1 systemd System and Service Manager
3 CHANGES WITH 239 in spe:
5 * NETWORK INTERFACE DEVICE NAMING CHANGES: elogind-udevd's "net_id"
6 builtin may name network interfaces differently than in previous
7 versions. SR-IOV virtual functions and NPAR partitions with PCI
8 function numbers of 8 and above will be named more predictably,
9 and udev may generate names based on PCI slot number in some cases
10 where it previously did not.
12 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
13 elogind.service. Since v235, IPAddressDeny=any has been set to
14 the unit. So, it is expected that the default behavior of elogind
15 is not changed. However, if distribution packagers or administrators
16 disabled or modified IPAddressDeny= setting by a drop-in config file,
17 then it may be necessary to update the file to re-enable AF_INET and
18 AF_INET6 to support network user name services, e.g. NIS.
20 * When the RestrictNamespaces= unit property is specified multiple times,
21 then the specified types are merged now. Previously, only the last
22 assignment was used. So, if distribution packagers or administrators
23 modified the setting by a drop-in config file, then it may be necessary
26 * When OnFailure= is used in combination with Restart= on a service
27 unit, then the specified units will no longer be triggered on
28 failures that result in restarting. Previously, the specified units
29 would be activated each time the unit failed, even when the unit was
30 going to be restarted automatically. This behaviour contradicted the
31 documentation. With this release the code is adjusted to match the
34 * elogind-tmpfiles will now print a notice whenever it encounters
35 tmpfiles.d/ lines referencing the /var/run/ directory. It will
36 recommend reworking them to use the /run/ directory instead (for
37 which /var/run/ is simply a symlinked compatibility alias). This is
38 recommended, so that elogind-tmpfiles can properly detect line
39 conflicts and merge lines referencing the same files by both paths,
40 without having to access them.
42 * systemctl disable/unmask/preset/preset-all cannot be used with
43 --runtime. Previously this was allowed, but resulted in unintuitive
44 behaviour that wasn't useful.
46 * sd-boot acquire new loader configuration settings for optionally
47 turning off Windows and MacOS boot partition discovery as well as
48 reboot-into-firmware menu items. It also should be able to pick a
49 better screen resolution for HiDPI systems, and now provides loader
50 configuration settings to change the resolution explicitly.
52 * The elogind-resolve tool has been renamed to resolvectl (it also
53 remains available under the old name, for compatibility), and its
54 interface is now verb-based, similar in style to the other <xyz>ctl
55 tools, such as systemctl or loginctl. Also, the tool may now be
56 linked to the "resolveconf" name, in which case it will take
57 arguments and input compatible with the Debian and FreeBSD resolvconf
60 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
61 where the system initially suspends, and after a time-out resumes and
64 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
65 set the client will only send a DUID as client identifier.
67 * The nss-elogind glibc NSS module will now enumerate dynamic users and
68 groups in effect. Previously, it could resolve UIDs/GIDs to user
69 names/groups and vice versa, but did not support enumeration.
71 * journald's Compress= configuration setting now optionally accepts a
72 byte threshold value. All journal objects larger than this threshold
73 will be compressed, smaller ones will not. Previously this threshold
74 was not configurable and set to 512.
76 * A new system.conf NoNewPrivileges= is now available which may be used
77 to turn off acquiring of new privileges system-wide (i.e. set Linux'
78 PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also all its
79 children). Note that turning this option on means setuid binaries and
80 file system capabilities lose their special powers. While turning on
81 this option is a big step towards a more secure system, doing so is
82 likely to break numerous pre-existing UNIX tools, in particular su
85 * A new service elogind-time-sync-wait.service has been added. If
86 enabled it will delay the time-sync.target unit at boot until time
87 synchronization has been recieved from the network. This
88 functionality is useful on systems lacking a local RTC or where it is
89 acceptable that the boot process shall be delayed by external network
92 * When hibernating, elogind will now inform the kernel of the image
93 write offset, on kernels new enough to support this. This means swap
94 files should work for hibernation now.
96 * When loading unit files, elogind will now look for drop-in unit file
97 extension at additional places. Previously, for a unit file name
98 "foo-bar-baz.service" it would look for dropin files in
99 "foo-bar-baz.service.d/*.conf". Now, it will also look in
100 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
101 service name truncated after all inner dashes. This scheme allows
102 writing drop-ins easily that apply to a whole set of unit files at
103 once. It's particularly useful for mount and slice units (as their
104 naming is prefix based anyway), but is also useful for service and
105 other units, for packages that install multiple unit files at once,
106 following a strict naming regime of beginning the unit file name with
107 the package's name. Two new specifiers are now supported while
108 parsing unit files, to match this: %j and %J are replaced by the part
109 of the unit name following the last dash.
111 * The ExecStart= lines of unit files are no longer required to
112 reference absolute paths. If non-absolute paths are specified the
113 specified binary name is searched within the service manager's
114 built-in $PATH. It's generally recommended to continue to use
115 absolute paths for all binaries specified in unit files.
117 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
118 support alternative debuggers, for example lldb. The old name
119 continues to be available however, for compatibility reasons. Use the
120 new --debugger= switch and $SYSTEMD_DEBUGGER environment variable to
121 pick an alternative debugger instead of the default of gdb.
123 * systemctl and the other tools will now output escape sequences that
124 generate proper clickable hyperlinks in various terminal emulators
125 where useful (for example, in the "systemctl status" output you can
126 now click on the unit file name to quickly open it in the
127 editor/viewer of your choice). Note that not all terminal emulators
128 support this functionality yet, but many do. Unfortunately, the
129 "less" pager doesn't support this yet, hence this functionality is
130 currently automatically turned off when a pager is started (which
131 happens quite often due to auto-paging). We hope to remove this
132 limitation as soon as "less" learns these escape sequences too. This
133 new behaviour may also be turned off explicitly with the
134 $SYSTEMD_URLIFY environment variable. For details on these escape
136 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
138 * networkd's .network files now support a new IPv6MTUBytes= option for
139 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
140 option in the [Route] section to configure the MTU to use for
141 specific routes. It also gained support for configuration of the DHCP
142 "UserClass" option through the new UserClass= setting. It gained
143 three new options in the new [CAN] section for configuring CAN
144 networks. The MULTICAST and ALLMULTI interface flags may now be
145 controlled explicitly with the new Multicast= and AllMulticast=
148 * networkd will now automatically make use of the kernel's route
149 expiration feature, if it is available.
151 * udevd's .link files now support setting the number of Rx/Tx channels,
152 using the RxChannels=, TxChannels=, OtherChannels=,
153 CombinedChannels=. Support for UDPSegmentationOffload= has been
154 removed, given its limited support in hardware, and waning software
157 * networkd's .netdev files now support creating "netdevsim" interfaces.
159 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
160 to query the unit belonging to a specific kernel control group.
162 * elogind-analyze learnt a new verb "cat-config", which may be used to
163 dump the contents of any configuration file, with all its matching
164 drop-in files added in, and honouring the usual search and masking
165 logic applied to elogind configuration files. For example use
166 "elogind-analyze cat-config elogind/system.conf" to get the complete
167 system configuration file of elogind how it would be loaded by PID 1
168 itself. Similar to this various tools, such as elogind-tmpfiles or
169 elogind-sysusers learnt new option "--cat-config", which do
170 corresponding operation for their own configuration settings. For
171 example, "elogind-tmpfiles --cat-config" will now output the full
172 list of tmpfiles.d/ lines in place.
174 * timedatectl gained two new verbs "timesync-status" (to show the
175 current NTP synchronization state of elogind-timesyncd) and
176 "show-timesync" (to show bus properties of elogind-timesyncd).
178 * elogind-timesyncd gained a bus interface on which it exposes details
181 * elogind-nspawn gained a new --rlimit= switch for setting initial
182 resource limits for the container payload. There's a new switch
183 --hostname= for explicitly overriding the container's hostname. A new
184 --no-new-privileges= switch may be used to control the
185 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
186 --oom-score-adjust= switch controls the OOM scoring adjustment value
187 for the payload. The new --cpu-affinity= switch controls the CPU
188 affinity of the container payload. The new --resolv-conf= switch
189 allows more detailed control of /etc/resolv.conf handling of the
190 container. Similar, the new --timezone= switch allows more detailed
191 control of /etc/localtime handling of the container.
193 * elogind-detect-virt learnt a new --list switch, which will print a
194 list of all currently known VM and container environments.
196 * Support for the new "Portable Services" concept has been added, see
197 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
198 experimental, but this is expected to change soon. Reflecting the
199 experimental state the "portablectl" binary is not installed into
200 /usr/bin yet. The binary has to be called with the full path
201 /usr/lib/elogind/portablectl instead.
203 * journalctl's and systemctl's -o switch now knows a new log output
204 mode "with-unit". The output it generates is very similar to the
205 regular "short" mode, but displays the unit name instead of the
206 syslog tag for each log line. Also, the date is shown with timezone
207 information. This mode is probably more useful than the classic
208 "short" output mode for most purposes, except where pixel-perfect
209 compatibility with classic /var/log/messages formatting is required.
211 * A new --dump-bus-properties switch has been added to the elogind
212 binary, which may be used to dump all supported D-Bus properties.
214 * Unit files and other configuration files that support specifier
215 expansion now understand two new specifiers: %T and %V will resolve
216 to /tmp and /var/tmp respectively, or whatever temporary directory
217 has been set for the calling user.
219 * sd-bus gained a set of new calls:
220 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
221 enable/disable the "floating" state of a bus slot object,
222 i.e. whether the slot object pins the bus it is allocated for into
223 memory or if the bus slot object gets disconnected when the bus goes
224 away. sd_bus_open_with_description(),
225 sd_bus_open_user_with_description(),
226 sd_bus_open_system_with_description() may be used to allocate bus
227 objects and set their description string already during allocation.
229 * sd-event gained support for watching inotify events from the event
230 loop, in an efficient way, sharing inotify handles between multiple
231 users. For this a new function sd_event_add_inotify() has been added.
233 * sd-event and sd-bus gained support for calling special user-supplied
234 destructor functions for userdata pointers associated with
235 sd_event_source, sd_bus_slot and sd_bus_track objects.
237 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
239 * PID 1 will now automatically reschedule .timer units whenever the
240 local timezone changes. (They previously got rescheduled already
241 automatically when the system clock changed.)
243 * New documentation has been added to document cgroups delegation,
244 portable services and the various code quality tools we have set up:
246 https://github.com/systemd/systemd/blob/master/doc/CGROUP_DELEGATION.md
247 https://github.com/systemd/systemd/blob/master/doc/PORTABLE_SERVICES.md
248 https://github.com/systemd/systemd/blob/master/doc/CODE_QUALITY.md
250 * pam_elogind will now look for PAM userdata fields elogind.memory_max,
251 elogind.tasks_max, elogind.cpu_weight, elogind.io_weight set by
252 earlier PAM modules. The data in these fields is used to initialize
253 the session scope's resource properties. Thus external PAM modules
254 may now configure per-session limits, for example sourced from
255 external user databases.
257 * socket units with Accept=yes will now maintain a "refused" counter in
258 addition to the existing "accepted" counter, counting connections
259 refused due to the enforced limits.
261 * The "elogind-path search-binaries-default" command may now be use to
262 query the default, built-in $PATH PID 1 will pass to the services it
265 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
266 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Antique, Arnaud
267 Rebillout, Brian J. Murrell, Bruno Vernay, Chris Lesiak, Christian
268 Brauner, Christian Hesse, Daniel Dao, Daniel Lin, Danylo Korostil,
269 Davide Cavalca, David Tardon, Dimitri John Ledkov, Dmitriy Geels,
270 Douglas Christman, Elia Geretto, emelenas, Evegeny Vereshchagin, Evgeny
271 Vereshchagin, Felipe Sateler, Feng Sun, Filipe Brandenburger, Franck
272 Bui, futpib, Giuseppe Scrivano, Guillem Jover, guixxx, Hans de Goede,
273 Henrique Dante de Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko,
274 Ivan Shapovalov, James Cowgill, Jan Janssen, Jan Synacek, Jared
275 Kazimir, João Paulo Rechi Vita, Joost Heitbrink, juergbi, Jui-Chi Ricky
276 Liang, Kai-Heng Feng, Karol Augustin, Krzysztof Nowicki, Lauri
277 Tirkkonen, Lennart Poettering, Leonard, Long Li, Luca Boccassi, Lucas
278 Werkmeister, Marcel Hoppe, Marc Kleine-Budde, Mario Limonciello, Martin
279 Jansa, Martin Wilck, Mathieu Malaterre, Matteo F. Vescovi, Michael
280 Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal Sekletar,
281 Mike Gilbert, Mikhail Kasimov, Milan, Milan Broz, mourikwa, Muhammet
282 Kara, Nicolas Boichat, Omer Katz, Paride Legovini, Paul Menzel, Paul
283 Milliken, Peter A. Bigot, Peter Hutterer, Peter Jones, Philip Sequeira,
284 Philip Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de
285 Araujo, Rosen Penev, rubensa, Ryan Gonzalez, Salvo 'LtWorf' Tomaselli,
286 Sebastian Reichel, Sergio Lindo Mansilla, Stefan Schweter, Stephen
287 Hemminger, Stuart Hayes, Susant Sahani, Sylvain Plantefève, Thomas
288 H. P. Andersen, Tobias Jungel, Tomasz Torcz, Vito Caputo, Will Dietz,
289 Will Thompson, xginn8, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
295 * The MemoryAccounting= unit property now defaults to on. After
296 discussions with the upstream control group maintainers we learnt
297 that the negative impact of cgroup memory accounting on current
298 kernels is finally relatively minimal, so that it should be safe to
299 enable this by default without affecting system performance. Besides
300 memory accounting only task accounting is turned on by default, all
301 other forms of resource accounting (CPU, IO, IP) remain off for now,
302 because it's not clear yet that their impact is small enough to move
303 from opt-in to opt-out. We recommend downstreams to leave memory
304 accounting on by default if kernel 4.14 or higher is primarily
305 used. On very resource constrained systems or when support for old
306 kernels is a necessity, -Dmemory-accounting-default=false can be used
307 to revert this change.
309 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
310 %udev_rules_update) and the journal catalog (%journal_catalog_update)
311 from the upgrade scriptlets of individual packages now do nothing.
312 Transfiletriggers have been added which will perform those updates
313 once at the end of the transaction.
315 Similar transfiletriggers have been added to execute any sysctl.d
316 and binfmt.d rules. Thus, it should be unnecessary to provide any
317 scriptlets to execute this configuration from package installation
320 * elogind-sysusers gained a mode where the configuration to execute is
321 specified on the command line, but this configuration is not executed
322 directly, but instead it is merged with the configuration on disk,
323 and the result is executed. This is useful for package installation
324 scripts which want to create the user before installing any files on
325 disk (in case some of those files are owned by that user), while
326 still allowing local admin overrides.
328 This functionality is exposed to rpm scriptlets through a new
329 %sysusers_create_package macro. Old %sysusers_create and
330 %sysusers_create_inline macros are deprecated.
332 A transfiletrigger for sysusers.d configuration is now installed,
333 which means that it should be unnecessary to call elogind-sysusers from
334 package installation scripts, unless the package installs any files
335 owned by those newly-created users, in which case
336 %sysusers_create_package should be used.
338 * Analogous change has been done for elogind-tmpfiles: it gained a mode
339 where the command-line configuration is merged with the configuration
340 on disk. This is exposed as the new %tmpfiles_create_package macro,
341 and %tmpfiles_create is deprecated. A transfiletrigger is installed
342 for tmpfiles.d, hence it should be unnecessary to call elogind-tmpfiles
343 from package installation scripts.
345 * sysusers.d configuration for a user may now also specify the group
346 number, in addition to the user number ("u username 123:456"), or
347 without the user number ("u username -:456").
349 * Configution items for elogind-sysusers can now be specified as
350 positional arguments when the new --inline switch is used.
352 * The login shell of users created through sysusers.d may now be
353 specified (previously, it was always /bin/sh for root and
354 /sbin/nologin for other users).
356 * elogind-analyze gained a new --global switch to look at global user
357 configuration. It also gained a unit-paths verb to list the unit load
358 paths that are compiled into elogind (which can be used with
359 --elogind, --user, or --global).
361 * udevadm trigger gained a new --settle/-w option to wait for any
362 triggered events to finish (but just those, and not any other events
363 which are triggered meanwhile).
365 * The action that elogind takes when the lid is closed and the
366 machine is connected to external power can now be configured using
367 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
368 was determined by HandleLidSwitch=, and, for backwards compatibility,
369 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
371 * journalctl will periodically call sd_journal_process() to make it
372 resilient against inotify queue overruns when journal files are
373 rotated very quickly.
375 * Two new functions in libelogind — sd_bus_get_n_queued_read and
376 sd_bus_get_n_queued_write — may be used to check the number of
377 pending bus messages.
379 * elogind gained a new
380 org.freedesktop.elogind1.Manager.AttachProcessesToUnit dbus call
381 which can be used to migrate foreign processes to scope and service
382 units. The primary user for this new API is elogind itself: the
383 elogind --user instance uses this call of the elogind --system
384 instance to migrate processes if it itself gets the request to
385 migrate processes and the kernel refuses this due to access
386 restrictions. Thanks to this "elogind-run --scope --user …" works
387 again in pure cgroups v2 environments when invoked from the user
390 * A new TemporaryFileSystem= setting can be used to mask out part of
391 the real file system tree with tmpfs mounts. This may be combined
392 with BindPaths= and BindReadOnlyPaths= to hide files or directories
393 not relevant to the unit, while still allowing some paths lower in
394 the tree to be accessed.
396 ProtectHome=tmpfs may now be used to hide user home and runtime
397 directories from units, in a way that is mostly equivalent to
398 "TemporaryFileSystem=/home /run/user /root".
400 * Non-service units are now started with KeyringMode=shared by default.
401 This means that mount and swapon and other mount tools have access
402 to keys in the main keyring.
404 * /sys/fs/bpf is now mounted automatically.
406 * QNX virtualization is now detected by elogind-detect-virt and may
407 be used in ConditionVirtualization=.
409 * IPAccounting= may now be enabled also for slice units.
411 * A new -Dsplit-bin= build configuration switch may be used to specify
412 whether bin and sbin directories are merged, or if they should be
413 included separately in $PATH and various listings of executable
414 directories. The build configuration scripts will try to autodetect
415 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
416 system, but distributions are encouraged to configure this
419 * A new -Dok-color= build configuration switch may be used to change
420 the colour of "OK" status messages.
422 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
423 PrivateNetwork=yes was buggy in previous versions of elogind. This
424 means that after the upgrade and daemon-reexec, any such units must
427 * INCOMPATIBILITY: as announced in the NEWS for 237, elogind-tmpfiles
428 will not exclude read-only files owned by root from cleanup.
430 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
431 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
432 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
433 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
434 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
435 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
436 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
437 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
438 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
439 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
440 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
441 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
442 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
443 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
444 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
445 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
451 * Some keyboards come with a zoom see-saw or rocker which until now got
452 mapped to the Linux "zoomin/out" keys in hwdb. However, these
453 keycodes are not recognized by any major desktop. They now produce
454 Up/Down key events so that they can be used for scrolling.
456 * INCOMPATIBILITY: elogind-tmpfiles' "f" lines changed behaviour
457 slightly: previously, if an argument was specified for lines of this
458 type (i.e. the right-most column was set) this string was appended to
459 existing files each time elogind-tmpfiles was run. This behaviour was
460 different from what the documentation said, and not particularly
461 useful, as repeated elogind-tmpfiles invocations would not be
462 idempotent and grow such files without bounds. With this release
463 behaviour has been altered slightly, to match what the documentation
464 says: lines of this type only have an effect if the indicated files
465 don't exist yet, and only then the argument string is written to the
468 * FUTURE INCOMPATIBILITY: In elogind v238 we intend to slightly change
469 elogind-tmpfiles behaviour: previously, read-only files owned by root
470 were always excluded from the file "aging" algorithm (i.e. the
471 automatic clean-up of directories like /tmp based on
472 atime/mtime/ctime). We intend to drop this restriction, and age files
473 by default even when owned by root and read-only. This behaviour was
474 inherited from older tools, but there have been requests to remove
475 it, and it's not obvious why this restriction was made in the first
476 place. Please speak up now, if you are aware of software that reqires
477 this behaviour, otherwise we'll remove the restriction in v238.
479 * A new environment variable $SYSTEMD_OFFLINE is now understood by
480 systemctl. It takes a boolean argument. If on, systemctl assumes it
481 operates on an "offline" OS tree, and will not attempt to talk to the
482 service manager. Previously, this mode was implicitly enabled if a
483 chroot() environment was detected, and this new environment variable
484 now provides explicit control.
486 * .path and .socket units may now be created transiently, too.
487 Previously only service, mount, automount and timer units were
488 supported as transient units. The elogind-run tool has been updated
489 to expose this new functionality, you may hence use it now to bind
490 arbitrary commands to path or socket activation on-the-fly from the
491 command line. Moreover, almost all properties are now exposed for the
492 unit types that already supported transient operation.
494 * The elogind-mount command gained support for a new --owner= parameter
495 which takes a user name, which is then resolved and included in uid=
496 and gid= mount options string of the file system to mount.
498 * A new unit condition ConditionControlGroupController= has been added
499 that checks whether a specific cgroup controller is available.
501 * Unit files, udev's .link files, and elogind-networkd's .netdev and
502 .network files all gained support for a new condition
503 ConditionKernelVersion= for checking against specific kernel
506 * In elogind-networkd, the [IPVLAN] section in .netdev files gained
507 support for configuring device flags in the Flags= setting. In the
508 same files, the [Tunnel] section gained support for configuring
509 AllowLocalRemote=. The [Route] section in .network files gained
510 support for configuring InitialCongestionWindow=,
511 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
512 understands RapidCommit=.
514 * elogind-networkd's DHCPv6 support gained support for Prefix
517 * sd-bus gained support for a new "watch-bind" feature. When this
518 feature is enabled, an sd_bus connection may be set up to connect to
519 an AF_UNIX socket in the file system as soon as it is created. This
520 functionality is useful for writing early-boot services that
521 automatically connect to the system bus as soon as it is started,
522 without ugly time-based polling. elogind-networkd and
523 elogind-resolved have been updated to make use of this
524 functionality. busctl exposes this functionality in a new
525 --watch-bind= command line switch.
527 * sd-bus will now optionally synthesize a local "Connected" signal as
528 soon as a D-Bus connection is set up fully. This message mirrors the
529 already existing "Disconnected" signal which is synthesized when the
530 connection is terminated. This signal is generally useful but
531 particularly handy in combination with the "watch-bind" feature
532 described above. Synthesizing of this message has to be requested
533 explicitly through the new API call sd_bus_set_connected_signal(). In
534 addition a new call sd_bus_is_ready() has been added that checks
535 whether a connection is fully set up (i.e. between the "Connected" and
536 "Disconnected" signals).
538 * sd-bus gained two new calls sd_bus_request_name_async() and
539 sd_bus_release_name_async() for asynchronously registering bus
540 names. Similar, there is now sd_bus_add_match_async() for installing
541 a signal match asynchronously. All of elogind's own services have
542 been updated to make use of these calls. Doing these operations
543 asynchronously has two benefits: it reduces the risk of deadlocks in
544 case of cyclic dependencies between bus services, and it speeds up
545 service initialization since synchronization points for bus
546 round-trips are removed.
548 * sd-bus gained two new calls sd_bus_match_signal() and
549 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
550 and sd_bus_add_match_async() but instead of taking a D-Bus match
551 string take match fields as normal function parameters.
553 * sd-bus gained two new calls sd_bus_set_sender() and
554 sd_bus_message_set_sender() for setting the sender name of outgoing
555 messages (either for all outgoing messages or for just one specific
556 one). These calls are only useful in direct connections as on
557 brokered connections the broker fills in the sender anyway,
558 overwriting whatever the client filled in.
560 * sd-event gained a new pseudo-handle that may be specified on all API
561 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
562 used this refers to the default event loop object of the calling
563 thread. Note however that this does not implicitly allocate one —
564 which has to be done prior by using sd_event_default(). Similarly
565 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
566 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
567 to the default bus of the specified type of the calling thread. Here
568 too this does not implicitly allocate bus connection objects, this
569 has to be done prior with sd_bus_default() and friends.
571 * sd-event gained a new call pair
572 sd_event_source_{get|set}_io_fd_own(). This may be used to request
573 automatic closure of the file descriptor an IO event source watches
574 when the event source is destroyed.
576 * elogind-networkd gained support for natively configuring WireGuard
579 * In previous versions elogind synthesized user records both for the
580 "nobody" (UID 65534) and "root" (UID 0) users in nss-elogind and
581 internally. In order to simplify distribution-wide renames of the
582 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
583 new transitional flag file has been added: if
584 /etc/elogind/dont-synthesize-nobody exists synthesizing of the 65534
585 user and group record within the elogind codebase is disabled.
587 * elogind-notify gained a new --uid= option for selecting the source
588 user/UID to use for notification messages sent to the service
591 * journalctl gained a new --grep= option to list only entries in which
592 the message matches a certain pattern. By default matching is case
593 insensitive if the pattern is lowercase, and case sensitive
594 otherwise. Option --case-sensitive=yes|no can be used to override
595 this an specify case sensitivity or case insensitivity.
597 * There's now a "elogind-analyze service-watchdogs" command for printing
598 the current state of the service runtime watchdog, and optionally
599 enabling or disabling the per-service watchdogs system-wide if given a
600 boolean argument (i.e. the concept you configure in WatchdogSec=), for
601 debugging purposes. There's also a kernel command line option
602 elogind.service_watchdogs= for controlling the same.
604 * Two new "log-level" and "log-target" options for elogind-analyze were
605 addded that merge the now deprecated get-log-level, set-log-level and
606 get-log-target, set-log-target pairs. The deprecated options are still
607 understood for backwards compatibility. The two new options print the
608 current value when no arguments are given, and set them when a
609 level/target is given as an argument.
611 * sysusers.d's "u" lines now optionally accept both a UID and a GID
612 specification, separated by a ":" character, in order to create users
613 where UID and GID do not match.
615 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
616 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
617 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
618 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
619 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
620 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
621 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
622 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
623 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
624 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
625 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
626 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
627 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
628 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
629 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
630 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
631 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
632 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
633 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
634 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
641 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
642 in v235 has been extended to also set the dummy.ko module option
643 numdummies=0, preventing the kernel from automatically creating
644 dummy0. All dummy interfaces must now be explicitly created.
646 * Unknown '%' specifiers in configuration files are now rejected. This
647 applies to units and tmpfiles.d configuration. Any percent characters
648 that are followed by a letter or digit that are not supposed to be
649 interpreted as the beginning of a specifier should be escaped by
650 doubling ("%%"). (So "size=5%" is still accepted, as well as
651 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
652 valid specifiers today.)
654 * systemd-resolved now maintains a new dynamic
655 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
656 recommended to make /etc/resolv.conf a symlink to it. This file
657 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
658 includes dynamically acquired search domains, achieving more correct
659 DNS resolution by software that bypasses local DNS APIs such as NSS.
661 * The "uaccess" udev tag has been dropped from /dev/kvm and
662 /dev/dri/renderD*. These devices now have the 0666 permissions by
663 default (but this may be changed at build-time). /dev/dri/renderD*
664 will now be owned by the "render" group along with /dev/kfd.
666 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
667 systemd-journal-gatewayd.service and
668 systemd-journal-upload.service. This means "nss-systemd" must be
669 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
670 services are resolved properly.
672 * In /etc/fstab two new mount options are now understood:
673 x-systemd.makefs and x-systemd.growfs. The former has the effect that
674 the configured file system is formatted before it is mounted, the
675 latter that the file system is resized to the full block device size
676 after it is mounted (i.e. if the file system is smaller than the
677 partition it resides on, it's grown). This is similar to the fsck
678 logic in /etc/fstab, and pulls in systemd-makefs@.service and
679 systemd-growfs@.service as necessary, similar to
680 systemd-fsck@.service. Resizing is currently only supported on ext4
683 * In systemd-networkd, the IPv6 RA logic now optionally may announce
684 DNS server and domain information.
686 * Support for the LUKS2 on-disk format for encrypted partitions has
687 been added. This requires libcryptsetup2 during compilation and
690 * The systemd --user instance will now signal "readiness" when its
691 basic.target unit has been reached, instead of when the run queue ran
692 empty for the first time.
694 * Tmpfiles.d with user configuration are now also supported.
695 systemd-tmpfiles gained a new --user switch, and snippets placed in
696 ~/.config/user-tmpfiles.d/ and corresponding directories will be
697 executed by systemd-tmpfiles --user running in the new
698 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
699 running in the user session.
701 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
702 %S resolves to the top-level state directory (/var/lib for the system
703 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
704 top-level cache directory (/var/cache for the system instance,
705 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
706 logs directory (/var/log for the system instance,
707 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
708 existing %t specifier, that resolves to the top-level runtime
709 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
712 * journalctl learnt a new parameter --output-fields= for limiting the
713 set of journal fields to output in verbose and JSON output modes.
715 * systemd-timesyncd's configuration file gained a new option
716 RootDistanceMaxSec= for setting the maximum root distance of servers
717 it'll use, as well as the new options PollIntervalMinSec= and
718 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
720 * bootctl gained a new command "list" for listing all available boot
721 menu items on systems that follow the boot loader specification.
723 * systemctl gained a new --dry-run switch that shows what would be done
724 instead of doing it, and is currently supported by the shutdown and
727 * ConditionSecurity= can now detect the TOMOYO security module.
729 * Unit file [Install] sections are now also respected in unit drop-in
730 files. This is intended to be used by drop-ins under /usr/lib/.
732 * systemd-firstboot may now also set the initial keyboard mapping.
734 * Udev "changed" events for devices which are exposed as systemd
735 .device units are now propagated to units specified in
736 ReloadPropagatedFrom= as reload requests.
738 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
739 unit template name (i.e. a name in the form of 'foobar@.service',
740 without the instance component between the '@' and - the '.'), then
741 the escaped sysfs path of the device is automatically used as the
744 * SystemCallFilter= in unit files has been extended so that an "errno"
745 can be specified individually for each system call. Example:
746 SystemCallFilter=~uname:EILSEQ.
748 * The cgroup delegation logic has been substantially updated. Delegate=
749 now optionally takes a list of controllers (instead of a boolean, as
750 before), which lists the controllers to delegate at least.
752 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
754 * A new LogLevelMax= setting configures the maximum log level any
755 process of the service may log at (i.e. anything with a lesser
756 priority than what is specified is automatically dropped). A new
757 LogExtraFields= setting allows configuration of additional journal
758 fields to attach to all log records generated by any of the unit's
761 * New StandardInputData= and StandardInputText= settings along with the
762 new option StandardInput=data may be used to configure textual or
763 binary data that shall be passed to the executed service process via
764 standard input, encoded in-line in the unit file.
766 * StandardInput=, StandardOutput= and StandardError= may now be used to
767 connect stdin/stdout/stderr of executed processes directly with a
768 file or AF_UNIX socket in the file system, using the new "file:" option.
770 * A new unit file option CollectMode= has been added, that allows
771 tweaking the garbage collection logic for units. It may be used to
772 tell systemd to garbage collect units that have failed automatically
773 (normally it only GCs units that exited successfully). systemd-run
774 and systemd-mount expose this new functionality with a new -G option.
776 * "machinectl bind" may now be used to bind mount non-directories
777 (i.e. regularfiles, devices, fifos, sockets).
779 * systemd-analyze gained a new verb "calendar" for validating and
780 testing calendar time specifications to use for OnCalendar= in timer
781 units. Besides validating the expression it will calculate the next
782 time the specified expression would elapse.
784 * In addition to the pre-existing FailureAction= unit file setting
785 there's now SuccessAction=, for configuring a shutdown action to
786 execute when a unit completes successfully. This is useful in
787 particular inside containers that shall terminate after some workload
788 has been completed. Also, both options are now supported for all unit
789 types, not just services.
791 * networkds's IP rule support gained two new options
792 IncomingInterface= and OutgoingInterface= for configuring the incoming
793 and outgoing interfaces of configured rules. systemd-networkd also
794 gained support for "vxcan" network devices.
796 * networkd gained a new setting RequiredForOnline=, taking a
797 boolean. If set, systemd-wait-online will take it into consideration
798 when determining that the system is up, otherwise it will ignore the
799 interface for this purpose.
801 * The sd_notify() protocol gained support for a new operation: with
802 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
803 store again, ahead of POLLHUP or POLLERR when they are removed
806 * A new document doc/UIDS-GIDS.md has been added to the source tree,
807 that documents the UID/GID range and assignment assumptions and
808 requirements of systemd.
810 * The watchdog device PID 1 will ping may now be configured through the
811 WatchdogDevice= configuration file setting, or by setting the
812 systemd.watchdog_service= kernel commandline option.
814 * systemd-resolved's gained support for registering DNS-SD services on
815 the local network using MulticastDNS. Services may either be
816 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
817 the same dir below /run, /usr/lib), or through its D-Bus API.
819 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
820 extend the effective start, runtime, and stop time. The service must
821 continue to send EXTEND_TIMEOUT_USEC within the period specified to
822 prevent the service manager from making the service as timedout.
824 * elogind-resolved's DNSSEC support gained support for RFC 8080
825 (Ed25519 keys and signatures).
827 * The elogind-resolve command line tool gained a new set of options
828 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
829 --set-nta= and --revert to configure per-interface DNS configuration
830 dynamically during runtime. It's useful for pushing DNS information
831 into elogind-resolved from DNS hook scripts that various interface
832 managing software supports (such as pppd).
834 * elogind-nspawn gained a new --network-namespace-path= command line
835 option, which may be used to make a container join an existing
836 network namespace, by specifying a path to a "netns" file.
838 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
839 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
840 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
841 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
842 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
843 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
844 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
845 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
846 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
847 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
848 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
849 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
850 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
851 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
852 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
853 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
854 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
855 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
856 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
857 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
858 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
859 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
860 Jędrzejewski-Szmek, Zeal Jagannatha
866 * INCOMPATIBILITY: systemd-logind.service and other long-running
867 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
868 communication with the outside. This generally improves security of
869 the system, and is in almost all cases a safe and good choice, as
870 these services do not and should not provide any network-facing
871 functionality. However, systemd-logind uses the glibc NSS API to
872 query the user database. This creates problems on systems where NSS
873 is set up to directly consult network services for user database
874 lookups. In particular, this creates incompatibilities with the
875 "nss-nis" module, which attempts to directly contact the NIS/YP
876 network servers it is configured for, and will now consistently
877 fail. In such cases, it is possible to turn off IP sandboxing for
878 systemd-logind.service (set IPAddressDeny= in its [Service] section
879 to the empty string, via a .d/ unit file drop-in). Downstream
880 distributions might want to update their nss-nis packaging to include
881 such a drop-in snippet, accordingly, to hide this incompatibility
882 from the user. Another option is to make use of glibc's nscd service
883 to proxy such network requests through a privilege-separated, minimal
884 local caching daemon, or to switch to more modern technologies such
885 sssd, whose NSS hook-ups generally do not involve direct network
886 access. In general, we think it's definitely time to question the
887 implementation choices of nss-nis, i.e. whether it's a good idea
888 today to embed a network-facing loadable module into all local
889 processes that need to query the user database, including the most
890 trivial and benign ones, such as "ls". For more details about
891 IPAddressDeny= see below.
893 * A new modprobe.d drop-in is now shipped by default that sets the
894 bonding module option max_bonds=0. This overrides the kernel default,
895 to avoid conflicts and ambiguity as to whether or not bond0 should be
896 managed by systemd-networkd or not. This resolves multiple issues
897 with bond0 properties not being applied, when bond0 is configured
898 with systemd-networkd. Distributors may choose to not package this,
899 however in that case users will be prevented from correctly managing
900 bond0 interface using systemd-networkd.
902 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
903 which print the logging level and target of the system manager. They
904 complement the existing "set-log-level" and "set-log-target" verbs
905 used to change those values.
907 * journald.conf gained a new boolean setting ReadKMsg= which defaults
908 to on. If turned off kernel log messages will not be read by
909 systemd-journald or included in the logs. It also gained a new
910 setting LineMax= for configuring the maximum line length in
911 STDOUT/STDERR log streams. The new default for this value is 48K, up
912 from the previous hardcoded 2048.
914 * A new unit setting RuntimeDirectoryPreserve= has been added, which
915 allows more detailed control of what to do with a runtime directory
916 configured with RuntimeDirectory= (i.e. a directory below /run or
917 $XDG_RUNTIME_DIR) after a unit is stopped.
919 * The RuntimeDirectory= setting for units gained support for creating
920 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
921 one top-level directory.
923 * Units gained new options StateDirectory=, CacheDirectory=,
924 LogsDirectory= and ConfigurationDirectory= which are closely related
925 to RuntimeDirectory= but manage per-service directories below
926 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
927 possible to write unit files which when activated automatically gain
928 properly owned service specific directories in these locations, thus
929 making unit files self-contained and increasing compatibility with
930 stateless systems and factory reset where /etc or /var are
931 unpopulated at boot. Matching these new settings there's also
932 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
933 ConfigurationDirectoryMode= for configuring the access mode of these
934 directories. These settings are particularly useful in combination
935 with DynamicUser=yes as they provide secure, properly-owned,
936 writable, and stateful locations for storage, excluded from the
937 sandbox that such services live in otherwise.
939 * Automake support has been removed from this release. systemd is now
942 * systemd-journald will now aggressively cache client metadata during
943 runtime, speeding up log write performance under pressure. This comes
944 at a small price though: as much of the metadata is read
945 asynchronously from /proc/ (and isn't implicitly attached to log
946 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
947 metadata stored alongside a log entry might be slightly
948 out-of-date. Previously it could only be slightly newer than the log
949 message. The time window is small however, and given that the kernel
950 is unlikely to be improved anytime soon in this regard, this appears
953 * nss-myhostname/systemd-resolved will now by default synthesize an
954 A/AAAA resource record for the "_gateway" hostname, pointing to the
955 current default IP gateway. Previously it did that for the "gateway"
956 name, hampering adoption, as some distributions wanted to leave that
957 host name open for local use. The old behaviour may still be
958 requested at build time.
