713 name of the container/VM a specific process belongs to.
6715 * bootchart can now store its data in the journal.
6717 * libsystemd-journal gained a new call
6718 sd_journal_add_conjunction() for AND expressions to the
6719 matching logic. This can be used to express more complex
6720 logical expressions.
6722 * journactl can now take multiple --unit= and --user-unit=
6725 * The cryptsetup logic now understands the "luks.key=" kernel
6726 command line switch for specifying a file to read the
6727 decryption key from. Also, if a configured key file is not
6728 found the tool will now automatically fall back to prompting
6731 * Python systemd.journal module was updated to wrap recently
6732 added functions from libsystemd-journal. The interface was
6733 changed to bring the low level interface in s.j._Reader
6734 closer to the C API, and the high level interface in
6735 s.j.Reader was updated to wrap and convert all data about
6738 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
6739 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
6740 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
6741 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
6742 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
6743 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6747 * journalctl --update-catalog now understands a new --root=
6748 option to operate on catalogs found in a different root
6751 * During shutdown after systemd has terminated all running
6752 services a final killing loop kills all remaining left-over
6753 processes. We will now print the name of these processes
6754 when we send SIGKILL to them, since this usually indicates a
6757 * If /etc/crypttab refers to password files stored on
6758 configured mount points automatic dependencies will now be
6759 generated to ensure the specific mount is established first
6760 before the key file is attempted to be read.
6762 * 'systemctl status' will now show information about the
6763 network sockets a socket unit is listening on.
6765 * 'systemctl status' will also shown information about any
6766 drop-in configuration file for units. (Drop-In configuration
6767 files in this context are files such as
6768 /etc/systemd/systemd/foobar.service.d/*.conf)
6770 * systemd-cgtop now optionally shows summed up CPU times of
6771 cgroups. Press '%' while running cgtop to switch between
6772 percentage and absolute mode. This is useful to determine
6773 which cgroups use up the most CPU time over the entire
6774 runtime of the system. systemd-cgtop has also been updated
6775 to be 'pipeable' for processing with further shell tools.
6777 * 'hostnamectl set-hostname' will now allow setting of FQDN
6780 * The formatting and parsing of time span values has been
6781 changed. The parser now understands fractional expressions
6782 such as "5.5h". The formatter will now output fractional
6783 expressions for all time spans under 1min, i.e. "5.123456s"
6784 rather than "5s 123ms 456us". For time spans under 1s
6785 millisecond values are shown, for those under 1ms
6786 microsecond values are shown. This should greatly improve
6787 all time-related output of systemd.
6789 * libsystemd-login and libsystemd-journal gained new
6790 functions for querying the poll() events mask and poll()
6791 timeout value for integration into arbitrary event
6794 * localectl gained the ability to list available X11 keymaps
6795 (models, layouts, variants, options).
6797 * 'systemd-analyze dot' gained the ability to filter for
6798 specific units via shell-style globs, to create smaller,
6799 more useful graphs. I.e. it is now possible to create simple
6800 graphs of all the dependencies between only target units, or
6801 of all units that Avahi has dependencies with.
6803 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
6804 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
6805 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
6806 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
6807 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
6808 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
6809 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
6813 * The boot-time readahead implementation for rotating media
6814 will now read the read-ahead data in multiple passes which
6815 consist of all read requests made in equidistant time
6816 intervals. This means instead of strictly reading read-ahead
6817 data in its physical order on disk we now try to find a
6818 middle ground between physical and access time order.
6820 * /etc/os-release files gained a new BUILD_ID= field for usage
6821 on operating systems that provide continuous builds of OS
6824 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
6825 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
6826 William Douglas, Zbigniew Jędrzejewski-Szmek
6830 * systemd-python gained an API exposing libsystemd-daemon.
6832 * The SMACK setup logic gained support for uploading CIPSO
6835 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
6836 ReadOnlyDirectories= and InaccessibleDirectories= has
6837 changed. The private /tmp and /var/tmp directories are now
6838 shared by all processes of a service (which means
6839 ExecStartPre= may now leave data in /tmp that ExecStart= of
6840 the same service can still access). When a service is
6841 stopped its temporary directories are immediately deleted
6842 (normal clean-up with tmpfiles is still done in addition to
6845 * By default, systemd will now set a couple of sysctl
6846 variables in the kernel: the safe sysrq options are turned
6847 on, IP route verification is turned on, and source routing
6848 disabled. The recently added hardlink and softlink
6849 protection of the kernel is turned on. These settings should
6850 be reasonably safe, and good defaults for all new systems.
6852 * The predictable network naming logic may now be turned off
6853 with a new kernel command line switch: net.ifnames=0.
6855 * A new libsystemd-bus module has been added that implements a
6856 pretty complete D-Bus client library. For details see:
6858 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
6860 * journald will now explicitly flush the journal files to disk
6861 at the latest 5min after each write. The file will then also
6862 be marked offline until the next write. This should increase
6863 reliability in case of a crash. The synchronization delay
6864 can be configured via SyncIntervalSec= in journald.conf.
6866 * There's a new remote-fs-setup.target unit that can be used
6867 to pull in specific services when at least one remote file
6868 system is to be mounted.
6870 * There are new targets timers.target and paths.target as
6871 canonical targets to pull user timer and path units in
6872 from. This complements sockets.target with a similar
6873 purpose for socket units.
6875 * libudev gained a new call udev_device_set_attribute_value()
6876 to set sysfs attributes of a device.
6878 * The udev daemon now sets the default number of worker
6879 processes executed in parallel based on the number of available
6880 CPUs instead of the amount of available RAM. This is supposed
6881 to provide a more reliable default and limit a too aggressive
6882 parallelism for setups with 1000s of devices connected.
6884 Contributions from: Auke Kok, Colin Walters, Cristian
6885 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
6886 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
6887 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
6888 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
6889 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
6890 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
6891 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
6892 Zbigniew Jędrzejewski-Szmek
6896 * Configuration of unit files may now be extended via drop-in
6897 files without having to edit/override the unit files
6898 themselves. More specifically, if the administrator wants to
6899 change one value for a service file foobar.service he can
6900 now do so by dropping in a configuration snippet into
6901 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
6902 will load all these snippets and apply them on top of the
6903 main unit configuration file, possibly extending or
6904 overriding its settings. Using these drop-in snippets is
6905 generally nicer than the two earlier options for changing
6906 unit files locally: copying the files from
6907 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
6908 them there; or creating a new file in /etc/systemd/system/
6909 that incorporates the original one via ".include". Drop-in
6910 snippets into these .d/ directories can be placed in any
6911 directory systemd looks for units in, and the usual
6912 overriding semantics between /usr/lib, /etc and /run apply
6915 * Most unit file settings which take lists of items can now be
6916 reset by assigning the empty string to them. For example,
6917 normally, settings such as Environment=FOO=BAR append a new
6918 environment variable assignment to the environment block,
6919 each time they are used. By assigning Environment= the empty
6920 string the environment block can be reset to empty. This is
6921 particularly useful with the .d/*.conf drop-in snippets
6922 mentioned above, since this adds the ability to reset list
6923 settings from vendor unit files via these drop-ins.
6925 * systemctl gained a new "list-dependencies" command for
6926 listing the dependencies of a unit recursively.
