1 systemd System and Service Manager
3 CHANGES WITH 239 in spe:
5 * NETWORK INTERFACE DEVICE NAMING CHANGES: elogind-udevd's "net_id"
6 builtin may name network interfaces differently than in previous
7 versions. SR-IOV virtual functions and NPAR partitions with PCI
8 function numbers of 8 and above will be named more predictably, and
9 udev may generate names based on the PCI slot number in some cases
10 where it previously did not.
12 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
13 elogind.service. Since v235, IPAddressDeny=any has been set to
14 the unit. So, it is expected that the default behavior of elogind
15 is not changed. However, if distribution packagers or administrators
16 disabled or modified IPAddressDeny= setting by a drop-in config file,
17 then it may be necessary to update the file to re-enable AF_INET and
18 AF_INET6 to support network user name services, e.g. NIS.
20 * When the RestrictNamespaces= unit property is specified multiple times,
21 then the specified types are merged now. Previously, only the last
22 assignment was used. So, if distribution packagers or administrators
23 modified the setting by a drop-in config file, then it may be necessary
26 * When OnFailure= is used in combination with Restart= on a service
27 unit, then the specified units will no longer be triggered on
28 failures that result in restarting. Previously, the specified units
29 would be activated each time the unit failed, even when the unit was
30 going to be restarted automatically. This behaviour contradicted the
31 documentation. With this release the code is adjusted to match the
34 * elogind-tmpfiles will now print a notice whenever it encounters
35 tmpfiles.d/ lines referencing the /var/run/ directory. It will
36 recommend reworking them to use the /run/ directory instead (for
37 which /var/run/ is simply a symlinked compatibility alias). This way
38 elogind-tmpfiles can properly detect line conflicts and merge lines
39 referencing the same file by two paths, without having to access
42 * systemctl disable/unmask/preset/preset-all cannot be used with
43 --runtime. Previously this was allowed, but resulted in unintuitive
44 behaviour that wasn't useful. systemctl disable/unmask will now
45 undo both runtime and persistent enablement/masking, i.e. it will
46 remove any relevant symlinks both in /run and /etc.
48 * sd-boot acquired new loader configuration settings to optionally turn
49 off Windows and MacOS boot partition discovery as well as
50 reboot-into-firmware menu items. It is also able to pick a better
51 screen resolution for HiDPI systems, and now provides loader
52 configuration settings to change the resolution explicitly.
54 * The elogind-resolve tool has been renamed to resolvectl (it also
55 * elogind-resolved now supports DNS-over-TLS ("PrivateDNS"). It's still
56 turned off by default, use PrivateDNS=yes to turn it on in
57 resolved.conf. We intend to make this the default as soon as couple
58 of additional techniques for optimizing the initial latency caused by
59 establishing a TLS/TCP connection are implemented.
61 remains available under the old name, for compatibility), and its
62 interface is now verb-based, similar in style to the other <xyz>ctl
63 tools, such as systemctl or loginctl.
65 * The resolvectl/elogind-resolve tool also provides 'resolveconf'
66 compatibility. It may be symlinked under the 'resolveconf' name, in
67 which case it will take arguments and input compatible with the
68 Debian and FreeBSD resolvconf tool.
70 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
71 where the system initially suspends, and after a time-out resumes and
74 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
75 set the client will only send a DUID as client identifier.
77 * The nss-elogind glibc NSS module will now enumerate dynamic users and
78 groups in effect. Previously, it could resolve UIDs/GIDs to user
79 names/groups and vice versa, but did not support enumeration.
81 * journald's Compress= configuration setting now optionally accepts a
82 byte threshold value. All journal objects larger than this threshold
83 will be compressed, smaller ones will not. Previously this threshold
84 was not configurable and set to 512.
86 * A new system.conf setting NoNewPrivileges= is now available which may
87 be used to turn off acquisition of new privileges system-wide
88 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
89 for all its children). Note that turning this option on means setuid
90 binaries and file system capabilities lose their special powers.
91 While turning on this option is a big step towards a more secure
92 system, doing so is likely to break numerous pre-existing UNIX tools,
93 in particular su and sudo.
95 * A new service elogind-time-sync-wait.service has been added. If
96 enabled it will delay the time-sync.target unit at boot until time
97 synchronization has been received from the network. This
98 functionality is useful on systems lacking a local RTC or where it is
99 acceptable that the boot process shall be delayed by external network
102 * When hibernating, elogind will now inform the kernel of the image
103 write offset, on kernels new enough to support this. This means swap
104 files should work for hibernation now.