960 * systemd-networkd's [Address] section in .network files gained a new
961 Scope= setting for configuring the IP address scope. The [Network]
962 section gained a new boolean setting ConfigureWithoutCarrier= that
963 tells systemd-networkd to ignore link sensing when configuring the
964 device. The [DHCP] section gained a new Anonymize= boolean option for
965 turning on a number of options suggested in RFC 7844. A new
966 [RoutingPolicyRule] section has been added for configuring the IP
967 routing policy. The [Route] section has gained support for a new
968 Type= setting which permits configuring
969 blackhole/unreachable/prohibit routes.
971 * The [VRF] section in .netdev files gained a new Table= setting for
972 configuring the routing table to use. The [Tunnel] section gained a
973 new Independent= boolean field for configuring tunnels independent of
974 an underlying network interface. The [Bridge] section gained a new
975 GroupForwardMask= option for configuration of propagation of link
976 local frames between bridge ports.
978 * The WakeOnLan= setting in .link files gained support for a number of
979 new modes. A new TCP6SegmentationOffload= setting has been added for
980 configuring TCP/IPv6 hardware segmentation offload.
982 * The IPv6 RA sender implementation may now optionally send out RDNSS
983 and RDNSSL records to supply DNS configuration to peers.
985 * systemd-nspawn gained support for a new --system-call-filter= command
986 line option for adding and removing entries in the default system
987 call filter it applies. Moreover systemd-nspawn has been changed to
988 implement a system call whitelist instead of a blacklist.
990 * systemd-run gained support for a new --pipe command line option. If
991 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
992 are directly passed on to the activated transient service
993 executable. This allows invoking arbitrary processes as systemd
994 services (for example to take benefit of dependency management,
995 accounting management, resource management or log management that is
996 done automatically for services) — while still allowing them to be
997 integrated in a classic UNIX shell pipeline.
999 * When a service sends RELOAD=1 via sd_notify() and reload propagation
1000 using ReloadPropagationTo= is configured, a reload is now propagated
1001 to configured units. (Previously this was only done on explicitly
1002 requested reloads, using "systemctl reload" or an equivalent
1005 * For each service unit a restart counter is now kept: it is increased
1006 each time the service is restarted due to Restart=, and may be
1007 queried using "systemctl show -p NRestarts …".
1009 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
1010 @signal and @timer have been added, for usage with SystemCallFilter=
1011 in unit files and the new --system-call-filter= command line option
1012 of systemd-nspawn (see above).
1014 * ExecStart= lines in unit files gained two new modifiers: when a
1015 command line is prefixed with "!" the command will be executed as
1016 configured, except for the credentials applied by
1017 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
1018 "+", but does still apply namespacing options unlike "+". There's
1019 also "!!" now, which is mostly identical, but becomes a NOP on
1020 systems that support ambient capabilities. This is useful to write
1021 unit files that work with ambient capabilities where possible but
1022 automatically fall back to traditional privilege dropping mechanisms
1023 on systems where this is not supported.
1025 * ListenNetlink= settings in socket units now support RDMA netlink
1028 * A new unit file setting LockPersonality= has been added which permits
1029 locking down the chosen execution domain ("personality") of a service
1032 * A new special target "getty-pre.target" has been added, which is
1033 ordered before all text logins, and may be used to order services
1034 before textual logins acquire access to the console.
1036 * systemd will now attempt to load the virtio-rng.ko kernel module very
1037 early on if a VM environment supporting this is detected. This should
1038 improve entropy during early boot in virtualized environments.
1040 * A _netdev option is now supported in /etc/crypttab that operates in a
1041 similar way as the same option in /etc/fstab: it permits configuring
1042 encrypted devices that need to be ordered after the network is up.
1043 Following this logic, two new special targets
1044 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
1045 added that are to cryptsetup.target what remote-fs.target and
1046 remote-fs-pre.target are to local-fs.target.
1048 * Service units gained a new UnsetEnvironment= setting which permits
1049 unsetting specific environment variables for services that are
1050 normally passed to it (for example in order to mask out locale
1051 settings for specific services that can't deal with it).
1053 * Units acquired a new boolean option IPAccounting=. When turned on, IP
1054 traffic accounting (packet count as well as byte count) is done for
1055 the service, and shown as part of "systemctl status" or "systemd-run
1058 * Service units acquired two new options IPAddressAllow= and
1059 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
1060 for configuring a simple IP access control list for all sockets of
1061 the unit. These options are available also on .slice and .socket
1062 units, permitting flexible access list configuration for individual
1063 services as well as groups of services (as defined by a slice unit),
1064 including system-wide. Note that IP ACLs configured this way are
1065 enforced on every single IPv4 and IPv6 socket created by any process
1066 of the service unit, and apply to ingress as well as egress traffic.
1068 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
1069 structured log message is generated each time the unit is stopped,
1070 containing information about the consumed resources of this
1073 * A new setting KeyringMode= has been added to unit files, which may be
1074 used to control how the kernel keyring is set up for executed
1077 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
1078 "systemctl kexec" and "systemctl exit" are now always asynchronous in
1079 behaviour (that is: these commands return immediately after the
1080 operation was enqueued instead of waiting for the operation to
1081 complete). Previously, "systemctl poweroff" and "systemctl reboot"
1082 were asynchronous on systems using systemd-logind (i.e. almost
1083 always, and like they were on sysvinit), and the other three commands
1084 were unconditionally synchronous. With this release this is cleaned
1085 up, and callers will see the same asynchronous behaviour on all
1086 systems for all five operations.
1088 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
1091 * .timer units now accept calendar specifications in other timezones
1092 than UTC or the local timezone.
1094 * The tmpfiles snippet var.conf has been changed to create
1095 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
1096 the "utmp" group already, and it appears to be generally understood
1097 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
1098 databases. Previously this was implemented correctly for all these
1099 databases excepts btmp, which has been opened up like this now
1100 too. Note that while the other databases are world-readable
1101 (i.e. 0644), btmp is not and remains more restrictive.
1103 * The systemd-resolve tool gained a new --reset-server-features
1104 switch. When invoked like this systemd-resolved will forget
1105 everything it learnt about the features supported by the configured
1106 upstream DNS servers, and restarts the feature probing logic on the
1107 next resolver look-up for them at the highest feature level
1110 * The status dump systemd-resolved sends to the logs upon receiving
1111 SIGUSR1 now also includes information about all DNS servers it is
1112 configured to use, and the features levels it probed for them.
1114 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
1115 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
1116 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
1117 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
1118 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
1119 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
1120 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
1121 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
1122 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
1123 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
1124 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
1125 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
1126 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
1127 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
1128 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
1129 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
1130 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
1131 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
1132 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
1133 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1135 — Berlin, 2017-10-06
1139 * Meson is now supported as build system in addition to Automake. It is
1140 our plan to remove Automake in one of our next releases, so that
1141 Meson becomes our exclusive build system. Hence, please start using
1142 the Meson build system in your downstream packaging. There's plenty
1143 of documentation around how to use Meson, the extremely brief
1146 ./autogen.sh && ./configure && make && sudo make install
1150 meson build && ninja -C build && sudo ninja -C build install
1152 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
1153 which permits configuring a timeout on the time a job is
1154 running. This is particularly useful for setting timeouts on jobs for
1157 * Unit files gained two new options ConditionUser= and ConditionGroup=
1158 for conditionalizing units based on the identity of the user/group
1159 running a systemd user instance.
1161 * systemd-networkd now understands a new FlowLabel= setting in the
1162 [VXLAN] section of .network files, as well as a Priority= in
1163 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
1164 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
1165 gained support for configuration of GENEVE links, and IPv6 address
1166 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
1168 * .link files now understand a new Port= setting.
1170 * systemd-networkd's DHCP support gained support for DHCP option 119
1171 (domain search list).
1173 * systemd-networkd gained support for serving IPv6 address ranges using
1174 the Router Advertisment protocol. The new .network configuration
1175 section [IPv6Prefix] may be used to configure the ranges to
1176 serve. This is implemented based on a new, minimal, native server
1177 implementation of RA.
1179 * journalctl's --output= switch gained support for a new parameter
1180 "short-iso-precise" for a mode where timestamps are shown as precise
1183 * systemd-udevd's "net_id" builtin may now generate stable network
1184 interface names from IBM PowerVM VIO devices as well as ACPI platform
1187 * MulticastDNS support in systemd-resolved may now be explicitly
1188 enabled/disabled using the new MulticastDNS= configuration file
1191 * systemd-resolved may now optionally use libidn2 instead of the libidn
1192 for processing internationalized domain names. Support for libidn2
1193 should be considered experimental and should not be enabled by
1196 * "machinectl pull-tar" and related call may now do verification of
1197 downloaded images using SUSE-style .sha256 checksum files in addition
1198 to the already existing support for validating using Ubuntu-style
1201 * sd-bus gained support for a new sd_bus_message_appendv() call which
1202 is va_list equivalent of sd_bus_message_append().
1204 * sd-boot gained support for validating images using SHIM/MOK.
1206 * The SMACK code learnt support for "onlycap".
1208 * systemd-mount --umount is now much smarter in figuring out how to
1209 properly unmount a device given its mount or device path.
1211 * The code to call libnss_dns as a fallback from libnss_resolve when
1212 the communication with systemd-resolved fails was removed. This
1213 fallback was redundant and interfered with the [!UNAVAIL=return]
1214 suffix. See nss-resolve(8) for the recommended configuration.
1216 * systemd-logind may now be restarted without losing state. It stores
1217 the file descriptors for devices it manages in the system manager
1218 using the FDSTORE= mechanism. Please note that further changes in
1219 other components may be required to make use of this (for example
1220 Xorg has code to listen for stops of systemd-logind and terminate
1221 itself when logind is stopped or restarted, in order to avoid using
1222 stale file descriptors for graphical devices, which is now
1223 counterproductive and must be reverted in order for restarts of
1224 systemd-logind to be safe. See
1225 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
1227 * All kernel install plugins are called with the environment variable
1228 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
1229 /etc/machine-id. If the file is missing or empty, the variable is
1230 empty and BOOT_DIR_ABS is the path of a temporary directory which is
1231 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
1232 is empty, all plugins should not put anything in BOOT_DIR_ABS.
1234 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
1235 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
1236 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
1237 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
1238 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
1239 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
1240 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
1241 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
1242 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
1243 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
1244 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
1245 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
1246 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
1247 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
1248 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
1249 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
1250 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
1251 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
1252 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
1253 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
1254 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
1255 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
1256 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
1257 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
1258 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
1259 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
1260 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
1261 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
1262 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
1265 — Berlin, 2017-07-12
1269 * The "hybrid" control group mode has been modified to improve
1270 compatibility with "legacy" cgroups-v1 setups. Specifically, the
1271 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
1272 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
1273 cgroups-v1 hierarchy), the only externally visible change being that
1274 the cgroups-v2 hierarchy is also mounted, to
1275 /sys/fs/cgroup/unified. This should provide a large degree of
1276 compatibility with "legacy" cgroups-v1, while taking benefit of the
1277 better management capabilities of cgroups-v2.
1279 * The default control group setup mode may be selected both a boot-time
1280 via a set of kernel command line parameters (specifically:
1281 systemd.unified_cgroup_hierarchy= and
1282 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
1283 default selected on the configure command line
1284 (--with-default-hierarchy=). The upstream default is "hybrid"
1285 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
1286 this will change in a future systemd version to be "unified" (pure
1287 cgroups-v2 mode). The third option for the compile time option is
1288 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
1289 distributions to default to "hybrid" mode for release distributions,
1290 starting with v233. We recommend "unified" for development
1291 distributions (specifically: distributions such as Fedora's rawhide)
1292 as that's where things are headed in the long run. Use "legacy" for
1293 greatest stability and compatibility only.
1295 * Note one current limitation of "unified" and "hybrid" control group
1296 setup modes: the kernel currently does not permit the systemd --user
1297 instance (i.e. unprivileged code) to migrate processes between two
1298 disconnected cgroup subtrees, even if both are managed and owned by
1299 the user. This effectively means "systemd-run --user --scope" doesn't
1300 work when invoked from outside of any "systemd --user" service or
1301 scope. Specifically, it is not supported from session scopes. We are
1302 working on fixing this in a future systemd version. (See #3388 for
1303 further details about this.)
1305 * DBus policy files are now installed into /usr rather than /etc. Make
1306 sure your system has dbus >= 1.9.18 running before upgrading to this
1307 version, or override the install path with --with-dbuspolicydir= .
1309 * All python scripts shipped with systemd (specifically: the various
1310 tests written in Python) now require Python 3.
1312 * systemd unit tests can now run standalone (without the source or
1313 build directories), and can be installed into /usr/lib/systemd/tests/
1314 with 'make install-tests'.
1316 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
1317 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
1320 * Support for the %c, %r, %R specifiers in unit files has been
1321 removed. Specifiers are not supposed to be dependent on configuration
1322 in the unit file itself (so that they resolve the same regardless
1323 where used in the unit files), but these specifiers were influenced
1324 by the Slice= option.
1326 * The shell invoked by debug-shell.service now defaults to /bin/sh in
1327 all cases. If distributions want to use a different shell for this
1328 purpose (for example Fedora's /sbin/sushell) they need to specify
1329 this explicitly at configure time using --with-debug-shell=.
1331 * The confirmation spawn prompt has been reworked to offer the
1334 (c)ontinue, proceed without asking anymore
1335 (D)ump, show the state of the unit
1336 (f)ail, don't execute the command and pretend it failed
1338 (i)nfo, show a short summary of the unit
1339 (j)obs, show jobs that are in progress
1340 (s)kip, don't execute the command and pretend it succeeded
1341 (y)es, execute the command
1343 The 'n' choice for the confirmation spawn prompt has been removed,
1344 because its meaning was confusing.
1346 The prompt may now also be redirected to an alternative console by
1347 specifying the console as parameter to systemd.confirm_spawn=.
1349 * Services of Type=notify require a READY=1 notification to be sent
1350 during startup. If no such message is sent, the service now fails,
1351 even if the main process exited with a successful exit code.
1353 * Services that fail to start up correctly now always have their
1354 ExecStopPost= commands executed. Previously, they'd enter "failed"
1355 state directly, without executing these commands.
1357 * The option MulticastDNS= of network configuration files has acquired
1358 an actual implementation. With MulticastDNS=yes a host can resolve
1359 names of remote hosts and reply to mDNS A and AAAA requests.
1361 * When units are about to be started an additional check is now done to
1362 ensure that all dependencies of type BindsTo= (when used in
1363 combination with After=) have been started.
1365 * systemd-analyze gained a new verb "syscall-filter" which shows which
1366 system call groups are defined for the SystemCallFilter= unit file
1367 setting, and which system calls they contain.
1369 * A new system call filter group "@filesystem" has been added,
1370 consisting of various file system related system calls. Group
1371 "@reboot" has been added, covering reboot, kexec and shutdown related
1372 calls. Finally, group "@swap" has been added covering swap
1373 configuration related calls.
1375 * A new unit file option RestrictNamespaces= has been added that may be
1376 used to restrict access to the various process namespace types the
1377 Linux kernel provides. Specifically, it may be used to take away the
1378 right for a service unit to create additional file system, network,
1379 user, and other namespaces. This sandboxing option is particularly
1380 relevant due to the high amount of recently discovered namespacing
1381 related vulnerabilities in the kernel.
1383 * systemd-udev's .link files gained support for a new AutoNegotiation=
1384 setting for configuring Ethernet auto-negotiation.
1386 * systemd-networkd's .network files gained support for a new
1387 ListenPort= setting in the [DHCP] section to explicitly configure the
1388 UDP client port the DHCP client shall listen on.
1390 * .network files gained a new Unmanaged= boolean setting for explicitly
1391 excluding one or more interfaces from management by systemd-networkd.
1393 * The systemd-networkd ProxyARP= option has been renamed to
1394 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
1395 renamed to ReduceARPProxy=. The old names continue to be available
1398 * systemd-networkd gained support for configuring IPv6 Proxy NDP
1399 addresses via the new IPv6ProxyNDPAddress= .network file setting.
1401 * systemd-networkd's bonding device support gained support for two new
1402 configuration options ActiveSlave= and PrimarySlave=.
1404 * The various options in the [Match] section of .network files gained
1405 support for negative matching.
1407 * New systemd-specific mount options are now understood in /etc/fstab:
1409 x-systemd.mount-timeout= may be used to configure the maximum
1410 permitted runtime of the mount command.
1412 x-systemd.device-bound may be set to bind a mount point to its
1413 backing device unit, in order to automatically remove a mount point
1414 if its backing device is unplugged. This option may also be
1415 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
1416 on the block device, which is now automatically set for all CDROM
1417 drives, so that mounted CDs are automatically unmounted when they are
1418 removed from the drive.
1420 x-systemd.after= and x-systemd.before= may be used to explicitly
1421 order a mount after or before another unit or mount point.
1423 * Enqueued start jobs for device units are now automatically garbage
1424 collected if there are no jobs waiting for them anymore.
1426 * systemctl list-jobs gained two new switches: with --after, for every
1427 queued job the jobs it's waiting for are shown; with --before the
1428 jobs which it's blocking are shown.
1430 * systemd-nspawn gained support for ephemeral boots from disk images
1431 (or in other words: --ephemeral and --image= may now be
1432 combined). Moreover, ephemeral boots are now supported for normal
1433 directories, even if the backing file system is not btrfs. Of course,
1434 if the file system does not support file system snapshots or
1435 reflinks, the initial copy operation will be relatively expensive, but
1436 this should still be suitable for many use cases.
1438 * Calendar time specifications in .timer units now support
1439 specifications relative to the end of a month by using "~" instead of
1440 "-" as separator between month and day. For example, "*-02~03" means
1441 "the third last day in February". In addition a new syntax for
1442 repeated events has been added using the "/" character. For example,
1443 "9..17/2:00" means "every two hours from 9am to 5pm".
1445 * systemd-socket-proxyd gained a new parameter --connections-max= for
1446 configuring the maximum number of concurrent connections.
1448 * sd-id128 gained a new API for generating unique IDs for the host in a
1449 way that does not leak the machine ID. Specifically,
1450 sd_id128_get_machine_app_specific() derives an ID based on the
1451 machine ID a in well-defined, non-reversible, stable way. This is
1452 useful whenever an identifier for the host is needed but where the
1453 identifier shall not be useful to identify the system beyond the
1454 scope of the application itself. (Internally this uses HMAC-SHA256 as
1455 keyed hash function using the machine ID as input.)
1457 * NotifyAccess= gained a new supported value "exec". When set
1458 notifications are accepted from all processes systemd itself invoked,
1459 including all control processes.
1461 * .nspawn files gained support for defining overlay mounts using the
1462 Overlay= and OverlayReadOnly= options. Previously this functionality
1463 was only available on the systemd-nspawn command line.
1465 * systemd-nspawn's --bind= and --overlay= options gained support for
1466 bind/overlay mounts whose source lies within the container tree by
1467 prefixing the source path with "+".
1469 * systemd-nspawn's --bind= and --overlay= options gained support for
1470 automatically allocating a temporary source directory in /var/tmp
1471 that is removed when the container dies. Specifically, if the source
1472 directory is specified as empty string this mechanism is selected. An
1473 example usage is --overlay=+/var::/var, which creates an overlay
1474 mount based on the original /var contained in the image, overlayed
1475 with a temporary directory in the host's /var/tmp. This way changes
1476 to /var are automatically flushed when the container shuts down.
1478 * systemd-nspawn --image= option does now permit raw file system block
1479 devices (in addition to images containing partition tables, as
1482 * The disk image dissection logic in systemd-nspawn gained support for
1483 automatically setting up LUKS encrypted as well as Verity protected
1484 partitions. When a container is booted from an encrypted image the
1485 passphrase is queried at start-up time. When a container with Verity
1486 data is started, the root hash is search in a ".roothash" file
1487 accompanying the disk image (alternatively, pass the root hash via
1488 the new --root-hash= command line option).
1490 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
1491 be used to dissect disk images the same way as systemd-nspawn does
1492 it, following the Bootable Partition Specification. It may even be
1493 used to mount disk images with complex partition setups (including
1494 LUKS and Verity partitions) to a local host directory, in order to
1495 inspect them. This tool is not considered public API (yet), and is
1496 thus not installed into /usr/bin. Please do not rely on its
1497 existence, since it might go away or be changed in later systemd
1500 * A new generator "systemd-verity-generator" has been added, similar in
1501 style to "systemd-cryptsetup-generator", permitting automatic setup of
1502 Verity root partitions when systemd boots up. In order to make use of
1503 this your partition setup should follow the Discoverable Partitions
1504 Specification, and the GPT partition ID of the root file system
1505 partition should be identical to the upper 128bit of the Verity root
1506 hash. The GPT partition ID of the Verity partition protecting it
1507 should be the lower 128bit of the Verity root hash. If the partition
1508 image follows this model it is sufficient to specify a single
1509 "roothash=" kernel command line argument to both configure which root
1510 image and verity partition to use as well as the root hash for
1511 it. Note that systemd-nspawn's Verity support follows the same
1512 semantics, meaning that disk images with proper Verity data in place
1513 may be booted in containers with systemd-nspawn as well as on
1514 physical systems via the verity generator. Also note that the "mkosi"
1515 tool available at https://github.com/systemd/mkosi has been updated
1516 to generate Verity protected disk images following this scheme. In
1517 fact, it has been updated to generate disk images that optionally
1518 implement a complete UEFI SecureBoot trust chain, involving a signed
1519 kernel and initrd image that incorporates such a root hash as well as
1520 a Verity-enabled root partition.
1522 * The hardware database (hwdb) udev supports has been updated to carry
1523 accelerometer quirks.
1525 * All system services are now run with a fresh kernel keyring set up
1526 for them. The invocation ID is stored by default in it, thus
1527 providing a safe, non-overridable way to determine the invocation
1530 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
1531 options for bind mounting arbitrary paths in a service-specific
1532 way. When these options are used, arbitrary host or service files and
1533 directories may be mounted to arbitrary locations in the service's
1536 * Documentation has been added that lists all of systemd's low-level
1537 environment variables:
1539 https://github.com/systemd/systemd/blob/master/doc/ENVIRONMENT.md
1541 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
1542 whether a specific socket file descriptor matches a specified socket
1545 * systemd-firstboot has been updated to check for the
1546 systemd.firstboot= kernel command line option. It accepts a boolean
1547 and when set to false the first boot questions are skipped.
1549 * systemd-fstab-generator has been updated to check for the
1550 systemd.volatile= kernel command line option, which either takes an
1551 optional boolean parameter or the special value "state". If used the
1552 system may be booted in a "volatile" boot mode. Specifically,
1553 "systemd.volatile" is used, the root directory will be mounted as
1554 tmpfs, and only /usr is mounted from the actual root file system. If
1555 "systemd.volatile=state" is used, the root directory will be mounted
1556 as usual, but /var is mounted as tmpfs. This concept provides similar
1557 functionality as systemd-nspawn's --volatile= option, but provides it
1558 on physical boots. Use this option for implementing stateless
1559 systems, or testing systems with all state and/or configuration reset
1560 to the defaults. (Note though that many distributions are not
1561 prepared to boot up without a populated /etc or /var, though.)
1563 * systemd-gpt-auto-generator gained support for LUKS encrypted root
1564 partitions. Previously it only supported LUKS encrypted partitions
1565 for all other uses, except for the root partition itself.
1567 * Socket units gained support for listening on AF_VSOCK sockets for
1568 communication in virtualized QEMU environments.
1570 * The "configure" script gained a new option --with-fallback-hostname=
1571 for specifying the fallback hostname to use if none is configured in
1572 /etc/hostname. For example, by specifying
1573 --with-fallback-hostname=fedora it is possible to default to a
1574 hostname of "fedora" on pristine installations.
1576 * systemd-cgls gained support for a new --unit= switch for listing only
1577 the control groups of a specific unit. Similar --user-unit= has been
1578 added for listing only the control groups of a specific user unit.
1580 * systemd-mount gained a new --umount switch for unmounting a mount or
1581 automount point (and all mount/automount points below it).
1583 * systemd will now refuse full configuration reloads (via systemctl
1584 daemon-reload and related calls) unless at least 16MiB of free space
1585 are available in /run. This is a safety precaution in order to ensure
1586 that generators can safely operate after the reload completed.
1588 * A new unit file option RootImage= has been added, which has a similar
1589 effect as RootDirectory= but mounts the service's root directory from
1590 a disk image instead of plain directory. This logic reuses the same
1591 image dissection and mount logic that systemd-nspawn already uses,
1592 and hence supports any disk images systemd-nspawn supports, including
1593 those following the Discoverable Partition Specification, as well as
1594 Verity enabled images. This option enables systemd to run system
1595 services directly off disk images acting as resource bundles,
1596 possibly even including full integrity data.
1598 * A new MountAPIVFS= unit file option has been added, taking a boolean
1599 argument. If enabled /proc, /sys and /dev (collectively called the
1600 "API VFS") will be mounted for the service. This is only relevant if
1601 RootDirectory= or RootImage= is used for the service, as these mounts
1602 are of course in place in the host mount namespace anyway.
1604 * systemd-nspawn gained support for a new --pivot-root= switch. If
1605 specified the root directory within the container image is pivoted to
1606 the specified mount point, while the original root disk is moved to a
1607 different place. This option enables booting of ostree images
1608 directly with systemd-nspawn.
1610 * The systemd build scripts will no longer complain if the NTP server
1611 addresses are not changed from the defaults. Google now supports
1612 these NTP servers officially. We still recommend downstreams to
1613 properly register an NTP pool with the NTP pool project though.
1615 * coredumpctl gained a new "--reverse" option for printing the list
1616 of coredumps in reverse order.
1618 * coredumpctl will now show additional information about truncated and
1619 inaccessible coredumps, as well as coredumps that are still being
1620 processed. It also gained a new --quiet switch for suppressing
1621 additional informational message in its output.
1623 * coredumpctl gained support for only showing coredumps newer and/or
1624 older than specific timestamps, using the new --since= and --until=
1625 options, reminiscent of journalctl's options by the same name.
1627 * The systemd-coredump logic has been improved so that it may be reused
1628 to collect backtraces in non-compiled languages, for example in
1629 scripting languages such as Python.
1631 * machinectl will now show the UID shift of local containers, if user
1632 namespacing is enabled for them.
1634 * systemd will now optionally run "environment generator" binaries at
1635 configuration load time. They may be used to add environment
1636 variables to the environment block passed to services invoked. One
1637 user environment generator is shipped by default that sets up
1638 environment variables based on files dropped into /etc/environment.d
1639 and ~/.config/environment.d/.
1641 * systemd-resolved now includes the new, recently published 2017 DNSSEC
1644 * hostnamed has been updated to report a new chassis type of
1645 "convertible" to cover "foldable" laptops that can both act as a
1646 tablet and as a laptop, such as various Lenovo Yoga devices.
1648 Contributions from: Adrián López, Alexander Galanin, Alexander
1649 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
1650 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
1651 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
1652 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
1653 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
1654 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
1655 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
1656 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
1657 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
1658 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
1659 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
1660 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
1661 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
1662 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
1663 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
1664 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
1665 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
1666 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
1667 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
1668 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
1669 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
1670 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
1671 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
1672 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
1673 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
1674 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
1675 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
1678 — Berlin, 2017-03-01
1682 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
1683 RestrictAddressFamilies= enabled. These sandboxing options should
1684 generally be compatible with the various external udev call-out
1685 binaries we are aware of, however there may be exceptions, in
1686 particular when exotic languages for these call-outs are used. In
1687 this case, consider turning off these settings locally.
1689 * The new RemoveIPC= option can be used to remove IPC objects owned by
1690 the user or group of a service when that service exits.
1692 * The new ProtectKernelModules= option can be used to disable explicit
1693 load and unload operations of kernel modules by a service. In
1694 addition access to /usr/lib/modules is removed if this option is set.
1696 * ProtectSystem= option gained a new value "strict", which causes the
1697 whole file system tree with the exception of /dev, /proc, and /sys,
1698 to be remounted read-only for a service.
1700 * The new ProtectKernelTunables= option can be used to disable
1701 modification of configuration files in /sys and /proc by a service.
1702 Various directories and files are remounted read-only, so access is
1703 restricted even if the file permissions would allow it.
1705 * The new ProtectControlGroups= option can be used to disable write
1706 access by a service to /sys/fs/cgroup.
1708 * Various systemd services have been hardened with
1709 ProtectKernelTunables=yes, ProtectControlGroups=yes,
1710 RestrictAddressFamilies=.
1712 * Support for dynamically creating users for the lifetime of a service
1713 has been added. If DynamicUser=yes is specified, user and group IDs
1714 will be allocated from the range 61184..65519 for the lifetime of the
1715 service. They can be resolved using the new nss-systemd.so NSS
1716 module. The module must be enabled in /etc/nsswitch.conf. Services
1717 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
1718 any resources allocated by the service will be cleaned up when the
1719 service exits. They also have ProtectHome=read-only and
1720 ProtectSystem=strict enabled, so they are not able to make any
1721 permanent modifications to the system.
1723 * The nss-systemd module also always resolves root and nobody, making
1724 it possible to have no /etc/passwd or /etc/group files in minimal
1725 container or chroot environments.
1727 * Services may be started with their own user namespace using the new
1728 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
1729 under which the service is running are mapped. All other users are
1732 * Support for the cgroup namespace has been added to systemd-nspawn. If
1733 supported by kernel, the container system started by systemd-nspawn
1734 will have its own view of the cgroup hierarchy. This new behaviour
1735 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
1737 * The new MemorySwapMax= option can be used to limit the maximum swap
1738 usage under the unified cgroup hierarchy.
1740 * Support for the CPU controller in the unified cgroup hierarchy has
1741 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
1742 options. This controller requires out-of-tree patches for the kernel
1743 and the support is provisional.
1745 * Mount and automount units may now be created transiently
1746 (i.e. dynamically at runtime via the bus API, instead of requiring
1747 unit files in the file system).
1749 * systemd-mount is a new tool which may mount file systems – much like
1750 mount(8), optionally pulling in additional dependencies through
1751 transient .mount and .automount units. For example, this tool
1752 automatically runs fsck on a backing block device before mounting,
1753 and allows the automount logic to be used dynamically from the
1754 command line for establishing mount points. This tool is particularly
1755 useful when dealing with removable media, as it will ensure fsck is
1756 run – if necessary – before the first access and that the file system
1757 is quickly unmounted after each access by utilizing the automount
1758 logic. This maximizes the chance that the file system on the
1759 removable media stays in a clean state, and if it isn't in a clean
1760 state is fixed automatically.
1762 * LazyUnmount=yes option for mount units has been added to expose the
1763 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
1766 * /efi will be used as the mount point of the EFI boot partition, if
1767 the directory is present, and the mount point was not configured
1768 through other means (e.g. fstab). If /efi directory does not exist,
1769 /boot will be used as before. This makes it easier to automatically
1770 mount the EFI partition on systems where /boot is used for something
1773 * When operating on GPT disk images for containers, systemd-nspawn will
1774 now mount the ESP to /boot or /efi according to the same rules as PID
1775 1 running on a host. This allows tools like "bootctl" to operate
1776 correctly within such containers, in order to make container images
1777 bootable on physical systems.
1779 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
1781 * Two new user session targets have been added to support running
1782 graphical sessions under the systemd --user instance:
1783 graphical-session.target and graphical-session-pre.target. See
1784 systemd.special(7) for a description of how those targets should be
1787 * The vconsole initialization code has been significantly reworked to
1788 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
1789 support unicode keymaps. Font and keymap configuration will now be
1790 copied to all allocated virtual consoles.
1792 * FreeBSD's bhyve virtualization is now detected.
1794 * Information recorded in the journal for core dumps now includes the
1795 contents of /proc/mountinfo and the command line of the process at
1796 the top of the process hierarchy (which is usually the init process
1799 * systemd-journal-gatewayd learned the --directory= option to serve
1800 files from the specified location.
1802 * journalctl --root=… can be used to peruse the journal in the
1803 /var/log/ directories inside of a container tree. This is similar to
1804 the existing --machine= option, but does not require the container to
1807 * The hardware database has been extended to support
1808 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
1811 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
1812 specify the click rate for mice which include a horizontal wheel with
1813 a click rate that is different than the one for the vertical wheel.
1815 * systemd-run gained a new --wait option that makes service execution
1816 synchronous. (Specifically, the command will not return until the
1817 specified service binary exited.)
1819 * systemctl gained a new --wait option that causes the start command to
1820 wait until the units being started have terminated again.
1822 * A new journal output mode "short-full" has been added which displays
1823 timestamps with abbreviated English day names and adds a timezone
1824 suffix. Those timestamps include more information than the default
1825 "short" output mode, and can be passed directly to journalctl's
1826 --since= and --until= options.
1828 * /etc/resolv.conf will be bind-mounted into containers started by
1829 systemd-nspawn, if possible, so any changes to resolv.conf contents
1830 are automatically propagated to the container.
1832 * The number of instances for socket-activated services originating
1833 from a single IP address can be limited with
1834 MaxConnectionsPerSource=, extending the existing setting of
1837 * systemd-networkd gained support for vcan ("Virtual CAN") interface
1840 * .netdev and .network configuration can now be extended through
1843 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
1844 Segmentation Offload, Generic Receive Offload, Large Receive Offload
1845 can be enabled and disabled using the new UDPSegmentationOffload=,
1846 TCPSegmentationOffload=, GenericSegmentationOffload=,
1847 GenericReceiveOffload=, LargeReceiveOffload= options in the
1848 [Link] section of .link files.
1850 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
1851 Port VLAN ID can be configured for bridge devices using the new STP=,
1852 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
1853 section of .netdev files.
1855 * The route table to which routes received over DHCP or RA should be
1856 added can be configured with the new RouteTable= option in the [DHCP]
1857 and [IPv6AcceptRA] sections of .network files.
1859 * The Address Resolution Protocol can be disabled on links managed by
1860 systemd-networkd using the ARP=no setting in the [Link] section of
1863 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
1864 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
1865 encode information about the result and exit codes of the current
1866 service runtime cycle.
1868 * systemd-sysctl will now configure kernel parameters in the order
1869 they occur in the configuration files. This matches what sysctl
1870 has been traditionally doing.
1872 * kernel-install "plugins" that are executed to perform various
1873 tasks after a new kernel is added and before an old one is removed
1874 can now return a special value to terminate the procedure and
1875 prevent any later plugins from running.
1877 * Journald's SplitMode=login setting has been deprecated. It has been
1878 removed from documentation, and its use is discouraged. In a future
1879 release it will be completely removed, and made equivalent to current
1880 default of SplitMode=uid.
1882 * Storage=both option setting in /etc/systemd/coredump.conf has been
1883 removed. With fast LZ4 compression storing the core dump twice is not
1886 * The --share-system systemd-nspawn option has been replaced with an
1887 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
1888 this functionality is discouraged. In addition the variables
1889 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
1890 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
1891 individual namespaces.
1893 * "machinectl list" now shows the IP address of running containers in
1894 the output, as well as OS release information.
1896 * "loginctl list" now shows the TTY of each session in the output.
1898 * sd-bus gained new API calls sd_bus_track_set_recursive(),
1899 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
1900 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
1901 tracking objects in a "recursive" mode, where a single client can be
1902 counted multiple times, if it takes multiple references.
1904 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
1905 sd_bus_get_exit_on_disconnect(). They may be used to to make a
1906 process using sd-bus automatically exit if the bus connection is
1909 * Bus clients of the service manager may now "pin" loaded units into
1910 memory, by taking an explicit reference on them. This is useful to
1911 ensure the client can retrieve runtime data about the service even
1912 after the service completed execution. Taking such a reference is
1913 available only for privileged clients and should be helpful to watch
1914 running services in a race-free manner, and in particular collect
1915 information about exit statuses and results.
1917 * The nss-resolve module has been changed to strictly return UNAVAIL
1918 when communication via D-Bus with resolved failed, and NOTFOUND when
1919 a lookup completed but was negative. This means it is now possible to
1920 neatly configure fallbacks using nsswitch.conf result checking
1921 expressions. Taking benefit of this, the new recommended
1922 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
1924 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
1926 * A new setting CtrlAltDelBurstAction= has been added to
1927 /etc/systemd/system.conf which may be used to configure the precise
1928 behaviour if the user on the console presses Ctrl-Alt-Del more often
1929 than 7 times in 2s. Previously this would unconditionally result in
1930 an expedited, immediate reboot. With this new setting the precise
1931 operation may be configured in more detail, and also turned off
1934 * In .netdev files two new settings RemoteChecksumTx= and
1935 RemoteChecksumRx= are now understood that permit configuring the
1936 remote checksumming logic for VXLAN networks.
1938 * The service manager learnt a new "invocation ID" concept for invoked
1939 services. Each runtime cycle of a service will get a new invocation
1940 ID (a 128bit random UUID) assigned that identifies the current
1941 run of the service uniquely and globally. A new invocation ID
1942 is generated each time a service starts up. The journal will store
1943 the invocation ID of a service along with any logged messages, thus
1944 making the invocation ID useful for matching the online runtime of a
1945 service with the offline log data it generated in a safe way without
1946 relying on synchronized timestamps. In many ways this new service
1947 invocation ID concept is similar to the kernel's boot ID concept that
1948 uniquely and globally identifies the runtime of each boot. The
1949 invocation ID of a service is passed to the service itself via an
1950 environment variable ($INVOCATION_ID). A new bus call
1951 GetUnitByInvocationID() has been added that is similar to GetUnit()
1952 but instead of retrieving the bus path for a unit by its name
1953 retrieves it by its invocation ID. The returned path is valid only as
1954 long as the passed invocation ID is current.
1956 * systemd-resolved gained a new "DNSStubListener" setting in
1957 resolved.conf. It either takes a boolean value or the special values
1958 "udp" and "tcp", and configures whether to enable the stub DNS
1959 listener on 127.0.0.53:53.