6928 * Inhibitors are now honored and listed by "systemctl
6929 suspend", "systemctl poweroff" (and similar) too, not only
6930 GNOME. These commands will also list active sessions by
6933 * Resource limits (as exposed by the various control group
6934 controllers) can now be controlled dynamically at runtime
6935 for all units. More specifically, you can now use a command
6936 like "systemctl set-cgroup-attr foobar.service cpu.shares
6937 2000" to alter the CPU shares a specific service gets. These
6938 settings are stored persistently on disk, and thus allow the
6939 administrator to easily adjust the resource usage of
6940 services with a few simple commands. This dynamic resource
6941 management logic is also available to other programs via the
6942 bus. Almost any kernel cgroup attribute and controller is
6945 * systemd-vconsole-setup will now copy all font settings to
6946 all allocated VTs, where it previously applied them only to
6949 * libsystemd-login gained the new sd_session_get_tty() API
6952 * This release drops support for a few legacy or
6953 distribution-specific LSB facility names when parsing init
6954 scripts: $x-display-manager, $mail-transfer-agent,
6955 $mail-transport-agent, $mail-transfer-agent, $smtp,
6956 $null. Also, the mail-transfer-agent.target unit backing
6957 this has been removed. Distributions which want to retain
6958 compatibility with this should carry the burden for
6959 supporting this themselves and patch support for these back
6960 in, if they really need to. Also, the facilities $syslog and
6961 $local_fs are now ignored, since systemd does not support
6962 early-boot LSB init scripts anymore, and these facilities
6963 are implied anyway for normal services. syslog.target has
6966 * There are new bus calls on PID1's Manager object for
6967 cancelling jobs, and removing snapshot units. Previously,
6968 both calls were only available on the Job and Snapshot
6971 * systemd-journal-gatewayd gained SSL support.
6973 * The various "environment" files, such as /etc/locale.conf
6974 now support continuation lines with a backslash ("\") as
6975 last character in the line, similarly in style (but different)
6976 to how this is supported in shells.
6978 * For normal user processes the _SYSTEMD_USER_UNIT= field is
6979 now implicitly appended to every log entry logged. systemctl
6980 has been updated to filter by this field when operating on a
6981 user systemd instance.
6983 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
6984 CAP_AUDIT_CONTROL capabilities to the capabilities set for
6985 the container. This makes it easier to boot unmodified
6986 Fedora systems in a container, which however still requires
6987 audit=0 to be passed on the kernel command line. Auditing in
6988 kernel and userspace is unfortunately still too broken in
6989 context of containers, hence we recommend compiling it out
6990 of the kernel or using audit=0. Hopefully this will be fixed
6991 one day for good in the kernel.
6993 * nspawn gained the new --bind= and --bind-ro= parameters to
6994 bind mount specific directories from the host into the
6997 * nspawn will now mount its own devpts file system instance
6998 into the container, in order not to leak pty devices from
6999 the host into the container.
7001 * systemd will now read the firmware boot time performance
7002 information from the EFI variables, if the used boot loader
7003 supports this, and takes it into account for boot performance
7004 analysis via "systemd-analyze". This is currently supported
7005 only in conjunction with Gummiboot, but could be supported
7006 by other boot loaders too. For details see:
7008 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
7010 * A new generator has been added that automatically mounts the
7011 EFI System Partition (ESP) to /boot, if that directory
7012 exists, is empty, and no other file system has been
7013 configured to be mounted there.
7015 * logind will now send out PrepareForSleep(false) out
7016 unconditionally, after coming back from suspend. This may be
7017 used by applications as asynchronous notification for
7018 system resume events.
7020 * "systemctl unlock-sessions" has been added, that allows
7021 unlocking the screens of all user sessions at once, similar
7022 to how "systemctl lock-sessions" already locked all users
7023 sessions. This is backed by a new D-Bus call UnlockSessions().
7025 * "loginctl seat-status" will now show the master device of a
7026 seat. (i.e. the device of a seat that needs to be around for
7027 the seat to be considered available, usually the graphics
7030 * tmpfiles gained a new "X" line type, that allows
7031 configuration of files and directories (with wildcards) that
7032 shall be excluded from automatic cleanup ("aging").
7034 * udev default rules set the device node permissions now only
7035 at "add" events, and do not change them any longer with a
7036 later "change" event.
7038 * The log messages for lid events and power/sleep keypresses
7039 now carry a message ID.
7041 * We now have a substantially larger unit test suite, but this
7042 continues to be work in progress.
7044 * udevadm hwdb gained a new --root= parameter to change the
7045 root directory to operate relative to.
7047 * logind will now issue a background sync() request to the kernel
7048 early at shutdown, so that dirty buffers are flushed to disk early
7049 instead of at the last moment, in order to optimize shutdown
7052 * A new bootctl tool has been added that is an interface for
7053 certain boot loader operations. This is currently a preview
7054 and is likely to be extended into a small mechanism daemon
7055 like timedated, localed, hostnamed, and can be used by
7056 graphical UIs to enumerate available boot options, and
7057 request boot into firmware operations.
7059 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
7060 the rest of the package. It also has been updated to work
7061 correctly in initrds.
7063 * Policykit previously has been runtime optional, and is now
7064 also compile time optional via a configure switch.
7066 * systemd-analyze has been reimplemented in C. Also "systemctl
7067 dot" has moved into systemd-analyze.
7069 * "systemctl status" with no further parameters will now print
7070 the status of all active or failed units.
7072 * Operations such as "systemctl start" can now be executed
7073 with a new mode "--irreversible" which may be used to queue
7074 operations that cannot accidentally be reversed by a later
7075 job queuing. This is by default used to make shutdown
7076 requests more robust.
7078 * The Python API of systemd now gained a new module for
7079 reading journal files.
7081 * A new tool kernel-install has been added that can install
7082 kernel images according to the Boot Loader Specification:
7084 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
7086 * Boot time console output has been improved to provide
7087 animated boot time output for hanging jobs.
7089 * A new tool systemd-activate has been added which can be used
7090 to test socket activation with, directly from the command
7091 line. This should make it much easier to test and debug
7092 socket activation in daemons.
7094 * journalctl gained a new "--reverse" (or -r) option to show
7095 journal output in reverse order (i.e. newest line first).
7097 * journalctl gained a new "--pager-end" (or -e) option to jump
7098 to immediately jump to the end of the journal in the
7099 pager. This is only supported in conjunction with "less".
7101 * journalctl gained a new "--user-unit=" option, that works
7102 similarly to "--unit=" but filters for user units rather than
7105 * A number of unit files to ease adoption of systemd in
7106 initrds has been added. This moves some minimal logic from
7107 the various initrd implementations into systemd proper.
7109 * The journal files are now owned by a new group
7110 "systemd-journal", which exists specifically to allow access
7111 to the journal, and nothing else. Previously, we used the
7112 "adm" group for that, which however possibly covers more
7113 than just journal/log file access. This new group is now
7114 already used by systemd-journal-gatewayd to ensure this
7115 daemon gets access to the journal files and as little else
7116 as possible. Note that "make install" will also set FS ACLs
7117 up for /var/log/journal to give "adm" and "wheel" read
7118 access to it, in addition to "systemd-journal" which owns
7119 the journal files. We recommend that packaging scripts also
7120 add read access to "adm" + "wheel" to /var/log/journal, and
7121 all existing/future journal files. To normal users and
7122 administrators little changes, however packagers need to
7123 ensure to create the "systemd-journal" system group at
7124 package installation time.
7126 * The systemd-journal-gatewayd now runs as unprivileged user
7127 systemd-journal-gateway:systemd-journal-gateway. Packaging
7128 scripts need to create these system user/group at
7131 * timedated now exposes a new boolean property CanNTP that
7132 indicates whether a local NTP service is available or not.