106 * When loading unit files, elogind will now look for drop-in unit files
107 extensions in additional places. Previously, for a unit file name
108 "foo-bar-baz.service" it would look for dropin files in
109 "foo-bar-baz.service.d/*.conf". Now, it will also look in
110 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
111 service name truncated after all inner dashes. This scheme allows
112 writing drop-ins easily that apply to a whole set of unit files at
113 once. It's particularly useful for mount and slice units (as their
114 naming is prefix based), but is also useful for service and other
115 units, for packages that install multiple unit files at once,
116 following a strict naming regime of beginning the unit file name with
117 the package's name. Two new specifiers are now supported in unit
118 files to match this: %j and %J are replaced by the part of the unit
119 name following the last dash.
121 * Unit files and other configuration files that support specifier
122 expansion now understand another two new specifiers: %T and %V will
123 resolve to /tmp and /var/tmp respectively, or whatever temporary
124 directory has been set for the calling user.
126 * The ExecStart= lines of unit files are no longer required to
127 reference absolute paths. If non-absolute paths are specified the
128 specified binary name is searched within the service manager's
129 built-in $PATH, which may be queried with 'elogind-path
130 search-binaries-default'. It's generally recommended to continue to
131 use absolute paths for all binaries specified in unit files.
133 * Units gained a new load state "bad-setting", which is used when a
134 unit file was loaded, but contained fatal errors which prevent it
135 from being started (for example, an ExecStart= path which references
136 a non-existent executable).
138 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
139 support alternative debuggers, for example lldb. The old name
140 continues to be available however, for compatibility reasons. Use the
141 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
142 to pick an alternative debugger instead of the default gdb.
144 * systemctl and the other tools will now output escape sequences that
145 generate proper clickable hyperlinks in various terminal emulators
146 where useful (for example, in the "systemctl status" output you can
147 now click on the unit file name to quickly open it in the
148 editor/viewer of your choice). Note that not all terminal emulators
149 support this functionality yet, but many do. Unfortunately, the
150 "less" pager doesn't support this yet, hence this functionality is
151 currently automatically turned off when a pager is started (which
152 happens quite often due to auto-paging). We hope to remove this
153 limitation as soon as "less" learns these escape sequences. This new
154 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
155 environment variable. For details on these escape sequences see:
156 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
158 * networkd's .network files now support a new IPv6MTUBytes= option for
159 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
160 option in the [Route] section to configure the MTU to use for
161 specific routes. It also gained support for configuration of the DHCP
162 "UserClass" option through the new UserClass= setting. It gained
163 three new options in the new [CAN] section for configuring CAN
164 networks. The MULTICAST and ALLMULTI interface flags may now be
165 controlled explicitly with the new Multicast= and AllMulticast=
168 * networkd will now automatically make use of the kernel's route
169 expiration feature, if it is available.
171 * udevd's .link files now support setting the number of receive and
172 transmit channels, using the RxChannels=, TxChannels=,
173 OtherChannels=, CombinedChannels= settings.
175 * Support for UDPSegmentationOffload= has been removed, given its
176 limited support in hardware, and waning software support.
178 * networkd's .netdev files now support creating "netdevsim" interfaces.
180 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
181 to query the unit belonging to a specific kernel control group.
183 * elogind-analyze gained a new verb "cat-config", which may be used to
184 dump the contents of any configuration file, with all its matching
185 drop-in files added in, and honouring the usual search and masking
186 logic applied to elogind configuration files. For example use
187 "elogind-analyze cat-config elogind/system.conf" to get the complete
188 system configuration file of elogind how it would be loaded by PID 1
189 itself. Similar to this, various tools such as elogind-tmpfiles or
190 elogind-sysusers, gained a new option "--cat-config", which does the
191 corresponding operation for their own configuration settings. For
192 example, "elogind-tmpfiles --cat-config" will now output the full
193 list of tmpfiles.d/ lines in place.
195 * timedatectl gained two new verbs "timesync-status" (to show the
196 current NTP synchronization state of elogind-timesyncd) and
197 "show-timesync" (to show bus properties of elogind-timesyncd).