1961 * IP addresses configured via networkd may now carry additional
1962 configuration settings supported by the kernel. New options include:
1963 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
1964 PrefixRoute=, AutoJoin=.
1966 * The PAM configuration fragment file for "user@.service" shipped with
1967 systemd (i.e. the --user instance of systemd) has been stripped to
1968 the minimum necessary to make the system boot. Previously, it
1969 contained Fedora-specific stanzas that did not apply to other
1970 distributions. It is expected that downstream distributions add
1971 additional configuration lines, matching their needs to this file,
1972 using it only as rough template of what systemd itself needs. Note
1973 that this reduced fragment does not even include an invocation of
1974 pam_limits which most distributions probably want to add, even though
1975 systemd itself does not need it. (There's also the new build time
1976 option --with-pamconfdir=no to disable installation of the PAM
1979 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
1980 capability is now also dropped from its set (in addition to
1981 CAP_SYS_MKNOD as before).
1983 * In service unit files it is now possible to connect a specific named
1984 file descriptor with stdin/stdout/stdout of an executed service. The
1985 name may be specified in matching .socket units using the
1986 FileDescriptorName= setting.
1988 * A number of journal settings may now be configured on the kernel
1989 command line. Specifically, the following options are now understood:
1990 systemd.journald.max_level_console=,
1991 systemd.journald.max_level_store=,
1992 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
1993 systemd.journald.max_level_wall=.
1995 * "systemctl is-enabled --full" will now show by which symlinks a unit
1996 file is enabled in the unit dependency tree.
1998 * Support for VeraCrypt encrypted partitions has been added to the
1999 "cryptsetup" logic and /etc/crypttab.
2001 * systemd-detect-virt gained support for a new --private-users switch
2002 that checks whether the invoking processes are running inside a user
2003 namespace. Similar, a new special value "private-users" for the
2004 existing ConditionVirtualization= setting has been added, permitting
2005 skipping of specific units in user namespace environments.
2007 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
2008 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
2009 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
2010 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
2011 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
2012 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
2013 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
2014 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
2015 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
2016 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
2017 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
2018 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
2019 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
2020 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
2021 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
2022 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
2023 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
2024 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
2025 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
2026 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
2027 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
2028 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
2029 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
2030 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
2031 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
2032 Jędrzejewski-Szmek, Zeal Jagannatha
2034 — Santa Fe, 2016-11-03
2038 * In service units the various ExecXYZ= settings have been extended
2039 with an additional special character as first argument of the
2040 assigned value: if the character '+' is used the specified command
2041 line it will be run with full privileges, regardless of User=,
2042 Group=, CapabilityBoundingSet= and similar options. The effect is
2043 similar to the existing PermissionsStartOnly= option, but allows
2044 configuration of this concept for each executed command line
2047 * Services may now alter the service watchdog timeout at runtime by
2048 sending a WATCHDOG_USEC= message via sd_notify().
2050 * MemoryLimit= and related unit settings now optionally take percentage
2051 specifications. The percentage is taken relative to the amount of
2052 physical memory in the system (or in case of containers, the assigned
2053 amount of memory). This allows scaling service resources neatly with
2054 the amount of RAM available on the system. Similarly, systemd-logind's
2055 RuntimeDirectorySize= option now also optionally takes percentage
2058 * In similar fashion TasksMax= takes percentage values now, too. The
2059 value is taken relative to the configured maximum number of processes
2060 on the system. The per-service task maximum has been changed to 15%
2061 using this functionality. (Effectively this is an increase of 512 →
2062 4915 for service units, given the kernel's default pid_max setting.)
2064 * Calendar time specifications in .timer units now understand a ".."
2065 syntax for time ranges. Example: "4..7:10" may now be used for
2066 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
2069 * The InaccessableDirectories=, ReadOnlyDirectories= and
2070 ReadWriteDirectories= unit file settings have been renamed to
2071 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
2072 applied to all kinds of file nodes, and not just directories, with
2073 the exception of symlinks. Specifically these settings may now be
2074 used on block and character device nodes, UNIX sockets and FIFOS as
2075 well as regular files. The old names of these settings remain
2076 available for compatibility.
2078 * systemd will now log about all service processes it kills forcibly
2079 (using SIGKILL) because they remained after the clean shutdown phase
2080 of the service completed. This should help identifying services that
2081 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
2082 systemd-logind's configuration a similar log message is generated for
2083 processes killed at the end of each session due to this setting.
2085 * systemd will now set the $JOURNAL_STREAM environment variable for all
2086 services whose stdout/stderr are connected to the Journal (which
2087 effectively means by default: all services). The variable contains
2088 the device and inode number of the file descriptor used for
2089 stdout/stderr. This may be used by invoked programs to detect whether
2090 their stdout/stderr is connected to the Journal, in which case they
2091 can switch over to direct Journal communication, thus being able to
2092 pass extended, structured metadata along with their log messages. As
2093 one example, this is now used by glib's logging primitives.
2095 * When using systemd's default tmp.mount unit for /tmp, the mount point
2096 will now be established with the "nosuid" and "nodev" options. This
2097 avoids privilege escalation attacks that put traps and exploits into
2098 /tmp. However, this might cause problems if you e. g. put container
2099 images or overlays into /tmp; if you need this, override tmp.mount's
2100 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
2103 * systemd now supports the "memory" cgroup controller also on
2106 * The systemd-cgtop tool now optionally takes a control group path as
2107 command line argument. If specified, the control group list shown is
2108 limited to subgroups of that group.
2110 * The SystemCallFilter= unit file setting gained support for
2111 pre-defined, named system call filter sets. For example
2112 SystemCallFilter=@clock is now an effective way to make all clock
2113 changing-related system calls unavailable to a service. A number of
2114 similar pre-defined groups are defined. Writing system call filters
2115 for system services is simplified substantially with this new
2116 concept. Accordingly, all of systemd's own, long-running services now
2117 enable system call filtering based on this, by default.
2119 * A new service setting MemoryDenyWriteExecute= has been added, taking
2120 a boolean value. If turned on, a service may no longer create memory
2121 mappings that are writable and executable at the same time. This
2122 enhances security for services where this is enabled as it becomes
2123 harder to dynamically write and then execute memory in exploited
2124 service processes. This option has been enabled for all of systemd's
2125 own long-running services.
2127 * A new RestrictRealtime= service setting has been added, taking a
2128 boolean argument. If set the service's processes may no longer
2129 acquire realtime scheduling. This improves security as realtime
2130 scheduling may otherwise be used to easily freeze the system.
2132 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
2133 value. This may be used for requesting that the system manager inside
2134 of the container reports start-up completion to nspawn which then
2135 propagates this notification further to the service manager
2136 supervising nspawn itself. A related option NotifyReady= in .nspawn
2137 files has been added too. This functionality allows ordering of the
2138 start-up of multiple containers using the usual systemd ordering
2141 * machinectl gained a new command "stop" that is an alias for
2144 * systemd-resolved gained support for contacting DNS servers on
2145 link-local IPv6 addresses.
2147 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
2148 its caches. A method call for requesting the same operation has been
2149 added to the bus API too, and is made available via "systemd-resolve
2152 * systemd-resolve gained a new --status switch. If passed a brief
2153 summary of the used DNS configuration with per-interface information
2156 * resolved.conf gained a new Cache= boolean option, defaulting to
2157 on. If turned off local DNS caching is disabled. This comes with a
2158 performance penalty in particular when DNSSEC is enabled. Note that
2159 resolved disables its internal caching implicitly anyway, when the
2160 configured DNS server is on a host-local IP address such as ::1 or
2161 127.0.0.1, thus automatically avoiding double local caching.
2163 * systemd-resolved now listens on the local IP address 127.0.0.53:53
2164 for DNS requests. This improves compatibility with local programs
2165 that do not use the libc NSS or systemd-resolved's bus APIs for name
2166 resolution. This minimal DNS service is only available to local
2167 programs and does not implement the full DNS protocol, but enough to
2168 cover local DNS clients. A new, static resolv.conf file, listing just
2169 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
2170 now recommended to make /etc/resolv.conf a symlink to this file in
2171 order to route all DNS lookups to systemd-resolved, regardless if
2172 done via NSS, the bus API or raw DNS packets. Note that this local
2173 DNS service is not as fully featured as the libc NSS or
2174 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
2175 used to deliver link-local address information (as this implies
2176 sending a local interface index along), LLMNR/mDNS support via this
2177 interface is severely restricted. It is thus strongly recommended for
2178 all applications to use the libc NSS API or native systemd-resolved
2181 * systemd-networkd's bridge support learned a new setting
2182 VLANFiltering= for controlling VLAN filtering. Moreover a new section
2183 in .network files has been added for configuring VLAN bridging in
2184 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
2186 * systemd-networkd's IPv6 Router Advertisement code now makes use of
2187 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
2188 now be acquired without relying on DHCPv6. Two new options
2189 UseDomains= and UseDNS= have been added to configure this behaviour.
2191 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
2192 renamed IPv6AcceptRA=, without altering its behaviour. The old
2193 setting name remains available for compatibility reasons.
2195 * The systemd-networkd VTI/VTI6 tunneling support gained new options
2196 Key=, InputKey= and OutputKey=.
2198 * systemd-networkd gained support for VRF ("Virtual Routing Function")
2199 interface configuration.
2201 * "systemctl edit" may now be used to create new unit files by
2202 specifying the --force switch.
2204 * sd-event gained a new function sd_event_get_iteration() for
2205 requesting the current iteration counter of the event loop. It starts
2206 at zero and is increased by one with each event loop iteration.
2208 * A new rpm macro %systemd_ordering is provided by the macros.systemd
2209 file. It can be used in lieu of %systemd_requires in packages which
2210 don't use any systemd functionality and are intended to be installed
2211 in minimal containers without systemd present. This macro provides
2212 ordering dependencies to ensure that if the package is installed in
2213 the same rpm transaction as systemd, systemd will be installed before
2214 the scriptlets for the package are executed, allowing unit presets
2217 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
2218 been added to simplify packaging of generators.
2220 * The os-release file gained VERSION_CODENAME field for the
2221 distribution nickname (e.g. VERSION_CODENAME=woody).
2223 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
2224 can be set to disable parsing of metadata and the creation
2225 of persistent symlinks for that device.
2227 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
2228 to make them available to logged-in users has been reverted.
2230 * Much of the common code of the various systemd components is now
2231 built into an internal shared library libsystemd-shared-231.so
2232 (incorporating the systemd version number in the name, to be updated
2233 with future releases) that the components link to. This should
2234 decrease systemd footprint both in memory during runtime and on
2235 disk. Note that the shared library is not for public use, and is
2236 neither API nor ABI stable, but is likely to change with every new
2237 released update. Packagers need to make sure that binaries
2238 linking to libsystemd-shared.so are updated in step with the
2241 * Configuration for "mkosi" is now part of the systemd
2242 repository. mkosi is a tool to easily build legacy-free OS images,
2243 and is available on github: https://github.com/systemd/mkosi. If
2244 "mkosi" is invoked in the build tree a new raw OS image is generated
2245 incorporating the systemd sources currently being worked on and a
2246 clean, fresh distribution installation. The generated OS image may be
2247 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
2248 UEFI PC. This functionality is particularly useful to easily test
2249 local changes made to systemd in a pristine, defined environment. See
2250 doc/HACKING for details.
2252 * configure learned the --with-support-url= option to specify the
2253 distribution's bugtracker.
2255 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
2256 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
2257 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
2258 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
2259 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
2260 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
2261 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
2262 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
2263 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
2264 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
2265 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
2266 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
2267 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
2268 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
2269 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
2270 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
2271 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
2272 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
2273 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
2275 — Berlin, 2016-07-25
2279 * DNSSEC is now turned on by default in systemd-resolved (in
2280 "allow-downgrade" mode), but may be turned off during compile time by
2281 passing "--with-default-dnssec=no" to "configure" (and of course,
2282 during runtime with DNSSEC= in resolved.conf). We recommend
2283 downstreams to leave this on at least during development cycles and
2284 report any issues with the DNSSEC logic upstream. We are very
2285 interested in collecting feedback about the DNSSEC validator and its
2286 limitations in the wild. Note however, that DNSSEC support is
2287 probably nothing downstreams should turn on in stable distros just
2288 yet, as it might create incompatibilities with a few DNS servers and
2289 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
2290 automatically whenever we detect such incompatible setups, but there
2291 might be systems we do not cover yet. Hence: please help us testing
2292 the DNSSEC code, leave this on where you can, report back, but then
2293 again don't consider turning this on in your stable, LTS or
2294 production release just yet. (Note that you have to enable
2295 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
2296 and its DNSSEC mode for host name resolution from local
2299 * systemd-resolve conveniently resolves DANE records with the --tlsa
2300 option and OPENPGPKEY records with the --openpgp option. It also
2301 supports dumping raw DNS record data via the new --raw= switch.
2303 * systemd-logind will now by default terminate user processes that are
2304 part of the user session scope unit (session-XX.scope) when the user
2305 logs out. This behavior is controlled by the KillUserProcesses=
2306 setting in logind.conf, and the previous default of "no" is now
2307 changed to "yes". This means that user sessions will be properly
2308 cleaned up after, but additional steps are necessary to allow
2309 intentionally long-running processes to survive logout.
2311 While the user is logged in at least once, user@.service is running,
2312 and any service that should survive the end of any individual login
2313 session can be started at a user service or scope using systemd-run.
2314 systemd-run(1) man page has been extended with an example which shows
2315 how to run screen in a scope unit underneath user@.service. The same
2316 command works for tmux.
2318 After the user logs out of all sessions, user@.service will be
2319 terminated too, by default, unless the user has "lingering" enabled.
2320 To effectively allow users to run long-term tasks even if they are
2321 logged out, lingering must be enabled for them. See loginctl(1) for
2322 details. The default polkit policy was modified to allow users to
2323 set lingering for themselves without authentication.
2325 Previous defaults can be restored at compile time by the
2326 --without-kill-user-processes option to "configure".
2328 * systemd-logind gained new configuration settings SessionsMax= and
2329 InhibitorsMax=, both with a default of 8192. It will not register new
2330 user sessions or inhibitors above this limit.
2332 * systemd-logind will now reload configuration on SIGHUP.
2334 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
2335 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
2336 enable. Also, support for the "io" cgroup controller in the unified
2337 hierarchy has been added, so that the "memory", "pids" and "io" are
2338 now the controllers that are supported on the unified hierarchy.
2340 WARNING: it is not possible to use previous systemd versions with
2341 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
2342 is necessary to also update systemd in the initramfs if using the
2343 unified hierarchy. An updated SELinux policy is also required.
2345 * LLDP support has been extended, and both passive (receive-only) and
2346 active (sender) modes are supported. Passive mode ("routers-only") is
2347 enabled by default in systemd-networkd. Active LLDP mode is enabled
2348 by default for containers on the internal network. The "networkctl
2349 lldp" command may be used to list information gathered. "networkctl
2350 status" will also show basic LLDP information on connected peers now.
2352 * The IAID and DUID unique identifier sent in DHCP requests may now be
2353 configured for the system and each .network file managed by
2354 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
2356 * systemd-networkd gained support for configuring proxy ARP support for
2357 each interface, via the ProxyArp= setting in .network files. It also
2358 gained support for configuring the multicast querier feature of
2359 bridge devices, via the new MulticastQuerier= setting in .netdev
2360 files. Similarly, snooping on the IGMP traffic can be controlled
2361 via the new setting MulticastSnooping=.
2363 A new setting PreferredLifetime= has been added for addresses
2364 configured in .network file to configure the lifetime intended for an
2367 The systemd-networkd DHCP server gained the option EmitRouter=, which
2368 defaults to yes, to configure whether the DHCP Option 3 (Router)
2371 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
2372 systemd-socket-activate and installed into /usr/bin. It is now fully
2375 * systemd-journald now uses separate threads to flush changes to disk
2376 when closing journal files, thus reducing impact of slow disk I/O on
2377 logging performance.
2379 * The sd-journal API gained two new calls
2380 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
2381 can be used to open journal files using file descriptors instead of
2382 file or directory paths. sd_journal_open_container() has been
2383 deprecated, sd_journal_open_directory_fd() should be used instead
2384 with the flag SD_JOURNAL_OS_ROOT.
2386 * journalctl learned a new output mode "-o short-unix" that outputs log
2387 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
2388 UTC). It also gained support for a new --no-hostname setting to
2389 suppress the hostname column in the family of "short" output modes.
2391 * systemd-ask-password now optionally skips printing of the password to
2392 stdout with --no-output which can be useful in scripts.
2394 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
2395 (devices tagged with ID_MAKER_TOOL) are now tagged with
2396 "uaccess" and are available to logged in users.
2398 * The DeviceAllow= unit setting now supports specifiers (with "%").
2400 * "systemctl show" gained a new --value switch, which allows print a
2401 only the contents of a specific unit property, without also printing
2402 the property's name. Similar support was added to "show*" verbs
2403 of loginctl and machinectl that output "key=value" lists.
2405 * A new unit type "generated" was added for files dynamically generated
2406 by generator tools. Similarly, a new unit type "transient" is used
2407 for unit files created using the runtime API. "systemctl enable" will
2408 refuse to operate on such files.
2410 * A new command "systemctl revert" has been added that may be used to
2411 revert to the vendor version of a unit file, in case local changes
2412 have been made by adding drop-ins or overriding the unit file.
2414 * "machinectl clean" gained a new verb to automatically remove all or
2415 just hidden container images.
2417 * systemd-tmpfiles gained support for a new line type "e" for emptying
2418 directories, if they exist, without creating them if they don't.
2420 * systemd-nspawn gained support for automatically patching the UID/GIDs
2421 of the owners and the ACLs of all files and directories in a
2422 container tree to match the UID/GID user namespacing range selected
2423 for the container invocation. This mode is enabled via the new
2424 --private-users-chown switch. It also gained support for
2425 automatically choosing a free, previously unused UID/GID range when
2426 starting a container, via the new --private-users=pick setting (which
2427 implies --private-users-chown). Together, these options for the first
2428 time make user namespacing for nspawn containers fully automatic and
2429 thus deployable. The systemd-nspawn@.service template unit file has
2430 been changed to use this functionality by default.
2432 * systemd-nspawn gained a new --network-zone= switch, that allows
2433 creating ad-hoc virtual Ethernet links between multiple containers,
2434 that only exist as long as at least one container referencing them is
2435 running. This allows easy connecting of multiple containers with a
2436 common link that implements an Ethernet broadcast domain. Each of
2437 these network "zones" may be named relatively freely by the user, and
2438 may be referenced by any number of containers, but each container may
2439 only reference one of these "zones". On the lower level, this is
2440 implemented by an automatically managed bridge network interface for
2441 each zone, that is created when the first container referencing its
2442 zone is created and removed when the last one referencing its zone
2445 * The default start timeout may now be configured on the kernel command
2446 line via systemd.default_timeout_start_sec=. It was already
2447 configurable via the DefaultTimeoutStartSec= option in
2448 /etc/systemd/system.conf.
2450 * Socket units gained a new TriggerLimitIntervalSec= and
2451 TriggerLimitBurst= setting to configure a limit on the activation
2452 rate of the socket unit.
2454 * The LimitNICE= setting now optionally takes normal UNIX nice values
2455 in addition to the raw integer limit value. If the specified
2456 parameter is prefixed with "+" or "-" and is in the range -20..19 the
2457 value is understood as UNIX nice value. If not prefixed like this it
2458 is understood as raw RLIMIT_NICE limit.
2460 * Note that the effect of the PrivateDevices= unit file setting changed
2461 slightly with this release: the per-device /dev file system will be
2462 mounted read-only from this version on, and will have "noexec"
2463 set. This (minor) change of behavior might cause some (exceptional)
2464 legacy software to break, when PrivateDevices=yes is set for its
2465 service. Please leave PrivateDevices= off if you run into problems
2468 * systemd-bootchart has been split out to a separate repository:
2469 https://github.com/systemd/systemd-bootchart
2471 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
2472 merged into the kernel in its current form.
2474 * The compatibility libraries libsystemd-daemon.so,
2475 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
2476 which have been deprecated since systemd-209 have been removed along
2477 with the corresponding pkg-config files. All symbols provided by
2478 those libraries are provided by libsystemd.so.
2480 * The Capabilities= unit file setting has been removed (it is ignored
2481 for backwards compatibility). AmbientCapabilities= and
2482 CapabilityBoundingSet= should be used instead.
2484 * A new special target has been added, initrd-root-device.target,
2485 which creates a synchronization point for dependencies of the root
2486 device in early userspace. Initramfs builders must ensure that this
2487 target is now included in early userspace.
2489 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
2490 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
2491 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
2492 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
2493 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
2494 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
2495 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
2496 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
2497 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
2498 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
2499 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
2500 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
2501 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
2502 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
2503 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
2504 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
2505 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
2506 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
2507 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
2508 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
2509 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
2510 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
2511 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
2512 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
2513 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
2516 — Fairfax, 2016-05-21
2520 * The systemd-resolved DNS resolver service has gained a substantial
2521 set of new features, most prominently it may now act as a DNSSEC
2522 validating stub resolver. DNSSEC mode is currently turned off by
2523 default, but is expected to be turned on by default in one of the
2524 next releases. For now, we invite everybody to test the DNSSEC logic
2525 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
2526 service also gained a full set of D-Bus interfaces, including calls
2527 to configure DNS and DNSSEC settings per link (for use by external
2528 network management software). systemd-resolved and systemd-networkd
2529 now distinguish between "search" and "routing" domains. The former
2530 are used to qualify single-label names, the latter are used purely
2531 for routing lookups within certain domains to specific links.
2532 resolved now also synthesizes RRs for all entries from /etc/hosts.
2534 * The systemd-resolve tool (which is a client utility for
2535 systemd-resolved) has been improved considerably and is now fully
2536 supported and documented. Hence it has moved from /usr/lib/systemd to
2539 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
2542 * The coredump collection logic has been reworked: when a coredump is
2543 collected it is now written to disk, compressed and processed
2544 (including stacktrace extraction) from a new instantiated service
2545 systemd-coredump@.service, instead of directly from the
2546 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
2547 processing large coredumps can take up a substantial amount of
2548 resources and time, and this previously happened entirely outside of
2549 systemd's service supervision. With the new logic the core_pattern
2550 hook only does minimal metadata collection before passing off control
2551 to the new instantiated service, which is configured with a time
2552 limit, a nice level and other settings to minimize negative impact on
2553 the rest of the system. Also note that the new logic will honour the
2554 RLIMIT_CORE setting of the crashed process, which now allows users
2555 and processes to turn off coredumping for their processes by setting
2558 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
2559 and all forked processes by default. Previously, PID 1 would leave
2560 the setting at "0" for all processes, as set by the kernel. Note that
2561 the resource limit traditionally has no effect on the generated
2562 coredumps on the system if the /proc/sys/kernel/core_pattern hook
2563 logic is used. Since the limit is now honoured (see above) its
2564 default has been changed so that the coredumping logic is enabled by
2565 default for all processes, while allowing specific opt-out.
2567 * When the stacktrace is extracted from processes of system users, this
2568 is now done as "systemd-coredump" user, in order to sandbox this
2569 potentially security sensitive parsing operation. (Note that when
2570 processing coredumps of normal users this is done under the user ID
2571 of process that crashed, as before.) Packagers should take notice
2572 that it is now necessary to create the "systemd-coredump" system user
2573 and group at package installation time.
2575 * The systemd-activate socket activation testing tool gained support
2576 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
2577 and --seqpacket switches. It also has been extended to support both
2578 new-style and inetd-style file descriptor passing. Use the new
2579 --inetd switch to request inetd-style file descriptor passing.
2581 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
2582 variable, which takes a boolean value. If set to false, ANSI color
2583 output is disabled in the tools even when run on a terminal that
2586 * The VXLAN support in networkd now supports two new settings
2587 DestinationPort= and PortRange=.
2589 * A new systemd.machine_id= kernel command line switch has been added,
2590 that may be used to set the machine ID in /etc/machine-id if it is
2591 not initialized yet. This command line option has no effect if the
2592 file is already initialized.
2594 * systemd-nspawn gained a new --as-pid2 switch that invokes any
2595 specified command line as PID 2 rather than PID 1 in the
2596 container. In this mode PID 1 is a minimal stub init process that
2597 implements the special POSIX and Linux semantics of PID 1 regarding
2598 signal and child process management. Note that this stub init process
2599 is implemented in nspawn itself and requires no support from the
2600 container image. This new logic is useful to support running
2601 arbitrary commands in the container, as normal processes are
2602 generally not prepared to run as PID 1.
2604 * systemd-nspawn gained a new --chdir= switch for setting the current
2605 working directory for the process started in the container.
2607 * "journalctl /dev/sda" will now output all kernel log messages for
2608 specified device from the current boot, in addition to all devices
2609 that are parents of it. This should make log output about devices
2610 pretty useful, as long as kernel drivers attach enough metadata to
2611 the log messages. (The usual SATA drivers do.)
2613 * The sd-journal API gained two new calls
2614 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
2615 that report whether log data from /run or /var has been found.
2617 * journalctl gained a new switch "--fields" that prints all journal
2618 record field names currently in use in the journal. This is backed
2619 by two new sd-journal API calls sd_journal_enumerate_fields() and
2620 sd_journal_restart_fields().
2622 * Most configurable timeouts in systemd now expect an argument of
2623 "infinity" to turn them off, instead of "0" as before. The semantics
2624 from now on is that a timeout of "0" means "now", and "infinity"
2625 means "never". To maintain backwards compatibility, "0" continues to
2626 turn off previously existing timeout settings.
2628 * "systemctl reload-or-try-restart" has been renamed to "systemctl
2629 try-reload-or-restart" to clarify what it actually does: the "try"
2630 logic applies to both reloading and restarting, not just restarting.
2631 The old name continues to be accepted for compatibility.
2633 * On boot-up, when PID 1 detects that the system clock is behind the
2634 release date of the systemd version in use, the clock is now set
2635 to the latter. Previously, this was already done in timesyncd, in order
2636 to avoid running with clocks set to the various clock epochs such as
2637 1902, 1938 or 1970. With this change the logic is now done in PID 1
2638 in addition to timesyncd during early boot-up, so that it is enforced
2639 before the first process is spawned by systemd. Note that the logic
2640 in timesyncd remains, as it is more comprehensive and ensures
2641 clock monotonicity by maintaining a persistent timestamp file in
2642 /var. Since /var is generally not available in earliest boot or the
2643 initrd, this part of the logic remains in timesyncd, and is not done
2646 * Support for tweaking details in net_cls.class_id through the
2647 NetClass= configuration directive has been removed, as the kernel
2648 people have decided to deprecate that controller in cgroup v2.
2649 Userspace tools such as nftables are moving over to setting rules
2650 that are specific to the full cgroup path of a task, which obsoletes
2651 these controllers anyway. The NetClass= directive is kept around for
2652 legacy compatibility reasons. For a more in-depth description of the
2653 kernel change, please refer to the respective upstream commit:
2655 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
2657 * A new service setting RuntimeMaxSec= has been added that may be used
2658 to specify a maximum runtime for a service. If the timeout is hit, the
2659 service is terminated and put into a failure state.
2661 * A new service setting AmbientCapabilities= has been added. It allows
2662 configuration of additional Linux process capabilities that are
2663 passed to the activated processes. This is only available on very
2666 * The process resource limit settings in service units may now be used
2667 to configure hard and soft limits individually.
2669 * The various libsystemd APIs such as sd-bus or sd-event now publicly
2670 expose support for gcc's __attribute__((cleanup())) C extension.
2671 Specifically, for many object destructor functions alternative
2672 versions have been added that have names suffixed with "p" and take a
2673 pointer to a pointer to the object to destroy, instead of just a
2674 pointer to the object itself. This is useful because these destructor
2675 functions may be used directly as parameters to the cleanup
2676 construct. Internally, systemd has been a heavy user of this GCC
2677 extension for a long time, and with this change similar support is
2678 now available to consumers of the library outside of systemd. Note
2679 that by using this extension in your sources compatibility with old
2680 and strictly ANSI compatible C compilers is lost. However, all gcc or
2681 LLVM versions of recent years support this extension.
2683 * Timer units gained support for a new setting RandomizedDelaySec= that
2684 allows configuring some additional randomized delay to the configured
2685 time. This is useful to spread out timer events to avoid load peaks in
2686 clusters or larger setups.
2688 * Calendar time specifications now support sub-second accuracy.
2690 * Socket units now support listening on SCTP and UDP-lite protocol
2693 * The sd-event API now comes with a full set of man pages.
2695 * Older versions of systemd contained experimental support for
2696 compressing journal files and coredumps with the LZ4 compressor that
2697 was not compatible with the lz4 binary (due to API limitations of the
2698 lz4 library). This support has been removed; only support for files
2699 compatible with the lz4 binary remains. This LZ4 logic is now
2700 officially supported and no longer considered experimental.
2702 * The dkr image import logic has been removed again from importd. dkr's
2703 micro-services focus doesn't fit into the machine image focus of
2704 importd, and quickly got out of date with the upstream dkr API.
2706 * Creation of the /run/lock/lockdev/ directory was dropped from
2707 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
2708 been available for many years. If you still need this, you need to
2709 create your own tmpfiles.d config file with:
2711 d /run/lock/lockdev 0775 root lock -
2713 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
2714 and RebootArgument= have been moved from the [Service] section of
2715 unit files to [Unit], and they are now supported on all unit types,
2716 not just service units. Of course, elogind will continue to
2717 understand these settings also at the old location, in order to
2718 maintain compatibility.
2720 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
2721 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
2722 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
2723 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
2724 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
2725 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
2726 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
2727 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
2728 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
2729 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
2730 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
2731 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
2732 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
2733 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
2734 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
2735 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
2736 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
2737 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
2738 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2740 — Berlin, 2016-02-11
2744 * A number of properties previously only settable in unit
2745 files are now also available as properties to set when
2746 creating transient units programmatically via the bus, as it
2747 is exposed with systemd-run's --property=
2748 setting. Specifically, these are: SyslogIdentifier=,
2749 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
2750 EnvironmentFile=, ReadWriteDirectories=,
2751 ReadOnlyDirectories=, InaccessibleDirectories=,
2752 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
2754 * When creating transient services via the bus API it is now
2755 possible to pass in a set of file descriptors to use as
2756 STDIN/STDOUT/STDERR for the invoked process.
2758 * Slice units may now be created transiently via the bus APIs,
2759 similar to the way service and scope units may already be
2760 created transiently.
2762 * Wherever systemd expects a calendar timestamp specification
2763 (like in journalctl's --since= and --until= switches) UTC
2764 timestamps are now supported. Timestamps suffixed with "UTC"
2765 are now considered to be in Universal Time Coordinated
2766 instead of the local timezone. Also, timestamps may now
2767 optionally be specified with sub-second accuracy. Both of
2768 these additions also apply to recurring calendar event
2769 specification, such as OnCalendar= in timer units.
2771 * journalctl gained a new "--sync" switch that asks the
2772 journal daemon to write all so far unwritten log messages to
2773 disk and sync the files, before returning.
2775 * systemd-tmpfiles learned two new line types "q" and "Q" that
2776 operate like "v", but also set up a basic btrfs quota
2777 hierarchy when used on a btrfs file system with quota
2780 * tmpfiles' "v", "q" and "Q" will now create a plain directory
2781 instead of a subvolume (even on a btrfs file system) if the
2782 root directory is a plain directory, and not a
2783 subvolume. This should simplify things with certain chroot()
2784 environments which are not aware of the concept of btrfs
2787 * systemd-detect-virt gained a new --chroot switch to detect
2788 whether execution takes place in a chroot() environment.
2790 * CPUAffinity= now takes CPU index ranges in addition to
2793 * The various memory-related resource limit settings (such as
2794 LimitAS=) now understand the usual K, M, G, ... suffixes to
2795 the base of 1024 (IEC). Similar, the time-related resource
2796 limit settings understand the usual min, h, day, ...
2799 * There's a new system.conf setting DefaultTasksMax= to
2800 control the default TasksMax= setting for services and
2801 scopes running on the system. (TasksMax= is the primary
2802 setting that exposes the "pids" cgroup controller on systemd
2803 and was introduced in the previous systemd release.) The
2804 setting now defaults to 512, which means services that are
2805 not explicitly configured otherwise will only be able to
2806 create 512 processes or threads at maximum, from this
2807 version on. Note that this means that thread- or
2808 process-heavy services might need to be reconfigured to set
2809 TasksMax= to a higher value. It is sufficient to set
2810 TasksMax= in these specific unit files to a higher value, or
2811 even "infinity". Similar, there's now a logind.conf setting
2812 UserTasksMax= that defaults to 4096 and limits the total
2813 number of processes or tasks each user may own
2814 concurrently. nspawn containers also have the TasksMax=
2815 value set by default now, to 8192. Note that all of this
2816 only has an effect if the "pids" cgroup controller is
2817 enabled in the kernel. The general benefit of these changes
2818 should be a more robust and safer system, that provides a
2819 certain amount of per-service fork() bomb protection.
2821 * systemd-nspawn gained the new --network-veth-extra= switch
2822 to define additional and arbitrarily-named virtual Ethernet
2823 links between the host and the container.
2825 * A new service execution setting PassEnvironment= has been
2826 added that allows importing select environment variables
2827 from PID1's environment block into the environment block of
2830 * Timer units gained support for a new RemainAfterElapse=
2831 setting which takes a boolean argument. It defaults to on,
2832 exposing behaviour unchanged to previous releases. If set to
2833 off, timer units are unloaded after they elapsed if they
2834 cannot elapse again. This is particularly useful for
2835 transient timer units, which shall not stay around longer
2836 than until they first elapse.
2838 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
2839 default now (the kernel default is 16). This is beneficial
2840 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
2841 allows substantially larger numbers of queued
2842 datagrams. This should increase the capability of systemd to
2843 parallelize boot-up, as logging and sd_notify() are unlikely
2844 to stall execution anymore. If you need to change the value
2845 from the new defaults, use the usual sysctl.d/ snippets.
2847 * The compression framing format used by the journal or
2848 coredump processing has changed to be in line with what the
2849 official LZ4 tools generate. LZ4 compression support in
2850 systemd was considered unsupported previously, as the format
2851 was not compatible with the normal tools. With this release
2852 this has changed now, and it is hence safe for downstream
2853 distributions to turn it on. While not compressing as well
2854 as the XZ, LZ4 is substantially faster, which makes
2855 it a good default choice for the compression logic in the
2856 journal and in coredump handling.
2858 * Any reference to /etc/mtab has been dropped from
2859 systemd. The file has been obsolete since a while, but
2860 systemd refused to work on systems where it was incorrectly
2861 set up (it should be a symlink or non-existent). Please make
2862 sure to update to util-linux 2.27.1 or newer in conjunction
2863 with this systemd release, which also drops any reference to
2864 /etc/mtab. If you maintain a distribution make sure that no
2865 software you package still references it, as this is a
2866 likely source of bugs. There's also a glibc bug pending,
2867 asking for removal of any reference to this obsolete file:
2869 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
2871 Note that only util-linux versions built with
2872 --enable-libmount-force-mountinfo are supported.
2874 * Support for the ".snapshot" unit type has been removed. This
2875 feature turned out to be little useful and little used, and
2876 has now been removed from the core and from systemctl.
2878 * The dependency types RequiresOverridable= and
2879 RequisiteOverridable= have been removed from systemd. They
2880 have been used only very sparingly to our knowledge and
2881 other options that provide a similar effect (such as
2882 systemctl --mode=ignore-dependencies) are much more useful
2883 and commonly used. Moreover, they were only half-way
2884 implemented as the option to control behaviour regarding
2885 these dependencies was never added to systemctl. By removing
2886 these dependency types the execution engine becomes a bit
2887 simpler. Unit files that use these dependencies should be
2888 changed to use the non-Overridable dependency types
2889 instead. In fact, when parsing unit files with these
2890 options, that's what systemd will automatically convert them
2891 too, but it will also warn, asking users to fix the unit
2892 files accordingly. Removal of these dependency types should
2893 only affect a negligible number of unit files in the wild.
2895 * Behaviour of networkd's IPForward= option changed
2896 (again). It will no longer maintain a per-interface setting,
2897 but propagate one way from interfaces where this is enabled
2898 to the global kernel setting. The global setting will be
2899 enabled when requested by a network that is set up, but
2900 never be disabled again. This change was made to make sure
2901 IPv4 and IPv6 behaviour regarding packet forwarding is
2902 similar (as the Linux IPv6 stack does not support
2903 per-interface control of this setting) and to minimize
2906 * In unit files the behaviour of %u, %U, %h, %s has
2907 changed. These specifiers will now unconditionally resolve
2908 to the various user database fields of the user that the
2909 systemd instance is running as, instead of the user
2910 configured in the specific unit via User=. Note that this
2911 effectively doesn't change much, as resolving of these
2912 specifiers was already turned off in the --system instance
2913 of systemd, as we cannot do NSS lookups from PID 1. In the
2914 --user instance of systemd these specifiers where correctly
2915 resolved, but hardly made any sense, since the user instance
2916 lacks privileges to do user switches anyway, and User= is
2917 hence useless. Moreover, even in the --user instance of
2918 systemd behaviour was awkward as it would only take settings
2919 from User= assignment placed before the specifier into
2920 account. In order to unify and simplify the logic around
2921 this the specifiers will now always resolve to the
2922 credentials of the user invoking the manager (which in case
2923 of PID 1 is the root user).
2925 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
2926 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
2927 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
2928 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
2929 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
2930 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
2931 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
2932 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
2933 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
2934 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
2935 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
2936 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
2937 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
2938 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
2941 — Berlin, 2015-11-18
2945 * systemd now depends on util-linux v2.27. More specifically,
2946 the newly added mount monitor feature in libmount now
2947 replaces systemd's former own implementation.
2949 * libmount mandates /etc/mtab not to be regular file, and
2950 systemd now enforces this condition at early boot.
2951 /etc/mtab has been deprecated and warned about for a very
2952 long time, so systems running systemd should already have
2953 stopped having this file around as anything else than a
2954 symlink to /proc/self/mounts.
2956 * Support for the "pids" cgroup controller has been added. It
2957 allows accounting the number of tasks in a cgroup and
2958 enforcing limits on it. This adds two new setting
2959 TasksAccounting= and TasksMax= to each unit, as well as a
2960 global option DefaultTasksAccounting=.
2962 * Support for the "net_cls" cgroup controller has been added.
2963 It allows assigning a net class ID to each task in the
2964 cgroup, which can then be used in firewall rules and traffic
2965 shaping configurations. Note that the kernel netfilter net
2966 class code does not currently work reliably for ingress
2967 packets on unestablished sockets.
2969 This adds a new config directive called NetClass= to CGroup
2970 enabled units. Allowed values are positive numbers for fixed
2971 assignments and "auto" for picking a free value
2974 * 'systemctl is-system-running' now returns 'offline' if the
2975 system is not booted with systemd. This command can now be
2976 used as a substitute for 'systemd-notify --booted'.