7134 * systemd-detect-virt will now also detect xen PVs
7136 * The pstore file system is now mounted by default, if it is
7139 * In addition to the SELinux and IMA policies we will now also
7140 load SMACK policies at early boot.
7142 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
7143 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
7144 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
7145 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
7146 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
7147 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
7148 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
7149 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
7150 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
7151 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
7152 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
7153 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
7154 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
7155 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
7159 * Timer units now support calendar time events in addition to
7160 monotonic time events. That means you can now trigger a unit
7161 based on a calendar time specification such as "Thu,Fri
7162 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
7163 or fifth day of any month of the year 2013, given that it is
7164 a thursday or friday. This brings timer event support
7165 considerably closer to cron's capabilities. For details on
7166 the supported calendar time specification language see
7169 * udev now supports a number of different naming policies for
7170 network interfaces for predictable names, and a combination
7171 of these policies is now the default. Please see this wiki
7172 document for details:
7174 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
7176 * Auke Kok's bootchart implementation has been added to the
7177 systemd tree. It is an optional component that can graph the
7178 boot in quite some detail. It is one of the best bootchart
7179 implementations around and minimal in its code and
7182 * nss-myhostname has been integrated into the systemd source
7183 tree. nss-myhostname guarantees that the local hostname
7184 always stays resolvable via NSS. It has been a weak
7185 requirement of systemd-hostnamed since a long time, and
7186 since its code is actually trivial we decided to just
7187 include it in systemd's source tree. It can be turned off
7188 with a configure switch.
7190 * The read-ahead logic is now capable of properly detecting
7191 whether a btrfs file system is on SSD or rotating media, in
7192 order to optimize the read-ahead scheme. Previously, it was
7193 only capable of detecting this on traditional file systems
7196 * In udev, additional device properties are now read from the
7197 IAB in addition to the OUI database. Also, Bluetooth company
7198 identities are attached to the devices as well.
7200 * In service files %U may be used as specifier that is
7201 replaced by the configured user name of the service.
7203 * nspawn may now be invoked without a controlling TTY. This
7204 makes it suitable for invocation as its own service. This
7205 may be used to set up a simple containerized server system
7206 using only core OS tools.
7208 * systemd and nspawn can now accept socket file descriptors
7209 when they are started for socket activation. This enables
7210 implementation of socket activated nspawn
7211 containers. i.e. think about autospawning an entire OS image
7212 when the first SSH or HTTP connection is received. We expect
7213 that similar functionality will also be added to libvirt-lxc
7216 * journalctl will now suppress ANSI color codes when
7217 presenting log data.
7219 * systemctl will no longer show control group information for
7220 a unit if the control group is empty anyway.
7222 * logind can now automatically suspend/hibernate/shutdown the
7225 * /etc/machine-info and hostnamed now also expose the chassis
7226 type of the system. This can be used to determine whether
7227 the local system is a laptop, desktop, handset or
7228 tablet. This information may either be configured by the
7229 user/vendor or is automatically determined from ACPI and DMI
7230 information if possible.
7232 * A number of PolicyKit actions are now bound together with
7233 "imply" rules. This should simplify creating UIs because
7234 many actions will now authenticate similar ones as well.
7236 * Unit files learnt a new condition ConditionACPower= which
7237 may be used to conditionalize a unit depending on whether an
7238 AC power source is connected or not, of whether the system
7239 is running on battery power.
7241 * systemctl gained a new "is-failed" verb that may be used in
7242 shell scripts and suchlike to check whether a specific unit
7243 is in the "failed" state.
7245 * The EnvironmentFile= setting in unit files now supports file
7246 globbing, and can hence be used to easily read a number of
7247 environment files at once.
7249 * systemd will no longer detect and recognize specific
7250 distributions. All distribution-specific #ifdeffery has been
7251 removed, systemd is now fully generic and
7252 distribution-agnostic. Effectively, not too much is lost as
7253 a lot of the code is still accessible via explicit configure
7254 switches. However, support for some distribution specific
7255 legacy configuration file formats has been dropped. We
7256 recommend distributions to simply adopt the configuration
7257 files everybody else uses now and convert the old
7258 configuration from packaging scripts. Most distributions
7259 already did that. If that's not possible or desirable,
7260 distributions are welcome to forward port the specific
7261 pieces of code locally from the git history.
7263 * When logging a message about a unit systemd will now always
7264 log the unit name in the message meta data.
7266 * localectl will now also discover system locale data that is
7267 not stored in locale archives, but directly unpacked.
7269 * logind will no longer unconditionally use framebuffer
7270 devices as seat masters, i.e. as devices that are required
7271 to be existing before a seat is considered preset. Instead,
7272 it will now look for all devices that are tagged as
7273 "seat-master" in udev. By default, framebuffer devices will
7274 be marked as such, but depending on local systems, other
7275 devices might be marked as well. This may be used to
7276 integrate graphics cards using closed source drivers (such
7277 as NVidia ones) more nicely into logind. Note however, that
7278 we recommend using the open source NVidia drivers instead,
7279 and no udev rules for the closed-source drivers will be
7280 shipped from us upstream.
7282 Contributions from: Adam Williamson, Alessandro Crismani, Auke
7283 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
7284 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
7285 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
7286 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
7287 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
7288 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
7289 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
7290 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
7291 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
7292 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
7293 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
7298 * udev gained support for loading additional device properties
7299 from an indexed database that is keyed by vendor/product IDs
7300 and similar device identifiers. For the beginning this
7301 "hwdb" is populated with data from the well-known PCI and
7302 USB database, but also includes PNP, ACPI and OID data. In
7303 the longer run this indexed database shall grow into
7304 becoming the one central database for non-essential
7305 userspace device metadata. Previously, data from the PCI/USB
7306 database was only attached to select devices, since the
7307 lookup was a relatively expensive operation due to O(n) time
7308 complexity (with n being the number of entries in the
7309 database). Since this is now O(1), we decided to add in this
7310 data for all devices where this is available, by
7311 default. Note that the indexed database needs to be rebuilt
7312 when new data files are installed. To achieve this you need
7313 to update your packaging scripts to invoke "udevadm hwdb
7314 --update" after installation of hwdb data files. For
7315 RPM-based distributions we introduced the new
7316 %udev_hwdb_update macro for this purpose.
7318 * The Journal gained support for the "Message Catalog", an
7319 indexed database to link up additional information with
7320 journal entries. For further details please check:
7322 https://www.freedesktop.org/wiki/Software/systemd/catalog
7324 The indexed message catalog database also needs to be
7325 rebuilt after installation of message catalog files. Use
7326 "journalctl --update-catalog" for this. For RPM-based
7327 distributions we introduced the %journal_catalog_update
7328 macro for this purpose.
7330 * The Python Journal bindings gained support for the standard
7331 Python logging framework.
7333 * The Journal API gained new functions for checking whether
7334 the underlying file system of a journal file is capable of
7335 properly reporting file change notifications, or whether
7336 applications that want to reflect journal changes "live"
7337 need to recheck journal files continuously in appropriate
7340 * It is now possible to set the "age" field for tmpfiles
7341 entries to 0, indicating that files matching this entry
7342 shall always be removed when the directories are cleaned up.
7344 * coredumpctl gained a new "gdb" verb which invokes gdb
7345 right-away on the selected coredump.
7347 * There's now support for "hybrid sleep" on kernels that
7348 support this, in addition to "suspend" and "hibernate". Use
7349 "systemctl hybrid-sleep" to make use of this.
7351 * logind's HandleSuspendKey= setting (and related settings)
7352 now gained support for a new "lock" setting to simply
7353 request the screen lock on all local sessions, instead of
7354 actually executing a suspend or hibernation.