199 * elogind-timesyncd gained a bus interface on which it exposes details
202 * elogind-nspawn gained a new --rlimit= switch for setting initial
203 resource limits for the container payload. There's a new switch
204 --hostname= to explicitly override the container's hostname. A new
205 --no-new-privileges= switch may be used to control the
206 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
207 --oom-score-adjust= switch controls the OOM scoring adjustment value
208 for the payload. The new --cpu-affinity= switch controls the CPU
209 affinity of the container payload. The new --resolv-conf= switch
210 allows more detailed control of /etc/resolv.conf handling of the
211 container. Similarly, the new --timezone= switch allows more detailed
212 control of /etc/localtime handling of the container.
214 * elogind-detect-virt gained a new --list switch, which will print a
215 list of all currently known VM and container environments.
217 * Support for "Portable Services" has been added, see
218 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
219 experimental, but this is expected to change soon. Reflecting this
220 experimental state, the "portablectl" binary is not installed into
221 /usr/bin yet. The binary has to be called with the full path
222 /usr/lib/elogind/portablectl instead.
224 * journalctl's and systemctl's -o switch now knows a new log output
225 mode "with-unit". The output it generates is very similar to the
226 regular "short" mode, but displays the unit name instead of the
227 syslog tag for each log line. Also, the date is shown with timezone
228 information. This mode is probably more useful than the classic
229 "short" output mode for most purposes, except where pixel-perfect
230 compatibility with classic /var/log/messages formatting is required.
232 * A new --dump-bus-properties switch has been added to the elogind
233 binary, which may be used to dump all supported D-Bus properties.
234 (Options which are still supported, but are deprecated, are *not*
237 * sd-bus gained a set of new calls:
238 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
239 enable/disable the "floating" state of a bus slot object,
240 i.e. whether the slot object pins the bus it is allocated for into
241 memory or if the bus slot object gets disconnected when the bus goes
242 away. sd_bus_open_with_description(),
243 sd_bus_open_user_with_description(),
244 sd_bus_open_system_with_description() may be used to allocate bus
245 objects and set their description string already during allocation.
247 * sd-event gained support for watching inotify events from the event
248 loop, in an efficient way, sharing inotify handles between multiple
249 users. For this a new function sd_event_add_inotify() has been added.
251 * sd-event and sd-bus gained support for calling special user-supplied
252 destructor functions for userdata pointers associated with
253 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
254 functions sd_bus_slot_set_destroy_callback,
255 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
256 sd_bus_track_get_destroy_callback,
257 sd_event_source_set_destroy_callback,
258 sd_event_source_get_destroy_callback have been added.
260 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
262 * PID 1 will now automatically reschedule .timer units whenever the
263 local timezone changes. (They previously got rescheduled
264 automatically when the system clock changed.)
266 * New documentation has been added to document cgroups delegation,
267 portable services and the various code quality tools we have set up:
269 https://github.com/systemd/systemd/blob/master/doc/CGROUP_DELEGATION.md
270 https://github.com/systemd/systemd/blob/master/doc/PORTABLE_SERVICES.md
271 https://github.com/systemd/systemd/blob/master/doc/CODE_QUALITY.md
273 * pam_elogind will now look for PAM userdata fields elogind.memory_max,
274 elogind.tasks_max, elogind.cpu_weight, elogind.io_weight set by
275 earlier PAM modules. The data in these fields is used to initialize
276 the session scope's resource properties. Thus external PAM modules
277 may now configure per-session limits, for example sourced from
278 external user databases.
280 * socket units with Accept=yes will now maintain a "refused" counter in
281 addition to the existing "accepted" counter, counting connections
282 refused due to the enforced limits.