2978 * Watchdog timeouts have been increased to 3 minutes for all
2979 in-tree service files. Apparently, disk IO issues are more
2980 frequent than we hoped, and user reported >1 minute waiting
2983 * 'machine-id-commit' functionality has been merged into
2984 'machine-id-setup --commit'. The separate binary has been
2987 * The WorkingDirectory= directive in unit files may now be set
2988 to the special value '~'. In this case, the working
2989 directory is set to the home directory of the user
2990 configured in User=.
2992 * "machinectl shell" will now open the shell in the home
2993 directory of the selected user by default.
2995 * The CrashChVT= configuration file setting is renamed to
2996 CrashChangeVT=, following our usual logic of not
2997 abbreviating unnecessarily. The old directive is still
2998 supported for compat reasons. Also, this directive now takes
2999 an integer value between 1 and 63, or a boolean value. The
3000 formerly supported '-1' value for disabling stays around for
3003 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
3004 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
3005 RootDirectory= properties can now be set for transient
3008 * The systemd-analyze tool gained a new "set-log-target" verb
3009 to change the logging target the system manager logs to
3010 dynamically during runtime. This is similar to how
3011 "systemd-analyze set-log-level" already changes the log
3014 * In nspawn /sys is now mounted as tmpfs, with only a selected
3015 set of subdirectories mounted in from the real sysfs. This
3016 enhances security slightly, and is useful for ensuring user
3017 namespaces work correctly.
3019 * Support for USB FunctionFS activation has been added. This
3020 allows implementation of USB gadget services that are
3021 activated as soon as they are requested, so that they don't
3022 have to run continuously, similar to classic socket
3025 * The "systemctl exit" command now optionally takes an
3026 additional parameter that sets the exit code to return from
3027 the systemd manager when exiting. This is only relevant when
3028 running the systemd user instance, or when running the
3029 system instance in a container.
3031 * sd-bus gained the new API calls sd_bus_path_encode_many()
3032 and sd_bus_path_decode_many() that allow easy encoding and
3033 decoding of multiple identifier strings inside a D-Bus
3034 object path. Another new call sd_bus_default_flush_close()
3035 has been added to flush and close per-thread default
3038 * systemd-cgtop gained support for a -M/--machine= switch to
3039 show the control groups within a certain container only.
3041 * "systemctl kill" gained support for an optional --fail
3042 switch. If specified the requested operation will fail of no
3043 processes have been killed, because the unit had no
3044 processes attached, or similar.
3046 * A new systemd.crash_reboot=1 kernel command line option has
3047 been added that triggers a reboot after crashing. This can
3048 also be set through CrashReboot= in systemd.conf.
3050 * The RuntimeDirectory= setting now understands unit
3051 specifiers like %i or %f.
3053 * A new (still internal) library API sd-ipv4acd has been added,
3054 that implements address conflict detection for IPv4. It's
3055 based on code from sd-ipv4ll, and will be useful for
3056 detecting DHCP address conflicts.
3058 * File descriptors passed during socket activation may now be
3059 named. A new API sd_listen_fds_with_names() is added to
3060 access the names. The default names may be overridden,
3061 either in the .socket file using the FileDescriptorName=
3062 parameter, or by passing FDNAME= when storing the file
3063 descriptors using sd_notify().
3065 * systemd-networkd gained support for:
3067 - Setting the IPv6 Router Advertisement settings via
3068 IPv6AcceptRouterAdvertisements= in .network files.
3070 - Configuring the HelloTimeSec=, MaxAgeSec= and
3071 ForwardDelaySec= bridge parameters in .netdev files.
3073 - Configuring PreferredSource= for static routes in
3076 * The "ask-password" framework used to query for LUKS harddisk
3077 passwords or SSL passwords during boot gained support for
3078 caching passwords in the kernel keyring, if it is
3079 available. This makes sure that the user only has to type in
3080 a passphrase once if there are multiple objects to unlock
3081 with the same one. Previously, such password caching was
3082 available only when Plymouth was used; this moves the
3083 caching logic into the systemd codebase itself. The
3084 "systemd-ask-password" utility gained a new --keyname=
3085 switch to control which kernel keyring key to use for
3086 caching a password in. This functionality is also useful for
3087 enabling display managers such as gdm to automatically
3088 unlock the user's GNOME keyring if its passphrase, the
3089 user's password and the harddisk password are the same, if
3090 gdm-autologin is used.
3092 * When downloading tar or raw images using "machinectl
3093 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
3094 file is now also downloaded, if it is available and stored
3095 next to the image file.
3097 * Units of type ".socket" gained a new boolean setting
3098 Writable= which is only useful in conjunction with
3099 ListenSpecial=. If true, enables opening the specified
3100 special file in O_RDWR mode rather than O_RDONLY mode.
3102 * systemd-rfkill has been reworked to become a singleton
3103 service that is activated through /dev/rfkill on each rfkill
3104 state change and saves the settings to disk. This way,
3105 systemd-rfkill is now compatible with devices that exist
3106 only intermittendly, and even restores state if the previous
3107 system shutdown was abrupt rather than clean.
3109 * The journal daemon gained support for vacuuming old journal
3110 files controlled by the number of files that shall remain,
3111 in addition to the already existing control by size and by
3112 date. This is useful as journal interleaving performance
3113 degrades with too many separate journal files, and allows
3114 putting an effective limit on them. The new setting defaults
3115 to 100, but this may be changed by setting SystemMaxFiles=
3116 and RuntimeMaxFiles= in journald.conf. Also, the
3117 "journalctl" tool gained the new --vacuum-files= switch to
3118 manually vacuum journal files to leave only the specified
3119 number of files in place.
3121 * udev will now create /dev/disk/by-path links for ATA devices
3122 on kernels where that is supported.
3124 * Galician, Serbian, Turkish and Korean translations were added.
3126 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
3127 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
3128 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
3129 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
3130 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
3131 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
3132 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
3133 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
3134 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
3135 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
3136 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
3137 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
3138 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
3139 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
3140 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
3141 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
3142 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
3143 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
3145 — Berlin, 2015-10-07
3149 * The DHCP implementation of systemd-networkd gained a set of
3152 - The DHCP server now supports emitting DNS and NTP
3153 information. It may be enabled and configured via
3154 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
3155 and NTP information is enabled, but no servers are
3156 configured, the corresponding uplink information (if there
3157 is any) is propagated.
3159 - Server and client now support transmission and reception
3160 of timezone information. It can be configured via the
3161 newly introduced network options UseTimezone=,
3162 EmitTimezone=, and Timezone=. Transmission of timezone
3163 information is enabled between host and containers by
3164 default now: the container will change its local timezone
3165 to what the host has set.
3167 - Lease timeouts can now be configured via
3168 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
3170 - The DHCP server improved on the stability of
3171 leases. Clients are more likely to get the same lease
3172 information back, even if the server loses state.
3174 - The DHCP server supports two new configuration options to
3175 control the lease address pool metrics, PoolOffset= and
3178 * The encapsulation limit of tunnels in systemd-networkd may
3179 now be configured via 'EncapsulationLimit='. It allows
3180 modifying the maximum additional levels of encapsulation
3181 that are permitted to be prepended to a packet.
3183 * systemd now supports the concept of user buses replacing
3184 session buses, if used with dbus-1.10 (and enabled via dbus
3185 --enable-user-session). It previously only supported this on
3186 kdbus-enabled systems, and this release expands this to
3187 'dbus-daemon' systems.
3189 * systemd-networkd now supports predictable interface names
3192 * systemd now optionally supports the new Linux kernel
3193 "unified" control group hierarchy. If enabled via the kernel
3194 command-line option 'systemd.unified_cgroup_hierarchy=1',
3195 systemd will try to mount the unified cgroup hierarchy
3196 directly on /sys/fs/cgroup. If not enabled, or not
3197 available, systemd will fall back to the legacy cgroup
3198 hierarchy setup, as before. Host system and containers can
3199 mix and match legacy and unified hierarchies as they
3200 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
3201 environment variable to individually select the hierarchy to
3202 use for executed containers. By default, nspawn will use the
3203 unified hierarchy for the containers if the host uses the
3204 unified hierarchy, and the legacy hierarchy otherwise.
3205 Please note that at this point the unified hierarchy is an
3206 experimental kernel feature and is likely to change in one
3207 of the next kernel releases. Therefore, it should not be
3208 enabled by default in downstream distributions yet. The
3209 minimum required kernel version for the unified hierarchy to
3210 work is 4.2. Note that when the unified hierarchy is used
3211 for the first time delegated access to controllers is
3212 safe. Because of this systemd-nspawn containers will get
3213 access to controllers now, as will systemd user
3214 sessions. This means containers and user sessions may now
3215 manage their own resources, partitioning up what the system
3218 * A new special scope unit "init.scope" has been introduced
3219 that encapsulates PID 1 of the system. It may be used to
3220 determine resource usage and enforce resource limits on PID
3221 1 itself. PID 1 hence moved out of the root of the control
3224 * The cgtop tool gained support for filtering out kernel
3225 threads when counting tasks in a control group. Also, the
3226 count of processes is now recursively summed up by
3227 default. Two options -k and --recursive= have been added to
3228 revert to old behaviour. The tool has also been updated to
3229 work correctly in containers now.
3231 * systemd-nspawn's --bind= and --bind-ro= options have been
3232 extended to allow creation of non-recursive bind mounts.
3234 * libsystemd gained two new calls sd_pid_get_cgroup() and
3235 sd_peer_get_cgroup() which return the control group path of
3236 a process or peer of a connected AF_UNIX socket. This
3237 function call is particularly useful when implementing
3238 delegated subtrees support in the control group hierarchy.
3240 * The "sd-event" event loop API of libsystemd now supports
3241 correct dequeuing of real-time signals, without losing
3244 * When systemd requests a PolicyKit decision when managing
3245 units it will now add additional fields to the request,
3246 including unit name and desired operation. This enables more
3247 powerful PolicyKit policies, that make decisions depending
3248 on these parameters.
3250 * nspawn learnt support for .nspawn settings files, that may
3251 accompany the image files or directories of containers, and
3252 may contain additional settings for the container. This is
3253 an alternative to configuring container parameters via the
3254 nspawn command line.
3256 Contributions from: Cristian Rodríguez, Daniel Mack, David
3257 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
3258 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
3259 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
3260 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
3261 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
3262 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
3263 Andersen, Tom Gundersen, Torstein Husebø
3265 — Berlin, 2015-09-08
3269 * machinectl gained a new verb 'shell' which opens a fresh
3270 shell on the target container or the host. It is similar to
3271 the existing 'login' command of machinectl, but spawns the
3272 shell directly without prompting for username or
3273 password. The pseudo machine '.host' now refers to the local
3274 host and is used by default. Hence, 'machinectl shell' can
3275 be used as replacement for 'su -' which spawns a session as
3276 a fresh systemd unit in a way that is fully isolated from
3277 the originating session.
3279 * systemd-networkd learned to cope with private-zone DHCP
3280 options and allows other programs to query the values.
3282 * SELinux access control when enabling/disabling units is no
3283 longer enforced with this release. The previous
3284 implementation was incorrect, and a new corrected
3285 implementation is not yet available. As unit file operations
3286 are still protected via PolicyKit and D-Bus policy this is
3287 not a security problem. Yet, distributions which care about
3288 optimal SELinux support should probably not stabilize on
3291 * sd-bus gained support for matches of type "arg0has=", that
3292 test for membership of strings in string arrays sent in bus
3295 * systemd-resolved now dumps the contents of its DNS and LLMNR
3296 caches to the logs on reception of the SIGUSR1 signal. This
3297 is useful to debug DNS behaviour.
3299 * The coredumpctl tool gained a new --directory= option to
3300 operate on journal files in a specific directory.
3302 * "systemctl reboot" and related commands gained a new
3303 "--message=" option which may be used to set a free-text
3304 wall message when shutting down or rebooting the
3305 system. This message is also logged, which is useful for
3306 figuring out the reason for a reboot or shutdown a
3309 * The "systemd-resolve-host" tool's -i switch now takes
3310 network interface numbers as alternative to interface names.
3312 * A new unit file setting for services has been introduced:
3313 UtmpMode= allows configuration of how precisely systemd
3314 handles utmp and wtmp entries for the service if this is
3315 enabled. This allows writing services that appear similar to
3316 user sessions in the output of the "w", "who", "last" and
3319 * systemd-resolved will now locally synthesize DNS resource
3320 records for the "localhost" and "gateway" domains as well as
3321 the local hostname. This should ensure that clients querying
3322 RRs via resolved will get similar results as those going via
3323 NSS, if nss-myhostname is enabled.
3325 Contributions from: Alastair Hughes, Alex Crawford, Daniel
3326 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
3327 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
3328 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
3329 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
3330 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
3331 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
3332 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
3333 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
3334 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
3335 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
3336 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3338 — Berlin, 2015-08-27
3342 * The systemd-efi-boot-generator functionality was merged into
3343 systemd-gpt-auto-generator.
3345 * systemd-networkd now supports Group Policy for vxlan
3346 devices. It can be enabled via the new boolean configuration
3347 option called 'GroupPolicyExtension='.
3349 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
3350 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
3351 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
3353 — Berlin, 2015-07-31
3357 * The python-systemd code has been removed from the systemd repository.
3358 A new repository has been created which accommodates the code from
3359 now on, and we kindly ask distributions to create a separate package
3360 for this: https://github.com/systemd/python-systemd
3362 * The systemd daemon will now reload its main configuration
3363 (/etc/systemd/system.conf) on daemon-reload.
3365 * sd-dhcp now exposes vendor specific extensions via
3366 sd_dhcp_lease_get_vendor_specific().
3368 * systemd-networkd gained a number of new configuration options.
3370 - A new boolean configuration option for TAP devices called
3371 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
3372 device, thus allowing to send and receive GSO packets.
3374 - A new tunnel configuration option called 'CopyDSCP='.
3375 If enabled, the DSCP field of ip6 tunnels is copied into the
3376 decapsulated packet.
3378 - A set of boolean bridge configuration options were added.
3379 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
3380 and 'UnicastFlood=' are now parsed by networkd and applied to the
3381 respective bridge link device via the respective IFLA_BRPORT_*
3384 - A new string configuration option to override the hostname sent
3385 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
3386 is true, networkd will use the configured hostname instead of the
3387 system hostname when sending DHCP requests.
3389 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
3390 networkd will configure the IPv6 flow-label of the tunnel device
3391 according to RFC2460.
3393 - The 'macvtap' virtual network devices are now supported, similar to
3394 the already supported 'macvlan' devices.
3396 * systemd-resolved now implements RFC5452 to improve resilience against
3397 cache poisoning. Additionally, source port randomization is enabled
3398 by default to further protect against DNS spoofing attacks.
3400 * nss-mymachines now supports translating UIDs and GIDs of running
3401 containers with user-namespaces enabled. If a container 'foo'
3402 translates a host uid 'UID' to the container uid 'TUID', then
3403 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
3404 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
3405 mapped as 'vg-foo-TGID'.
3407 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
3408 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
3409 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
3410 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
3411 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
3412 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
3413 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
3414 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
3415 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
3416 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
3418 — Berlin, 2015-07-29
3422 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
3423 There are no known issues with current sysfs, and udev does not need
3424 or should be used to work around such bugs.
3426 * udev does no longer enable USB HID power management. Several reports
3427 indicate, that some devices cannot handle that setting.
3429 * The udev accelerometer helper was removed. The functionality
3430 is now fully included in iio-sensor-proxy. But this means,
3431 older iio-sensor-proxy versions will no longer provide
3432 accelerometer/orientation data with this systemd version.
3433 Please upgrade iio-sensor-proxy to version 1.0.
3435 * networkd gained a new configuration option IPv6PrivacyExtensions=
3436 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
3437 for Stateless Address") on selected networks.
3439 * For the sake of fewer build-time dependencies and less code in the
3440 main repository, the python bindings are about to be removed in the
3441 next release. A new repository has been created which accommodates
3442 the code from now on, and we kindly ask distributions to create a
3443 separate package for this. The removal will take place in v223.
3445 https://github.com/systemd/python-systemd
3447 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
3448 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
3449 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
3450 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
3451 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
3452 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
3453 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
3454 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
3455 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3456 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3458 — Berlin, 2015-07-07
3462 * The sd-bus.h and sd-event.h APIs have now been declared
3463 stable and have been added to the official interface of
3464 libsystemd.so. sd-bus implements an alternative D-Bus client
3465 library, that is relatively easy to use, very efficient and
3466 supports both classic D-Bus as well as kdbus as transport
3467 backend. sd-event is a generic event loop abstraction that
3468 is built around Linux epoll, but adds features such as event
3469 prioritization or efficient timer handling. Both APIs are good
3470 choices for C programs looking for a bus and/or event loop
3471 implementation that is minimal and does not have to be
3472 portable to other kernels.
3474 * kdbus support is no longer compile-time optional. It is now
3475 always built-in. However, it can still be disabled at
3476 runtime using the kdbus=0 kernel command line setting, and
3477 that setting may be changed to default to off, by specifying
3478 --disable-kdbus at build-time. Note though that the kernel
3479 command line setting has no effect if the kdbus.ko kernel
3480 module is not installed, in which case kdbus is (obviously)
3481 also disabled. We encourage all downstream distributions to
3482 begin testing kdbus by adding it to the kernel images in the
3483 development distributions, and leaving kdbus support in
3486 * The minimal required util-linux version has been bumped to
3489 * Support for chkconfig (--enable-chkconfig) was removed in
3490 favor of calling an abstraction tool
3491 /lib/systemd/systemd-sysv-install. This needs to be
3492 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
3493 in README for details.
3495 * If there's a systemd unit and a SysV init script for the
3496 same service name, and the user executes "systemctl enable"
3497 for it (or a related call), then this will now enable both
3498 (or execute the related operation on both), not just the
3501 * The libudev API documentation has been converted from gtkdoc
3504 * gudev has been removed from the systemd tree, it is now an
3507 * The systemd-cgtop tool learnt a new --raw switch to generate
3508 "raw" (machine parsable) output.
3510 * networkd's IPForwarding= .network file setting learnt the
3511 new setting "kernel", which ensures that networkd does not
3512 change the IP forwarding sysctl from the default kernel
3515 * The systemd-logind bus API now exposes a new boolean
3516 property "Docked" that reports whether logind considers the
3517 system "docked", i.e. connected to a docking station or not.
3519 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
3520 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
3521 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
3522 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
3523 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
3524 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
3525 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
3526 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
3527 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
3528 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
3529 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
3530 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
3531 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
3532 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3533 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
3534 Fink, Zbigniew Jędrzejewski-Szmek
3536 — Berlin, 2015-06-19
3540 * The gudev library has been extracted into a separate repository
3541 available at: https://git.gnome.org/browse/libgudev/
3542 It is now managed as part of the Gnome project. Distributions
3543 are recommended to pass --disable-gudev to systemd and use
3544 gudev from the Gnome project instead. gudev is still included
3545 in systemd, for now. It will be removed soon, though. Please
3546 also see the announcement-thread on systemd-devel:
3547 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
3549 * systemd now exposes a CPUUsageNSec= property for each
3550 service unit on the bus, that contains the overall consumed
3551 CPU time of a service (the sum of what each process of the
3552 service consumed). This value is only available if
3553 CPUAccounting= is turned on for a service, and is then shown
3554 in the "systemctl status" output.
3556 * Support for configuring alternative mappings of the old SysV
3557 runlevels to systemd targets has been removed. They are now
3558 hardcoded in a way that runlevels 2, 3, 4 all map to
3559 multi-user.target and 5 to graphical.target (which
3560 previously was already the default behaviour).
3562 * The auto-mounter logic gained support for mount point
3563 expiry, using a new TimeoutIdleSec= setting in .automount
3564 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
3566 * The EFI System Partition (ESP) as mounted to /boot by
3567 systemd-efi-boot-generator will now be unmounted
3568 automatically after 2 minutes of not being used. This should
3569 minimize the risk of ESP corruptions.
3571 * New /etc/fstab options x-systemd.requires= and
3572 x-systemd.requires-mounts-for= are now supported to express
3573 additional dependencies for mounts. This is useful for
3574 journalling file systems that support external journal
3575 devices or overlay file systems that require underlying file
3576 systems to be mounted.
3578 * systemd does not support direct live-upgrades (via systemctl
3579 daemon-reexec) from versions older than v44 anymore. As no
3580 distribution we are aware of shipped such old versions in a
3581 stable release this should not be problematic.
3583 * When systemd forks off a new per-connection service instance
3584 it will now set the $REMOTE_ADDR environment variable to the
3585 remote IP address, and $REMOTE_PORT environment variable to
3586 the remote IP port. This behaviour is similar to the
3587 corresponding environment variables defined by CGI.
3589 * systemd-networkd gained support for uplink failure
3590 detection. The BindCarrier= option allows binding interface
3591 configuration dynamically to the link sense of other
3592 interfaces. This is useful to achieve behaviour like in
3595 * systemd-networkd gained support for configuring the DHCP
3596 client identifier to use when requesting leases.
3598 * systemd-networkd now has a per-network UseNTP= option to
3599 configure whether NTP server information acquired via DHCP
3600 is passed on to services like systemd-timesyncd.
3602 * systemd-networkd gained support for vti6 tunnels.
3604 * Note that systemd-networkd manages the sysctl variable
3605 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
3606 it is configured for since v219. The variable controls IP
3607 forwarding, and is a per-interface alternative to the global
3608 /proc/sys/net/ipv[46]/ip_forward. This setting is
3609 configurable in the IPForward= option, which defaults to
3610 "no". This means if networkd is used for an interface it is
3611 no longer sufficient to set the global sysctl option to turn
3612 on IP forwarding! Instead, the .network file option
3613 IPForward= needs to be turned on! Note that the
3614 implementation of this behaviour was broken in v219 and has
3617 * Many bonding and vxlan options are now configurable in
3620 * systemd-nspawn gained a new --property= setting to set unit
3621 properties for the container scope. This is useful for
3622 setting resource parameters (e.g. "CPUShares=500") on
3623 containers started from the command line.
3625 * systemd-nspawn gained a new --private-users= switch to make
3626 use of user namespacing available on recent Linux kernels.
3628 * systemd-nspawn may now be called as part of a shell pipeline
3629 in which case the pipes used for stdin and stdout are passed
3630 directly to the process invoked in the container, without
3631 indirection via a pseudo tty.
3633 * systemd-nspawn gained a new switch to control the UNIX
3634 signal to use when killing the init process of the container
3637 * systemd-nspawn gained a new --overlay= switch for mounting
3638 overlay file systems into the container using the new kernel
3641 * When a container image is imported via systemd-importd and
3642 the host file system is not btrfs, a loopback block device
3643 file is created in /var/lib/machines.raw with a btrfs file
3644 system inside. It is then mounted to /var/lib/machines to
3645 enable btrfs features for container management. The loopback
3646 file and btrfs file system is grown as needed when container
3647 images are imported via systemd-importd.
3649 * systemd-machined/systemd-importd gained support for btrfs
3650 quota, to enforce container disk space limits on disk. This
3651 is exposed in "machinectl set-limit".
3653 * systemd-importd now can import containers from local .tar,
3654 .raw and .qcow2 images, and export them to .tar and .raw. It
3655 can also import dkr v2 images now from the network (on top
3658 * systemd-importd gained support for verifying downloaded
3659 images with gpg2 (previously only gpg1 was supported).
3661 * systemd-machined, systemd-logind, systemd: most bus calls
3662 are now accessible to unprivileged processes via
3663 PolicyKit. Also, systemd-logind will now allow users to kill
3664 their own sessions without further privileges or
3667 * systemd-shutdownd has been removed. This service was
3668 previously responsible for implementing scheduled shutdowns
3669 as exposed in /usr/bin/shutdown's time parameter. This
3670 functionality has now been moved into systemd-logind and is
3671 accessible via a bus interface.
3673 * "systemctl reboot" gained a new switch --firmware-setup that
3674 can be used to reboot into the EFI firmware setup, if that
3675 is available. systemd-logind now exposes an API on the bus
3676 to trigger such reboots, in case graphical desktop UIs want
3677 to cover this functionality.
3679 * "systemctl enable", "systemctl disable" and "systemctl mask"
3680 now support a new "--now" switch. If specified the units
3681 that are enabled will also be started, and the ones
3682 disabled/masked also stopped.
3684 * The Gummiboot EFI boot loader tool has been merged into
3685 systemd, and renamed to "systemd-boot". The bootctl tool has been
3686 updated to support systemd-boot.
3688 * An EFI kernel stub has been added that may be used to create
3689 kernel EFI binaries that contain not only the actual kernel,
3690 but also an initrd, boot splash, command line and OS release
3691 information. This combined binary can then be signed as a
3692 single image, so that the firmware can verify it all in one
3693 step. systemd-boot has special support for EFI binaries created
3694 like this and can extract OS release information from them
3695 and show them in the boot menu. This functionality is useful
3696 to implement cryptographically verified boot schemes.
3698 * Optional support has been added to systemd-fsck to pass
3699 fsck's progress report to an AF_UNIX socket in the file
3702 * udev will no longer create device symlinks for all block
3703 devices by default. A blacklist for excluding special block
3704 devices from this logic has been turned into a whitelist
3705 that requires picking block devices explicitly that require
3708 * A new (currently still internal) API sd-device.h has been
3709 added to libsystemd. This modernized API is supposed to
3710 replace libudev eventually. In fact, already much of libudev
3711 is now just a wrapper around sd-device.h.
3713 * A new hwdb database for storing metadata about pointing
3714 stick devices has been added.
3716 * systemd-tmpfiles gained support for setting file attributes
3717 similar to the "chattr" tool with new 'h' and 'H' lines.
3719 * systemd-journald will no longer unconditionally set the
3720 btrfs NOCOW flag on new journal files. This is instead done
3721 with tmpfiles snippet using the new 'h' line type. This
3722 allows easy disabling of this logic, by masking the
3723 journal-nocow.conf tmpfiles file.
3725 * systemd-journald will now translate audit message types to
3726 human readable identifiers when writing them to the
3727 journal. This should improve readability of audit messages.
3729 * The LUKS logic gained support for the offset= and skip=
3730 options in /etc/crypttab, as previously implemented by
3733 * /usr/lib/os-release gained a new optional field VARIANT= for
3734 distributions that support multiple variants (such as a
3735 desktop edition, a server edition, ...)
3737 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
3738 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
3739 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
3740 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
3741 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
3742 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
3743 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
3744 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
3745 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
3746 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
3747 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
3748 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
3749 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
3750 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
3751 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
3752 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
3753 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
3754 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
3755 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
3756 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
3757 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
3758 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
3759 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
3760 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
3761 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
3762 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
3763 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
3765 — Berlin, 2015-05-22
3769 * Introduce a new API "sd-hwdb.h" for querying the hardware
3770 metadata database. With this minimal interface one can query
3771 and enumerate the udev hwdb, decoupled from the old libudev
3772 library. libudev's interface for this is now only a wrapper
3773 around sd-hwdb. A new tool systemd-hwdb has been added to
3774 interface with and update the database.
3776 * When any of systemd's tools copies files (for example due to
3777 tmpfiles' C lines) a btrfs reflink will attempted first,
3778 before bytewise copying is done.
3780 * systemd-nspawn gained a new --ephemeral switch. When
3781 specified a btrfs snapshot is taken of the container's root
3782 directory, and immediately removed when the container
3783 terminates again. Thus, a container can be started whose
3784 changes never alter the container's root directory, and are
3785 lost on container termination. This switch can also be used
3786 for starting a container off the root file system of the
3787 host without affecting the host OS. This switch is only
3788 available on btrfs file systems.
3790 * systemd-nspawn gained a new --template= switch. It takes the
3791 path to a container tree to use as template for the tree
3792 specified via --directory=, should that directory be
3793 missing. This allows instantiating containers dynamically,
3794 on first run. This switch is only available on btrfs file
3797 * When a .mount unit refers to a mount point on which multiple
3798 mounts are stacked, and the .mount unit is stopped all of
3799 the stacked mount points will now be unmounted until no
3800 mount point remains.
3802 * systemd now has an explicit notion of supported and
3803 unsupported unit types. Jobs enqueued for unsupported unit
3804 types will now fail with an "unsupported" error code. More
3805 specifically .swap, .automount and .device units are not
3806 supported in containers, .busname units are not supported on
3807 non-kdbus systems. .swap and .automount are also not
3808 supported if their respective kernel compile time options
3811 * machinectl gained support for two new "copy-from" and
3812 "copy-to" commands for copying files from a running
3813 container to the host or vice versa.
3815 * machinectl gained support for a new "bind" command to bind
3816 mount host directories into local containers. This is
3817 currently only supported for nspawn containers.
3819 * networkd gained support for configuring bridge forwarding
3820 database entries (fdb) from .network files.
3822 * A new tiny daemon "systemd-importd" has been added that can
3823 download container images in tar, raw, qcow2 or dkr formats,
3824 and make them available locally in /var/lib/machines, so
3825 that they can run as nspawn containers. The daemon can GPG
3826 verify the downloads (not supported for dkr, since it has no
3827 provisions for verifying downloads). It will transparently
3828 decompress bz2, xz, gzip compressed downloads if necessary,
3829 and restore sparse files on disk. The daemon uses privilege
3830 separation to ensure the actual download logic runs with
3831 fewer privileges than the daemon itself. machinectl has
3832 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
3833 make the functionality of importd available to the
3834 user. With this in place the Fedora and Ubuntu "Cloud"
3835 images can be downloaded and booted as containers unmodified
3836 (the Fedora images lack the appropriate GPG signature files
3837 currently, so they cannot be verified, but this will change
3838 soon, hopefully). Note that downloading images is currently
3839 only fully supported on btrfs.
3841 * machinectl is now able to list container images found in
3842 /var/lib/machines, along with some metadata about sizes of
3843 disk and similar. If the directory is located on btrfs and
3844 quota is enabled, this includes quota display. A new command
3845 "image-status" has been added that shows additional
3846 information about images.
3848 * machinectl is now able to clone container images
3849 efficiently, if the underlying file system (btrfs) supports
3850 it, with the new "machinectl clone" command. It also
3851 gained commands for renaming and removing images, as well as
3852 marking them read-only or read-write (supported also on
3853 legacy file systems).
3855 * networkd gained support for collecting LLDP network
3856 announcements, from hardware that supports this. This is
3857 shown in networkctl output.
3859 * systemd-run gained support for a new -t (--pty) switch for
3860 invoking a binary on a pty whose input and output is
3861 connected to the invoking terminal. This allows executing
3862 processes as system services while interactively
3863 communicating with them via the terminal. Most interestingly
3864 this is supported across container boundaries. Invoking
3865 "systemd-run -t /bin/bash" is an alternative to running a
3866 full login session, the difference being that the former
3867 will not register a session, nor go through the PAM session
3870 * tmpfiles gained support for a new "v" line type for creating
3871 btrfs subvolumes. If the underlying file system is a legacy
3872 file system, this automatically degrades to creating a
3873 normal directory. Among others /var/lib/machines is now
3874 created like this at boot, should it be missing.
3876 * The directory /var/lib/containers/ has been deprecated and
3877 been replaced by /var/lib/machines. The term "machines" has
3878 been used in the systemd context as generic term for both
3879 VMs and containers, and hence appears more appropriate for
3880 this, as the directory can also contain raw images bootable
3883 * systemd-nspawn when invoked with -M but without --directory=
3884 or --image= is now capable of searching for the container
3885 root directory, subvolume or disk image automatically, in
3886 /var/lib/machines. systemd-nspawn@.service has been updated
3887 to make use of this, thus allowing it to be used for raw
3890 * A new machines.target unit has been introduced that is
3891 supposed to group all containers/VMs invoked as services on
3892 the system. systemd-nspawn@.service has been updated to
3893 integrate with that.
3895 * machinectl gained a new "start" command, for invoking a
3896 container as a service. "machinectl start foo" is mostly
3897 equivalent to "systemctl start systemd-nspawn@foo.service",
3898 but handles escaping in a nicer way.
3900 * systemd-nspawn will now mount most of the cgroupfs tree
3901 read-only into each container, with the exception of the
3902 container's own subtree in the name=systemd hierarchy.
3904 * journald now sets the special FS_NOCOW file flag for its
3905 journal files. This should improve performance on btrfs, by
3906 avoiding heavy fragmentation when journald's write-pattern
3907 is used on COW file systems. It degrades btrfs' data
3908 integrity guarantees for the files to the same levels as for
3909 ext3/ext4 however. This should be OK though as journald does
3910 its own data integrity checks and all its objects are
3911 checksummed on disk. Also, journald should handle btrfs disk
3912 full events a lot more gracefully now, by processing SIGBUS
3913 errors, and not relying on fallocate() anymore.
3915 * When journald detects that journal files it is writing to
3916 have been deleted it will immediately start new journal
3919 * systemd now provides a way to store file descriptors
3920 per-service in PID 1. This is useful for daemons to ensure
3921 that fds they require are not lost during a daemon
3922 restart. The fds are passed to the daemon on the next
3923 invocation in the same way socket activation fds are
3924 passed. This is now used by journald to ensure that the
3925 various sockets connected to all the system's stdout/stderr
3926 are not lost when journald is restarted. File descriptors
3927 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
3928 an extension to sd_notify(). Note that a limit is enforced
3929 on the number of fds a service can store in PID 1, and it
3930 defaults to 0, so that no fds may be stored, unless this is
3931 explicitly turned on.
3933 * The default TERM variable to use for units connected to a
3934 terminal, when no other value is explicitly is set is now
3935 vt220 rather than vt102. This should be fairly safe still,
3936 but allows PgUp/PgDn work.
3938 * The /etc/crypttab option header= as known from Debian is now
3941 * "loginctl user-status" and "loginctl session-status" will
3942 now show the last 10 lines of log messages of the
3943 user/session following the status output. Similar,
3944 "machinectl status" will show the last 10 log lines
3945 associated with a virtual machine or container
3946 service. (Note that this is usually not the log messages
3947 done in the VM/container itself, but simply what the
3948 container manager logs. For nspawn this includes all console
3951 * "loginctl session-status" without further argument will now
3952 show the status of the session of the caller. Similar,
3953 "lock-session", "unlock-session", "activate",
3954 "enable-linger", "disable-linger" may now be called without
3955 session/user parameter in which case they apply to the
3956 caller's session/user.
3958 * An X11 session scriptlet is now shipped that uploads
3959 $DISPLAY and $XAUTHORITY into the environment of the systemd
3960 --user daemon if a session begins. This should improve
3961 compatibility with X11 enabled applications run as systemd
3964 * Generators are now subject to masking via /etc and /run, the
3965 same way as unit files.
3967 * networkd .network files gained support for configuring
3968 per-link IPv4/IPv6 packet forwarding as well as IPv4
3969 masquerading. This is by default turned on for veth links to
3970 containers, as registered by systemd-nspawn. This means that
3971 nspawn containers run with --network-veth will now get
3972 automatic routed access to the host's networks without any
3973 further configuration or setup, as long as networkd runs on
3976 * systemd-nspawn gained the --port= (-p) switch to expose TCP
3977 or UDP posts of a container on the host. With this in place
3978 it is possible to run containers with private veth links
3979 (--network-veth), and have their functionality exposed on
3980 the host as if their services were running directly on the
3983 * systemd-nspawn's --network-veth switch now gained a short
3984 version "-n", since with the changes above it is now truly
3985 useful out-of-the-box. The systemd-nspawn@.service has been
3986 updated to make use of it too by default.
3988 * systemd-nspawn will now maintain a per-image R/W lock, to
3989 ensure that the same image is not started more than once
3990 writable. (It's OK to run an image multiple times
3991 simultaneously in read-only mode.)
3993 * systemd-nspawn's --image= option is now capable of
3994 dissecting and booting MBR and GPT disk images that contain
3995 only a single active Linux partition. Previously it
3996 supported only GPT disk images with proper GPT type
3997 IDs. This allows running cloud images from major
3998 distributions directly with systemd-nspawn, without
4001 * In addition to collecting mouse dpi data in the udev
4002 hardware database, there's now support for collecting angle
4003 information for mouse scroll wheels. The database is
4004 supposed to guarantee similar scrolling behavior on mice
4005 that it knows about. There's also support for collecting
4006 information about Touchpad types.
4008 * udev's input_id built-in will now also collect touch screen
4009 dimension data and attach it to probed devices.
4011 * /etc/os-release gained support for a Distribution Privacy
4014 * networkd gained support for creating "ipvlan", "gretap",
4015 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
4017 * systemd-tmpfiles gained support for "a" lines for setting
4020 * systemd-nspawn will now mount /tmp in the container to
4021 tmpfs, automatically.
4023 * systemd now exposes the memory.usage_in_bytes cgroup
4024 attribute and shows it for each service in the "systemctl
4025 status" output, if available.
4027 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
4028 immediate reboot is triggered. This useful if shutdown is
4029 hung and is unable to complete, to expedite the
4030 operation. Note that this kind of reboot will still unmount
4031 all file systems, and hence should not result in fsck being
4034 * A .device unit for an optical block device will now be
4035 considered active only when a medium is in the drive. Also,
4036 mount units are now bound to their backing devices thus
4037 triggering automatic unmounting when devices become
4038 unavailable. With this in place systemd will now
4039 automatically unmount left-over mounts when a CD-ROM is
4040 ejected or an USB stick is yanked from the system.
4042 * networkd-wait-online now has support for waiting for
4043 specific interfaces only (with globbing), and for giving up
4044 after a configurable timeout.
4046 * networkd now exits when idle. It will be automatically
4047 restarted as soon as interfaces show up, are removed or
4048 change state. networkd will stay around as long as there is
4049 at least one DHCP state machine or similar around, that keep
4052 * networkd may now configure IPv6 link-local addressing in
4053 addition to IPv4 link-local addressing.
4055 * The IPv6 "token" for use in SLAAC may now be configured for
4056 each .network interface in networkd.
4058 * Routes configured with networkd may now be assigned a scope
4061 * networkd's [Match] sections now support globbing and lists
4062 of multiple space-separated matches per item.