7356 * systemd will now mount the EFI variables file system by
7359 * Socket units now gained support for configuration of the
7360 SMACK security label.
7362 * timedatectl will now output the time of the last and next
7363 daylight saving change.
7365 * We dropped support for various legacy and distro-specific
7366 concepts, such as insserv, early-boot SysV services
7367 (i.e. those for non-standard runlevels such as 'b' or 'S')
7368 or ArchLinux /etc/rc.conf support. We recommend the
7369 distributions who still need support this to either continue
7370 to maintain the necessary patches downstream, or find a
7371 different solution. (Talk to us if you have questions!)
7373 * Various systemd components will now bypass PolicyKit checks
7374 for root and otherwise handle properly if PolicyKit is not
7375 found to be around. This should fix most issues for
7376 PolicyKit-less systems. Quite frankly this should have been
7377 this way since day one. It is absolutely our intention to
7378 make systemd work fine on PolicyKit-less systems, and we
7379 consider it a bug if something does not work as it should if
7380 PolicyKit is not around.
7382 * For embedded systems it is now possible to build udev and
7383 systemd without blkid and/or kmod support.
7385 * "systemctl switch-root" is now capable of switching root
7386 more than once. I.e. in addition to transitions from the
7387 initrd to the host OS it is now possible to transition to
7388 further OS images from the host. This is useful to implement
7389 offline updating tools.
7391 * Various other additions have been made to the RPM macros
7392 shipped with systemd. Use %udev_rules_update() after
7393 installing new udev rules files. %_udevhwdbdir,
7394 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
7395 %_sysctldir are now available which resolve to the right
7396 directories for packages to place various data files in.
7398 * journalctl gained the new --full switch (in addition to
7399 --all, to disable ellipsation for long messages.
7401 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
7402 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
7403 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
7404 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
7405 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
7406 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
7407 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
7408 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
7409 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
7413 * journalctl gained new --since= and --until= switches to
7414 filter by time. It also now supports nice filtering for
7415 units via --unit=/-u.
7417 * Type=oneshot services may use ExecReload= and do the
7420 * The journal daemon now supports time-based rotation and
7421 vacuuming, in addition to the usual disk-space based
7424 * The journal will now index the available field values for
7425 each field name. This enables clients to show pretty drop
7426 downs of available match values when filtering. The bash
7427 completion of journalctl has been updated
7428 accordingly. journalctl gained a new switch -F to list all
7429 values a certain field takes in the journal database.
7431 * More service events are now written as structured messages
7432 to the journal, and made recognizable via message IDs.
7434 * The timedated, localed and hostnamed mini-services which
7435 previously only provided support for changing time, locale
7436 and hostname settings from graphical DEs such as GNOME now
7437 also have a minimal (but very useful) text-based client
7438 utility each. This is probably the nicest way to changing
7439 these settings from the command line now, especially since
7440 it lists available options and is fully integrated with bash
7443 * There's now a new tool "systemd-coredumpctl" to list and
7444 extract coredumps from the journal.
7446 * We now install a README each in /var/log/ and
7447 /etc/rc.d/init.d explaining where the system logs and init
7448 scripts went. This hopefully should help folks who go to
7449 that dirs and look into the otherwise now empty void and
7450 scratch their heads.
7452 * When user-services are invoked (by systemd --user) the
7453 $MANAGERPID env var is set to the PID of systemd.
7455 * SIGRTMIN+24 when sent to a --user instance will now result
7456 in immediate termination of systemd.
7458 * gatewayd received numerous feature additions such as a
7459 "follow" mode, for live syncing and filtering.
7461 * browse.html now allows filtering and showing detailed
7462 information on specific entries. Keyboard navigation and
7463 mouse screen support has been added.
7465 * gatewayd/journalctl now supports HTML5/JSON
7466 Server-Sent-Events as output.
7468 * The SysV init script compatibility logic will now
7469 heuristically determine whether a script supports the
7470 "reload" verb, and only then make this available as
7473 * "systemctl status --follow" has been removed, use "journalctl
7476 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
7477 have been removed since they are hardly useful to be
7480 * And I'd like to take the opportunity to specifically mention
7481 Zbigniew for his great contributions. Zbigniew, you rock!
7483 Contributions from: Andrew Eikum, Christian Hesse, Colin
7484 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
7485 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
7486 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
7487 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
7488 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
7489 Jędrzejewski-Szmek, Сковорода Никита Андреевич
7493 * If /etc/vconsole.conf is non-existent or empty we will no
7494 longer load any console font or key map at boot by
7495 default. Instead the kernel defaults will be left
7496 intact. This is definitely the right thing to do, as no
7497 configuration should mean no configuration, and hard-coding
7498 font names that are different on all archs is probably a bad
7499 idea. Also, the kernel default key map and font should be
7500 good enough for most cases anyway, and mostly identical to
7501 the userspace fonts/key maps we previously overloaded them
7502 with. If distributions want to continue to default to a
7503 non-kernel font or key map they should ship a default
7504 /etc/vconsole.conf with the appropriate contents.
7506 Contributions from: Colin Walters, Daniel J Walsh, Dave
7507 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
7508 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
7512 * journalctl gained a new --cursor= switch to show entries
7513 starting from the specified location in the journal.
7515 * We now enforce a size limit on journal entry fields exported
7516 with "-o json" in journalctl. Fields larger than 4K will be
7517 assigned null. This can be turned off with --all.
7519 * An (optional) journal gateway daemon is now available as
7520 "systemd-journal-gatewayd.service". This service provides
7521 access to the journal via HTTP and JSON. This functionality
7522 will be used to implement live log synchronization in both
7523 pull and push modes, but has various other users too, such
7524 as easy log access for debugging of embedded devices. Right
7525 now it is already useful to retrieve the journal via HTTP:
7527 # systemctl start systemd-journal-gatewayd.service
7528 # wget http://localhost:19531/entries
7530 This will download the journal contents in a
7531 /var/log/messages compatible format. The same as JSON:
7533 # curl -H"Accept: application/json" http://localhost:19531/entries
7535 This service is also accessible via a web browser where a
7536 single static HTML5 app is served that uses the JSON logic
7537 to enable the user to do some basic browsing of the
7538 journal. This will be extended later on. Here's an example
7539 screenshot of this app in its current state:
7541 http://0pointer.de/public/journal-gatewayd
7543 Contributions from: Kay Sievers, Lennart Poettering, Robert
7544 Milasan, Tom Gundersen
7548 * The bash completion logic is now available for journalctl
7551 * We do not mount the "cpuset" controller anymore together with
7552 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
7553 started if no parameters are assigned to it. "cpuset" hence
7554 broke code that assumed it could create "cpu" groups and
7557 * journalctl -f will now subscribe to terminal size changes,
7558 and line break accordingly.
7560 Contributions from: Dave Reisner, Kay Sievers, Lennart
7561 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
7565 * nspawn will now create a symlink /etc/localtime in the
7566 container environment, copying the host's timezone
7567 setting. Previously this has been done via a bind mount, but
7568 since symlinks cannot be bind mounted this has now been
7569 changed to create/update the appropriate symlink.
7571 * journalctl -n's line number argument is now optional, and
7572 will default to 10 if omitted.
7574 * journald will now log the maximum size the journal files may
7575 take up on disk. This is particularly useful if the default
7576 built-in logic of determining this parameter from the file
7577 system size is used. Use "systemctl status
7578 systemd-journald.service" to see this information.