284 * The "elogind-path search-binaries-default" command may now be use to
285 query the default, built-in $PATH PID 1 will pass to the services it
288 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
289 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Antique, Arnaud
290 Rebillout, Brian J. Murrell, Bruno Vernay, Chris Lesiak, Christian
291 Brauner, Christian Hesse, Daniel Dao, Daniel Lin, Danylo Korostil,
292 Davide Cavalca, David Tardon, Dimitri John Ledkov, Dmitriy Geels,
293 Douglas Christman, Elia Geretto, emelenas, Evegeny Vereshchagin, Evgeny
294 Vereshchagin, Felipe Sateler, Feng Sun, Filipe Brandenburger, Franck
295 Bui, futpib, Giuseppe Scrivano, Guillem Jover, guixxx, Hans de Goede,
296 Henrique Dante de Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko,
297 Ivan Shapovalov, James Cowgill, Jan Janssen, Jan Synacek, Jared
298 Kazimir, João Paulo Rechi Vita, Joost Heitbrink, juergbi, Jui-Chi Ricky
299 Liang, Kai-Heng Feng, Karol Augustin, Krzysztof Nowicki, Lauri
300 Tirkkonen, Lennart Poettering, Leonard, Long Li, Luca Boccassi, Lucas
301 Werkmeister, Marcel Hoppe, Marc Kleine-Budde, Mario Limonciello, Martin
302 Jansa, Martin Wilck, Mathieu Malaterre, Matteo F. Vescovi, Michael
303 Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal Sekletar,
304 Mike Gilbert, Mikhail Kasimov, Milan, Milan Broz, mourikwa, Muhammet
305 Kara, Nicolas Boichat, Omer Katz, Paride Legovini, Paul Menzel, Paul
306 Milliken, Peter A. Bigot, Peter Hutterer, Peter Jones, Philip Sequeira,
307 Philip Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de
308 Araujo, Rosen Penev, rubensa, Ryan Gonzalez, Salvo 'LtWorf' Tomaselli,
309 Sebastian Reichel, Sergio Lindo Mansilla, Stefan Schweter, Stephen
310 Hemminger, Stuart Hayes, Susant Sahani, Sylvain Plantefève, Thomas
311 H. P. Andersen, Tobias Jungel, Tomasz Torcz, Vito Caputo, Will Dietz,
312 Will Thompson, xginn8, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
318 * The MemoryAccounting= unit property now defaults to on. After
319 discussions with the upstream control group maintainers we learnt
320 that the negative impact of cgroup memory accounting on current
321 kernels is finally relatively minimal, so that it should be safe to
322 enable this by default without affecting system performance. Besides
323 memory accounting only task accounting is turned on by default, all
324 other forms of resource accounting (CPU, IO, IP) remain off for now,
325 because it's not clear yet that their impact is small enough to move
326 from opt-in to opt-out. We recommend downstreams to leave memory
327 accounting on by default if kernel 4.14 or higher is primarily
328 used. On very resource constrained systems or when support for old
329 kernels is a necessity, -Dmemory-accounting-default=false can be used
330 to revert this change.
332 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
333 %udev_rules_update) and the journal catalog (%journal_catalog_update)
334 from the upgrade scriptlets of individual packages now do nothing.
335 Transfiletriggers have been added which will perform those updates
336 once at the end of the transaction.
338 Similar transfiletriggers have been added to execute any sysctl.d
339 and binfmt.d rules. Thus, it should be unnecessary to provide any
340 scriptlets to execute this configuration from package installation
343 * elogind-sysusers gained a mode where the configuration to execute is
344 specified on the command line, but this configuration is not executed
345 directly, but instead it is merged with the configuration on disk,
346 and the result is executed. This is useful for package installation
347 scripts which want to create the user before installing any files on
348 disk (in case some of those files are owned by that user), while
349 still allowing local admin overrides.
351 This functionality is exposed to rpm scriptlets through a new
352 %sysusers_create_package macro. Old %sysusers_create and
353 %sysusers_create_inline macros are deprecated.
355 A transfiletrigger for sysusers.d configuration is now installed,
356 which means that it should be unnecessary to call elogind-sysusers from
357 package installation scripts, unless the package installs any files
358 owned by those newly-created users, in which case
359 %sysusers_create_package should be used.
361 * Analogous change has been done for elogind-tmpfiles: it gained a mode
362 where the command-line configuration is merged with the configuration
363 on disk. This is exposed as the new %tmpfiles_create_package macro,
364 and %tmpfiles_create is deprecated. A transfiletrigger is installed
365 for tmpfiles.d, hence it should be unnecessary to call elogind-tmpfiles
366 from package installation scripts.
368 * sysusers.d configuration for a user may now also specify the group
369 number, in addition to the user number ("u username 123:456"), or
370 without the user number ("u username -:456").
372 * Configution items for elogind-sysusers can now be specified as
373 positional arguments when the new --inline switch is used.
375 * The login shell of users created through sysusers.d may now be
376 specified (previously, it was always /bin/sh for root and
377 /sbin/nologin for other users).
379 * elogind-analyze gained a new --global switch to look at global user
380 configuration. It also gained a unit-paths verb to list the unit load
381 paths that are compiled into elogind (which can be used with
382 --elogind, --user, or --global).
384 * udevadm trigger gained a new --settle/-w option to wait for any
385 triggered events to finish (but just those, and not any other events
386 which are triggered meanwhile).