4064 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
4065 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
4066 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
4067 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
4068 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
4069 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
4070 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
4071 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
4072 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
4073 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
4074 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
4075 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
4076 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
4077 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
4078 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
4079 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
4080 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
4081 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
4082 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
4083 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
4084 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
4085 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
4086 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
4087 Hoffmann, Zbigniew Jędrzejewski-Szmek
4089 — Berlin, 2015-02-16
4093 * When querying unit file enablement status (for example via
4094 "systemctl is-enabled"), a new state "indirect" is now known
4095 which indicates that a unit might not be enabled itself, but
4096 another unit listed in its Also= setting might be.
4098 * Similar to the various existing ConditionXYZ= settings for
4099 units, there are now matching AssertXYZ= settings. While
4100 failing conditions cause a unit to be skipped, but its job
4101 to succeed, failing assertions declared like this will cause
4102 a unit start operation and its job to fail.
4104 * hostnamed now knows a new chassis type "embedded".
4106 * systemctl gained a new "edit" command. When used on a unit
4107 file, this allows extending unit files with .d/ drop-in
4108 configuration snippets or editing the full file (after
4109 copying it from /usr/lib to /etc). This will invoke the
4110 user's editor (as configured with $EDITOR), and reload the
4111 modified configuration after editing.
4113 * "systemctl status" now shows the suggested enablement state
4114 for a unit, as declared in the (usually vendor-supplied)
4115 system preset files.
4117 * nss-myhostname will now resolve the single-label host name
4118 "gateway" to the locally configured default IP routing
4119 gateways, ordered by their metrics. This assigns a stable
4120 name to the used gateways, regardless which ones are
4121 currently configured. Note that the name will only be
4122 resolved after all other name sources (if nss-myhostname is
4123 configured properly) and should hence not negatively impact
4124 systems that use the single-label host name "gateway" in
4127 * systemd-inhibit now allows filtering by mode when listing
4130 * Scope and service units gained a new "Delegate" boolean
4131 property, which, when set, allows processes running inside the
4132 unit to further partition resources. This is primarily
4133 useful for systemd user instances as well as container
4136 * journald will now pick up audit messages directly from
4137 the kernel, and log them like any other log message. The
4138 audit fields are split up and fully indexed. This means that
4139 journalctl in many ways is now a (nicer!) alternative to
4140 ausearch, the traditional audit client. Note that this
4141 implements only a minimal audit client. If you want the
4142 special audit modes like reboot-on-log-overflow, please use
4143 the traditional auditd instead, which can be used in
4144 parallel to journald.
4146 * The ConditionSecurity= unit file option now understands the
4147 special string "audit" to check whether auditing is
4150 * journalctl gained two new commands --vacuum-size= and
4151 --vacuum-time= to delete old journal files until the
4152 remaining ones take up no more than the specified size on disk,
4153 or are not older than the specified time.
4155 * A new, native PPPoE library has been added to sd-network,
4156 systemd's library of light-weight networking protocols. This
4157 library will be used in a future version of networkd to
4158 enable PPPoE communication without an external pppd daemon.
4160 * The busctl tool now understands a new "capture" verb that
4161 works similar to "monitor", but writes a packet capture
4162 trace to STDOUT that can be redirected to a file which is
4163 compatible with libcap's capture file format. This can then
4164 be loaded in Wireshark and similar tools to inspect bus
4167 * The busctl tool now understands a new "tree" verb that shows
4168 the object trees of a specific service on the bus, or of all
4171 * The busctl tool now understands a new "introspect" verb that
4172 shows all interfaces and members of objects on the bus,
4173 including their signature and values. This is particularly
4174 useful to get more information about bus objects shown by
4175 the new "busctl tree" command.
4177 * The busctl tool now understands new verbs "call",
4178 "set-property" and "get-property" for invoking bus method
4179 calls, setting and getting bus object properties in a
4182 * busctl gained a new --augment-creds= argument that controls
4183 whether the tool shall augment credential information it
4184 gets from the bus with data from /proc, in a possibly
4187 * nspawn's --link-journal= switch gained two new values
4188 "try-guest" and "try-host" that work like "guest" and
4189 "host", but do not fail if the host has no persistent
4190 journalling enabled. -j is now equivalent to
4191 --link-journal=try-guest.
4193 * macvlan network devices created by nspawn will now have
4194 stable MAC addresses.
4196 * A new SmackProcessLabel= unit setting has been added, which
4197 controls the SMACK security label processes forked off by
4198 the respective unit shall use.
4200 * If compiled with --enable-xkbcommon, systemd-localed will
4201 verify x11 keymap settings by compiling the given keymap. It
4202 will spew out warnings if the compilation fails. This
4203 requires libxkbcommon to be installed.
4205 * When a coredump is collected, a larger number of metadata
4206 fields is now collected and included in the journal records
4207 created for it. More specifically, control group membership,
4208 environment variables, memory maps, working directory,
4209 chroot directory, /proc/$PID/status, and a list of open file
4210 descriptors is now stored in the log entry.
4212 * The udev hwdb now contains DPI information for mice. For
4215 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
4217 * All systemd programs that read standalone configuration
4218 files in /etc now also support a corresponding series of
4219 .conf.d configuration directories in /etc/, /run/,
4220 /usr/local/lib/, /usr/lib/, and (if configured with
4221 --enable-split-usr) /lib/. In particular, the following
4222 configuration files now have corresponding configuration
4223 directories: system.conf user.conf, logind.conf,
4224 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
4225 resolved.conf, timesyncd.conf, journal-remote.conf, and
4226 journal-upload.conf. Note that distributions should use the
4227 configuration directories in /usr/lib/; the directories in
4228 /etc/ are reserved for the system administrator.
4230 * systemd-rfkill will no longer take the rfkill device name
4231 into account when storing rfkill state on disk, as the name
4232 might be dynamically assigned and not stable. Instead, the
4233 ID_PATH udev variable combined with the rfkill type (wlan,
4234 bluetooth, ...) is used.
4236 * A new service systemd-machine-id-commit.service has been
4237 added. When used on systems where /etc is read-only during
4238 boot, and /etc/machine-id is not initialized (but an empty
4239 file), this service will copy the temporary machine ID
4240 created as replacement into /etc after the system is fully
4241 booted up. This is useful for systems that are freshly
4242 installed with a non-initialized machine ID, but should get
4243 a fixed machine ID for subsequent boots.
4245 * networkd's .netdev files now provide a large set of
4246 configuration parameters for VXLAN devices. Similarly, the
4247 bridge port cost parameter is now configurable in .network
4248 files. There's also new support for configuring IP source
4249 routing. networkd .link files gained support for a new
4250 OriginalName= match that is useful to match against the
4251 original interface name the kernel assigned. .network files
4252 may include MTU= and MACAddress= fields for altering the MTU
4253 and MAC address while being connected to a specific network
4256 * The LUKS logic gained supported for configuring
4257 UUID-specific key files. There's also new support for naming
4258 LUKS device from the kernel command line, using the new
4259 luks.name= argument.
4261 * Timer units may now be transiently created via the bus API
4262 (this was previously already available for scope and service
4263 units). In addition it is now possible to create multiple
4264 transient units at the same time with a single bus call. The
4265 "systemd-run" tool has been updated to make use of this for
4266 running commands on a specified time, in at(1)-style.
4268 * tmpfiles gained support for "t" lines, for assigning
4269 extended attributes to files. Among other uses this may be
4270 used to assign SMACK labels to files.
4272 Contributions from: Alin Rauta, Alison Chaiken, Andrej
4273 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
4274 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
4275 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
4276 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
4277 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
4278 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
4279 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
4280 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
4281 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
4282 Michael Biebl, Michael Chapman, Michael Marineau, Michal
4283 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
4284 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
4285 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
4286 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
4287 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4288 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
4289 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
4291 — Berlin, 2014-12-10
4295 * journalctl gained the new options -t/--identifier= to match
4296 on the syslog identifier (aka "tag"), as well as --utc to
4297 show log timestamps in the UTC timezone. journalctl now also
4298 accepts -n/--lines=all to disable line capping in a pager.
4300 * journalctl gained a new switch, --flush, that synchronously
4301 flushes logs from /run/log/journal to /var/log/journal if
4302 persistent storage is enabled. systemd-journal-flush.service
4303 now waits until the operation is complete.
4305 * Services can notify the manager before they start a reload
4306 (by sending RELOADING=1) or shutdown (by sending
4307 STOPPING=1). This allows the manager to track and show the
4308 internal state of daemons and closes a race condition when
4309 the process is still running but has closed its D-Bus
4312 * Services with Type=oneshot do not have to have any ExecStart
4315 * User units are now loaded also from
4316 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
4317 /run/systemd/user directory that was already previously
4318 supported, but is under the control of the user.
4320 * Job timeouts (i.e. time-outs on the time a job that is
4321 queued stays in the run queue) can now optionally result in
4322 immediate reboot or power-off actions (JobTimeoutAction= and
4323 JobTimeoutRebootArgument=). This is useful on ".target"
4324 units, to limit the maximum time a target remains
4325 undispatched in the run queue, and to trigger an emergency
4326 operation in such a case. This is now used by default to
4327 turn off the system if boot-up (as defined by everything in
4328 basic.target) hangs and does not complete for at least
4329 15min. Also, if power-off or reboot hang for at least 30min
4330 an immediate power-off/reboot operation is triggered. This
4331 functionality is particularly useful to increase reliability
4332 on embedded devices, but also on laptops which might
4333 accidentally get powered on when carried in a backpack and
4334 whose boot stays stuck in a hard disk encryption passphrase
4337 * systemd-logind can be configured to also handle lid switch
4338 events even when the machine is docked or multiple displays
4339 are attached (HandleLidSwitchDocked= option).
4341 * A helper binary and a service have been added which can be
4342 used to resume from hibernation in the initramfs. A
4343 generator will parse the resume= option on the kernel
4344 command line to trigger resume.
4346 * A user console daemon systemd-consoled has been
4347 added. Currently, it is a preview, and will so far open a
4348 single terminal on each session of the user marked as
4349 Desktop=systemd-console.
4351 * Route metrics can be specified for DHCP routes added by
4354 * The SELinux context of socket-activated services can be set
4355 from the information provided by the networking stack
4356 (SELinuxContextFromNet= option).
4358 * Userspace firmware loading support has been removed and
4359 the minimum supported kernel version is thus bumped to 3.7.
4361 * Timeout for udev workers has been increased from 1 to 3
4362 minutes, but a warning will be printed after 1 minute to
4363 help diagnose kernel modules that take a long time to load.
4365 * Udev rules can now remove tags on devices with TAG-="foobar".
4367 * systemd's readahead implementation has been removed. In many
4368 circumstances it didn't give expected benefits even for
4369 rotational disk drives and was becoming less relevant in the
4370 age of SSDs. As none of the developers has been using
4371 rotating media anymore, and nobody stepped up to actively
4372 maintain this component of systemd it has now been removed.
4374 * Swap units can use Options= to specify discard options.
4375 Discard options specified for swaps in /etc/fstab are now
4378 * Docker containers are now detected as a separate type of
4381 * The Password Agent protocol gained support for queries where
4382 the user input is shown, useful e.g. for user names.
4383 systemd-ask-password gained a new --echo option to turn that
4386 * The default sysctl.d/ snippets will now set:
4388 net.core.default_qdisc = fq_codel
4390 This selects Fair Queuing Controlled Delay as the default
4391 queuing discipline for network interfaces. fq_codel helps
4392 fight the network bufferbloat problem. It is believed to be
4393 a good default with no tuning required for most workloads.
4394 Downstream distributions may override this choice. On 10Gbit
4395 servers that do not do forwarding, "fq" may perform better.
4396 Systems without a good clocksource should use "pfifo_fast".
4398 * If kdbus is enabled during build a new option BusPolicy= is
4399 available for service units, that allows locking all service
4400 processes into a stricter bus policy, in order to limit
4401 access to various bus services, or even hide most of them
4402 from the service's view entirely.
4404 * networkctl will now show the .network and .link file
4405 networkd has applied to a specific interface.
4407 * sd-login gained a new API call sd_session_get_desktop() to
4408 query which desktop environment has been selected for a
4411 * UNIX utmp support is now compile-time optional to support
4412 legacy-free systems.
4414 * systemctl gained two new commands "add-wants" and
4415 "add-requires" for pulling in units from specific targets
4418 * If the word "rescue" is specified on the kernel command line
4419 the system will now boot into rescue mode (aka
4420 rescue.target), which was previously available only by
4421 specifying "1" or "systemd.unit=rescue.target" on the kernel
4422 command line. This new kernel command line option nicely
4423 mirrors the already existing "emergency" kernel command line
4426 * New kernel command line options mount.usr=, mount.usrflags=,
4427 mount.usrfstype= have been added that match root=, rootflags=,
4428 rootfstype= but allow mounting a specific file system to
4431 * The $NOTIFY_SOCKET is now also passed to control processes of
4432 services, not only the main process.
4434 * This version reenables support for fsck's -l switch. This
4435 means at least version v2.25 of util-linux is required for
4436 operation, otherwise dead-locks on device nodes may
4437 occur. Again: you need to update util-linux to at least
4438 v2.25 when updating systemd to v217.
4440 * The "multi-seat-x" tool has been removed from systemd, as
4441 its functionality has been integrated into X servers 1.16,
4442 and the tool is hence redundant. It is recommended to update
4443 display managers invoking this tool to simply invoke X
4444 directly from now on, again.
4446 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
4447 message flag has been added for all of systemd's PolicyKit
4448 authenticated method calls has been added. In particular
4449 this now allows optional interactive authorization via
4450 PolicyKit for many of PID1's privileged operations such as
4451 unit file enabling and disabling.
4453 * "udevadm hwdb --update" learnt a new switch "--usr" for
4454 placing the rebuilt hardware database in /usr instead of
4455 /etc. When used only hardware database entries stored in
4456 /usr will be used, and any user database entries in /etc are
4457 ignored. This functionality is useful for vendors to ship a
4458 pre-built database on systems where local configuration is
4459 unnecessary or unlikely.
4461 * Calendar time specifications in .timer units now also
4462 understand the strings "semi-annually", "quarterly" and
4463 "minutely" as shortcuts (in addition to the preexisting
4464 "anually", "hourly", ...).
4466 * systemd-tmpfiles will now correctly create files in /dev
4467 at boot which are marked for creation only at boot. It is
4468 recommended to always create static device nodes with 'c!'
4469 and 'b!', so that they are created only at boot and not
4470 overwritten at runtime.
4472 * When the watchdog logic is used for a service (WatchdogSec=)
4473 and the watchdog timeout is hit the service will now be
4474 terminated with SIGABRT (instead of just SIGTERM), in order
4475 to make sure a proper coredump and backtrace is
4476 generated. This ensures that hanging services will result in
4477 similar coredump/backtrace behaviour as services that hit a
4480 Contributions from: Andreas Henriksson, Andrei Borzenkov,
4481 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
4482 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
4483 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
4484 Herrmann, David Sommerseth, David Strauss, Emil Renner
4485 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
4486 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
4487 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
4488 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
4489 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
4490 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
4491 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
4492 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
4493 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
4494 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
4495 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
4496 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
4497 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
4498 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
4499 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
4500 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
4503 — Berlin, 2014-10-28
4507 * timedated no longer reads NTP implementation unit names from
4508 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
4509 implementations should add a
4511 Conflicts=systemd-timesyncd.service
4513 to their unit files to take over and replace systemd's NTP
4514 default functionality.
4516 * systemd-sysusers gained a new line type "r" for configuring
4517 which UID/GID ranges to allocate system users/groups
4518 from. Lines of type "u" may now add an additional column
4519 that specifies the home directory for the system user to be
4520 created. Also, systemd-sysusers may now optionally read user
4521 information from STDIN instead of a file. This is useful for
4522 invoking it from RPM preinst scriptlets that need to create
4523 users before the first RPM file is installed since these
4524 files might need to be owned by them. A new
4525 %sysusers_create_inline RPM macro has been introduced to do
4526 just that. systemd-sysusers now updates the shadow files as
4527 well as the user/group databases, which should enhance
4528 compatibility with certain tools like grpck.
4530 * A number of bus APIs of PID 1 now optionally consult
4531 PolicyKit to permit access for otherwise unprivileged
4532 clients under certain conditions. Note that this currently
4533 doesn't support interactive authentication yet, but this is
4534 expected to be added eventually, too.
4536 * /etc/machine-info now has new fields for configuring the
4537 deployment environment of the machine, as well as the
4538 location of the machine. hostnamectl has been updated with
4539 new command to update these fields.
4541 * systemd-timesyncd has been updated to automatically acquire
4542 NTP server information from systemd-networkd, which might
4543 have been discovered via DHCP.
4545 * systemd-resolved now includes a caching DNS stub resolver
4546 and a complete LLMNR name resolution implementation. A new
4547 NSS module "nss-resolve" has been added which can be used
4548 instead of glibc's own "nss-dns" to resolve hostnames via
4549 systemd-resolved. Hostnames, addresses and arbitrary RRs may
4550 be resolved via systemd-resolved D-Bus APIs. In contrast to
4551 the glibc internal resolver systemd-resolved is aware of
4552 multi-homed system, and keeps DNS server and caches separate
4553 and per-interface. Queries are sent simultaneously on all
4554 interfaces that have DNS servers configured, in order to
4555 properly handle VPNs and local LANs which might resolve
4556 separate sets of domain names. systemd-resolved may acquire
4557 DNS server information from systemd-networkd automatically,
4558 which in turn might have discovered them via DHCP. A tool
4559 "systemd-resolve-host" has been added that may be used to
4560 query the DNS logic in resolved. systemd-resolved implements
4561 IDNA and automatically uses IDNA or UTF-8 encoding depending
4562 on whether classic DNS or LLMNR is used as transport. In the
4563 next releases we intend to add a DNSSEC and mDNS/DNS-SD
4564 implementation to systemd-resolved.
4566 * A new NSS module nss-mymachines has been added, that
4567 automatically resolves the names of all local registered
4568 containers to their respective IP addresses.
4570 * A new client tool "networkctl" for systemd-networkd has been
4571 added. It currently is entirely passive and will query
4572 networking configuration from udev, rtnetlink and networkd,
4573 and present it to the user in a very friendly
4574 way. Eventually, we hope to extend it to become a full
4575 control utility for networkd.
4577 * .socket units gained a new DeferAcceptSec= setting that
4578 controls the kernels' TCP_DEFER_ACCEPT sockopt for
4579 TCP. Similarly, support for controlling TCP keep-alive
4580 settings has been added (KeepAliveTimeSec=,
4581 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
4582 turning off Nagle's algorithm on TCP has been added
4585 * logind learned a new session type "web", for use in projects
4586 like Cockpit which register web clients as PAM sessions.
4588 * timer units with at least one OnCalendar= setting will now
4589 be started only after time-sync.target has been
4590 reached. This way they will not elapse before the system
4591 clock has been corrected by a local NTP client or
4592 similar. This is particular useful on RTC-less embedded
4593 machines, that come up with an invalid system clock.
4595 * systemd-nspawn's --network-veth= switch should now result in
4596 stable MAC addresses for both the outer and the inner side
4599 * systemd-nspawn gained a new --volatile= switch for running
4600 container instances with /etc or /var unpopulated.
4602 * The kdbus client code has been updated to use the new Linux
4603 3.17 memfd subsystem instead of the old kdbus-specific one.
4605 * systemd-networkd's DHCP client and server now support
4606 FORCERENEW. There are also new configuration options to
4607 configure the vendor client identifier and broadcast mode
4610 * systemd will no longer inform the kernel about the current
4611 timezone, as this is necessarily incorrect and racy as the
4612 kernel has no understanding of DST and similar
4613 concepts. This hence means FAT timestamps will be always
4614 considered UTC, similar to what Android is already
4615 doing. Also, when the RTC is configured to the local time
4616 (rather than UTC) systemd will never synchronize back to it,
4617 as this might confuse Windows at a later boot.
4619 * systemd-analyze gained a new command "verify" for offline
4620 validation of unit files.
4622 * systemd-networkd gained support for a couple of additional
4623 settings for bonding networking setups. Also, the metric for
4624 statically configured routes may now be configured. For
4625 network interfaces where this is appropriate the peer IP
4626 address may now be configured.
4628 * systemd-networkd's DHCP client will no longer request
4629 broadcasting by default, as this tripped up some networks.
4630 For hardware where broadcast is required the feature should
4631 be switched back on using RequestBroadcast=yes.
4633 * systemd-networkd will now set up IPv4LL addresses (when
4634 enabled) even if DHCP is configured successfully.
4636 * udev will now default to respect network device names given
4637 by the kernel when the kernel indicates that these are
4638 predictable. This behavior can be tweaked by changing
4639 NamePolicy= in the relevant .link file.
4641 * A new library systemd-terminal has been added that
4642 implements full TTY stream parsing and rendering. This
4643 library is supposed to be used later on for implementing a
4644 full userspace VT subsystem, replacing the current kernel
4647 * A new tool systemd-journal-upload has been added to push
4648 journal data to a remote system running
4649 systemd-journal-remote.
4651 * journald will no longer forward all local data to another
4652 running syslog daemon. This change has been made because
4653 rsyslog (which appears to be the most commonly used syslog
4654 implementation these days) no longer makes use of this, and
4655 instead pulls the data out of the journal on its own. Since
4656 forwarding the messages to a non-existent syslog server is
4657 more expensive than we assumed we have now turned this
4658 off. If you run a syslog server that is not a recent rsyslog
4659 version, you have to turn this option on again
4660 (ForwardToSyslog= in journald.conf).
4662 * journald now optionally supports the LZ4 compressor for
4663 larger journal fields. This compressor should perform much
4664 better than XZ which was the previous default.
4666 * machinectl now shows the IP addresses of local containers,
4667 if it knows them, plus the interface name of the container.
4669 * A new tool "systemd-escape" has been added that makes it
4670 easy to escape strings to build unit names and similar.
4672 * sd_notify() messages may now include a new ERRNO= field
4673 which is parsed and collected by systemd and shown among the
4674 "systemctl status" output for a service.
4676 * A new component "systemd-firstboot" has been added that
4677 queries the most basic systemd information (timezone,
4678 hostname, root password) interactively on first
4679 boot. Alternatively it may also be used to provision these
4680 things offline on OS images installed into directories.
4682 * The default sysctl.d/ snippets will now set
4684 net.ipv4.conf.default.promote_secondaries=1
4686 This has the benefit of no flushing secondary IP addresses
4687 when primary addresses are removed.
4689 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
4690 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
4691 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
4692 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
4693 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
4694 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
4695 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
4696 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
4697 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
4698 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
4699 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
4700 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
4701 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
4702 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
4703 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
4705 — Berlin, 2014-08-19
4709 * A new tool systemd-sysusers has been added. This tool
4710 creates system users and groups in /etc/passwd and
4711 /etc/group, based on static declarative system user/group
4712 definitions in /usr/lib/sysusers.d/. This is useful to
4713 enable factory resets and volatile systems that boot up with
4714 an empty /etc directory, and thus need system users and
4715 groups created during early boot. systemd now also ships
4716 with two default sysusers.d/ files for the most basic
4717 users and groups systemd and the core operating system
4720 * A new tmpfiles snippet has been added that rebuilds the
4721 essential files in /etc on boot, should they be missing.
4723 * A directive for ensuring automatic clean-up of
4724 /var/cache/man/ has been removed from the default
4725 configuration. This line should now be shipped by the man
4726 implementation. The necessary change has been made to the
4727 man-db implementation. Note that you need to update your man
4728 implementation to one that ships this line, otherwise no
4729 automatic clean-up of /var/cache/man will take place.
4731 * A new condition ConditionNeedsUpdate= has been added that
4732 may conditionalize services to only run when /etc or /var
4733 are "older" than the vendor operating system resources in
4734 /usr. This is useful for reconstructing or updating /etc
4735 after an offline update of /usr or a factory reset, on the
4736 next reboot. Services that want to run once after such an
4737 update or reset should use this condition and order
4738 themselves before the new systemd-update-done.service, which
4739 will mark the two directories as fully updated. A number of
4740 service files have been added making use of this, to rebuild
4741 the udev hardware database, the journald message catalog and
4742 dynamic loader cache (ldconfig). The systemd-sysusers tool
4743 described above also makes use of this now. With this in
4744 place it is now possible to start up a minimal operating
4745 system with /etc empty cleanly. For more information on the
4746 concepts involved see this recent blog story:
4748 http://0pointer.de/blog/projects/stateless.html
4750 * A new system group "input" has been introduced, and all
4751 input device nodes get this group assigned. This is useful
4752 for system-level software to get access to input devices. It
4753 complements what is already done for "audio" and "video".
4755 * systemd-networkd learnt minimal DHCPv4 server support in
4756 addition to the existing DHCPv4 client support. It also
4757 learnt DHCPv6 client and IPv6 Router Solicitation client
4758 support. The DHCPv4 client gained support for static routes
4759 passed in from the server. Note that the [DHCPv4] section
4760 known in older systemd-networkd versions has been renamed to
4761 [DHCP] and is now also used by the DHCPv6 client. Existing
4762 .network files using settings of this section should be
4763 updated, though compatibility is maintained. Optionally, the
4764 client hostname may now be sent to the DHCP server.
4766 * networkd gained support for vxlan virtual networks as well
4767 as tun/tap and dummy devices.
4769 * networkd gained support for automatic allocation of address
4770 ranges for interfaces from a system-wide pool of
4771 addresses. This is useful for dynamically managing a large
4772 number of interfaces with a single network configuration
4773 file. In particular this is useful to easily assign
4774 appropriate IP addresses to the veth links of a large number
4775 of nspawn instances.
4777 * RPM macros for processing sysusers, sysctl and binfmt
4778 drop-in snippets at package installation time have been
4781 * The /etc/os-release file should now be placed in
4782 /usr/lib/os-release. The old location is automatically
4783 created as symlink. /usr/lib is the more appropriate
4784 location of this file, since it shall actually describe the
4785 vendor operating system shipped in /usr, and not the
4786 configuration stored in /etc.
4788 * .mount units gained a new boolean SloppyOptions= setting
4789 that maps to mount(8)'s -s option which enables permissive
4790 parsing of unknown mount options.
4792 * tmpfiles learnt a new "L+" directive which creates a symlink
4793 but (unlike "L") deletes a pre-existing file first, should
4794 it already exist and not already be the correct
4795 symlink. Similarly, "b+", "c+" and "p+" directives have been
4796 added as well, which create block and character devices, as
4797 well as fifos in the filesystem, possibly removing any
4798 pre-existing files of different types.
4800 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
4801 'argument' field (which so far specified the source to
4802 symlink/copy the files from) is now optional. If omitted the
4803 same file os copied from /usr/share/factory/ suffixed by the
4804 full destination path. This is useful for populating /etc
4805 with essential files, by copying them from vendor defaults
4806 shipped in /usr/share/factory/etc.
4808 * A new command "systemctl preset-all" has been added that
4809 applies the service preset settings to all installed unit
4810 files. A new switch --preset-mode= has been added that
4811 controls whether only enable or only disable operations
4814 * A new command "systemctl is-system-running" has been added
4815 that allows checking the overall state of the system, for
4816 example whether it is fully up and running.
4818 * When the system boots up with an empty /etc, the equivalent
4819 to "systemctl preset-all" is executed during early boot, to
4820 make sure all default services are enabled after a factory
4823 * systemd now contains a minimal preset file that enables the
4824 most basic services systemd ships by default.
4826 * Unit files' [Install] section gained a new DefaultInstance=
4827 field for defining the default instance to create if a
4828 template unit is enabled with no instance specified.
4830 * A new passive target cryptsetup-pre.target has been added
4831 that may be used by services that need to make they run and
4832 finish before the first LUKS cryptographic device is set up.
4834 * The /dev/loop-control and /dev/btrfs-control device nodes
4835 are now owned by the "disk" group by default, opening up
4836 access to this group.
4838 * systemd-coredump will now automatically generate a
4839 stack trace of all core dumps taking place on the system,
4840 based on elfutils' libdw library. This stack trace is logged
4843 * systemd-coredump may now optionally store coredumps directly
4844 on disk (in /var/lib/systemd/coredump, possibly compressed),
4845 instead of storing them unconditionally in the journal. This
4846 mode is the new default. A new configuration file
4847 /etc/systemd/coredump.conf has been added to configure this
4848 and other parameters of systemd-coredump.
4850 * coredumpctl gained a new "info" verb to show details about a
4851 specific coredump. A new switch "-1" has also been added
4852 that makes sure to only show information about the most
4853 recent entry instead of all entries. Also, as the tool is
4854 generally useful now the "systemd-" prefix of the binary
4855 name has been removed. Distributions that want to maintain
4856 compatibility with the old name should add a symlink from
4857 the old name to the new name.
4859 * journald's SplitMode= now defaults to "uid". This makes sure
4860 that unprivileged users can access their own coredumps with
4861 coredumpctl without restrictions.
4863 * New kernel command line options "systemd.wants=" (for
4864 pulling an additional unit during boot), "systemd.mask="
4865 (for masking a specific unit for the boot), and
4866 "systemd.debug-shell" (for enabling the debug shell on tty9)
4867 have been added. This is implemented in the new generator
4868 "systemd-debug-generator".
4870 * systemd-nspawn will now by default filter a couple of
4871 syscalls for containers, among them those required for
4872 kernel module loading, direct x86 IO port access, swap
4873 management, and kexec. Most importantly though
4874 open_by_handle_at() is now prohibited for containers,
4875 closing a hole similar to a recently discussed vulnerability
4876 in docker regarding access to files on file hierarchies the
4877 container should normally not have access to. Note that, for
4878 nspawn, we generally make no security claims anyway (and
4879 this is explicitly documented in the man page), so this is
4880 just a fix for one of the most obvious problems.
4882 * A new man page file-hierarchy(7) has been added that
4883 contains a minimized, modernized version of the file system
4884 layout systemd expects, similar in style to the FHS
4885 specification or hier(5). A new tool systemd-path(1) has
4886 been added to query many of these paths for the local
4889 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
4890 longer done. Since the directory now has a per-user size
4891 limit, and is cleaned on logout this appears unnecessary,
4892 in particular since this now brings the lifecycle of this
4893 directory closer in line with how IPC objects are handled.
4895 * systemd.pc now exports a number of additional directories,
4896 including $libdir (which is useful to identify the library
4897 path for the primary architecture of the system), and a
4898 couple of drop-in directories.
4900 * udev's predictable network interface names now use the dev_port
4901 sysfs attribute, introduced in linux 3.15 instead of dev_id to
4902 distinguish between ports of the same PCI function. dev_id should
4903 only be used for ports using the same HW address, hence the need
4906 * machined has been updated to export the OS version of a
4907 container (read from /etc/os-release and
4908 /usr/lib/os-release) on the bus. This is now shown in
4909 "machinectl status" for a machine.
4911 * A new service setting RestartForceExitStatus= has been
4912 added. If configured to a set of exit signals or process
4913 return values, the service will be restarted when the main
4914 daemon process exits with any of them, regardless of the
4917 * systemctl's -H switch for connecting to remote systemd
4918 machines has been extended so that it may be used to
4919 directly connect to a specific container on the
4920 host. "systemctl -H root@foobar:waldi" will now connect as
4921 user "root" to host "foobar", and then proceed directly to
4922 the container named "waldi". Note that currently you have to
4923 authenticate as user "root" for this to work, as entering
4924 containers is a privileged operation.
4926 Contributions from: Andreas Henriksson, Benjamin Steinwender,
4927 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
4928 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
4929 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
4930 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
4931 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
4932 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
4933 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
4934 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
4935 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
4936 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
4937 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
4939 — Berlin, 2014-07-03
4943 * As an experimental feature, udev now tries to lock the
4944 disk device node (flock(LOCK_SH|LOCK_NB)) while it
4945 executes events for the disk or any of its partitions.
4946 Applications like partitioning programs can lock the
4947 disk device node (flock(LOCK_EX)) and claim temporary
4948 device ownership that way; udev will entirely skip all event
4949 handling for this disk and its partitions. If the disk
4950 was opened for writing, the close will trigger a partition
4951 table rescan in udev's "watch" facility, and if needed
4952 synthesize "change" events for the disk and all its partitions.
4953 This is now unconditionally enabled, and if it turns out to
4954 cause major problems, we might turn it on only for specific
4955 devices, or might need to disable it entirely. Device Mapper
4956 devices are excluded from this logic.
4958 * We temporarily dropped the "-l" switch for fsck invocations,
4959 since they collide with the flock() logic above. util-linux
4960 upstream has been changed already to avoid this conflict,
4961 and we will readd "-l" as soon as util-linux with this
4962 change has been released.
4964 * The dependency on libattr has been removed. Since a long
4965 time, the extended attribute calls have moved to glibc, and
4966 libattr is thus unnecessary.
4968 * Virtualization detection works without privileges now. This
4969 means the systemd-detect-virt binary no longer requires
4970 CAP_SYS_PTRACE file capabilities, and our daemons can run
4971 with fewer privileges.
4973 * systemd-networkd now runs under its own "systemd-network"
4974 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
4975 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
4976 loses the ability to write to files owned by root this way.
4978 * Similarly, systemd-resolved now runs under its own
4979 "systemd-resolve" user with no capabilities remaining.
4981 * Similarly, systemd-bus-proxyd now runs under its own
4982 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
4984 * systemd-networkd gained support for setting up "veth"
4985 virtual Ethernet devices for container connectivity, as well
4986 as GRE and VTI tunnels.
4988 * systemd-networkd will no longer automatically attempt to
4989 manually load kernel modules necessary for certain tunnel
4990 transports. Instead, it is assumed the kernel loads them
4991 automatically when required. This only works correctly on
4992 very new kernels. On older kernels, please consider adding
4993 the kernel modules to /etc/modules-load.d/ as a work-around.
4995 * The resolv.conf file systemd-resolved generates has been
4996 moved to /run/systemd/resolve/. If you have a symlink from
4997 /etc/resolv.conf, it might be necessary to correct it.
4999 * Two new service settings, ProtectHome= and ProtectSystem=,
5000 have been added. When enabled, they will make the user data
5001 (such as /home) inaccessible or read-only and the system
5002 (such as /usr) read-only, for specific services. This allows
5003 very light-weight per-service sandboxing to avoid
5004 modifications of user data or system files from
5005 services. These two new switches have been enabled for all
5006 of systemd's long-running services, where appropriate.
5008 * Socket units gained new SocketUser= and SocketGroup=
5009 settings to set the owner user and group of AF_UNIX sockets
5010 and FIFOs in the file system.
5012 * Socket units gained a new RemoveOnStop= setting. If enabled,
5013 all FIFOS and sockets in the file system will be removed
5014 when the specific socket unit is stopped.
5016 * Socket units gained a new Symlinks= setting. It takes a list
5017 of symlinks to create to file system sockets or FIFOs
5018 created by the specific Unix sockets. This is useful to
5019 manage symlinks to socket nodes with the same life-cycle as
5022 * The /dev/log socket and /dev/initctl FIFO have been moved to
5023 /run, and have been replaced by symlinks. This allows
5024 connecting to these facilities even if PrivateDevices=yes is
5025 used for a service (which makes /dev/log itself unavailable,
5026 but /run is left). This also has the benefit of ensuring
5027 that /dev only contains device nodes, directories and
5028 symlinks, and nothing else.
5030 * sd-daemon gained two new calls sd_pid_notify() and
5031 sd_pid_notifyf(). They are similar to sd_notify() and
5032 sd_notifyf(), but allow overriding of the source PID of
5033 notification messages if permissions permit this. This is
5034 useful to send notify messages on behalf of a different
5035 process (for example, the parent process). The
5036 systemd-notify tool has been updated to make use of this
5037 when sending messages (so that notification messages now
5038 originate from the shell script invoking systemd-notify and
5039 not the systemd-notify process itself. This should minimize
5040 a race where systemd fails to associate notification
5041 messages to services when the originating process already
5044 * A new "on-abnormal" setting for Restart= has been added. If
5045 set, it will result in automatic restarts on all "abnormal"
5046 reasons for a process to exit, which includes unclean
5047 signals, core dumps, timeouts and watchdog timeouts, but
5048 does not include clean and unclean exit codes or clean
5049 signals. Restart=on-abnormal is an alternative for
5050 Restart=on-failure for services that shall be able to
5051 terminate and avoid restarts on certain errors, by
5052 indicating so with an unclean exit code. Restart=on-failure
5053 or Restart=on-abnormal is now the recommended setting for
5054 all long-running services.
5056 * If the InaccessibleDirectories= service setting points to a
5057 mount point (or if there are any submounts contained within
5058 it), it is now attempted to completely unmount it, to make
5059 the file systems truly unavailable for the respective
5062 * The ReadOnlyDirectories= service setting and
5063 systemd-nspawn's --read-only parameter are now recursively
5064 applied to all submounts, too.
5066 * Mount units may now be created transiently via the bus APIs.
5068 * The support for SysV and LSB init scripts has been removed
5069 from the systemd daemon itself. Instead, it is now
5070 implemented as a generator that creates native systemd units
5071 from these scripts when needed. This enables us to remove a
5072 substantial amount of legacy code from PID 1, following the
5073 fact that many distributions only ship a very small number
5074 of LSB/SysV init scripts nowadays.
5076 * Privileged Xen (dom0) domains are not considered
5077 virtualization anymore by the virtualization detection
5078 logic. After all, they generally have unrestricted access to
5079 the hardware and usually are used to manage the unprivileged
5082 * systemd-tmpfiles gained a new "C" line type, for copying
5083 files or entire directories.
5085 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
5086 lines. So far, they have been non-globbing versions of the
5087 latter, and have thus been redundant. In future, it is
5088 recommended to only use "z". "m" has hence been removed
5089 from the documentation, even though it stays supported.
5091 * A tmpfiles snippet to recreate the most basic structure in
5092 /var has been added. This is enough to create the /var/run →
5093 /run symlink and create a couple of structural
5094 directories. This allows systems to boot up with an empty or
5095 volatile /var. Of course, while with this change, the core OS
5096 now is capable with dealing with a volatile /var, not all
5097 user services are ready for it. However, we hope that sooner
5098 or later, many service daemons will be changed upstream so
5099 that they are able to automatically create their necessary
5100 directories in /var at boot, should they be missing. This is
5101 the first step to allow state-less systems that only require
5102 the vendor image for /usr to boot.
5104 * systemd-nspawn has gained a new --tmpfs= switch to mount an
5105 empty tmpfs instance to a specific directory. This is
5106 particularly useful for making use of the automatic
5107 reconstruction of /var (see above), by passing --tmpfs=/var.