7580 * The multi-seat X wrapper tool has been stripped down. As X
7581 is now capable of enumerating graphics devices via udev in a
7582 seat-aware way the wrapper is not strictly necessary
7583 anymore. A stripped down temporary stop-gap is still shipped
7584 until the upstream display managers have been updated to
7585 fully support the new X logic. Expect this wrapper to be
7586 removed entirely in one of the next releases.
7588 * HandleSleepKey= in logind.conf has been split up into
7589 HandleSuspendKey= and HandleHibernateKey=. The old setting
7590 is not available anymore. X11 and the kernel are
7591 distinguishing between these keys and we should too. This
7592 also means the inhibition lock for these keys has been split
7595 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
7596 Poettering, Lukas Nykryn, Václav Pavlín
7600 * Whenever a unit changes state we will now log this to the
7601 journal and show along the unit's own log output in
7604 * ConditionPathIsMountPoint= can now properly detect bind
7605 mount points too. (Previously, a bind mount of one file
7606 system to another place in the same file system could not be
7607 detected as mount, since they shared struct stat's st_dev
7610 * We will now mount the cgroup controllers cpu, cpuacct,
7611 cpuset and the controllers net_cls, net_prio together by
7614 * nspawn containers will now have a virtualized boot
7615 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
7616 over with a randomized ID at container initialization). This
7617 has the effect of making "journalctl -b" do the right thing
7620 * The JSON output journal serialization has been updated not
7621 to generate "endless" list objects anymore, but rather one
7622 JSON object per line. This is more in line how most JSON
7623 parsers expect JSON objects. The new output mode
7624 "json-pretty" has been added to provide similar output, but
7625 neatly aligned for readability by humans.
7627 * We dropped all explicit sync() invocations in the shutdown
7628 code. The kernel does this implicitly anyway in the kernel
7629 reboot() syscall. halt(8)'s -n option is now a compatibility
7632 * We now support virtualized reboot() in containers, as
7633 supported by newer kernels. We will fall back to exit() if
7634 CAP_SYS_REBOOT is not available to the container. Also,
7635 nspawn makes use of this now and will actually reboot the
7636 container if the containerized OS asks for that.
7638 * journalctl will only show local log output by default
7639 now. Use --merge (-m) to show remote log output, too.
7641 * libsystemd-journal gained the new sd_journal_get_usage()
7642 call to determine the current disk usage of all journal
7643 files. This is exposed in the new "journalctl --disk-usage"
7646 * journald gained a new configuration setting SplitMode= in
7647 journald.conf which may be used to control how user journals
7648 are split off. See journald.conf(5) for details.
7650 * A new condition type ConditionFileNotEmpty= has been added.
7652 * tmpfiles' "w" lines now support file globbing, to write
7653 multiple files at once.
7655 * We added Python bindings for the journal submission
7656 APIs. More Python APIs for a number of selected APIs will
7657 likely follow. Note that we intend to add native bindings
7658 only for the Python language, as we consider it common
7659 enough to deserve bindings shipped within systemd. There are
7660 various projects outside of systemd that provide bindings
7661 for languages such as PHP or Lua.
7663 * Many conditions will now resolve specifiers such as %i. In
7664 addition, PathChanged= and related directives of .path units
7665 now support specifiers as well.
7667 * There's now a new RPM macro definition for the system preset
7670 * journald will now warn if it ca not forward a message to the
7671 syslog daemon because its socket is full.
7673 * timedated will no longer write or process /etc/timezone,
7674 except on Debian. As we do not support late mounted /usr
7675 anymore /etc/localtime always being a symlink is now safe,
7676 and hence the information in /etc/timezone is not necessary
7679 * logind will now always reserve one VT for a text getty (VT6
7680 by default). Previously if more than 6 X sessions where
7681 started they took up all the VTs with auto-spawned gettys,
7682 so that no text gettys were available anymore.
7684 * udev will now automatically inform the btrfs kernel logic
7685 about btrfs RAID components showing up. This should make
7686 simple hotplug based btrfs RAID assembly work.
7688 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
7689 (but not for its children which will stay at the kernel
7690 default). This should allow setups with a lot more listening
7693 * systemd will now always pass the configured timezone to the
7694 kernel at boot. timedated will do the same when the timezone
7697 * logind's inhibition logic has been updated. By default,
7698 logind will now handle the lid switch, the power and sleep
7699 keys all the time, even in graphical sessions. If DEs want
7700 to handle these events on their own they should take the new
7701 handle-power-key, handle-sleep-key and handle-lid-switch
7702 inhibitors during their runtime. A simple way to achieve
7703 that is to invoke the DE wrapped in an invocation of:
7705 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
7707 * Access to unit operations is now checked via SELinux taking
7708 the unit file label and client process label into account.
7710 * systemd will now notify the administrator in the journal
7711 when he over-mounts a non-empty directory.
7713 * There are new specifiers that are resolved in unit files,
7714 for the host name (%H), the machine ID (%m) and the boot ID
7717 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
7718 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
7719 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
7720 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
7721 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
7722 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
7723 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
7727 * Support for reading structured kernel messages from
7728 /dev/kmsg has now been added and is enabled by default.
7730 * Support for reading kernel messages from /proc/kmsg has now
7731 been removed. If you want kernel messages in the journal
7732 make sure to run a recent kernel (>= 3.5) that supports
7733 reading structured messages from /dev/kmsg (see
7734 above). /proc/kmsg is now exclusive property of classic
7735 syslog daemons again.
7737 * The libudev API gained the new
7738 udev_device_new_from_device_id() call.
7740 * The logic for file system namespace (ReadOnlyDirectory=,
7741 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
7742 require pivot_root() anymore. This means fewer temporary
7743 directories are created below /tmp for this feature.
7745 * nspawn containers will now see and receive all submounts
7746 made on the host OS below the root file system of the
7749 * Forward Secure Sealing is now supported for Journal files,
7750 which provide cryptographical sealing of journal files so
7751 that attackers cannot alter log history anymore without this
7752 being detectable. Lennart will soon post a blog story about
7753 this explaining it in more detail.
7755 * There are two new service settings RestartPreventExitStatus=
7756 and SuccessExitStatus= which allow configuration of exit
7757 status (exit code or signal) which will be excepted from the
7758 restart logic, resp. consider successful.
7760 * journalctl gained the new --verify switch that can be used
7761 to check the integrity of the structure of journal files and
7762 (if Forward Secure Sealing is enabled) the contents of
7765 * nspawn containers will now be run with /dev/stdin, /dev/fd/
7766 and similar symlinks pre-created. This makes running shells
7767 as container init process a lot more fun.
7769 * The fstab support can now handle PARTUUID= and PARTLABEL=
7772 * A new ConditionHost= condition has been added to match
7773 against the hostname (with globs) and machine ID. This is
7774 useful for clusters where a single OS image is used to
7775 provision a large number of hosts which shall run slightly
7776 different sets of services.
7778 * Services which hit the restart limit will now be placed in a
7781 Contributions from: Bertram Poettering, Dave Reisner, Huang
7782 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
7783 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
7787 * When running in --user mode systemd will now become a
7788 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
7789 tree a lot more organized.
7791 * A new PartOf= unit dependency type has been introduced that
7792 may be used to group services in a natural way.
7794 * "systemctl enable" may now be used to enable instances of
7797 * journalctl now prints error log levels in red, and
7798 warning/notice log levels in bright white. It also supports
7799 filtering by log level now.
7801 * cgtop gained a new -n switch (similar to top), to configure
7802 the maximum number of iterations to run for. It also gained
7803 -b, to run in batch mode (accepting no input).