388 * The action that elogind takes when the lid is closed and the
389 machine is connected to external power can now be configured using
390 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
391 was determined by HandleLidSwitch=, and, for backwards compatibility,
392 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
394 * journalctl will periodically call sd_journal_process() to make it
395 resilient against inotify queue overruns when journal files are
396 rotated very quickly.
398 * Two new functions in libelogind — sd_bus_get_n_queued_read and
399 sd_bus_get_n_queued_write — may be used to check the number of
400 pending bus messages.
402 * elogind gained a new
403 org.freedesktop.elogind1.Manager.AttachProcessesToUnit dbus call
404 which can be used to migrate foreign processes to scope and service
405 units. The primary user for this new API is elogind itself: the
406 elogind --user instance uses this call of the elogind --system
407 instance to migrate processes if it itself gets the request to
408 migrate processes and the kernel refuses this due to access
409 restrictions. Thanks to this "elogind-run --scope --user …" works
410 again in pure cgroups v2 environments when invoked from the user
413 * A new TemporaryFileSystem= setting can be used to mask out part of
414 the real file system tree with tmpfs mounts. This may be combined
415 with BindPaths= and BindReadOnlyPaths= to hide files or directories
416 not relevant to the unit, while still allowing some paths lower in
417 the tree to be accessed.
419 ProtectHome=tmpfs may now be used to hide user home and runtime
420 directories from units, in a way that is mostly equivalent to
421 "TemporaryFileSystem=/home /run/user /root".
423 * Non-service units are now started with KeyringMode=shared by default.
424 This means that mount and swapon and other mount tools have access
425 to keys in the main keyring.
427 * /sys/fs/bpf is now mounted automatically.
429 * QNX virtualization is now detected by elogind-detect-virt and may
430 be used in ConditionVirtualization=.
432 * IPAccounting= may now be enabled also for slice units.
434 * A new -Dsplit-bin= build configuration switch may be used to specify
435 whether bin and sbin directories are merged, or if they should be
436 included separately in $PATH and various listings of executable
437 directories. The build configuration scripts will try to autodetect
438 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
439 system, but distributions are encouraged to configure this
442 * A new -Dok-color= build configuration switch may be used to change
443 the colour of "OK" status messages.
445 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
446 PrivateNetwork=yes was buggy in previous versions of elogind. This
447 means that after the upgrade and daemon-reexec, any such units must
450 * INCOMPATIBILITY: as announced in the NEWS for 237, elogind-tmpfiles
451 will not exclude read-only files owned by root from cleanup.
453 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
454 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
455 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
456 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
457 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
458 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
459 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
460 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
461 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
462 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
463 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
464 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
465 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
466 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
467 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
468 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
474 * Some keyboards come with a zoom see-saw or rocker which until now got
475 mapped to the Linux "zoomin/out" keys in hwdb. However, these
476 keycodes are not recognized by any major desktop. They now produce
477 Up/Down key events so that they can be used for scrolling.
479 * INCOMPATIBILITY: elogind-tmpfiles' "f" lines changed behaviour
480 slightly: previously, if an argument was specified for lines of this
481 type (i.e. the right-most column was set) this string was appended to
482 existing files each time elogind-tmpfiles was run. This behaviour was
483 different from what the documentation said, and not particularly
484 useful, as repeated elogind-tmpfiles invocations would not be
485 idempotent and grow such files without bounds. With this release
486 behaviour has been altered slightly, to match what the documentation
487 says: lines of this type only have an effect if the indicated files
488 don't exist yet, and only then the argument string is written to the
491 * FUTURE INCOMPATIBILITY: In elogind v238 we intend to slightly change
492 elogind-tmpfiles behaviour: previously, read-only files owned by root
493 were always excluded from the file "aging" algorithm (i.e. the
494 automatic clean-up of directories like /tmp based on
495 atime/mtime/ctime). We intend to drop this restriction, and age files
496 by default even when owned by root and read-only. This behaviour was
497 inherited from older tools, but there have been requests to remove
498 it, and it's not obvious why this restriction was made in the first
499 place. Please speak up now, if you are aware of software that reqires
500 this behaviour, otherwise we'll remove the restriction in v238.
502 * A new environment variable $SYSTEMD_OFFLINE is now understood by
503 systemctl. It takes a boolean argument. If on, systemctl assumes it
504 operates on an "offline" OS tree, and will not attempt to talk to the
505 service manager. Previously, this mode was implicitly enabled if a
506 chroot() environment was detected, and this new environment variable
507 now provides explicit control.