5109 * Access modes specified in tmpfiles snippets may now be
5110 prefixed with "~", which indicates that they shall be masked
5111 by whether the existing file or directory is currently
5112 writable, readable or executable at all. Also, if specified,
5113 the sgid/suid/sticky bits will be masked for all
5116 * A new passive target unit "network-pre.target" has been
5117 added which is useful for services that shall run before any
5118 network is configured, for example firewall scripts.
5120 * The "floppy" group that previously owned the /dev/fd*
5121 devices is no longer used. The "disk" group is now used
5122 instead. Distributions should probably deprecate usage of
5125 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
5126 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
5127 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
5128 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
5129 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
5130 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
5133 — Berlin, 2014-06-11
5137 * A new "systemd-timesyncd" daemon has been added for
5138 synchronizing the system clock across the network. It
5139 implements an SNTP client. In contrast to NTP
5140 implementations such as chrony or the NTP reference server,
5141 this only implements a client side, and does not bother with
5142 the full NTP complexity, focusing only on querying time from
5143 one remote server and synchronizing the local clock to
5144 it. Unless you intend to serve NTP to networked clients or
5145 want to connect to local hardware clocks, this simple NTP
5146 client should be more than appropriate for most
5147 installations. The daemon runs with minimal privileges, and
5148 has been hooked up with networkd to only operate when
5149 network connectivity is available. The daemon saves the
5150 current clock to disk every time a new NTP sync has been
5151 acquired, and uses this to possibly correct the system clock
5152 early at bootup, in order to accommodate for systems that
5153 lack an RTC such as the Raspberry Pi and embedded devices,
5154 and to make sure that time monotonically progresses on these
5155 systems, even if it is not always correct. To make use of
5156 this daemon, a new system user and group "systemd-timesync"
5157 needs to be created on installation of systemd.
5159 * The queue "seqnum" interface of libudev has been disabled, as
5160 it was generally incompatible with device namespacing as
5161 sequence numbers of devices go "missing" if the devices are
5162 part of a different namespace.
5164 * "systemctl list-timers" and "systemctl list-sockets" gained
5165 a --recursive switch for showing units of these types also
5166 for all local containers, similar in style to the already
5167 supported --recursive switch for "systemctl list-units".
5169 * A new RebootArgument= setting has been added for service
5170 units, which may be used to specify a kernel reboot argument
5171 to use when triggering reboots with StartLimitAction=.
5173 * A new FailureAction= setting has been added for service
5174 units which may be used to specify an operation to trigger
5175 when a service fails. This works similarly to
5176 StartLimitAction=, but unlike it, controls what is done
5177 immediately rather than only after several attempts to
5178 restart the service in question.
5180 * hostnamed got updated to also expose the kernel name,
5181 release, and version on the bus. This is useful for
5182 executing commands like hostnamectl with the -H switch.
5183 systemd-analyze makes use of this to properly display
5184 details when running non-locally.
5186 * The bootchart tool can now show cgroup information in the
5187 graphs it generates.
5189 * The CFS CPU quota cgroup attribute is now exposed for
5190 services. The new CPUQuota= switch has been added for this
5191 which takes a percentage value. Setting this will have the
5192 result that a service may never get more CPU time than the
5193 specified percentage, even if the machine is otherwise idle.
5195 * systemd-networkd learned IPIP and SIT tunnel support.
5197 * LSB init scripts exposing a dependency on $network will now
5198 get a dependency on network-online.target rather than simply
5199 network.target. This should bring LSB handling closer to
5200 what it was on SysV systems.
5202 * A new fsck.repair= kernel option has been added to control
5203 how fsck shall deal with unclean file systems at boot.
5205 * The (.ini) configuration file parser will now silently
5206 ignore sections whose name begins with "X-". This may be
5207 used to maintain application-specific extension sections in unit
5210 * machined gained a new API to query the IP addresses of
5211 registered containers. "machinectl status" has been updated
5212 to show these addresses in its output.
5214 * A new call sd_uid_get_display() has been added to the
5215 sd-login APIs for querying the "primary" session of a
5216 user. The "primary" session of the user is elected from the
5217 user's sessions and generally a graphical session is
5218 preferred over a text one.
5220 * A minimal systemd-resolved daemon has been added. It
5221 currently simply acts as a companion to systemd-networkd and
5222 manages resolv.conf based on per-interface DNS
5223 configuration, possibly supplied via DHCP. In the long run
5224 we hope to extend this into a local DNSSEC enabled DNS and
5227 * The systemd-networkd-wait-online tool is now enabled by
5228 default. It will delay network-online.target until a network
5229 connection has been configured. The tool primarily integrates
5230 with networkd, but will also make a best effort to make sense
5231 of network configuration performed in some other way.
5233 * Two new service options StartupCPUShares= and
5234 StartupBlockIOWeight= have been added that work similarly to
5235 CPUShares= and BlockIOWeight= however only apply during
5236 system startup. This is useful to prioritize certain services
5237 differently during bootup than during normal runtime.
5239 * hostnamed has been changed to prefer the statically
5240 configured hostname in /etc/hostname (unless set to
5241 'localhost' or empty) over any dynamic one supplied by
5242 dhcp. With this change, the rules for picking the hostname
5243 match more closely the rules of other configuration settings
5244 where the local administrator's configuration in /etc always
5245 overrides any other settings.
5247 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
5248 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
5249 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
5250 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
5251 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
5252 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
5253 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
5254 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
5255 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
5256 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
5257 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
5258 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
5259 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
5260 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
5261 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
5262 Lindskog, WaLyong Cho, Will Woods, Zbigniew
5265 — Beijing, 2014-05-28
5269 * When restoring the screen brightness at boot, stay away from
5270 the darkest setting or from the lowest 5% of the available
5271 range, depending on which is the larger value of both. This
5272 should effectively protect the user from rebooting into a
5273 black screen, should the brightness have been set to minimum
5276 * sd-login gained a new sd_machine_get_class() call to
5277 determine the class ("vm" or "container") of a machine
5278 registered with machined.
5280 * sd-login gained new calls
5281 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
5282 to query the identity of the peer of a local AF_UNIX
5283 connection. They operate similarly to their sd_pid_get_xyz()
5286 * PID 1 will now maintain a system-wide system state engine
5287 with the states "starting", "running", "degraded",
5288 "maintenance", "stopping". These states are bound to system
5289 startup, normal runtime, runtime with at least one failed
5290 service, rescue/emergency mode and system shutdown. This
5291 state is shown in the "systemctl status" output when no unit
5292 name is passed. It is useful to determine system state, in
5293 particularly when doing so for many systems or containers at
5296 * A new command "list-machines" has been added to "systemctl"
5297 that lists all local OS containers and shows their system
5298 state (see above), if systemd runs inside of them.
5300 * systemctl gained a new "-r" switch to recursively enumerate
5301 units on all local containers, when used with the
5302 "list-unit" command (which is the default one that is
5303 executed when no parameters are specified).
5305 * The GPT automatic partition discovery logic will now honour
5306 two GPT partition flags: one may be set on a partition to
5307 cause it to be mounted read-only, and the other may be set
5308 on a partition to ignore it during automatic discovery.
5310 * Two new GPT type UUIDs have been added for automatic root
5311 partition discovery, for 32-bit and 64-bit ARM. This is not
5312 particularly useful for discovering the root directory on
5313 these architectures during bare-metal boots (since UEFI is
5314 not common there), but still very useful to allow booting of
5315 ARM disk images in nspawn with the -i option.
5317 * MAC addresses of interfaces created with nspawn's
5318 --network-interface= switch will now be generated from the
5319 machine name, and thus be stable between multiple invocations
5322 * logind will now automatically remove all IPC objects owned
5323 by a user if she or he fully logs out. This makes sure that
5324 users who are logged out cannot continue to consume IPC
5325 resources. This covers SysV memory, semaphores and message
5326 queues as well as POSIX shared memory and message
5327 queues. Traditionally, SysV and POSIX IPC had no life-cycle
5328 limits. With this functionality, that is corrected. This may
5329 be turned off by using the RemoveIPC= switch of logind.conf.
5331 * The systemd-machine-id-setup and tmpfiles tools gained a
5332 --root= switch to operate on a specific root directory,
5335 * journald can now forward logged messages to the TTYs of all
5336 logged in users ("wall"). This is the default for all
5337 emergency messages now.
5339 * A new tool systemd-journal-remote has been added to stream
5340 journal log messages across the network.
5342 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
5343 controller trees are mounted into it. Note that the
5344 directories mounted beneath it are not read-only. This is a
5345 security measure and is particularly useful because glibc
5346 actually includes a search logic to pick any tmpfs it can
5347 find to implement shm_open() if /dev/shm is not available
5348 (which it might very well be in namespaced setups).
5350 * machinectl gained a new "poweroff" command to cleanly power
5351 down a local OS container.
5353 * The PrivateDevices= unit file setting will now also drop the
5354 CAP_MKNOD capability from the capability bound set, and
5355 imply DevicePolicy=closed.
5357 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
5358 comprehensively on all long-running systemd services where
5359 this is appropriate.
5361 * systemd-udevd will now run in a disassociated mount
5362 namespace. To mount directories from udev rules, make sure to
5363 pull in mount units via SYSTEMD_WANTS properties.
5365 * The kdbus support gained support for uploading policy into
5366 the kernel. sd-bus gained support for creating "monitoring"
5367 connections that can eavesdrop into all bus communication
5368 for debugging purposes.
5370 * Timestamps may now be specified in seconds since the UNIX
5371 epoch Jan 1st, 1970 by specifying "@" followed by the value
5374 * Native tcpwrap support in systemd has been removed. tcpwrap
5375 is old code, not really maintained anymore and has serious
5376 shortcomings, and better options such as firewalls
5377 exist. For setups that require tcpwrap usage, please
5378 consider invoking your socket-activated service via tcpd,
5379 like on traditional inetd.
5381 * A new system.conf configuration option
5382 DefaultTimerAccuracySec= has been added that controls the
5383 default AccuracySec= setting of .timer units.
5385 * Timer units gained a new WakeSystem= switch. If enabled,
5386 timers configured this way will cause the system to resume
5387 from system suspend (if the system supports that, which most
5390 * Timer units gained a new Persistent= switch. If enabled,
5391 timers configured this way will save to disk when they have
5392 been last triggered. This information is then used on next
5393 reboot to possible execute overdue timer events, that
5394 could not take place because the system was powered off.
5395 This enables simple anacron-like behaviour for timer units.
5397 * systemctl's "list-timers" will now also list the time a
5398 timer unit was last triggered in addition to the next time
5399 it will be triggered.
5401 * systemd-networkd will now assign predictable IPv4LL
5402 addresses to its local interfaces.
5404 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
5405 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
5406 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
5407 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
5408 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
5409 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
5410 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
5411 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
5414 — Berlin, 2014-03-25
5418 * A new unit file setting RestrictAddressFamilies= has been
5419 added to restrict which socket address families unit
5420 processes gain access to. This takes address family names
5421 like "AF_INET" or "AF_UNIX", and is useful to minimize the
5422 attack surface of services via exotic protocol stacks. This
5423 is built on seccomp system call filters.
5425 * Two new unit file settings RuntimeDirectory= and
5426 RuntimeDirectoryMode= have been added that may be used to
5427 manage a per-daemon runtime directories below /run. This is
5428 an alternative for setting up directory permissions with
5429 tmpfiles snippets, and has the advantage that the runtime
5430 directory's lifetime is bound to the daemon runtime and that
5431 the daemon starts up with an empty directory each time. This
5432 is particularly useful when writing services that drop
5433 privileges using the User= or Group= setting.
5435 * The DeviceAllow= unit setting now supports globbing for
5436 matching against device group names.
5438 * The systemd configuration file system.conf gained new
5439 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
5440 DefaultMemoryAccounting= to globally turn on/off accounting
5441 for specific resources (cgroups) for all units. These
5442 settings may still be overridden individually in each unit
5445 * systemd-gpt-auto-generator is now able to discover /srv and
5446 root partitions in addition to /home and swap partitions. It
5447 also supports LUKS-encrypted partitions now. With this in
5448 place, automatic discovery of partitions to mount following
5449 the Discoverable Partitions Specification
5450 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
5451 is now a lot more complete. This allows booting without
5452 /etc/fstab and without root= on the kernel command line on
5453 systems prepared appropriately.
5455 * systemd-nspawn gained a new --image= switch which allows
5456 booting up disk images and Linux installations on any block
5457 device that follow the Discoverable Partitions Specification
5458 (see above). This means that installations made with
5459 appropriately updated installers may now be started and
5460 deployed using container managers, completely
5461 unmodified. (We hope that libvirt-lxc will add support for
5462 this feature soon, too.)
5464 * systemd-nspawn gained a new --network-macvlan= setting to
5465 set up a private macvlan interface for the
5466 container. Similarly, systemd-networkd gained a new
5467 Kind=macvlan setting in .netdev files.
5469 * systemd-networkd now supports configuring local addresses
5472 * A new tool systemd-network-wait-online has been added to
5473 synchronously wait for network connectivity using
5476 * The sd-bus.h bus API gained a new sd_bus_track object for
5477 tracking the life-cycle of bus peers. Note that sd-bus.h is
5478 still not a public API though (unless you specify
5479 --enable-kdbus on the configure command line, which however
5480 voids your warranty and you get no API stability guarantee).
5482 * The $XDG_RUNTIME_DIR runtime directories for each user are
5483 now individual tmpfs instances, which has the benefit of
5484 introducing separate pools for each user, with individual
5485 size limits, and thus making sure that unprivileged clients
5486 can no longer negatively impact the system or other users by
5487 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
5488 RuntimeDirectorySize= has been introduced that allows
5489 controlling the default size limit for all users. It
5490 defaults to 10% of the available physical memory. This is no
5491 replacement for quotas on tmpfs though (which the kernel
5492 still does not support), as /dev/shm and /tmp are still
5493 shared resources used by both the system and unprivileged
5496 * logind will now automatically turn off automatic suspending
5497 on laptop lid close when more than one display is
5498 connected. This was previously expected to be implemented
5499 individually in desktop environments (such as GNOME),
5500 however has been added to logind now, in order to fix a
5501 boot-time race where a desktop environment might not have
5502 been started yet and thus not been able to take an inhibitor
5503 lock at the time where logind already suspends the system
5504 due to a closed lid.
5506 * logind will now wait at least 30s after each system
5507 suspend/resume cycle, and 3min after system boot before
5508 suspending the system due to a closed laptop lid. This
5509 should give USB docking stations and similar enough time to
5510 be probed and configured after system resume and boot in
5511 order to then act as suspend blocker.
5513 * systemd-run gained a new --property= setting which allows
5514 initialization of resource control properties (and others)
5515 for the created scope or service unit. Example: "systemd-run
5516 --property=BlockIOWeight=10 updatedb" may be used to run
5517 updatedb at a low block IO scheduling weight.
5519 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
5520 now also work in --scope mode.
5522 * When systemd is compiled with kdbus support, basic support
5523 for enforced policies is now in place. (Note that enabling
5524 kdbus still voids your warranty and no API compatibility
5527 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
5528 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5529 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
5530 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
5531 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
5532 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
5533 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
5534 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
5535 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
5536 Zbigniew Jędrzejewski-Szmek
5538 — Berlin, 2014-03-12
5542 * systemd will now relabel /dev after loading the SMACK policy
5543 according to SMACK rules.
5545 * A new unit file option AppArmorProfile= has been added to
5546 set the AppArmor profile for the processes of a unit.
5548 * A new condition check ConditionArchitecture= has been added
5549 to conditionalize units based on the system architecture, as
5550 reported by uname()'s "machine" field.
5552 * systemd-networkd now supports matching on the system
5553 virtualization, architecture, kernel command line, host name
5556 * logind is now a lot more aggressive when suspending the
5557 machine due to a closed laptop lid. Instead of acting only
5558 on the lid close action, it will continuously watch the lid
5559 status and act on it. This is useful for laptops where the
5560 power button is on the outside of the chassis so that it can
5561 be reached without opening the lid (such as the Lenovo
5562 Yoga). On those machines, logind will now immediately
5563 re-suspend the machine if the power button has been
5564 accidentally pressed while the laptop was suspended and in a
5565 backpack or similar.
5567 * logind will now watch SW_DOCK switches and inhibit reaction
5568 to the lid switch if it is pressed. This means that logind
5569 will not suspend the machine anymore if the lid is closed
5570 and the system is docked, if the laptop supports SW_DOCK
5571 notifications via the input layer. Note that ACPI docking
5572 stations do not generate this currently. Also note that this
5573 logic is usually not fully sufficient and Desktop
5574 Environments should take a lid switch inhibitor lock when an
5575 external display is connected, as systemd will not watch
5578 * nspawn will now make use of the devices cgroup controller by
5579 default, and only permit creation of and access to the usual
5580 API device nodes like /dev/null or /dev/random, as well as
5581 access to (but not creation of) the pty devices.
5583 * We will now ship a default .network file for
5584 systemd-networkd that automatically configures DHCP for
5585 network interfaces created by nspawn's --network-veth or
5586 --network-bridge= switches.
5588 * systemd will now understand the usual M, K, G, T suffixes
5589 according to SI conventions (i.e. to the base 1000) when
5590 referring to throughput and hardware metrics. It will stay
5591 with IEC conventions (i.e. to the base 1024) for software
5592 metrics, according to what is customary according to
5593 Wikipedia. We explicitly document which base applies for
5594 each configuration option.
5596 * The DeviceAllow= setting in unit files now supports a syntax
5597 to whitelist an entire group of devices node majors at once,
5598 based on the /proc/devices listing. For example, with the
5599 string "char-pts", it is now possible to whitelist all
5600 current and future pseudo-TTYs at once.
5602 * sd-event learned a new "post" event source. Event sources of
5603 this type are triggered by the dispatching of any event
5604 source of a type that is not "post". This is useful for
5605 implementing clean-up and check event sources that are
5606 triggered by other work being done in the program.
5608 * systemd-networkd is no longer statically enabled, but uses
5609 the usual [Install] sections so that it can be
5610 enabled/disabled using systemctl. It still is enabled by
5613 * When creating a veth interface pair with systemd-nspawn, the
5614 host side will now be prefixed with "vb-" if
5615 --network-bridge= is used, and with "ve-" if --network-veth
5616 is used. This way, it is easy to distinguish these cases on
5617 the host, for example to apply different configuration to
5618 them with systemd-networkd.
5620 * The compatibility libraries for libsystemd-journal.so,
5621 libsystem-id128.so, libsystemd-login.so and
5622 libsystemd-daemon.so do not make use of IFUNC
5623 anymore. Instead, we now build libsystemd.so multiple times
5624 under these alternative names. This means that the footprint
5625 is drastically increased, but given that these are
5626 transitional compatibility libraries, this should not matter
5627 much. This change has been made necessary to support the ARM
5628 platform for these compatibility libraries, as the ARM
5629 toolchain is not really at the same level as the toolchain
5630 for other architectures like x86 and does not support
5631 IFUNC. Please make sure to use --enable-compat-libs only
5632 during a transitional period!
5634 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
5635 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5636 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
5637 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
5638 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
5639 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
5640 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
5641 Zbigniew Jędrzejewski-Szmek
5643 — Berlin, 2014-02-24
5647 * A new component "systemd-networkd" has been added that can
5648 be used to configure local network interfaces statically or
5649 via DHCP. It is capable of bringing up bridges, VLANs, and
5650 bonding. Currently, no hook-ups for interactive network
5651 configuration are provided. Use this for your initrd,
5652 container, embedded, or server setup if you need a simple,
5653 yet powerful, network configuration solution. This
5654 configuration subsystem is quite nifty, as it allows wildcard
5655 hotplug matching in interfaces. For example, with a single
5656 configuration snippet, you can configure that all Ethernet
5657 interfaces showing up are automatically added to a bridge,
5658 or similar. It supports link-sensing and more.
5660 * A new tool "systemd-socket-proxyd" has been added which can
5661 act as a bidirectional proxy for TCP sockets. This is
5662 useful for adding socket activation support to services that
5663 do not actually support socket activation, including virtual
5664 machines and the like.
5666 * Add a new tool to save/restore rfkill state on
5669 * Save/restore state of keyboard backlights in addition to
5670 display backlights on shutdown/boot.
5672 * udev learned a new SECLABEL{} construct to label device
5673 nodes with a specific security label when they appear. For
5674 now, only SECLABEL{selinux} is supported, but the syntax is
5675 prepared for additional security frameworks.
5677 * udev gained a new scheme to configure link-level attributes
5678 from files in /etc/systemd/network/*.link. These files can
5679 match against MAC address, device path, driver name and type,
5680 and will apply attributes like the naming policy, link speed,
5681 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
5682 address assignment policy (randomized, ...).
5684 * The configuration of network interface naming rules for
5685 "permanent interface names" has changed: a new NamePolicy=
5686 setting in the [Link] section of .link files determines the
5687 priority of possible naming schemes (onboard, slot, MAC,
5688 path). The default value of this setting is determined by
5689 /usr/lib/net/links/99-default.link. Old
5690 80-net-name-slot.rules udev configuration file has been
5691 removed, so local configuration overriding this file should
5692 be adapted to override 99-default.link instead.
5694 * When the User= switch is used in a unit file, also
5695 initialize $SHELL= based on the user database entry.
5697 * systemd no longer depends on libdbus. All communication is
5698 now done with sd-bus, systemd's low-level bus library
5701 * kdbus support has been added to PID 1 itself. When kdbus is
5702 enabled, this causes PID 1 to set up the system bus and
5703 enable support for a new ".busname" unit type that
5704 encapsulates bus name activation on kdbus. It works a little
5705 bit like ".socket" units, except for bus names. A new
5706 generator has been added that converts classic dbus1 service
5707 activation files automatically into native systemd .busname
5710 * sd-bus: add a light-weight vtable implementation that allows
5711 defining objects on the bus with a simple static const
5712 vtable array of its methods, signals and properties.
5714 * systemd will not generate or install static dbus
5715 introspection data anymore to /usr/share/dbus-1/interfaces,
5716 as the precise format of these files is unclear, and
5717 nothing makes use of it.
5719 * A proxy daemon is now provided to proxy clients connecting
5720 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
5721 compatibility with classic D-Bus.
5723 * A bus driver implementation has been added that supports the
5724 classic D-Bus bus driver calls on kdbus, also for
5725 compatibility purposes.
5727 * A new API "sd-event.h" has been added that implements a
5728 minimal event loop API built around epoll. It provides a
5729 couple of features that direct epoll usage is lacking:
5730 prioritization of events, scales to large numbers of timer
5731 events, per-event timer slack (accuracy), system-wide
5732 coalescing of timer events, exit handlers, watchdog
5733 supervision support using systemd's sd_notify() API, child
5736 * A new API "sd-rntl.h" has been added that provides an API
5737 around the route netlink interface of the kernel, similar in
5738 style to "sd-bus.h".
5740 * A new API "sd-dhcp-client.h" has been added that provides a
5741 small DHCPv4 client-side implementation. This is used by
5744 * There is a new kernel command line option
5745 "systemd.restore_state=0|1". When set to "0", none of the
5746 systemd tools will restore saved runtime state to hardware
5747 devices. More specifically, the rfkill and backlight states
5750 * The FsckPassNo= compatibility option in mount/service units
5751 has been removed. The fstab generator will now add the
5752 necessary dependencies automatically, and does not require
5753 PID1's support for that anymore.
5755 * journalctl gained a new switch, --list-boots, that lists
5756 recent boots with their times and boot IDs.
5758 * The various tools like systemctl, loginctl, timedatectl,
5759 busctl, systemd-run, ... have gained a new switch "-M" to
5760 connect to a specific, local OS container (as direct
5761 connection, without requiring SSH). This works on any
5762 container that is registered with machined, such as those
5763 created by libvirt-lxc or nspawn.
5765 * systemd-run and systemd-analyze also gained support for "-H"
5766 to connect to remote hosts via SSH. This is particularly
5767 useful for systemd-run because it enables queuing of jobs
5768 onto remote systems.
5770 * machinectl gained a new command "login" to open a getty
5771 login in any local container. This works with any container
5772 that is registered with machined (such as those created by
5773 libvirt-lxc or nspawn), and which runs systemd inside.
5775 * machinectl gained a new "reboot" command that may be used to
5776 trigger a reboot on a specific container that is registered
5777 with machined. This works on any container that runs an init
5778 system of some kind.
5780 * systemctl gained a new "list-timers" command to print a nice
5781 listing of installed timer units with the times they elapse
5784 * Alternative reboot() parameters may now be specified on the
5785 "systemctl reboot" command line and are passed to the
5786 reboot() system call.
5788 * systemctl gained a new --job-mode= switch to configure the
5789 mode to queue a job with. This is a more generic version of
5790 --fail, --irreversible, and --ignore-dependencies, which are
5791 still available but not advertised anymore.
5793 * /etc/systemd/system.conf gained new settings to configure
5794 various default timeouts of units, as well as the default
5795 start limit interval and burst. These may still be overridden
5798 * PID1 will now export on the bus profile data of the security
5799 policy upload process (such as the SELinux policy upload to
5802 * journald: when forwarding logs to the console, include
5803 timestamps (following the setting in
5804 /sys/module/printk/parameters/time).
5806 * OnCalendar= in timer units now understands the special
5807 strings "yearly" and "annually". (Both are equivalent)
5809 * The accuracy of timer units is now configurable with the new
5810 AccuracySec= setting. It defaults to 1min.
5812 * A new dependency type JoinsNamespaceOf= has been added that
5813 allows running two services within the same /tmp and network
5814 namespace, if PrivateNetwork= or PrivateTmp= are used.
5816 * A new command "cat" has been added to systemctl. It outputs
5817 the original unit file of a unit, and concatenates the
5818 contents of additional "drop-in" unit file snippets, so that
5819 the full configuration is shown.
5821 * systemctl now supports globbing on the various "list-xyz"
5822 commands, like "list-units" or "list-sockets", as well as on
5823 those commands which take multiple unit names.
5825 * journalctl's --unit= switch gained support for globbing.
5827 * All systemd daemons now make use of the watchdog logic so
5828 that systemd automatically notices when they hang.
5830 * If the $container_ttys environment variable is set,
5831 getty-generator will automatically spawn a getty for each
5832 listed tty. This is useful for container managers to request
5833 login gettys to be spawned on as many ttys as needed.
5835 * %h, %s, %U specifier support is not available anymore when
5836 used in unit files for PID 1. This is because NSS calls are
5837 not safe from PID 1. They stay available for --user
5838 instances of systemd, and as special case for the root user.
5840 * loginctl gained a new "--no-legend" switch to turn off output
5843 * The "sd-login.h" API gained three new calls:
5844 sd_session_is_remote(), sd_session_get_remote_user(),
5845 sd_session_get_remote_host() to query information about
5848 * The udev hardware database now also carries vendor/product
5849 information of SDIO devices.
5851 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
5852 determine whether watchdog notifications are requested by
5855 * Socket-activated per-connection services now include a
5856 short description of the connection parameters in the
5859 * tmpfiles gained a new "--boot" option. When this is not used,
5860 only lines where the command character is not suffixed with
5861 "!" are executed. When this option is specified, those
5862 options are executed too. This partitions tmpfiles
5863 directives into those that can be safely executed at any
5864 time, and those which should be run only at boot (for
5865 example, a line that creates /run/nologin).
5867 * A new API "sd-resolve.h" has been added which provides a simple
5868 asynchronous wrapper around glibc NSS host name resolution
5869 calls, such as getaddrinfo(). In contrast to glibc's
5870 getaddrinfo_a(), it does not use signals. In contrast to most
5871 other asynchronous name resolution libraries, this one does
5872 not reimplement DNS, but reuses NSS, so that alternate
5873 host name resolution systems continue to work, such as mDNS,
5874 LDAP, etc. This API is based on libasyncns, but it has been
5875 cleaned up for inclusion in systemd.
5877 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
5878 "sd-daemon.h" are no longer found in individual libraries
5879 libsystemd-journal.so, libsystemd-login.so,
5880 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
5881 merged them into a single library, libsystemd.so, which
5882 provides all symbols. The reason for this is cyclic
5883 dependencies, as these libraries tend to use each other's
5884 symbols. So far, we have managed to workaround that by linking
5885 a copy of a good part of our code into each of these
5886 libraries again and again, which, however, makes certain
5887 things hard to do, like sharing static variables. Also, it
5888 substantially increases footprint. With this change, there
5889 is only one library for the basic APIs systemd
5890 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
5891 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
5892 library as well, however are subject to the --enable-kdbus
5893 switch (see below). Note that "sd-dhcp-client.h" is not part
5894 of this library (this is because it only consumes, never
5895 provides, services of/to other APIs). To make the transition
5896 easy from the separate libraries to the unified one, we
5897 provide the --enable-compat-libs compile-time switch which
5898 will generate stub libraries that are compatible with the
5899 old ones but redirect all calls to the new one.
5901 * All of the kdbus logic and the new APIs "sd-bus.h",
5902 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
5903 and "sd-utf8.h" are compile-time optional via the
5904 "--enable-kdbus" switch, and they are not compiled in by
5905 default. To make use of kdbus, you have to explicitly enable
5906 the switch. Note however, that neither the kernel nor the
5907 userspace API for all of this is considered stable yet. We
5908 want to maintain the freedom to still change the APIs for
5909 now. By specifying this build-time switch, you acknowledge
5910 that you are aware of the instability of the current
5913 * Also, note that while kdbus is pretty much complete,
5914 it lacks one thing: proper policy support. This means you
5915 can build a fully working system with all features; however,
5916 it will be highly insecure. Policy support will be added in
5917 one of the next releases, at the same time that we will
5918 declare the APIs stable.
5920 * When the kernel command line argument "kdbus" is specified,
5921 systemd will automatically load the kdbus.ko kernel module. At
5922 this stage of development, it is only useful for testing kdbus
5923 and should not be used in production. Note: if "--enable-kdbus"
5924 is specified, and the kdbus.ko kernel module is available, and
5925 "kdbus" is added to the kernel command line, the entire system
5926 runs with kdbus instead of dbus-daemon, with the above mentioned
5927 problem of missing the system policy enforcement. Also a future
5928 version of kdbus.ko or a newer systemd will not be compatible with
5929 each other, and will unlikely be able to boot the machine if only
5930 one of them is updated.
5932 * systemctl gained a new "import-environment" command which
5933 uploads the caller's environment (or parts thereof) into the
5934 service manager so that it is inherited by services started
5935 by the manager. This is useful to upload variables like
5936 $DISPLAY into the user service manager.
5938 * A new PrivateDevices= switch has been added to service units
5939 which allows running a service with a namespaced /dev
5940 directory that does not contain any device nodes for
5941 physical devices. More specifically, it only includes devices
5942 such as /dev/null, /dev/urandom, and /dev/zero which are API
5945 * logind has been extended to support behaviour like VT
5946 switching on seats that do not support a VT. This makes
5947 multi-session available on seats that are not the first seat
5948 (seat0), and on systems where kernel support for VTs has
5949 been disabled at compile-time.
5951 * If a process holds a delay lock for system sleep or shutdown
5952 and fails to release it in time, we will now log its
5953 identity. This makes it easier to identify processes that
5954 cause slow suspends or power-offs.
5956 * When parsing /etc/crypttab, support for a new key-slot=
5957 option as supported by Debian is added. It allows indicating
5958 which LUKS slot to use on disk, speeding up key loading.
5960 * The sd_journal_sendv() API call has been checked and
5961 officially declared to be async-signal-safe so that it may
5962 be invoked from signal handlers for logging purposes.
5964 * Boot-time status output is now enabled automatically after a
5965 short timeout if boot does not progress, in order to give
5966 the user an indication what she or he is waiting for.
5968 * The boot-time output has been improved to show how much time
5969 remains until jobs expire.
5971 * The KillMode= switch in service units gained a new possible
5972 value "mixed". If set, and the unit is shut down, then the
5973 initial SIGTERM signal is sent only to the main daemon
5974 process, while the following SIGKILL signal is sent to
5975 all remaining processes of the service.
5977 * When a scope unit is registered, a new property "Controller"
5978 may be set. If set to a valid bus name, systemd will send a
5979 RequestStop() signal to this name when it would like to shut
5980 down the scope. This may be used to hook manager logic into
5981 the shutdown logic of scope units. Also, scope units may now
5982 be put in a special "abandoned" state, in which case the
5983 manager process which created them takes no further
5984 responsibilities for it.
5986 * When reading unit files, systemd will now verify
5987 the access mode of these files, and warn about certain
5988 suspicious combinations. This has been added to make it
5989 easier to track down packaging bugs where unit files are
5990 marked executable or world-writable.
5992 * systemd-nspawn gained a new "--setenv=" switch to set
5993 container-wide environment variables. The similar option in
5994 systemd-activate was renamed from "--environment=" to
5995 "--setenv=" for consistency.
5997 * systemd-nspawn has been updated to create a new kdbus domain
5998 for each container that is invoked, thus allowing each
5999 container to have its own set of system and user buses,
6000 independent of the host.
6002 * systemd-nspawn gained a new --drop-capability= switch to run
6003 the container with less capabilities than the default. Both
6004 --drop-capability= and --capability= now take the special
6005 string "all" for dropping or keeping all capabilities.
6007 * systemd-nspawn gained new switches for executing containers
6008 with specific SELinux labels set.
6010 * systemd-nspawn gained a new --quiet switch to not generate
6011 any additional output but the container's own console
6014 * systemd-nspawn gained a new --share-system switch to run a
6015 container without PID namespacing enabled.
6017 * systemd-nspawn gained a new --register= switch to control
6018 whether the container is registered with systemd-machined or
6019 not. This is useful for containers that do not run full
6020 OS images, but only specific apps.
6022 * systemd-nspawn gained a new --keep-unit which may be used
6023 when invoked as the only program from a service unit, and
6024 results in registration of the unit service itself in
6025 systemd-machined, instead of a newly opened scope unit.
6027 * systemd-nspawn gained a new --network-interface= switch for
6028 moving arbitrary interfaces to the container. The new
6029 --network-veth switch creates a virtual Ethernet connection
6030 between host and container. The new --network-bridge=
6031 switch then allows assigning the host side of this virtual
6032 Ethernet connection to a bridge device.
6034 * systemd-nspawn gained a new --personality= switch for
6035 setting the kernel personality for the container. This is
6036 useful when running a 32-bit container on a 64-bit host. A
6037 similar option Personality= is now also available for service
6040 * logind will now also track a "Desktop" identifier for each
6041 session which encodes the desktop environment of it. This is
6042 useful for desktop environments that want to identify
6043 multiple running sessions of itself easily.
6045 * A new SELinuxContext= setting for service units has been
6046 added that allows setting a specific SELinux execution
6047 context for a service.
6049 * Most systemd client tools will now honour $SYSTEMD_LESS for
6050 settings of the "less" pager. By default, these tools will
6051 override $LESS to allow certain operations to work, such as
6052 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
6053 influence this logic.
6055 * systemd's "seccomp" hook-up has been changed to make use of
6056 the libseccomp library instead of using its own
6057 implementation. This has benefits for portability among
6060 * For usage together with SystemCallFilter=, a new
6061 SystemCallErrorNumber= setting has been introduced that
6062 allows configuration of a system error number to be returned
6063 on filtered system calls, instead of immediately killing the
6064 process. Also, SystemCallArchitectures= has been added to
6065 limit access to system calls of a particular architecture
6066 (in order to turn off support for unused secondary
6067 architectures). There is also a global
6068 SystemCallArchitectures= setting in system.conf now to turn
6069 off support for non-native system calls system-wide.
6071 * systemd requires a kernel with a working name_to_handle_at(),
6072 please see the kernel config requirements in the README file.
6074 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
6075 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
6076 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
6077 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
6078 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
6079 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
6080 Elia Pinto, Florian Weimer, George McCollister, Goffredo
6081 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
6082 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
6083 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
6084 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
6085 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
6086 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
6087 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
6088 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
6089 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
6090 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
6091 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
6092 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
6093 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
6094 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
6095 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
6096 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
6097 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
6099 — Berlin, 2014-02-20
6103 * logind has gained support for facilitating privileged input
6104 and drm device access for unprivileged clients. This work is
6105 useful to allow Wayland display servers (and similar
6106 programs, such as kmscon) to run under the user's ID and
6107 access input and drm devices which are normally
6108 protected. When this is used (and the kernel is new enough)
6109 logind will "mute" IO on the file descriptors passed to
6110 Wayland as long as it is in the background and "unmute" it
6111 if it returns into the foreground. This allows secure
6112 session switching without allowing background sessions to
6113 eavesdrop on input and display data. This also introduces
6114 session switching support if VT support is turned off in the
6115 kernel, and on seats that are not seat0.
6117 * A new kernel command line option luks.options= is understood
6118 now which allows specifying LUKS options for usage for LUKS
6119 encrypted partitions specified with luks.uuid=.
6121 * tmpfiles.d(5) snippets may now use specifier expansion in
6122 path names. More specifically %m, %b, %H, %v, are now
6123 replaced by the local machine id, boot id, hostname, and
6124 kernel version number.
6126 * A new tmpfiles.d(5) command "m" has been introduced which
6127 may be used to change the owner/group/access mode of a file
6128 or directory if it exists, but do nothing if it does not.
6130 * This release removes high-level support for the
6131 MemorySoftLimit= cgroup setting. The underlying kernel
6132 cgroup attribute memory.soft_limit= is currently badly
6133 designed and likely to be removed from the kernel API in its
6134 current form, hence we should not expose it for now.
6136 * The memory.use_hierarchy cgroup attribute is now enabled for
6137 all cgroups systemd creates in the memory cgroup
6138 hierarchy. This option is likely to be come the built-in
6139 default in the kernel anyway, and the non-hierarchical mode
6140 never made much sense in the intrinsically hierarchical
6143 * A new field _SYSTEMD_SLICE= is logged along with all journal
6144 messages containing the slice a message was generated
6145 from. This is useful to allow easy per-customer filtering of
6146 logs among other things.