7805 * The suffix ".service" may now be omitted on most systemctl
7806 command lines involving service unit names.
7808 * There's a new bus call in logind to lock all sessions, as
7809 well as a loginctl verb for it "lock-sessions".
7811 * libsystemd-logind.so gained a new call sd_journal_perror()
7812 that works similar to libc perror() but logs to the journal
7813 and encodes structured information about the error number.
7815 * /etc/crypttab entries now understand the new keyfile-size=
7818 * shutdown(8) now can send a (configurable) wall message when
7819 a shutdown is cancelled.
7821 * The mount propagation mode for the root file system will now
7822 default to "shared", which is useful to make containers work
7823 nicely out-of-the-box so that they receive new mounts from
7824 the host. This can be undone locally by running "mount
7825 --make-rprivate /" if needed.
7827 * The prefdm.service file has been removed. Distributions
7828 should maintain this unit downstream if they intend to keep
7829 it around. However, we recommend writing normal unit files
7830 for display managers instead.
7832 * Since systemd is a crucial part of the OS we will now
7833 default to a number of compiler switches that improve
7834 security (hardening) such as read-only relocations, stack
7835 protection, and suchlike.
7837 * The TimeoutSec= setting for services is now split into
7838 TimeoutStartSec= and TimeoutStopSec= to allow configuration
7839 of individual time outs for the start and the stop phase of
7842 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
7843 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
7844 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
7845 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
7846 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
7847 Gundersen, Zbigniew Jędrzejewski-Szmek
7851 * The journal and id128 C APIs are now fully documented as man
7854 * Extra safety checks have been added when transitioning from
7855 the initial RAM disk to the main system to avoid accidental
7858 * /etc/crypttab entries now understand the new keyfile-offset=
7861 * systemctl -t can now be used to filter by unit load state.
7863 * The journal C API gained the new sd_journal_wait() call to
7864 make writing synchronous journal clients easier.
7866 * journalctl gained the new -D switch to show journals from a
7869 * journalctl now displays a special marker between log
7870 messages of two different boots.
7872 * The journal is now explicitly flushed to /var via a service
7873 systemd-journal-flush.service, rather than implicitly simply
7874 by seeing /var/log/journal to be writable.
7876 * journalctl (and the journal C APIs) can now match for much
7877 more complex expressions, with alternatives and
7880 * When transitioning from the initial RAM disk to the main
7881 system we will now kill all processes in a killing spree to
7882 ensure no processes stay around by accident.
7884 * Three new specifiers may be used in unit files: %u, %h, %s
7885 resolve to the user name, user home directory resp. user
7886 shell. This is useful for running systemd user instances.
7888 * We now automatically rotate journal files if their data
7889 object hash table gets a fill level > 75%. We also size the
7890 hash table based on the configured maximum file size. This
7891 together should lower hash collisions drastically and thus
7892 speed things up a bit.
7894 * journalctl gained the new "--header" switch to introspect
7895 header data of journal files.
7897 * A new setting SystemCallFilters= has been added to services
7898 which may be used to apply blacklists or whitelists to
7899 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
7901 * nspawn gained a new --link-journal= switch (and quicker: -j)
7902 to link the container journal with the host. This makes it
7903 very easy to centralize log viewing on the host for all
7904 guests while still keeping the journal files separated.
7906 * Many bugfixes and optimizations
7908 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
7909 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
7910 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
7915 * Several tools now understand kernel command line arguments,
7916 which are only read when run in an initial RAM disk. They
7917 usually follow closely their normal counterparts, but are
7920 * There's a new tool to analyze the readahead files that are
7921 automatically generated at boot. Use:
7923 /usr/lib/systemd/systemd-readahead analyze /.readahead
7925 * We now provide an early debug shell on tty9 if this enabled. Use:
7927 systemctl enable debug-shell.service
7929 * All plymouth related units have been moved into the Plymouth
7930 package. Please make sure to upgrade your Plymouth version
7933 * systemd-tmpfiles now supports getting passed the basename of
7934 a configuration file only, in which case it will look for it
7935 in all appropriate directories automatically.
7937 * udevadm info now takes a /dev or /sys path as argument, and
7938 does the right thing. Example:
7940 udevadm info /dev/sda
7941 udevadm info /sys/class/block/sda
7943 * systemctl now prints a warning if a unit is stopped but a
7944 unit that might trigger it continues to run. Example: a
7945 service is stopped but the socket that activates it is left
7948 * "systemctl status" will now mention if the log output was
7949 shortened due to rotation since a service has been started.
7951 * The journal API now exposes functions to determine the
7952 "cutoff" times due to rotation.
7954 * journald now understands SIGUSR1 and SIGUSR2 for triggering
7955 immediately flushing of runtime logs to /var if possible,
7956 resp. for triggering immediate rotation of the journal
7959 * It is now considered an error if a service is attempted to
7960 be stopped that is not loaded.
7962 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
7964 * systemd-analyze now supports Python 3
7966 * tmpfiles now supports cleaning up directories via aging
7967 where the first level dirs are always kept around but
7968 directories beneath it automatically aged. This is enabled
7969 by prefixing the age field with '~'.
7971 * Seat objects now expose CanGraphical, CanTTY properties
7972 which is required to deal with very fast bootups where the
7973 display manager might be running before the graphics drivers
7974 completed initialization.
7976 * Seat objects now expose a State property.
7978 * We now include RPM macros for service enabling/disabling
7979 based on the preset logic. We recommend RPM based
7980 distributions to make use of these macros if possible. This
7981 makes it simpler to reuse RPM spec files across
7984 * We now make sure that the collected systemd unit name is
7985 always valid when services log to the journal via
7988 * There's a new man page kernel-command-line(7) detailing all
7989 command line options we understand.
7991 * The fstab generator may now be disabled at boot by passing
7992 fstab=0 on the kernel command line.
7994 * A new kernel command line option modules-load= is now understood
7995 to load a specific kernel module statically, early at boot.
7997 * Unit names specified on the systemctl command line are now
7998 automatically escaped as needed. Also, if file system or
7999 device paths are specified they are automatically turned
8000 into the appropriate mount or device unit names. Example:
8002 systemctl status /home
8003 systemctl status /dev/sda
8005 * The SysVConsole= configuration option has been removed from
8006 system.conf parsing.
8008 * The SysV search path is no longer exported on the D-Bus
8011 * The Names= option has been removed from unit file parsing.
8013 * There's a new man page bootup(7) detailing the boot process.
8015 * Every unit and every generator we ship with systemd now
8016 comes with full documentation. The self-explanatory boot is
8019 * A couple of services gained "systemd-" prefixes in their
8020 name if they wrap systemd code, rather than only external
8021 code. Among them fsck@.service which is now
8022 systemd-fsck@.service.
8024 * The HaveWatchdog property has been removed from the D-Bus
8027 * systemd.confirm_spawn= on the kernel command line should now
8030 * There's a new man page crypttab(5) which details all options
8031 we actually understand.
8033 * systemd-nspawn gained a new --capability= switch to pass
8034 additional capabilities to the container.
8036 * timedated will now read known NTP implementation unit names
8037 from /usr/lib/systemd/ntp-units.d/*.list,
8038 systemd-timedated-ntp.target has been removed.
8040 * journalctl gained a new switch "-b" that lists log data of
8041 the current boot only.
8043 * The notify socket is in the abstract namespace again, in
8044 order to support daemons which chroot() at start-up.