509 * .path and .socket units may now be created transiently, too.
510 Previously only service, mount, automount and timer units were
511 supported as transient units. The elogind-run tool has been updated
512 to expose this new functionality, you may hence use it now to bind
513 arbitrary commands to path or socket activation on-the-fly from the
514 command line. Moreover, almost all properties are now exposed for the
515 unit types that already supported transient operation.
517 * The elogind-mount command gained support for a new --owner= parameter
518 which takes a user name, which is then resolved and included in uid=
519 and gid= mount options string of the file system to mount.
521 * A new unit condition ConditionControlGroupController= has been added
522 that checks whether a specific cgroup controller is available.
524 * Unit files, udev's .link files, and elogind-networkd's .netdev and
525 .network files all gained support for a new condition
526 ConditionKernelVersion= for checking against specific kernel
529 * In elogind-networkd, the [IPVLAN] section in .netdev files gained
530 support for configuring device flags in the Flags= setting. In the
531 same files, the [Tunnel] section gained support for configuring
532 AllowLocalRemote=. The [Route] section in .network files gained
533 support for configuring InitialCongestionWindow=,
534 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
535 understands RapidCommit=.
537 * elogind-networkd's DHCPv6 support gained support for Prefix
540 * sd-bus gained support for a new "watch-bind" feature. When this
541 feature is enabled, an sd_bus connection may be set up to connect to
542 an AF_UNIX socket in the file system as soon as it is created. This
543 functionality is useful for writing early-boot services that
544 automatically connect to the system bus as soon as it is started,
545 without ugly time-based polling. elogind-networkd and
546 elogind-resolved have been updated to make use of this
547 functionality. busctl exposes this functionality in a new
548 --watch-bind= command line switch.
550 * sd-bus will now optionally synthesize a local "Connected" signal as
551 soon as a D-Bus connection is set up fully. This message mirrors the
552 already existing "Disconnected" signal which is synthesized when the
553 connection is terminated. This signal is generally useful but
554 particularly handy in combination with the "watch-bind" feature
555 described above. Synthesizing of this message has to be requested
556 explicitly through the new API call sd_bus_set_connected_signal(). In
557 addition a new call sd_bus_is_ready() has been added that checks
558 whether a connection is fully set up (i.e. between the "Connected" and
559 "Disconnected" signals).
561 * sd-bus gained two new calls sd_bus_request_name_async() and
562 sd_bus_release_name_async() for asynchronously registering bus
563 names. Similar, there is now sd_bus_add_match_async() for installing
564 a signal match asynchronously. All of elogind's own services have
565 been updated to make use of these calls. Doing these operations
566 asynchronously has two benefits: it reduces the risk of deadlocks in
567 case of cyclic dependencies between bus services, and it speeds up
568 service initialization since synchronization points for bus
569 round-trips are removed.
571 * sd-bus gained two new calls sd_bus_match_signal() and
572 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
573 and sd_bus_add_match_async() but instead of taking a D-Bus match
574 string take match fields as normal function parameters.
576 * sd-bus gained two new calls sd_bus_set_sender() and
577 sd_bus_message_set_sender() for setting the sender name of outgoing
578 messages (either for all outgoing messages or for just one specific
579 one). These calls are only useful in direct connections as on
580 brokered connections the broker fills in the sender anyway,
581 overwriting whatever the client filled in.
583 * sd-event gained a new pseudo-handle that may be specified on all API
584 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
585 used this refers to the default event loop object of the calling
586 thread. Note however that this does not implicitly allocate one —
587 which has to be done prior by using sd_event_default(). Similarly
588 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
589 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
590 to the default bus of the specified type of the calling thread. Here
591 too this does not implicitly allocate bus connection objects, this
592 has to be done prior with sd_bus_default() and friends.
594 * sd-event gained a new call pair
595 sd_event_source_{get|set}_io_fd_own(). This may be used to request
596 automatic closure of the file descriptor an IO event source watches
597 when the event source is destroyed.
599 * elogind-networkd gained support for natively configuring WireGuard
602 * In previous versions elogind synthesized user records both for the
603 "nobody" (UID 65534) and "root" (UID 0) users in nss-elogind and
604 internally. In order to simplify distribution-wide renames of the
605 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
606 new transitional flag file has been added: if
607 /etc/elogind/dont-synthesize-nobody exists synthesizing of the 65534
608 user and group record within the elogind codebase is disabled.