6148 * systemd-journald will no longer adjust the group of journal
6149 files it creates to the "systemd-journal" group. Instead we
6150 rely on the journal directory to be owned by the
6151 "systemd-journal" group, and its setgid bit set, so that the
6152 kernel file system layer will automatically enforce that
6153 journal files inherit this group assignment. The reason for
6154 this change is that we cannot allow NSS look-ups from
6155 journald which would be necessary to resolve
6156 "systemd-journal" to a numeric GID, because this might
6157 create deadlocks if NSS involves synchronous queries to
6158 other daemons (such as nscd, or sssd) which in turn are
6159 logging clients of journald and might block on it, which
6160 would then dead lock. A tmpfiles.d(5) snippet included in
6161 systemd will make sure the setgid bit and group are
6162 properly set on the journal directory if it exists on every
6163 boot. However, we recommend adjusting it manually after
6164 upgrades too (or from RPM scriptlets), so that the change is
6165 not delayed until next reboot.
6167 * Backlight and random seed files in /var/lib/ have moved into
6168 the /var/lib/systemd/ directory, in order to centralize all
6169 systemd generated files in one directory.
6171 * Boot time performance measurements (as displayed by
6172 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
6173 performance information if that's available to determine how
6174 much time BIOS and boot loader initialization required. With
6175 a sufficiently new BIOS you hence no longer need to boot
6176 with Gummiboot to get access to such information.
6178 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
6179 Cristian Rodríguez, Dave Reisner, David Herrmann, David
6180 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
6181 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
6182 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
6183 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
6184 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6186 — Berlin, 2013-10-02
6190 * The Restart= option for services now understands a new
6191 on-watchdog setting, which will restart the service
6192 automatically if the service stops sending out watchdog keep
6193 alive messages (as configured with WatchdogSec=).
6195 * The getty generator (which is responsible for bringing up a
6196 getty on configured serial consoles) will no longer only
6197 start a getty on the primary kernel console but on all
6198 others, too. This makes the order in which console= is
6199 specified on the kernel command line less important.
6201 * libsystemd-logind gained a new sd_session_get_vt() call to
6202 retrieve the VT number of a session.
6204 * If the option "tries=0" is set for an entry of /etc/crypttab
6205 its passphrase is queried indefinitely instead of any
6206 maximum number of tries.
6208 * If a service with a configure PID file terminates its PID
6209 file will now be removed automatically if it still exists
6210 afterwards. This should put an end to stale PID files.
6212 * systemd-run will now also take relative binary path names
6213 for execution and no longer insists on absolute paths.
6215 * InaccessibleDirectories= and ReadOnlyDirectories= now take
6216 paths that are optionally prefixed with "-" to indicate that
6217 it should not be considered a failure if they do not exist.
6219 * journalctl -o (and similar commands) now understands a new
6220 output mode "short-precise", it is similar to "short" but
6221 shows timestamps with usec accuracy.
6223 * The option "discard" (as known from Debian) is now
6224 synonymous to "allow-discards" in /etc/crypttab. In fact,
6225 "discard" is preferred now (since it is easier to remember
6228 * Some licensing clean-ups were made, so that more code is now
6229 LGPL-2.1 licensed than before.
6231 * A minimal tool to save/restore the display backlight
6232 brightness across reboots has been added. It will store the
6233 backlight setting as late as possible at shutdown, and
6234 restore it as early as possible during reboot.
6236 * A logic to automatically discover and enable home and swap
6237 partitions on GPT disks has been added. With this in place
6238 /etc/fstab becomes optional for many setups as systemd can
6239 discover certain partitions located on the root disk
6240 automatically. Home partitions are recognized under their
6241 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
6242 partitions are recognized under their GPT type ID
6243 0657fd6da4ab43c484e50933c84b4f4f.
6245 * systemd will no longer pass any environment from the kernel
6246 or initrd to system services. If you want to set an
6247 environment for all services, do so via the kernel command
6248 line systemd.setenv= assignment.
6250 * The systemd-sysctl tool no longer natively reads the file
6251 /etc/sysctl.conf. If desired, the file should be symlinked
6252 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
6253 legacy support by a symlink rather than built-in code, it
6254 also makes the otherwise hidden order of application of the
6255 different files visible. (Note that this partly reverts to a
6256 pre-198 application order of sysctl knobs!)
6258 * The "systemctl set-log-level" and "systemctl dump" commands
6259 have been moved to systemd-analyze.
6261 * systemd-run learned the new --remain-after-exit switch,
6262 which causes the scope unit not to be cleaned up
6263 automatically after the process terminated.
6265 * tmpfiles learned a new --exclude-prefix= switch to exclude
6266 certain paths from operation.
6268 * journald will now automatically flush all messages to disk
6269 as soon as a message at the log level CRIT, ALERT or EMERG
6272 Contributions from: Andrew Cook, Brandon Philips, Christian
6273 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
6274 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
6275 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
6276 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
6277 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
6278 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
6279 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
6280 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
6281 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
6282 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
6283 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
6284 William Giokas, Zbigniew Jędrzejewski-Szmek
6286 — Berlin, 2013-09-13
6290 * The documentation has been updated to cover the various new
6291 concepts introduced with 205.
6293 * Unit files now understand the new %v specifier which
6294 resolves to the kernel version string as returned by "uname
6297 * systemctl now supports filtering the unit list output by
6298 load state, active state and sub state, using the new
6301 * "systemctl status" will now show the results of the
6302 condition checks (like ConditionPathExists= and similar) of
6303 the last start attempts of the unit. They are also logged to
6306 * "journalctl -b" may now be used to look for boot output of a
6307 specific boot. Try "journalctl -b -1" for the previous boot,
6308 but the syntax is substantially more powerful.
6310 * "journalctl --show-cursor" has been added which prints the
6311 cursor string the last shown log line. This may then be used
6312 with the new "journalctl --after-cursor=" switch to continue
6313 browsing logs from that point on.
6315 * "journalctl --force" may now be used to force regeneration
6318 * Creation of "dead" device nodes has been moved from udev
6319 into kmod and tmpfiles. Previously, udev would read the kmod
6320 databases to pre-generate dead device nodes based on meta
6321 information contained in kernel modules, so that these would
6322 be auto-loaded on access rather then at boot. As this
6323 does not really have much to do with the exposing actual
6324 kernel devices to userspace this has always been slightly
6325 alien in the udev codebase. Following the new scheme kmod
6326 will now generate a runtime snippet for tmpfiles from the
6327 module meta information and it now is tmpfiles' job to the
6328 create the nodes. This also allows overriding access and
6329 other parameters for the nodes using the usual tmpfiles
6330 facilities. As side effect this allows us to remove the
6331 CAP_SYS_MKNOD capability bit from udevd entirely.
6333 * logind's device ACLs may now be applied to these "dead"
6334 devices nodes too, thus finally allowing managed access to
6335 devices such as /dev/snd/sequencer without loading the
6336 backing module right-away.
6338 * A new RPM macro has been added that may be used to apply
6339 tmpfiles configuration during package installation.
6341 * systemd-detect-virt and ConditionVirtualization= now can
6342 detect User-Mode-Linux machines (UML).
6344 * journald will now implicitly log the effective capabilities
6345 set of processes in the message metadata.
6347 * systemd-cryptsetup has gained support for TrueCrypt volumes.
6349 * The initrd interface has been simplified (more specifically,
6350 support for passing performance data via environment
6351 variables and fsck results via files in /run has been
6352 removed). These features were non-essential, and are
6353 nowadays available in a much nicer way by having systemd in
6354 the initrd serialize its state and have the hosts systemd
6355 deserialize it again.
6357 * The udev "keymap" data files and tools to apply keyboard
6358 specific mappings of scan to key codes, and force-release
6359 scan code lists have been entirely replaced by a udev
6360 "keyboard" builtin and a hwdb data file.
6362 * systemd will now honour the kernel's "quiet" command line
6363 argument also during late shutdown, resulting in a
6364 completely silent shutdown when used.
6366 * There's now an option to control the SO_REUSEPORT socket
6367 option in .socket units.
6369 * Instance units will now automatically get a per-template
6370 subslice of system.slice unless something else is explicitly
6371 configured. For example, instances of sshd@.service will now
6372 implicitly be placed in system-sshd.slice rather than
6373 system.slice as before.
6375 * Test coverage support may now be enabled at build time.
6377 Contributions from: Dave Reisner, Frederic Crozat, Harald
6378 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
6379 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
6380 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
6381 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
6382 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
6383 Giokas, Zbigniew Jędrzejewski-Szmek
6385 — Berlin, 2013-07-23
6389 * Two new unit types have been introduced:
6391 Scope units are very similar to service units, however, are
6392 created out of pre-existing processes — instead of PID 1
6393 forking off the processes. By using scope units it is
6394 possible for system services and applications to group their
6395 own child processes (worker processes) in a powerful way
6396 which then maybe used to organize them, or kill them
6397 together, or apply resource limits on them.
6399 Slice units may be used to partition system resources in an
6400 hierarchical fashion and then assign other units to them. By
6401 default there are now three slices: system.slice (for all
6402 system services), user.slice (for all user sessions),
6403 machine.slice (for VMs and containers).
6405 Slices and scopes have been introduced primarily in
6406 context of the work to move cgroup handling to a
6407 single-writer scheme, where only PID 1
6408 creates/removes/manages cgroups.
6410 * There's a new concept of "transient" units. In contrast to
6411 normal units these units are created via an API at runtime,
6412 not from configuration from disk. More specifically this
6413 means it is now possible to run arbitrary programs as
6414 independent services, with all execution parameters passed
6415 in via bus APIs rather than read from disk. Transient units
6416 make systemd substantially more dynamic then it ever was,
6417 and useful as a general batch manager.
6419 * logind has been updated to make use of scope and slice units
6420 for managing user sessions. As a user logs in he will get
6421 his own private slice unit, to which all sessions are added
6422 as scope units. We also added support for automatically
6423 adding an instance of user@.service for the user into the
6424 slice. Effectively logind will no longer create cgroup
6425 hierarchies on its own now, it will defer entirely to PID 1
6426 for this by means of scope, service and slice units. Since
6427 user sessions this way become entities managed by PID 1
6428 the output of "systemctl" is now a lot more comprehensive.
6430 * A new mini-daemon "systemd-machined" has been added which
6431 may be used by virtualization managers to register local
6432 VMs/containers. nspawn has been updated accordingly, and
6433 libvirt will be updated shortly. machined will collect a bit
6434 of meta information about the VMs/containers, and assign
6435 them their own scope unit (see above). The collected
6436 meta-data is then made available via the "machinectl" tool,
6437 and exposed in "ps" and similar tools. machined/machinectl
6438 is compile-time optional.
6440 * As discussed earlier, the low-level cgroup configuration
6441 options ControlGroup=, ControlGroupModify=,
6442 ControlGroupPersistent=, ControlGroupAttribute= have been
6443 removed. Please use high-level attribute settings instead as
6444 well as slice units.
6446 * A new bus call SetUnitProperties() has been added to alter
6447 various runtime parameters of a unit. This is primarily
6448 useful to alter cgroup parameters dynamically in a nice way,
6449 but will be extended later on to make more properties
6450 modifiable at runtime. systemctl gained a new set-properties
6451 command that wraps this call.
6453 * A new tool "systemd-run" has been added which can be used to
6454 run arbitrary command lines as transient services or scopes,
6455 while configuring a number of settings via the command
6456 line. This tool is currently very basic, however already
6457 very useful. We plan to extend this tool to even allow
6458 queuing of execution jobs with time triggers from the
6459 command line, similar in fashion to "at".
6461 * nspawn will now inform the user explicitly that kernels with
6462 audit enabled break containers, and suggest the user to turn
6465 * Support for detecting the IMA and AppArmor security
6466 frameworks with ConditionSecurity= has been added.
6468 * journalctl gained a new "-k" switch for showing only kernel
6469 messages, mimicking dmesg output; in addition to "--user"
6470 and "--system" switches for showing only user's own logs
6473 * systemd-delta can now show information about drop-in
6474 snippets extending unit files.
6476 * libsystemd-bus has been substantially updated but is still
6477 not available as public API.
6479 * systemd will now look for the "debug" argument on the kernel
6480 command line and enable debug logging, similar to what
6481 "systemd.log_level=debug" already did before.
6483 * "systemctl set-default", "systemctl get-default" has been
6484 added to configure the default.target symlink, which
6485 controls what to boot into by default.
6487 * "systemctl set-log-level" has been added as a convenient
6488 way to raise and lower systemd logging threshold.
6490 * "systemd-analyze plot" will now show the time the various
6491 generators needed for execution, as well as information
6492 about the unit file loading.
6494 * libsystemd-journal gained a new sd_journal_open_files() call
6495 for opening specific journal files. journactl also gained a
6496 new switch to expose this new functionality. Previously we
6497 only supported opening all files from a directory, or all
6498 files from the system, as opening individual files only is
6499 racy due to journal file rotation.
6501 * systemd gained the new DefaultEnvironment= setting in
6502 /etc/systemd/system.conf to set environment variables for
6505 * If a privileged process logs a journal message with the
6506 OBJECT_PID= field set, then journald will automatically
6507 augment this with additional OBJECT_UID=, OBJECT_GID=,
6508 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
6509 system services want to log events about specific client
6510 processes. journactl/systemctl has been updated to make use
6511 of this information if all log messages regarding a specific
6514 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
6515 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
6516 Reisner, David Coppa, David King, David Strauss, Eelco
6517 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
6518 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
6519 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
6520 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
6521 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
6522 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
6523 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
6524 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
6525 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
6526 Łukasz Stelmach, 장동준
6530 * The Python bindings gained some minimal support for the APIs
6531 exposed by libsystemd-logind.
6533 * ConditionSecurity= gained support for detecting SMACK. Since
6534 this condition already supports SELinux and AppArmor we only
6535 miss IMA for this. Patches welcome!
6537 Contributions from: Karol Lewandowski, Lennart Poettering,
6538 Zbigniew Jędrzejewski-Szmek
6542 * systemd-nspawn will now create /etc/resolv.conf if
6543 necessary, before bind-mounting the host's file onto it.
6545 * systemd-nspawn will now store meta information about a
6546 container on the container's cgroup as extended attribute
6547 fields, including the root directory.
6549 * The cgroup hierarchy has been reworked in many ways. All
6550 objects any of the components systemd creates in the cgroup
6551 tree are now suffixed. More specifically, user sessions are
6552 now placed in cgroups suffixed with ".session", users in
6553 cgroups suffixed with ".user", and nspawn containers in
6554 cgroups suffixed with ".nspawn". Furthermore, all cgroup
6555 names are now escaped in a simple scheme to avoid collision
6556 of userspace object names with kernel filenames. This work
6557 is preparation for making these objects relocatable in the
6558 cgroup tree, in order to allow easy resource partitioning of
6559 these objects without causing naming conflicts.
6561 * systemctl list-dependencies gained the new switches
6562 --plain, --reverse, --after and --before.
6564 * systemd-inhibit now shows the process name of processes that
6565 have taken an inhibitor lock.
6567 * nss-myhostname will now also resolve "localhost"
6568 implicitly. This makes /etc/hosts an optional file and
6569 nicely handles that on IPv6 ::1 maps to both "localhost" and
6572 * libsystemd-logind.so gained a new call
6573 sd_get_machine_names() to enumerate running containers and
6574 VMs (currently only supported by very new libvirt and
6575 nspawn). sd_login_monitor can now be used to watch
6576 VMs/containers coming and going.
6578 * .include is not allowed recursively anymore, and only in
6579 unit files. Usually it is better to use drop-in snippets in
6580 .d/*.conf anyway, as introduced with systemd 198.
6582 * systemd-analyze gained a new "critical-chain" command that
6583 determines the slowest chain of units run during system
6584 boot-up. It is very useful for tracking down where
6585 optimizing boot time is the most beneficial.
6587 * systemd will no longer allow manipulating service paths in
6588 the name=systemd:/system cgroup tree using ControlGroup= in
6589 units. (But is still fine with it in all other dirs.)
6591 * There's a new systemd-nspawn@.service service file that may
6592 be used to easily run nspawn containers as system
6593 services. With the container's root directory in
6594 /var/lib/container/foobar it is now sufficient to run
6595 "systemctl start systemd-nspawn@foobar.service" to boot it.
6597 * systemd-cgls gained a new parameter "--machine" to list only
6598 the processes within a certain container.
6600 * ConditionSecurity= now can check for "apparmor". We still
6601 are lacking checks for SMACK and IMA for this condition
6602 check though. Patches welcome!
6604 * A new configuration file /etc/systemd/sleep.conf has been
6605 added that may be used to configure which kernel operation
6606 systemd is supposed to execute when "suspend", "hibernate"
6607 or "hybrid-sleep" is requested. This makes the new kernel
6608 "freeze" state accessible to the user.
6610 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
6611 the passed argument if applicable.
6613 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
6614 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
6615 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
6616 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
6617 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
6618 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
6619 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
6624 * The output of 'systemctl list-jobs' got some polishing. The
6625 '--type=' argument may now be passed more than once. A new
6626 command 'systemctl list-sockets' has been added which shows
6627 a list of kernel sockets systemd is listening on with the
6628 socket units they belong to, plus the units these socket
6631 * The experimental libsystemd-bus library got substantial
6632 updates to work in conjunction with the (also experimental)
6633 kdbus kernel project. It works well enough to exchange
6634 messages with some sophistication. Note that kdbus is not
6635 ready yet, and the library is mostly an elaborate test case
6636 for now, and not installable.
6638 * systemd gained a new unit 'systemd-static-nodes.service'
6639 that generates static device nodes earlier during boot, and
6640 can run in conjunction with udev.
6642 * libsystemd-login gained a new call sd_pid_get_user_unit()
6643 to retrieve the user systemd unit a process is running
6644 in. This is useful for systems where systemd is used as
6647 * systemd-nspawn now places all containers in the new /machine
6648 top-level cgroup directory in the name=systemd
6649 hierarchy. libvirt will soon do the same, so that we get a
6650 uniform separation of /system, /user and /machine for system
6651 services, user processes and containers/virtual
6652 machines. This new cgroup hierarchy is also useful to stick
6653 stable names to specific container instances, which can be
6654 recognized later this way (this name may be controlled
6655 via systemd-nspawn's new -M switch). libsystemd-login also
6656 gained a new call sd_pid_get_machine_name() to retrieve the
6657 name of the container/VM a specific process belongs to.
6659 * bootchart can now store its data in the journal.
6661 * libsystemd-journal gained a new call
6662 sd_journal_add_conjunction() for AND expressions to the
6663 matching logic. This can be used to express more complex
6664 logical expressions.
6666 * journactl can now take multiple --unit= and --user-unit=
6669 * The cryptsetup logic now understands the "luks.key=" kernel
6670 command line switch for specifying a file to read the
6671 decryption key from. Also, if a configured key file is not
6672 found the tool will now automatically fall back to prompting
6675 * Python systemd.journal module was updated to wrap recently
6676 added functions from libsystemd-journal. The interface was
6677 changed to bring the low level interface in s.j._Reader
6678 closer to the C API, and the high level interface in
6679 s.j.Reader was updated to wrap and convert all data about
6682 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
6683 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
6684 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
6685 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
6686 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
6687 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6691 * journalctl --update-catalog now understands a new --root=
6692 option to operate on catalogs found in a different root
6695 * During shutdown after systemd has terminated all running
6696 services a final killing loop kills all remaining left-over
6697 processes. We will now print the name of these processes
6698 when we send SIGKILL to them, since this usually indicates a
6701 * If /etc/crypttab refers to password files stored on
6702 configured mount points automatic dependencies will now be
6703 generated to ensure the specific mount is established first
6704 before the key file is attempted to be read.
6706 * 'systemctl status' will now show information about the
6707 network sockets a socket unit is listening on.
6709 * 'systemctl status' will also shown information about any
6710 drop-in configuration file for units. (Drop-In configuration
6711 files in this context are files such as
6712 /etc/systemd/systemd/foobar.service.d/*.conf)
6714 * systemd-cgtop now optionally shows summed up CPU times of
6715 cgroups. Press '%' while running cgtop to switch between
6716 percentage and absolute mode. This is useful to determine
6717 which cgroups use up the most CPU time over the entire
6718 runtime of the system. systemd-cgtop has also been updated
6719 to be 'pipeable' for processing with further shell tools.
6721 * 'hostnamectl set-hostname' will now allow setting of FQDN
6724 * The formatting and parsing of time span values has been
6725 changed. The parser now understands fractional expressions
6726 such as "5.5h". The formatter will now output fractional
6727 expressions for all time spans under 1min, i.e. "5.123456s"
6728 rather than "5s 123ms 456us". For time spans under 1s
6729 millisecond values are shown, for those under 1ms
6730 microsecond values are shown. This should greatly improve
6731 all time-related output of systemd.
6733 * libsystemd-login and libsystemd-journal gained new
6734 functions for querying the poll() events mask and poll()
6735 timeout value for integration into arbitrary event
6738 * localectl gained the ability to list available X11 keymaps
6739 (models, layouts, variants, options).
6741 * 'systemd-analyze dot' gained the ability to filter for
6742 specific units via shell-style globs, to create smaller,
6743 more useful graphs. I.e. it is now possible to create simple
6744 graphs of all the dependencies between only target units, or
6745 of all units that Avahi has dependencies with.
6747 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
6748 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
6749 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
6750 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
6751 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
6752 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
6753 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
6757 * The boot-time readahead implementation for rotating media
6758 will now read the read-ahead data in multiple passes which
6759 consist of all read requests made in equidistant time
6760 intervals. This means instead of strictly reading read-ahead
6761 data in its physical order on disk we now try to find a
6762 middle ground between physical and access time order.
6764 * /etc/os-release files gained a new BUILD_ID= field for usage
6765 on operating systems that provide continuous builds of OS
6768 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
6769 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
6770 William Douglas, Zbigniew Jędrzejewski-Szmek
6774 * systemd-python gained an API exposing libsystemd-daemon.
6776 * The SMACK setup logic gained support for uploading CIPSO
6779 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
6780 ReadOnlyDirectories= and InaccessibleDirectories= has
6781 changed. The private /tmp and /var/tmp directories are now
6782 shared by all processes of a service (which means
6783 ExecStartPre= may now leave data in /tmp that ExecStart= of
6784 the same service can still access). When a service is
6785 stopped its temporary directories are immediately deleted
6786 (normal clean-up with tmpfiles is still done in addition to
6789 * By default, systemd will now set a couple of sysctl
6790 variables in the kernel: the safe sysrq options are turned
6791 on, IP route verification is turned on, and source routing
6792 disabled. The recently added hardlink and softlink
6793 protection of the kernel is turned on. These settings should
6794 be reasonably safe, and good defaults for all new systems.
6796 * The predictable network naming logic may now be turned off
6797 with a new kernel command line switch: net.ifnames=0.
6799 * A new libsystemd-bus module has been added that implements a
6800 pretty complete D-Bus client library. For details see:
6802 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
6804 * journald will now explicitly flush the journal files to disk
6805 at the latest 5min after each write. The file will then also
6806 be marked offline until the next write. This should increase
6807 reliability in case of a crash. The synchronization delay
6808 can be configured via SyncIntervalSec= in journald.conf.
6810 * There's a new remote-fs-setup.target unit that can be used
6811 to pull in specific services when at least one remote file
6812 system is to be mounted.
6814 * There are new targets timers.target and paths.target as
6815 canonical targets to pull user timer and path units in
6816 from. This complements sockets.target with a similar
6817 purpose for socket units.
6819 * libudev gained a new call udev_device_set_attribute_value()
6820 to set sysfs attributes of a device.
6822 * The udev daemon now sets the default number of worker
6823 processes executed in parallel based on the number of available
6824 CPUs instead of the amount of available RAM. This is supposed
6825 to provide a more reliable default and limit a too aggressive
6826 parallelism for setups with 1000s of devices connected.
6828 Contributions from: Auke Kok, Colin Walters, Cristian
6829 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
6830 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
6831 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
6832 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
6833 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
6834 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
6835 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
6836 Zbigniew Jędrzejewski-Szmek
6840 * Configuration of unit files may now be extended via drop-in
6841 files without having to edit/override the unit files
6842 themselves. More specifically, if the administrator wants to
6843 change one value for a service file foobar.service he can
6844 now do so by dropping in a configuration snippet into
6845 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
6846 will load all these snippets and apply them on top of the
6847 main unit configuration file, possibly extending or
6848 overriding its settings. Using these drop-in snippets is
6849 generally nicer than the two earlier options for changing
6850 unit files locally: copying the files from
6851 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
6852 them there; or creating a new file in /etc/systemd/system/
6853 that incorporates the original one via ".include". Drop-in
6854 snippets into these .d/ directories can be placed in any
6855 directory systemd looks for units in, and the usual
6856 overriding semantics between /usr/lib, /etc and /run apply
6859 * Most unit file settings which take lists of items can now be
6860 reset by assigning the empty string to them. For example,
6861 normally, settings such as Environment=FOO=BAR append a new
6862 environment variable assignment to the environment block,
6863 each time they are used. By assigning Environment= the empty
6864 string the environment block can be reset to empty. This is
6865 particularly useful with the .d/*.conf drop-in snippets
6866 mentioned above, since this adds the ability to reset list
6867 settings from vendor unit files via these drop-ins.
6869 * systemctl gained a new "list-dependencies" command for
6870 listing the dependencies of a unit recursively.
6872 * Inhibitors are now honored and listed by "systemctl
6873 suspend", "systemctl poweroff" (and similar) too, not only
6874 GNOME. These commands will also list active sessions by
6877 * Resource limits (as exposed by the various control group
6878 controllers) can now be controlled dynamically at runtime
6879 for all units. More specifically, you can now use a command
6880 like "systemctl set-cgroup-attr foobar.service cpu.shares
6881 2000" to alter the CPU shares a specific service gets. These
6882 settings are stored persistently on disk, and thus allow the
6883 administrator to easily adjust the resource usage of
6884 services with a few simple commands. This dynamic resource
6885 management logic is also available to other programs via the
6886 bus. Almost any kernel cgroup attribute and controller is
6889 * systemd-vconsole-setup will now copy all font settings to
6890 all allocated VTs, where it previously applied them only to
6893 * libsystemd-login gained the new sd_session_get_tty() API
6896 * This release drops support for a few legacy or
6897 distribution-specific LSB facility names when parsing init
6898 scripts: $x-display-manager, $mail-transfer-agent,
6899 $mail-transport-agent, $mail-transfer-agent, $smtp,
6900 $null. Also, the mail-transfer-agent.target unit backing
6901 this has been removed. Distributions which want to retain
6902 compatibility with this should carry the burden for
6903 supporting this themselves and patch support for these back
6904 in, if they really need to. Also, the facilities $syslog and
6905 $local_fs are now ignored, since systemd does not support
6906 early-boot LSB init scripts anymore, and these facilities
6907 are implied anyway for normal services. syslog.target has
6910 * There are new bus calls on PID1's Manager object for
6911 cancelling jobs, and removing snapshot units. Previously,
6912 both calls were only available on the Job and Snapshot
6915 * systemd-journal-gatewayd gained SSL support.
6917 * The various "environment" files, such as /etc/locale.conf
6918 now support continuation lines with a backslash ("\") as
6919 last character in the line, similarly in style (but different)
6920 to how this is supported in shells.
6922 * For normal user processes the _SYSTEMD_USER_UNIT= field is
6923 now implicitly appended to every log entry logged. systemctl
6924 has been updated to filter by this field when operating on a
6925 user systemd instance.
6927 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
6928 CAP_AUDIT_CONTROL capabilities to the capabilities set for
6929 the container. This makes it easier to boot unmodified
6930 Fedora systems in a container, which however still requires
6931 audit=0 to be passed on the kernel command line. Auditing in
6932 kernel and userspace is unfortunately still too broken in
6933 context of containers, hence we recommend compiling it out
6934 of the kernel or using audit=0. Hopefully this will be fixed
6935 one day for good in the kernel.
6937 * nspawn gained the new --bind= and --bind-ro= parameters to
6938 bind mount specific directories from the host into the
6941 * nspawn will now mount its own devpts file system instance
6942 into the container, in order not to leak pty devices from
6943 the host into the container.
6945 * systemd will now read the firmware boot time performance
6946 information from the EFI variables, if the used boot loader
6947 supports this, and takes it into account for boot performance
6948 analysis via "systemd-analyze". This is currently supported
6949 only in conjunction with Gummiboot, but could be supported
6950 by other boot loaders too. For details see:
6952 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
6954 * A new generator has been added that automatically mounts the
6955 EFI System Partition (ESP) to /boot, if that directory
6956 exists, is empty, and no other file system has been
6957 configured to be mounted there.
6959 * logind will now send out PrepareForSleep(false) out
6960 unconditionally, after coming back from suspend. This may be
6961 used by applications as asynchronous notification for
6962 system resume events.
6964 * "systemctl unlock-sessions" has been added, that allows
6965 unlocking the screens of all user sessions at once, similar
6966 to how "systemctl lock-sessions" already locked all users
6967 sessions. This is backed by a new D-Bus call UnlockSessions().
6969 * "loginctl seat-status" will now show the master device of a
6970 seat. (i.e. the device of a seat that needs to be around for
6971 the seat to be considered available, usually the graphics
6974 * tmpfiles gained a new "X" line type, that allows
6975 configuration of files and directories (with wildcards) that
6976 shall be excluded from automatic cleanup ("aging").
6978 * udev default rules set the device node permissions now only
6979 at "add" events, and do not change them any longer with a
6980 later "change" event.
6982 * The log messages for lid events and power/sleep keypresses
6983 now carry a message ID.
6985 * We now have a substantially larger unit test suite, but this
6986 continues to be work in progress.
6988 * udevadm hwdb gained a new --root= parameter to change the
6989 root directory to operate relative to.
6991 * logind will now issue a background sync() request to the kernel
6992 early at shutdown, so that dirty buffers are flushed to disk early
6993 instead of at the last moment, in order to optimize shutdown
6996 * A new bootctl tool has been added that is an interface for
6997 certain boot loader operations. This is currently a preview
6998 and is likely to be extended into a small mechanism daemon
6999 like timedated, localed, hostnamed, and can be used by
7000 graphical UIs to enumerate available boot options, and
7001 request boot into firmware operations.
7003 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
7004 the rest of the package. It also has been updated to work
7005 correctly in initrds.
7007 * Policykit previously has been runtime optional, and is now
7008 also compile time optional via a configure switch.
7010 * systemd-analyze has been reimplemented in C. Also "systemctl
7011 dot" has moved into systemd-analyze.
7013 * "systemctl status" with no further parameters will now print
7014 the status of all active or failed units.
7016 * Operations such as "systemctl start" can now be executed
7017 with a new mode "--irreversible" which may be used to queue
7018 operations that cannot accidentally be reversed by a later
7019 job queuing. This is by default used to make shutdown
7020 requests more robust.
7022 * The Python API of systemd now gained a new module for
7023 reading journal files.
7025 * A new tool kernel-install has been added that can install
7026 kernel images according to the Boot Loader Specification:
7028 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
7030 * Boot time console output has been improved to provide
7031 animated boot time output for hanging jobs.
7033 * A new tool systemd-activate has been added which can be used
7034 to test socket activation with, directly from the command
7035 line. This should make it much easier to test and debug
7036 socket activation in daemons.
7038 * journalctl gained a new "--reverse" (or -r) option to show
7039 journal output in reverse order (i.e. newest line first).
7041 * journalctl gained a new "--pager-end" (or -e) option to jump
7042 to immediately jump to the end of the journal in the
7043 pager. This is only supported in conjunction with "less".
7045 * journalctl gained a new "--user-unit=" option, that works
7046 similarly to "--unit=" but filters for user units rather than
7049 * A number of unit files to ease adoption of systemd in
7050 initrds has been added. This moves some minimal logic from
7051 the various initrd implementations into systemd proper.
7053 * The journal files are now owned by a new group
7054 "systemd-journal", which exists specifically to allow access
7055 to the journal, and nothing else. Previously, we used the
7056 "adm" group for that, which however possibly covers more
7057 than just journal/log file access. This new group is now
7058 already used by systemd-journal-gatewayd to ensure this
7059 daemon gets access to the journal files and as little else
7060 as possible. Note that "make install" will also set FS ACLs
7061 up for /var/log/journal to give "adm" and "wheel" read
7062 access to it, in addition to "systemd-journal" which owns
7063 the journal files. We recommend that packaging scripts also
7064 add read access to "adm" + "wheel" to /var/log/journal, and
7065 all existing/future journal files. To normal users and
7066 administrators little changes, however packagers need to
7067 ensure to create the "systemd-journal" system group at
7068 package installation time.
7070 * The systemd-journal-gatewayd now runs as unprivileged user
7071 systemd-journal-gateway:systemd-journal-gateway. Packaging
7072 scripts need to create these system user/group at
7075 * timedated now exposes a new boolean property CanNTP that
7076 indicates whether a local NTP service is available or not.
7078 * systemd-detect-virt will now also detect xen PVs
7080 * The pstore file system is now mounted by default, if it is
7083 * In addition to the SELinux and IMA policies we will now also
7084 load SMACK policies at early boot.
7086 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
7087 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
7088 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
7089 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
7090 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
7091 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
7092 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
7093 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
7094 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
7095 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
7096 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
7097 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
7098 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
7099 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
7103 * Timer units now support calendar time events in addition to
7104 monotonic time events. That means you can now trigger a unit
7105 based on a calendar time specification such as "Thu,Fri
7106 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
7107 or fifth day of any month of the year 2013, given that it is
7108 a thursday or friday. This brings timer event support
7109 considerably closer to cron's capabilities. For details on
7110 the supported calendar time specification language see
7113 * udev now supports a number of different naming policies for
7114 network interfaces for predictable names, and a combination
7115 of these policies is now the default. Please see this wiki
7116 document for details:
7118 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
7120 * Auke Kok's bootchart implementation has been added to the
7121 systemd tree. It is an optional component that can graph the
7122 boot in quite some detail. It is one of the best bootchart
7123 implementations around and minimal in its code and
7126 * nss-myhostname has been integrated into the systemd source
7127 tree. nss-myhostname guarantees that the local hostname
7128 always stays resolvable via NSS. It has been a weak
7129 requirement of systemd-hostnamed since a long time, and
7130 since its code is actually trivial we decided to just
7131 include it in systemd's source tree. It can be turned off
7132 with a configure switch.
7134 * The read-ahead logic is now capable of properly detecting
7135 whether a btrfs file system is on SSD or rotating media, in
7136 order to optimize the read-ahead scheme. Previously, it was
7137 only capable of detecting this on traditional file systems
7140 * In udev, additional device properties are now read from the
7141 IAB in addition to the OUI database. Also, Bluetooth company
7142 identities are attached to the devices as well.
7144 * In service files %U may be used as specifier that is
7145 replaced by the configured user name of the service.
7147 * nspawn may now be invoked without a controlling TTY. This
7148 makes it suitable for invocation as its own service. This
7149 may be used to set up a simple containerized server system
7150 using only core OS tools.
7152 * systemd and nspawn can now accept socket file descriptors
7153 when they are started for socket activation. This enables
7154 implementation of socket activated nspawn
7155 containers. i.e. think about autospawning an entire OS image
7156 when the first SSH or HTTP connection is received. We expect
7157 that similar functionality will also be added to libvirt-lxc
7160 * journalctl will now suppress ANSI color codes when
7161 presenting log data.
7163 * systemctl will no longer show control group information for
7164 a unit if the control group is empty anyway.
7166 * logind can now automatically suspend/hibernate/shutdown the
7169 * /etc/machine-info and hostnamed now also expose the chassis
7170 type of the system. This can be used to determine whether
7171 the local system is a laptop, desktop, handset or
7172 tablet. This information may either be configured by the
7173 user/vendor or is automatically determined from ACPI and DMI
7174 information if possible.
7176 * A number of PolicyKit actions are now bound together with
7177 "imply" rules. This should simplify creating UIs because
7178 many actions will now authenticate similar ones as well.
7180 * Unit files learnt a new condition ConditionACPower= which
7181 may be used to conditionalize a unit depending on whether an
7182 AC power source is connected or not, of whether the system
7183 is running on battery power.
7185 * systemctl gained a new "is-failed" verb that may be used in
7186 shell scripts and suchlike to check whether a specific unit
7187 is in the "failed" state.
7189 * The EnvironmentFile= setting in unit files now supports file
7190 globbing, and can hence be used to easily read a number of
7191 environment files at once.
7193 * systemd will no longer detect and recognize specific
7194 distributions. All distribution-specific #ifdeffery has been
7195 removed, systemd is now fully generic and
7196 distribution-agnostic. Effectively, not too much is lost as
7197 a lot of the code is still accessible via explicit configure
7198 switches. However, support for some distribution specific
7199 legacy configuration file formats has been dropped. We
7200 recommend distributions to simply adopt the configuration
7201 files everybody else uses now and convert the old
7202 configuration from packaging scripts. Most distributions
7203 already did that. If that's not possible or desirable,
7204 distributions are welcome to forward port the specific
7205 pieces of code locally from the git history.
7207 * When logging a message about a unit systemd will now always
7208 log the unit name in the message meta data.
7210 * localectl will now also discover system locale data that is
7211 not stored in locale archives, but directly unpacked.
7213 * logind will no longer unconditionally use framebuffer
7214 devices as seat masters, i.e. as devices that are required
7215 to be existing before a seat is considered preset. Instead,
7216 it will now look for all devices that are tagged as
7217 "seat-master" in udev. By default, framebuffer devices will
7218 be marked as such, but depending on local systems, other
7219 devices might be marked as well. This may be used to
7220 integrate graphics cards using closed source drivers (such
7221 as NVidia ones) more nicely into logind. Note however, that
7222 we recommend using the open source NVidia drivers instead,
7223 and no udev rules for the closed-source drivers will be
7224 shipped from us upstream.
7226 Contributions from: Adam Williamson, Alessandro Crismani, Auke
7227 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
7228 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
7229 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
7230 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
7231 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
7232 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
7233 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
7234 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
7235 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
7236 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
7237 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
7242 * udev gained support for loading additional device properties
7243 from an indexed database that is keyed by vendor/product IDs
7244 and similar device identifiers. For the beginning this
7245 "hwdb" is populated with data from the well-known PCI and
7246 USB database, but also includes PNP, ACPI and OID data. In
7247 the longer run this indexed database shall grow into
7248 becoming the one central database for non-essential
7249 userspace device metadata. Previously, data from the PCI/USB
7250 database was only attached to select devices, since the
7251 lookup was a relatively expensive operation due to O(n) time
7252 complexity (with n being the number of entries in the
7253 database). Since this is now O(1), we decided to add in this
7254 data for all devices where this is available, by
7255 default. Note that the indexed database needs to be rebuilt
7256 when new data files are installed. To achieve this you need
7257 to update your packaging scripts to invoke "udevadm hwdb
7258 --update" after installation of hwdb data files. For
7259 RPM-based distributions we introduced the new
7260 %udev_hwdb_update macro for this purpose.