8046 * There is a new Storage= configuration option for journald
8047 which allows configuration of where log data should go. This
8048 also provides a way to disable journal logging entirely, so
8049 that data collected is only forwarded to the console, the
8050 kernel log buffer or another syslog implementation.
8052 * Many bugfixes and optimizations
8054 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
8055 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
8056 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
8057 Shawn Landden, Tom Gundersen
8061 * "systemctl help <unit>" now shows the man page if one is
8064 * Several new man pages have been added.
8066 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
8067 MaxLevelConsole= can now be specified in
8068 journald.conf. These options allow reducing the amount of
8069 data stored on disk or forwarded by the log level.
8071 * TimerSlackNSec= can now be specified in system.conf for
8072 PID1. This allows system-wide power savings.
8074 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
8075 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
8080 * logind is now capable of (optionally) handling power and
8081 sleep keys as well as the lid switch.
8083 * journalctl now understands the syntax "journalctl
8084 /usr/bin/avahi-daemon" to get all log output of a specific
8087 * CapabilityBoundingSet= in system.conf now also influences
8088 the capability bound set of usermode helpers of the kernel.
8090 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
8091 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
8092 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
8093 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
8097 * Note that we skipped 139 releases here in order to set the
8098 new version to something that is greater than both udev's
8099 and systemd's most recent version number.
8101 * udev: all udev sources are merged into the systemd source tree now.
8102 All future udev development will happen in the systemd tree. It
8103 is still fully supported to use the udev daemon and tools without
8104 systemd running, like in initramfs or other init systems. Building
8105 udev though, will require the *build* of the systemd tree, but
8106 udev can be properly *run* without systemd.
8108 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
8109 should be used to create dead device nodes as workarounds for broken
8112 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
8113 no longer supported. udev_monitor_new_from_netlink() needs to be
8114 used to subscribe to events.
8116 * udev: when udevd is started by systemd, processes which are left
8117 behind by forking them off of udev rules, are unconditionally cleaned
8118 up and killed now after the event handling has finished. Services or
8119 daemons must be started as systemd services. Services can be
8120 pulled-in by udev to get started, but they can no longer be directly
8121 forked by udev rules.
8123 * udev: the daemon binary is called systemd-udevd now and installed
8124 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
8125 to adapt to that, create symlink, or rename the binary after building
8128 * libudev no longer provides these symbols:
8129 udev_monitor_from_socket()
8130 udev_queue_get_failed_list_entry()
8131 udev_get_{dev,sys,run}_path()
8132 The versions number was bumped and symbol versioning introduced.
8134 * systemd-loginctl and systemd-journalctl have been renamed
8135 to loginctl and journalctl to match systemctl.
8137 * The config files: /etc/systemd/systemd-logind.conf and
8138 /etc/systemd/systemd-journald.conf have been renamed to
8139 logind.conf and journald.conf. Package updates should rename
8140 the files to the new names on upgrade.
8142 * For almost all files the license is now LGPL2.1+, changed
8143 from the previous GPL2.0+. Exceptions are some minor stuff
8144 of udev (which will be changed to LGPL2.1 eventually, too),
8145 and the MIT licensed sd-daemon.[ch] library that is suitable
8146 to be used as drop-in files.
8148 * systemd and logind now handle system sleep states, in
8149 particular suspending and hibernating.
8151 * logind now implements a sleep/shutdown/idle inhibiting logic
8152 suitable for a variety of uses. Soonishly Lennart will blog
8153 about this in more detail.
8155 * var-run.mount and var-lock.mount are no longer provided
8156 (which previously bind mounted these directories to their new
8157 places). Distributions which have not converted these
8158 directories to symlinks should consider stealing these files
8159 from git history and add them downstream.
8161 * We introduced the Documentation= field for units and added
8162 this to all our shipped units. This is useful to make it
8163 easier to explore the boot and the purpose of the various
8166 * All smaller setup units (such as
8167 systemd-vconsole-setup.service) now detect properly if they
8168 are run in a container and are skipped when
8169 appropriate. This guarantees an entirely noise-free boot in
8170 Linux container environments such as systemd-nspawn.
8172 * A framework for implementing offline system updates is now
8173 integrated, for details see:
8174 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
8176 * A new service type Type=idle is available now which helps us
8177 avoiding ugly interleaving of getty output and boot status
8180 * There's now a system-wide CapabilityBoundingSet= option to
8181 globally reduce the set of capabilities for the
8182 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
8183 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
8184 even CAP_NET_ADMIN system-wide for secure systems.
8186 * There are now system-wide DefaultLimitXXX= options to
8187 globally change the defaults of the various resource limits
8188 for all units started by PID 1.
8190 * Harald Hoyer's systemd test suite has been integrated into
8191 systemd which allows easy testing of systemd builds in qemu
8192 and nspawn. (This is really awesome! Ask us for details!)
8194 * The fstab parser is now implemented as generator, not inside
8197 * systemctl will now warn you if .mount units generated from
8198 /etc/fstab are out of date due to changes in fstab that
8199 have not been read by systemd yet.
8201 * systemd is now suitable for usage in initrds. Dracut has
8202 already been updated to make use of this. With this in place
8203 initrds get a slight bit faster but primarily are much
8204 easier to introspect and debug since "systemctl status" in
8205 the host system can be used to introspect initrd services,
8206 and the journal from the initrd is kept around too.
8208 * systemd-delta has been added, a tool to explore differences
8209 between user/admin configuration and vendor defaults.
8211 * PrivateTmp= now affects both /tmp and /var/tmp.
8213 * Boot time status messages are now much prettier and feature
8214 proper english language. Booting up systemd has never been
8217 * Read-ahead pack files now include the inode number of all
8218 files to pre-cache. When the inode changes the pre-caching
8219 is not attempted. This should be nicer to deal with updated
8220 packages which might result in changes of read-ahead
8223 * We now temporaritly lower the kernel's read_ahead_kb variable
8224 when collecting read-ahead data to ensure the kernel's
8225 built-in read-ahead does not add noise to our measurements
8226 of necessary blocks to pre-cache.
8228 * There's now RequiresMountsFor= to add automatic dependencies
8229 for all mounts necessary for a specific file system path.
8231 * MountAuto= and SwapAuto= have been removed from
8232 system.conf. Mounting file systems at boot has to take place
8235 * nspawn now learned a new switch --uuid= to set the machine
8236 ID on the command line.
8238 * nspawn now learned the -b switch to automatically search
8241 * vt102 is now the default TERM for serial TTYs, upgraded from
8244 * systemd-logind now works on VT-less systems.
8246 * The build tree has been reorganized. The individual
8247 components now have directories of their own.
8249 * A new condition type ConditionPathIsReadWrite= is now available.
8251 * nspawn learned the new -C switch to create cgroups for the
8252 container in other hierarchies.
8254 * We now have support for hardware watchdogs, configurable in
8257 * The scheduled shutdown logic now has a public API.
8259 * We now mount /tmp as tmpfs by default, but this can be
8260 masked and /etc/fstab can override it.
8262 * Since udisks does not make use of /media anymore we are not
8263 mounting a tmpfs on it anymore.
8265 * journalctl gained a new --local switch to only interleave
8266 locally generated journal files.
8268 * We can now load the IMA policy at boot automatically.
8270 * The GTK tools have been split off into a systemd-ui.