610 * elogind-notify gained a new --uid= option for selecting the source
611 user/UID to use for notification messages sent to the service
614 * journalctl gained a new --grep= option to list only entries in which
615 the message matches a certain pattern. By default matching is case
616 insensitive if the pattern is lowercase, and case sensitive
617 otherwise. Option --case-sensitive=yes|no can be used to override
618 this an specify case sensitivity or case insensitivity.
620 * There's now a "elogind-analyze service-watchdogs" command for printing
621 the current state of the service runtime watchdog, and optionally
622 enabling or disabling the per-service watchdogs system-wide if given a
623 boolean argument (i.e. the concept you configure in WatchdogSec=), for
624 debugging purposes. There's also a kernel command line option
625 elogind.service_watchdogs= for controlling the same.
627 * Two new "log-level" and "log-target" options for elogind-analyze were
628 added that merge the now deprecated get-log-level, set-log-level and
629 get-log-target, set-log-target pairs. The deprecated options are still
630 understood for backwards compatibility. The two new options print the
631 current value when no arguments are given, and set them when a
632 level/target is given as an argument.
634 * sysusers.d's "u" lines now optionally accept both a UID and a GID
635 specification, separated by a ":" character, in order to create users
636 where UID and GID do not match.
638 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
639 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
640 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
641 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
642 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
643 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
644 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
645 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
646 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
647 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
648 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
649 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
650 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
651 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
652 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
653 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
654 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
655 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
656 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
657 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
664 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
665 in v235 has been extended to also set the dummy.ko module option
666 numdummies=0, preventing the kernel from automatically creating
667 dummy0. All dummy interfaces must now be explicitly created.
669 * Unknown '%' specifiers in configuration files are now rejected. This
670 applies to units and tmpfiles.d configuration. Any percent characters
671 that are followed by a letter or digit that are not supposed to be
672 interpreted as the beginning of a specifier should be escaped by
673 doubling ("%%"). (So "size=5%" is still accepted, as well as
674 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
675 valid specifiers today.)
677 * systemd-resolved now maintains a new dynamic
678 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
679 recommended to make /etc/resolv.conf a symlink to it. This file
680 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
681 includes dynamically acquired search domains, achieving more correct
682 DNS resolution by software that bypasses local DNS APIs such as NSS.
684 * The "uaccess" udev tag has been dropped from /dev/kvm and
685 /dev/dri/renderD*. These devices now have the 0666 permissions by
686 default (but this may be changed at build-time). /dev/dri/renderD*
687 will now be owned by the "render" group along with /dev/kfd.
689 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
690 systemd-journal-gatewayd.service and
691 systemd-journal-upload.service. This means "nss-systemd" must be
692 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
693 services are resolved properly.
695 * In /etc/fstab two new mount options are now understood:
696 x-systemd.makefs and x-systemd.growfs. The former has the effect that
697 the configured file system is formatted before it is mounted, the
698 latter that the file system is resized to the full block device size
699 after it is mounted (i.e. if the file system is smaller than the
700 partition it resides on, it's grown). This is similar to the fsck
701 logic in /etc/fstab, and pulls in systemd-makefs@.service and
702 systemd-growfs@.service as necessary, similar to
703 systemd-fsck@.service. Resizing is currently only supported on ext4
706 * In systemd-networkd, the IPv6 RA logic now optionally may announce
707 DNS server and domain information.
709 * Support for the LUKS2 on-disk format for encrypted partitions has
710 been added. This requires libcryptsetup2 during compilation and
713 * The systemd --user instance will now signal "readiness" when its
714 basic.target unit has been reached, instead of when the run queue ran
715 empty for the first time.
717 * Tmpfiles.d with user configuration are now also supported.
718 systemd-tmpfiles gained a new --user switch, and snippets placed in
719 ~/.config/user-tmpfiles.d/ and corresponding directories will be
720 executed by systemd-tmpfiles --user running in the new
721 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
722 running in the user session.
724 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
725 %S resolves to the top-level state directory (/var/lib for the system
726 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
727 top-level cache directory (/var/cache for the system instance,
728 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
729 logs directory (/var/log for the system instance,
730 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
731 existing %t specifier, that resolves to the top-level runtime
732 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
735 * journalctl learnt a new parameter --output-fields= for limiting the
736 set of journal fields to output in verbose and JSON output modes.