7262 * The Journal gained support for the "Message Catalog", an
7263 indexed database to link up additional information with
7264 journal entries. For further details please check:
7266 https://www.freedesktop.org/wiki/Software/systemd/catalog
7268 The indexed message catalog database also needs to be
7269 rebuilt after installation of message catalog files. Use
7270 "journalctl --update-catalog" for this. For RPM-based
7271 distributions we introduced the %journal_catalog_update
7272 macro for this purpose.
7274 * The Python Journal bindings gained support for the standard
7275 Python logging framework.
7277 * The Journal API gained new functions for checking whether
7278 the underlying file system of a journal file is capable of
7279 properly reporting file change notifications, or whether
7280 applications that want to reflect journal changes "live"
7281 need to recheck journal files continuously in appropriate
7284 * It is now possible to set the "age" field for tmpfiles
7285 entries to 0, indicating that files matching this entry
7286 shall always be removed when the directories are cleaned up.
7288 * coredumpctl gained a new "gdb" verb which invokes gdb
7289 right-away on the selected coredump.
7291 * There's now support for "hybrid sleep" on kernels that
7292 support this, in addition to "suspend" and "hibernate". Use
7293 "systemctl hybrid-sleep" to make use of this.
7295 * logind's HandleSuspendKey= setting (and related settings)
7296 now gained support for a new "lock" setting to simply
7297 request the screen lock on all local sessions, instead of
7298 actually executing a suspend or hibernation.
7300 * systemd will now mount the EFI variables file system by
7303 * Socket units now gained support for configuration of the
7304 SMACK security label.
7306 * timedatectl will now output the time of the last and next
7307 daylight saving change.
7309 * We dropped support for various legacy and distro-specific
7310 concepts, such as insserv, early-boot SysV services
7311 (i.e. those for non-standard runlevels such as 'b' or 'S')
7312 or ArchLinux /etc/rc.conf support. We recommend the
7313 distributions who still need support this to either continue
7314 to maintain the necessary patches downstream, or find a
7315 different solution. (Talk to us if you have questions!)
7317 * Various systemd components will now bypass PolicyKit checks
7318 for root and otherwise handle properly if PolicyKit is not
7319 found to be around. This should fix most issues for
7320 PolicyKit-less systems. Quite frankly this should have been
7321 this way since day one. It is absolutely our intention to
7322 make systemd work fine on PolicyKit-less systems, and we
7323 consider it a bug if something does not work as it should if
7324 PolicyKit is not around.
7326 * For embedded systems it is now possible to build udev and
7327 systemd without blkid and/or kmod support.
7329 * "systemctl switch-root" is now capable of switching root
7330 more than once. I.e. in addition to transitions from the
7331 initrd to the host OS it is now possible to transition to
7332 further OS images from the host. This is useful to implement
7333 offline updating tools.
7335 * Various other additions have been made to the RPM macros
7336 shipped with systemd. Use %udev_rules_update() after
7337 installing new udev rules files. %_udevhwdbdir,
7338 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
7339 %_sysctldir are now available which resolve to the right
7340 directories for packages to place various data files in.
7342 * journalctl gained the new --full switch (in addition to
7343 --all, to disable ellipsation for long messages.
7345 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
7346 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
7347 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
7348 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
7349 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
7350 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
7351 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
7352 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
7353 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
7357 * journalctl gained new --since= and --until= switches to
7358 filter by time. It also now supports nice filtering for
7359 units via --unit=/-u.
7361 * Type=oneshot services may use ExecReload= and do the
7364 * The journal daemon now supports time-based rotation and
7365 vacuuming, in addition to the usual disk-space based
7368 * The journal will now index the available field values for
7369 each field name. This enables clients to show pretty drop
7370 downs of available match values when filtering. The bash
7371 completion of journalctl has been updated
7372 accordingly. journalctl gained a new switch -F to list all
7373 values a certain field takes in the journal database.
7375 * More service events are now written as structured messages
7376 to the journal, and made recognizable via message IDs.
7378 * The timedated, localed and hostnamed mini-services which
7379 previously only provided support for changing time, locale
7380 and hostname settings from graphical DEs such as GNOME now
7381 also have a minimal (but very useful) text-based client
7382 utility each. This is probably the nicest way to changing
7383 these settings from the command line now, especially since
7384 it lists available options and is fully integrated with bash
7387 * There's now a new tool "systemd-coredumpctl" to list and
7388 extract coredumps from the journal.
7390 * We now install a README each in /var/log/ and
7391 /etc/rc.d/init.d explaining where the system logs and init
7392 scripts went. This hopefully should help folks who go to
7393 that dirs and look into the otherwise now empty void and
7394 scratch their heads.
7396 * When user-services are invoked (by systemd --user) the
7397 $MANAGERPID env var is set to the PID of systemd.
7399 * SIGRTMIN+24 when sent to a --user instance will now result
7400 in immediate termination of systemd.
7402 * gatewayd received numerous feature additions such as a
7403 "follow" mode, for live syncing and filtering.
7405 * browse.html now allows filtering and showing detailed
7406 information on specific entries. Keyboard navigation and
7407 mouse screen support has been added.
7409 * gatewayd/journalctl now supports HTML5/JSON
7410 Server-Sent-Events as output.
7412 * The SysV init script compatibility logic will now
7413 heuristically determine whether a script supports the
7414 "reload" verb, and only then make this available as
7417 * "systemctl status --follow" has been removed, use "journalctl
7420 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
7421 have been removed since they are hardly useful to be
7424 * And I'd like to take the opportunity to specifically mention
7425 Zbigniew for his great contributions. Zbigniew, you rock!
7427 Contributions from: Andrew Eikum, Christian Hesse, Colin
7428 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
7429 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
7430 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
7431 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
7432 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
7433 Jędrzejewski-Szmek, Сковорода Никита Андреевич
7437 * If /etc/vconsole.conf is non-existent or empty we will no
7438 longer load any console font or key map at boot by
7439 default. Instead the kernel defaults will be left
7440 intact. This is definitely the right thing to do, as no
7441 configuration should mean no configuration, and hard-coding
7442 font names that are different on all archs is probably a bad
7443 idea. Also, the kernel default key map and font should be
7444 good enough for most cases anyway, and mostly identical to
7445 the userspace fonts/key maps we previously overloaded them
7446 with. If distributions want to continue to default to a
7447 non-kernel font or key map they should ship a default
7448 /etc/vconsole.conf with the appropriate contents.
7450 Contributions from: Colin Walters, Daniel J Walsh, Dave
7451 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
7452 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
7456 * journalctl gained a new --cursor= switch to show entries
7457 starting from the specified location in the journal.
7459 * We now enforce a size limit on journal entry fields exported
7460 with "-o json" in journalctl. Fields larger than 4K will be
7461 assigned null. This can be turned off with --all.
7463 * An (optional) journal gateway daemon is now available as
7464 "systemd-journal-gatewayd.service". This service provides
7465 access to the journal via HTTP and JSON. This functionality
7466 will be used to implement live log synchronization in both
7467 pull and push modes, but has various other users too, such
7468 as easy log access for debugging of embedded devices. Right
7469 now it is already useful to retrieve the journal via HTTP:
7471 # systemctl start systemd-journal-gatewayd.service
7472 # wget http://localhost:19531/entries
7474 This will download the journal contents in a
7475 /var/log/messages compatible format. The same as JSON:
7477 # curl -H"Accept: application/json" http://localhost:19531/entries
7479 This service is also accessible via a web browser where a
7480 single static HTML5 app is served that uses the JSON logic
7481 to enable the user to do some basic browsing of the
7482 journal. This will be extended later on. Here's an example
7483 screenshot of this app in its current state:
7485 http://0pointer.de/public/journal-gatewayd
7487 Contributions from: Kay Sievers, Lennart Poettering, Robert
7488 Milasan, Tom Gundersen
7492 * The bash completion logic is now available for journalctl
7495 * We do not mount the "cpuset" controller anymore together with
7496 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
7497 started if no parameters are assigned to it. "cpuset" hence
7498 broke code that assumed it could create "cpu" groups and
7501 * journalctl -f will now subscribe to terminal size changes,
7502 and line break accordingly.
7504 Contributions from: Dave Reisner, Kay Sievers, Lennart
7505 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
7509 * nspawn will now create a symlink /etc/localtime in the
7510 container environment, copying the host's timezone
7511 setting. Previously this has been done via a bind mount, but
7512 since symlinks cannot be bind mounted this has now been
7513 changed to create/update the appropriate symlink.
7515 * journalctl -n's line number argument is now optional, and
7516 will default to 10 if omitted.
7518 * journald will now log the maximum size the journal files may
7519 take up on disk. This is particularly useful if the default
7520 built-in logic of determining this parameter from the file
7521 system size is used. Use "systemctl status
7522 systemd-journald.service" to see this information.
7524 * The multi-seat X wrapper tool has been stripped down. As X
7525 is now capable of enumerating graphics devices via udev in a
7526 seat-aware way the wrapper is not strictly necessary
7527 anymore. A stripped down temporary stop-gap is still shipped
7528 until the upstream display managers have been updated to
7529 fully support the new X logic. Expect this wrapper to be
7530 removed entirely in one of the next releases.
7532 * HandleSleepKey= in logind.conf has been split up into
7533 HandleSuspendKey= and HandleHibernateKey=. The old setting
7534 is not available anymore. X11 and the kernel are
7535 distinguishing between these keys and we should too. This
7536 also means the inhibition lock for these keys has been split
7539 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
7540 Poettering, Lukas Nykryn, Václav Pavlín
7544 * Whenever a unit changes state we will now log this to the
7545 journal and show along the unit's own log output in
7548 * ConditionPathIsMountPoint= can now properly detect bind
7549 mount points too. (Previously, a bind mount of one file
7550 system to another place in the same file system could not be
7551 detected as mount, since they shared struct stat's st_dev
7554 * We will now mount the cgroup controllers cpu, cpuacct,
7555 cpuset and the controllers net_cls, net_prio together by
7558 * nspawn containers will now have a virtualized boot
7559 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
7560 over with a randomized ID at container initialization). This
7561 has the effect of making "journalctl -b" do the right thing
7564 * The JSON output journal serialization has been updated not
7565 to generate "endless" list objects anymore, but rather one
7566 JSON object per line. This is more in line how most JSON
7567 parsers expect JSON objects. The new output mode
7568 "json-pretty" has been added to provide similar output, but
7569 neatly aligned for readability by humans.
7571 * We dropped all explicit sync() invocations in the shutdown
7572 code. The kernel does this implicitly anyway in the kernel
7573 reboot() syscall. halt(8)'s -n option is now a compatibility
7576 * We now support virtualized reboot() in containers, as
7577 supported by newer kernels. We will fall back to exit() if
7578 CAP_SYS_REBOOT is not available to the container. Also,
7579 nspawn makes use of this now and will actually reboot the
7580 container if the containerized OS asks for that.
7582 * journalctl will only show local log output by default
7583 now. Use --merge (-m) to show remote log output, too.
7585 * libsystemd-journal gained the new sd_journal_get_usage()
7586 call to determine the current disk usage of all journal
7587 files. This is exposed in the new "journalctl --disk-usage"
7590 * journald gained a new configuration setting SplitMode= in
7591 journald.conf which may be used to control how user journals
7592 are split off. See journald.conf(5) for details.
7594 * A new condition type ConditionFileNotEmpty= has been added.
7596 * tmpfiles' "w" lines now support file globbing, to write
7597 multiple files at once.
7599 * We added Python bindings for the journal submission
7600 APIs. More Python APIs for a number of selected APIs will
7601 likely follow. Note that we intend to add native bindings
7602 only for the Python language, as we consider it common
7603 enough to deserve bindings shipped within systemd. There are
7604 various projects outside of systemd that provide bindings
7605 for languages such as PHP or Lua.
7607 * Many conditions will now resolve specifiers such as %i. In
7608 addition, PathChanged= and related directives of .path units
7609 now support specifiers as well.
7611 * There's now a new RPM macro definition for the system preset
7614 * journald will now warn if it ca not forward a message to the
7615 syslog daemon because its socket is full.
7617 * timedated will no longer write or process /etc/timezone,
7618 except on Debian. As we do not support late mounted /usr
7619 anymore /etc/localtime always being a symlink is now safe,
7620 and hence the information in /etc/timezone is not necessary
7623 * logind will now always reserve one VT for a text getty (VT6
7624 by default). Previously if more than 6 X sessions where
7625 started they took up all the VTs with auto-spawned gettys,
7626 so that no text gettys were available anymore.
7628 * udev will now automatically inform the btrfs kernel logic
7629 about btrfs RAID components showing up. This should make
7630 simple hotplug based btrfs RAID assembly work.
7632 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
7633 (but not for its children which will stay at the kernel
7634 default). This should allow setups with a lot more listening
7637 * systemd will now always pass the configured timezone to the
7638 kernel at boot. timedated will do the same when the timezone
7641 * logind's inhibition logic has been updated. By default,
7642 logind will now handle the lid switch, the power and sleep
7643 keys all the time, even in graphical sessions. If DEs want
7644 to handle these events on their own they should take the new
7645 handle-power-key, handle-sleep-key and handle-lid-switch
7646 inhibitors during their runtime. A simple way to achieve
7647 that is to invoke the DE wrapped in an invocation of:
7649 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
7651 * Access to unit operations is now checked via SELinux taking
7652 the unit file label and client process label into account.
7654 * systemd will now notify the administrator in the journal
7655 when he over-mounts a non-empty directory.
7657 * There are new specifiers that are resolved in unit files,
7658 for the host name (%H), the machine ID (%m) and the boot ID
7661 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
7662 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
7663 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
7664 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
7665 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
7666 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
7667 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
7671 * Support for reading structured kernel messages from
7672 /dev/kmsg has now been added and is enabled by default.
7674 * Support for reading kernel messages from /proc/kmsg has now
7675 been removed. If you want kernel messages in the journal
7676 make sure to run a recent kernel (>= 3.5) that supports
7677 reading structured messages from /dev/kmsg (see
7678 above). /proc/kmsg is now exclusive property of classic
7679 syslog daemons again.
7681 * The libudev API gained the new
7682 udev_device_new_from_device_id() call.
7684 * The logic for file system namespace (ReadOnlyDirectory=,
7685 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
7686 require pivot_root() anymore. This means fewer temporary
7687 directories are created below /tmp for this feature.
7689 * nspawn containers will now see and receive all submounts
7690 made on the host OS below the root file system of the
7693 * Forward Secure Sealing is now supported for Journal files,
7694 which provide cryptographical sealing of journal files so
7695 that attackers cannot alter log history anymore without this
7696 being detectable. Lennart will soon post a blog story about
7697 this explaining it in more detail.
7699 * There are two new service settings RestartPreventExitStatus=
7700 and SuccessExitStatus= which allow configuration of exit
7701 status (exit code or signal) which will be excepted from the
7702 restart logic, resp. consider successful.
7704 * journalctl gained the new --verify switch that can be used
7705 to check the integrity of the structure of journal files and
7706 (if Forward Secure Sealing is enabled) the contents of
7709 * nspawn containers will now be run with /dev/stdin, /dev/fd/
7710 and similar symlinks pre-created. This makes running shells
7711 as container init process a lot more fun.
7713 * The fstab support can now handle PARTUUID= and PARTLABEL=
7716 * A new ConditionHost= condition has been added to match
7717 against the hostname (with globs) and machine ID. This is
7718 useful for clusters where a single OS image is used to
7719 provision a large number of hosts which shall run slightly
7720 different sets of services.
7722 * Services which hit the restart limit will now be placed in a
7725 Contributions from: Bertram Poettering, Dave Reisner, Huang
7726 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
7727 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
7731 * When running in --user mode systemd will now become a
7732 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
7733 tree a lot more organized.
7735 * A new PartOf= unit dependency type has been introduced that
7736 may be used to group services in a natural way.
7738 * "systemctl enable" may now be used to enable instances of
7741 * journalctl now prints error log levels in red, and
7742 warning/notice log levels in bright white. It also supports
7743 filtering by log level now.
7745 * cgtop gained a new -n switch (similar to top), to configure
7746 the maximum number of iterations to run for. It also gained
7747 -b, to run in batch mode (accepting no input).
7749 * The suffix ".service" may now be omitted on most systemctl
7750 command lines involving service unit names.
7752 * There's a new bus call in logind to lock all sessions, as
7753 well as a loginctl verb for it "lock-sessions".
7755 * libsystemd-logind.so gained a new call sd_journal_perror()
7756 that works similar to libc perror() but logs to the journal
7757 and encodes structured information about the error number.
7759 * /etc/crypttab entries now understand the new keyfile-size=
7762 * shutdown(8) now can send a (configurable) wall message when
7763 a shutdown is cancelled.
7765 * The mount propagation mode for the root file system will now
7766 default to "shared", which is useful to make containers work
7767 nicely out-of-the-box so that they receive new mounts from
7768 the host. This can be undone locally by running "mount
7769 --make-rprivate /" if needed.
7771 * The prefdm.service file has been removed. Distributions
7772 should maintain this unit downstream if they intend to keep
7773 it around. However, we recommend writing normal unit files
7774 for display managers instead.
7776 * Since systemd is a crucial part of the OS we will now
7777 default to a number of compiler switches that improve
7778 security (hardening) such as read-only relocations, stack
7779 protection, and suchlike.
7781 * The TimeoutSec= setting for services is now split into
7782 TimeoutStartSec= and TimeoutStopSec= to allow configuration
7783 of individual time outs for the start and the stop phase of
7786 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
7787 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
7788 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
7789 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
7790 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
7791 Gundersen, Zbigniew Jędrzejewski-Szmek
7795 * The journal and id128 C APIs are now fully documented as man
7798 * Extra safety checks have been added when transitioning from
7799 the initial RAM disk to the main system to avoid accidental
7802 * /etc/crypttab entries now understand the new keyfile-offset=
7805 * systemctl -t can now be used to filter by unit load state.
7807 * The journal C API gained the new sd_journal_wait() call to
7808 make writing synchronous journal clients easier.
7810 * journalctl gained the new -D switch to show journals from a
7813 * journalctl now displays a special marker between log
7814 messages of two different boots.
7816 * The journal is now explicitly flushed to /var via a service
7817 systemd-journal-flush.service, rather than implicitly simply
7818 by seeing /var/log/journal to be writable.
7820 * journalctl (and the journal C APIs) can now match for much
7821 more complex expressions, with alternatives and
7824 * When transitioning from the initial RAM disk to the main
7825 system we will now kill all processes in a killing spree to
7826 ensure no processes stay around by accident.
7828 * Three new specifiers may be used in unit files: %u, %h, %s
7829 resolve to the user name, user home directory resp. user
7830 shell. This is useful for running systemd user instances.
7832 * We now automatically rotate journal files if their data
7833 object hash table gets a fill level > 75%. We also size the
7834 hash table based on the configured maximum file size. This
7835 together should lower hash collisions drastically and thus
7836 speed things up a bit.
7838 * journalctl gained the new "--header" switch to introspect
7839 header data of journal files.
7841 * A new setting SystemCallFilters= has been added to services
7842 which may be used to apply blacklists or whitelists to
7843 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
7845 * nspawn gained a new --link-journal= switch (and quicker: -j)
7846 to link the container journal with the host. This makes it
7847 very easy to centralize log viewing on the host for all
7848 guests while still keeping the journal files separated.
7850 * Many bugfixes and optimizations
7852 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
7853 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
7854 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
7859 * Several tools now understand kernel command line arguments,
7860 which are only read when run in an initial RAM disk. They
7861 usually follow closely their normal counterparts, but are
7864 * There's a new tool to analyze the readahead files that are
7865 automatically generated at boot. Use:
7867 /usr/lib/systemd/systemd-readahead analyze /.readahead
7869 * We now provide an early debug shell on tty9 if this enabled. Use:
7871 systemctl enable debug-shell.service
7873 * All plymouth related units have been moved into the Plymouth
7874 package. Please make sure to upgrade your Plymouth version
7877 * systemd-tmpfiles now supports getting passed the basename of
7878 a configuration file only, in which case it will look for it
7879 in all appropriate directories automatically.
7881 * udevadm info now takes a /dev or /sys path as argument, and
7882 does the right thing. Example:
7884 udevadm info /dev/sda
7885 udevadm info /sys/class/block/sda
7887 * systemctl now prints a warning if a unit is stopped but a
7888 unit that might trigger it continues to run. Example: a
7889 service is stopped but the socket that activates it is left
7892 * "systemctl status" will now mention if the log output was
7893 shortened due to rotation since a service has been started.
7895 * The journal API now exposes functions to determine the
7896 "cutoff" times due to rotation.
7898 * journald now understands SIGUSR1 and SIGUSR2 for triggering
7899 immediately flushing of runtime logs to /var if possible,
7900 resp. for triggering immediate rotation of the journal
7903 * It is now considered an error if a service is attempted to
7904 be stopped that is not loaded.
7906 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
7908 * systemd-analyze now supports Python 3
7910 * tmpfiles now supports cleaning up directories via aging
7911 where the first level dirs are always kept around but
7912 directories beneath it automatically aged. This is enabled
7913 by prefixing the age field with '~'.
7915 * Seat objects now expose CanGraphical, CanTTY properties
7916 which is required to deal with very fast bootups where the
7917 display manager might be running before the graphics drivers
7918 completed initialization.
7920 * Seat objects now expose a State property.
7922 * We now include RPM macros for service enabling/disabling
7923 based on the preset logic. We recommend RPM based
7924 distributions to make use of these macros if possible. This
7925 makes it simpler to reuse RPM spec files across
7928 * We now make sure that the collected systemd unit name is
7929 always valid when services log to the journal via
7932 * There's a new man page kernel-command-line(7) detailing all
7933 command line options we understand.
7935 * The fstab generator may now be disabled at boot by passing
7936 fstab=0 on the kernel command line.
7938 * A new kernel command line option modules-load= is now understood
7939 to load a specific kernel module statically, early at boot.
7941 * Unit names specified on the systemctl command line are now
7942 automatically escaped as needed. Also, if file system or
7943 device paths are specified they are automatically turned
7944 into the appropriate mount or device unit names. Example:
7946 systemctl status /home
7947 systemctl status /dev/sda
7949 * The SysVConsole= configuration option has been removed from
7950 system.conf parsing.
7952 * The SysV search path is no longer exported on the D-Bus
7955 * The Names= option has been removed from unit file parsing.
7957 * There's a new man page bootup(7) detailing the boot process.
7959 * Every unit and every generator we ship with systemd now
7960 comes with full documentation. The self-explanatory boot is
7963 * A couple of services gained "systemd-" prefixes in their
7964 name if they wrap systemd code, rather than only external
7965 code. Among them fsck@.service which is now
7966 systemd-fsck@.service.
7968 * The HaveWatchdog property has been removed from the D-Bus
7971 * systemd.confirm_spawn= on the kernel command line should now
7974 * There's a new man page crypttab(5) which details all options
7975 we actually understand.
7977 * systemd-nspawn gained a new --capability= switch to pass
7978 additional capabilities to the container.
7980 * timedated will now read known NTP implementation unit names
7981 from /usr/lib/systemd/ntp-units.d/*.list,
7982 systemd-timedated-ntp.target has been removed.
7984 * journalctl gained a new switch "-b" that lists log data of
7985 the current boot only.
7987 * The notify socket is in the abstract namespace again, in
7988 order to support daemons which chroot() at start-up.
7990 * There is a new Storage= configuration option for journald
7991 which allows configuration of where log data should go. This
7992 also provides a way to disable journal logging entirely, so
7993 that data collected is only forwarded to the console, the
7994 kernel log buffer or another syslog implementation.
7996 * Many bugfixes and optimizations
7998 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
7999 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
8000 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
8001 Shawn Landden, Tom Gundersen
8005 * "systemctl help <unit>" now shows the man page if one is
8008 * Several new man pages have been added.
8010 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
8011 MaxLevelConsole= can now be specified in
8012 journald.conf. These options allow reducing the amount of
8013 data stored on disk or forwarded by the log level.
8015 * TimerSlackNSec= can now be specified in system.conf for
8016 PID1. This allows system-wide power savings.
8018 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
8019 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
8024 * logind is now capable of (optionally) handling power and
8025 sleep keys as well as the lid switch.
8027 * journalctl now understands the syntax "journalctl
8028 /usr/bin/avahi-daemon" to get all log output of a specific
8031 * CapabilityBoundingSet= in system.conf now also influences
8032 the capability bound set of usermode helpers of the kernel.
8034 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
8035 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
8036 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
8037 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
8041 * Note that we skipped 139 releases here in order to set the
8042 new version to something that is greater than both udev's
8043 and systemd's most recent version number.
8045 * udev: all udev sources are merged into the systemd source tree now.
8046 All future udev development will happen in the systemd tree. It
8047 is still fully supported to use the udev daemon and tools without
8048 systemd running, like in initramfs or other init systems. Building
8049 udev though, will require the *build* of the systemd tree, but
8050 udev can be properly *run* without systemd.
8052 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
8053 should be used to create dead device nodes as workarounds for broken
8056 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
8057 no longer supported. udev_monitor_new_from_netlink() needs to be
8058 used to subscribe to events.
8060 * udev: when udevd is started by systemd, processes which are left
8061 behind by forking them off of udev rules, are unconditionally cleaned
8062 up and killed now after the event handling has finished. Services or
8063 daemons must be started as systemd services. Services can be
8064 pulled-in by udev to get started, but they can no longer be directly
8065 forked by udev rules.
8067 * udev: the daemon binary is called systemd-udevd now and installed
8068 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
8069 to adapt to that, create symlink, or rename the binary after building
8072 * libudev no longer provides these symbols:
8073 udev_monitor_from_socket()
8074 udev_queue_get_failed_list_entry()
8075 udev_get_{dev,sys,run}_path()
8076 The versions number was bumped and symbol versioning introduced.
8078 * systemd-loginctl and systemd-journalctl have been renamed
8079 to loginctl and journalctl to match systemctl.
8081 * The config files: /etc/systemd/systemd-logind.conf and
8082 /etc/systemd/systemd-journald.conf have been renamed to
8083 logind.conf and journald.conf. Package updates should rename
8084 the files to the new names on upgrade.
8086 * For almost all files the license is now LGPL2.1+, changed
8087 from the previous GPL2.0+. Exceptions are some minor stuff
8088 of udev (which will be changed to LGPL2.1 eventually, too),
8089 and the MIT licensed sd-daemon.[ch] library that is suitable
8090 to be used as drop-in files.
8092 * systemd and logind now handle system sleep states, in
8093 particular suspending and hibernating.
8095 * logind now implements a sleep/shutdown/idle inhibiting logic
8096 suitable for a variety of uses. Soonishly Lennart will blog
8097 about this in more detail.
8099 * var-run.mount and var-lock.mount are no longer provided
8100 (which previously bind mounted these directories to their new
8101 places). Distributions which have not converted these
8102 directories to symlinks should consider stealing these files
8103 from git history and add them downstream.
8105 * We introduced the Documentation= field for units and added
8106 this to all our shipped units. This is useful to make it
8107 easier to explore the boot and the purpose of the various
8110 * All smaller setup units (such as
8111 systemd-vconsole-setup.service) now detect properly if they
8112 are run in a container and are skipped when
8113 appropriate. This guarantees an entirely noise-free boot in
8114 Linux container environments such as systemd-nspawn.
8116 * A framework for implementing offline system updates is now
8117 integrated, for details see:
8118 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
8120 * A new service type Type=idle is available now which helps us
8121 avoiding ugly interleaving of getty output and boot status
8124 * There's now a system-wide CapabilityBoundingSet= option to
8125 globally reduce the set of capabilities for the
8126 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
8127 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
8128 even CAP_NET_ADMIN system-wide for secure systems.
8130 * There are now system-wide DefaultLimitXXX= options to
8131 globally change the defaults of the various resource limits
8132 for all units started by PID 1.
8134 * Harald Hoyer's systemd test suite has been integrated into
8135 systemd which allows easy testing of systemd builds in qemu
8136 and nspawn. (This is really awesome! Ask us for details!)
8138 * The fstab parser is now implemented as generator, not inside
8141 * systemctl will now warn you if .mount units generated from
8142 /etc/fstab are out of date due to changes in fstab that
8143 have not been read by systemd yet.
8145 * systemd is now suitable for usage in initrds. Dracut has
8146 already been updated to make use of this. With this in place
8147 initrds get a slight bit faster but primarily are much
8148 easier to introspect and debug since "systemctl status" in
8149 the host system can be used to introspect initrd services,
8150 and the journal from the initrd is kept around too.
8152 * systemd-delta has been added, a tool to explore differences
8153 between user/admin configuration and vendor defaults.
8155 * PrivateTmp= now affects both /tmp and /var/tmp.
8157 * Boot time status messages are now much prettier and feature
8158 proper english language. Booting up systemd has never been
8161 * Read-ahead pack files now include the inode number of all
8162 files to pre-cache. When the inode changes the pre-caching
8163 is not attempted. This should be nicer to deal with updated
8164 packages which might result in changes of read-ahead
8167 * We now temporaritly lower the kernel's read_ahead_kb variable
8168 when collecting read-ahead data to ensure the kernel's
8169 built-in read-ahead does not add noise to our measurements
8170 of necessary blocks to pre-cache.
8172 * There's now RequiresMountsFor= to add automatic dependencies
8173 for all mounts necessary for a specific file system path.
8175 * MountAuto= and SwapAuto= have been removed from
8176 system.conf. Mounting file systems at boot has to take place
8179 * nspawn now learned a new switch --uuid= to set the machine
8180 ID on the command line.
8182 * nspawn now learned the -b switch to automatically search
8185 * vt102 is now the default TERM for serial TTYs, upgraded from
8188 * systemd-logind now works on VT-less systems.
8190 * The build tree has been reorganized. The individual
8191 components now have directories of their own.
8193 * A new condition type ConditionPathIsReadWrite= is now available.
8195 * nspawn learned the new -C switch to create cgroups for the
8196 container in other hierarchies.
8198 * We now have support for hardware watchdogs, configurable in
8201 * The scheduled shutdown logic now has a public API.
8203 * We now mount /tmp as tmpfs by default, but this can be
8204 masked and /etc/fstab can override it.
8206 * Since udisks does not make use of /media anymore we are not
8207 mounting a tmpfs on it anymore.
8209 * journalctl gained a new --local switch to only interleave
8210 locally generated journal files.
8212 * We can now load the IMA policy at boot automatically.
8214 * The GTK tools have been split off into a systemd-ui.
8216 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
8217 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
8218 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
8219 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
8220 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
8221 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
8222 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
8223 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
8224 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
8229 * This is mostly a bugfix release
8231 * Support optional initialization of the machine ID from the
8232 KVM or container configured UUID.
8234 * Support immediate reboots with "systemctl reboot -ff"
8236 * Show /etc/os-release data in systemd-analyze output
8238 * Many bugfixes for the journal, including endianness fixes and
8239 ensuring that disk space enforcement works
8241 * sd-login.h is C++ compatible again
8243 * Extend the /etc/os-release format on request of the Debian
8246 * We now refuse non-UTF8 strings used in various configuration
8247 and unit files. This is done to ensure we do not pass invalid
8248 data over D-Bus or expose it elsewhere.
8250 * Register Mimo USB Screens as suitable for automatic seat
8253 * Read SELinux client context from journal clients in a race
8256 * Reorder configuration file lookup order. /etc now always
8257 overrides /run in order to allow the administrator to always
8258 and unconditionally override vendor-supplied or
8259 automatically generated data.
8261 * The various user visible bits of the journal now have man
8262 pages. We still lack man pages for the journal API calls
8265 * We now ship all man pages in HTML format again in the
8268 Contributions from: Dave Reisner, Dirk Eibach, Frederic
8269 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
8270 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
8275 * This is mostly a bugfix release
8277 * systems lacking /etc/os-release are no longer supported.
8279 * Various functionality updates to libsystemd-login.so
8281 * Track class of PAM logins to distinguish greeters from
8284 Contributions from: Kay Sievers, Lennart Poettering, Michael
8289 * This is an important bugfix release for v41.
8291 * Building man pages is now optional which should be useful
8292 for those building systemd from git but unwilling to install
8295 * Watchdog support for supervising services is now usable. In
8296 a future release support for hardware watchdogs
8297 (i.e. /dev/watchdog) will be added building on this.
8299 * Service start rate limiting is now configurable and can be
8300 turned off per service. When a start rate limit is hit a
8301 reboot can automatically be triggered.
8303 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
8305 Contributions from: Benjamin Franzke, Bill Nottingham,
8306 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
8307 Schmidt, Michał Górny, Piotr Drąg
8311 * The systemd binary is installed /usr/lib/systemd/systemd now;
8312 An existing /sbin/init symlink needs to be adapted with the
8315 * The code that loads kernel modules has been ported to invoke
8316 libkmod directly, instead of modprobe. This means we do not
8317 support systems with module-init-tools anymore.
8319 * Watchdog support is now already useful, but still not
8322 * A new kernel command line option systemd.setenv= is
8323 understood to set system wide environment variables
8324 dynamically at boot.
8326 * We now limit the set of capabilities of systemd-journald.
8328 * We now set SIGPIPE to ignore by default, since it only is
8329 useful in shell pipelines, and has little use in general
8330 code. This can be disabled with IgnoreSIPIPE=no in unit
8333 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
8334 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
8339 * This is mostly a bugfix release
8341 * We now expose the reason why a service failed in the
8342 "Result" D-Bus property.
8344 * Rudimentary service watchdog support (will be completed over
8345 the next few releases.)
8347 * When systemd forks off in order execute some service we will
8348 now immediately changes its argv[0] to reflect which process
8349 it will execute. This is useful to minimize the time window
8350 with a generic argv[0], which makes bootcharts more useful
8352 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
8353 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
8354 Mike Kazantsev, Ray Strode
8358 * This is mostly a test release, but incorporates many
8361 * New systemd-cgtop tool to show control groups by their
8364 * Linking against libacl for ACLs is optional again. If
8365 disabled, support tracking device access for active logins
8366 goes becomes unavailable, and so does access to the user
8367 journals by the respective users.
8369 * If a group "adm" exists, journal files are automatically
8370 owned by them, thus allow members of this group full access
8371 to the system journal as well as all user journals.
8373 * The journal now stores the SELinux context of the logging
8374 client for all entries.
8376 * Add C++ inclusion guards to all public headers
8378 * New output mode "cat" in the journal to print only text
8379 messages, without any meta data like date or time.
8381 * Include tiny X server wrapper as a temporary stop-gap to
8382 teach XOrg udev display enumeration. This is used by display
8383 managers such as gdm, and will go away as soon as XOrg
8384 learned native udev hotplugging for display devices.
8386 * Add new systemd-cat tool for executing arbitrary programs
8387 with STDERR/STDOUT connected to the journal. Can also act as
8388 BSD logger replacement, and does so by default.
8390 * Optionally store all locally generated coredumps in the
8391 journal along with meta data.
8393 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
8394 writing short strings to files (for usage for /sys), and for
8395 creating symlinks, character and block device nodes.
8397 * New unit file option ControlGroupPersistent= to make cgroups
8398 persistent, following the mechanisms outlined in
8399 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
8401 * Support multiple local RTCs in a sane way
8403 * No longer monopolize IO when replaying readahead data on
8404 rotating disks, since we might starve non-file-system IO to
8405 death, since fanotify() will not see accesses done by blkid,
8408 * Do not show kernel threads in systemd-cgls anymore, unless
8409 requested with new -k switch.
8411 Contributions from: Dan Horák, Kay Sievers, Lennart
8412 Poettering, Michal Schmidt
8416 * This is mostly a test release, but incorporates many
8419 * The git repository moved to:
8420 git://anongit.freedesktop.org/systemd/systemd
8421 ssh://git.freedesktop.org/git/systemd/systemd
8423 * First release with the journal
8424 http://0pointer.de/blog/projects/the-journal.html
8426 * The journal replaces both systemd-kmsg-syslogd and
8427 systemd-stdout-bridge.
8429 * New sd_pid_get_unit() API call in libsystemd-logind
8431 * Many systemadm clean-ups
8433 * Introduce remote-fs-pre.target which is ordered before all
8434 remote mounts and may be used to start services before all
8437 * Added Mageia support
8439 * Add bash completion for systemd-loginctl
8441 * Actively monitor PID file creation for daemons which exit in
8442 the parent process before having finished writing the PID
8443 file in the daemon process. Daemons which do this need to be
8444 fixed (i.e. PID file creation must have finished before the
8445 parent exits), but we now react a bit more gracefully to them.
8447 * Add colourful boot output, mimicking the well-known output
8448 of existing distributions.
8450 * New option PassCredentials= for socket units, for
8451 compatibility with a recent kernel ABI breakage.
8453 * /etc/rc.local is now hooked in via a generator binary, and
8454 thus will no longer act as synchronization point during
8457 * systemctl list-unit-files now supports --root=.
8459 * systemd-tmpfiles now understands two new commands: z, Z for
8460 relabelling files according to the SELinux database. This is
8461 useful to apply SELinux labels to specific files in /sys,
8464 * Output of SysV services is now forwarded to both the console
8465 and the journal by default, not only just the console.
8467 * New man pages for all APIs from libsystemd-login.
8469 * The build tree got reorganized and the build system is a
8470 lot more modular allowing embedded setups to specifically
8471 select the components of systemd they are interested in.
8473 * Support for Linux systems lacking the kernel VT subsystem is
8476 * configure's --with-rootdir= got renamed to
8477 --with-rootprefix= to follow the naming used by udev and
8480 * Unless specified otherwise we will now install to /usr instead
8481 of /usr/local by default.
8483 * Processes with '@' in argv[0][0] are now excluded from the
8484 final shut-down killing spree, following the logic explained
8486 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
8488 * All processes remaining in a service cgroup when we enter
8489 the START or START_PRE states are now killed with
8490 SIGKILL. That means it is no longer possible to spawn
8491 background processes from ExecStart= lines (which was never
8492 supported anyway, and bad style).
8494 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
8495 reloading of units together.
8497 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
8498 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
8499 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
8500 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
8501 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
~ianmdlvl / elogind.git / blob