8272 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
8273 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
8274 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
8275 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
8276 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
8277 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
8278 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
8279 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
8280 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
8285 * This is mostly a bugfix release
8287 * Support optional initialization of the machine ID from the
8288 KVM or container configured UUID.
8290 * Support immediate reboots with "systemctl reboot -ff"
8292 * Show /etc/os-release data in systemd-analyze output
8294 * Many bugfixes for the journal, including endianness fixes and
8295 ensuring that disk space enforcement works
8297 * sd-login.h is C++ compatible again
8299 * Extend the /etc/os-release format on request of the Debian
8302 * We now refuse non-UTF8 strings used in various configuration
8303 and unit files. This is done to ensure we do not pass invalid
8304 data over D-Bus or expose it elsewhere.
8306 * Register Mimo USB Screens as suitable for automatic seat
8309 * Read SELinux client context from journal clients in a race
8312 * Reorder configuration file lookup order. /etc now always
8313 overrides /run in order to allow the administrator to always
8314 and unconditionally override vendor-supplied or
8315 automatically generated data.
8317 * The various user visible bits of the journal now have man
8318 pages. We still lack man pages for the journal API calls
8321 * We now ship all man pages in HTML format again in the
8324 Contributions from: Dave Reisner, Dirk Eibach, Frederic
8325 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
8326 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
8331 * This is mostly a bugfix release
8333 * systems lacking /etc/os-release are no longer supported.
8335 * Various functionality updates to libsystemd-login.so
8337 * Track class of PAM logins to distinguish greeters from
8340 Contributions from: Kay Sievers, Lennart Poettering, Michael
8345 * This is an important bugfix release for v41.
8347 * Building man pages is now optional which should be useful
8348 for those building systemd from git but unwilling to install
8351 * Watchdog support for supervising services is now usable. In
8352 a future release support for hardware watchdogs
8353 (i.e. /dev/watchdog) will be added building on this.
8355 * Service start rate limiting is now configurable and can be
8356 turned off per service. When a start rate limit is hit a
8357 reboot can automatically be triggered.
8359 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
8361 Contributions from: Benjamin Franzke, Bill Nottingham,
8362 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
8363 Schmidt, Michał Górny, Piotr Drąg
8367 * The systemd binary is installed /usr/lib/systemd/systemd now;
8368 An existing /sbin/init symlink needs to be adapted with the
8371 * The code that loads kernel modules has been ported to invoke
8372 libkmod directly, instead of modprobe. This means we do not
8373 support systems with module-init-tools anymore.
8375 * Watchdog support is now already useful, but still not
8378 * A new kernel command line option systemd.setenv= is
8379 understood to set system wide environment variables
8380 dynamically at boot.
8382 * We now limit the set of capabilities of systemd-journald.
8384 * We now set SIGPIPE to ignore by default, since it only is
8385 useful in shell pipelines, and has little use in general
8386 code. This can be disabled with IgnoreSIPIPE=no in unit
8389 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
8390 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
8395 * This is mostly a bugfix release
8397 * We now expose the reason why a service failed in the
8398 "Result" D-Bus property.
8400 * Rudimentary service watchdog support (will be completed over
8401 the next few releases.)
8403 * When systemd forks off in order execute some service we will
8404 now immediately changes its argv[0] to reflect which process
8405 it will execute. This is useful to minimize the time window
8406 with a generic argv[0], which makes bootcharts more useful
8408 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
8409 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
8410 Mike Kazantsev, Ray Strode
8414 * This is mostly a test release, but incorporates many
8417 * New systemd-cgtop tool to show control groups by their
8420 * Linking against libacl for ACLs is optional again. If
8421 disabled, support tracking device access for active logins
8422 goes becomes unavailable, and so does access to the user
8423 journals by the respective users.
8425 * If a group "adm" exists, journal files are automatically
8426 owned by them, thus allow members of this group full access
8427 to the system journal as well as all user journals.
8429 * The journal now stores the SELinux context of the logging
8430 client for all entries.
8432 * Add C++ inclusion guards to all public headers
8434 * New output mode "cat" in the journal to print only text
8435 messages, without any meta data like date or time.
8437 * Include tiny X server wrapper as a temporary stop-gap to
8438 teach XOrg udev display enumeration. This is used by display
8439 managers such as gdm, and will go away as soon as XOrg
8440 learned native udev hotplugging for display devices.
8442 * Add new systemd-cat tool for executing arbitrary programs
8443 with STDERR/STDOUT connected to the journal. Can also act as
8444 BSD logger replacement, and does so by default.
8446 * Optionally store all locally generated coredumps in the
8447 journal along with meta data.
8449 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
8450 writing short strings to files (for usage for /sys), and for
8451 creating symlinks, character and block device nodes.
8453 * New unit file option ControlGroupPersistent= to make cgroups
8454 persistent, following the mechanisms outlined in
8455 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
8457 * Support multiple local RTCs in a sane way
8459 * No longer monopolize IO when replaying readahead data on
8460 rotating disks, since we might starve non-file-system IO to
8461 death, since fanotify() will not see accesses done by blkid,
8464 * Do not show kernel threads in systemd-cgls anymore, unless
8465 requested with new -k switch.
8467 Contributions from: Dan Horák, Kay Sievers, Lennart
8468 Poettering, Michal Schmidt
8472 * This is mostly a test release, but incorporates many
8475 * The git repository moved to:
8476 git://anongit.freedesktop.org/systemd/systemd
8477 ssh://git.freedesktop.org/git/systemd/systemd
8479 * First release with the journal
8480 http://0pointer.de/blog/projects/the-journal.html
8482 * The journal replaces both systemd-kmsg-syslogd and
8483 systemd-stdout-bridge.
8485 * New sd_pid_get_unit() API call in libsystemd-logind
8487 * Many systemadm clean-ups
8489 * Introduce remote-fs-pre.target which is ordered before all
8490 remote mounts and may be used to start services before all
8493 * Added Mageia support
8495 * Add bash completion for systemd-loginctl
8497 * Actively monitor PID file creation for daemons which exit in
8498 the parent process before having finished writing the PID
8499 file in the daemon process. Daemons which do this need to be
8500 fixed (i.e. PID file creation must have finished before the
8501 parent exits), but we now react a bit more gracefully to them.
8503 * Add colourful boot output, mimicking the well-known output
8504 of existing distributions.
8506 * New option PassCredentials= for socket units, for
8507 compatibility with a recent kernel ABI breakage.
8509 * /etc/rc.local is now hooked in via a generator binary, and
8510 thus will no longer act as synchronization point during
8513 * systemctl list-unit-files now supports --root=.
8515 * systemd-tmpfiles now understands two new commands: z, Z for
8516 relabelling files according to the SELinux database. This is
8517 useful to apply SELinux labels to specific files in /sys,
8520 * Output of SysV services is now forwarded to both the console
8521 and the journal by default, not only just the console.
8523 * New man pages for all APIs from libsystemd-login.
8525 * The build tree got reorganized and the build system is a
8526 lot more modular allowing embedded setups to specifically
8527 select the components of systemd they are interested in.
8529 * Support for Linux systems lacking the kernel VT subsystem is
8532 * configure's --with-rootdir= got renamed to
8533 --with-rootprefix= to follow the naming used by udev and
8536 * Unless specified otherwise we will now install to /usr instead
8537 of /usr/local by default.
8539 * Processes with '@' in argv[0][0] are now excluded from the
8540 final shut-down killing spree, following the logic explained
8542 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
8544 * All processes remaining in a service cgroup when we enter
8545 the START or START_PRE states are now killed with
8546 SIGKILL. That means it is no longer possible to spawn
8547 background processes from ExecStart= lines (which was never
8548 supported anyway, and bad style).
8550 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
8551 reloading of units together.
8553 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
8554 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
8555 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
8556 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
8557 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
~ianmdlvl / elogind.git / blob