738 * systemd-timesyncd's configuration file gained a new option
739 RootDistanceMaxSec= for setting the maximum root distance of servers
740 it'll use, as well as the new options PollIntervalMinSec= and
741 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
743 * bootctl gained a new command "list" for listing all available boot
744 menu items on systems that follow the boot loader specification.
746 * systemctl gained a new --dry-run switch that shows what would be done
747 instead of doing it, and is currently supported by the shutdown and
750 * ConditionSecurity= can now detect the TOMOYO security module.
752 * Unit file [Install] sections are now also respected in unit drop-in
753 files. This is intended to be used by drop-ins under /usr/lib/.
755 * systemd-firstboot may now also set the initial keyboard mapping.
757 * Udev "changed" events for devices which are exposed as systemd
758 .device units are now propagated to units specified in
759 ReloadPropagatedFrom= as reload requests.
761 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
762 unit template name (i.e. a name in the form of 'foobar@.service',
763 without the instance component between the '@' and - the '.'), then
764 the escaped sysfs path of the device is automatically used as the
767 * SystemCallFilter= in unit files has been extended so that an "errno"
768 can be specified individually for each system call. Example:
769 SystemCallFilter=~uname:EILSEQ.
771 * The cgroup delegation logic has been substantially updated. Delegate=
772 now optionally takes a list of controllers (instead of a boolean, as
773 before), which lists the controllers to delegate at least.
775 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
777 * A new LogLevelMax= setting configures the maximum log level any
778 process of the service may log at (i.e. anything with a lesser
779 priority than what is specified is automatically dropped). A new
780 LogExtraFields= setting allows configuration of additional journal
781 fields to attach to all log records generated by any of the unit's
784 * New StandardInputData= and StandardInputText= settings along with the
785 new option StandardInput=data may be used to configure textual or
786 binary data that shall be passed to the executed service process via
787 standard input, encoded in-line in the unit file.
789 * StandardInput=, StandardOutput= and StandardError= may now be used to
790 connect stdin/stdout/stderr of executed processes directly with a
791 file or AF_UNIX socket in the file system, using the new "file:" option.
793 * A new unit file option CollectMode= has been added, that allows
794 tweaking the garbage collection logic for units. It may be used to
795 tell systemd to garbage collect units that have failed automatically
796 (normally it only GCs units that exited successfully). systemd-run
797 and systemd-mount expose this new functionality with a new -G option.
799 * "machinectl bind" may now be used to bind mount non-directories
800 (i.e. regularfiles, devices, fifos, sockets).
802 * systemd-analyze gained a new verb "calendar" for validating and
803 testing calendar time specifications to use for OnCalendar= in timer
804 units. Besides validating the expression it will calculate the next
805 time the specified expression would elapse.
807 * In addition to the pre-existing FailureAction= unit file setting
808 there's now SuccessAction=, for configuring a shutdown action to
809 execute when a unit completes successfully. This is useful in
810 particular inside containers that shall terminate after some workload
811 has been completed. Also, both options are now supported for all unit
812 types, not just services.
814 * networkds's IP rule support gained two new options
815 IncomingInterface= and OutgoingInterface= for configuring the incoming
816 and outgoing interfaces of configured rules. systemd-networkd also
817 gained support for "vxcan" network devices.
819 * networkd gained a new setting RequiredForOnline=, taking a
820 boolean. If set, systemd-wait-online will take it into consideration
821 when determining that the system is up, otherwise it will ignore the
822 interface for this purpose.
824 * The sd_notify() protocol gained support for a new operation: with
825 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
826 store again, ahead of POLLHUP or POLLERR when they are removed
829 * A new document doc/UIDS-GIDS.md has been added to the source tree,
830 that documents the UID/GID range and assignment assumptions and
831 requirements of systemd.
833 * The watchdog device PID 1 will ping may now be configured through the
834 WatchdogDevice= configuration file setting, or by setting the
835 systemd.watchdog_service= kernel commandline option.
837 * systemd-resolved's gained support for registering DNS-SD services on
838 the local network using MulticastDNS. Services may either be
839 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
840 the same dir below /run, /usr/lib), or through its D-Bus API.
842 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
843 extend the effective start, runtime, and stop time. The service must
844 continue to send EXTEND_TIMEOUT_USEC within the period specified to
845 prevent the service manager from making the service as timedout.
847 * elogind-resolved's DNSSEC support gained support for RFC 8080
848 (Ed25519 keys and signatures).
850 * The elogind-resolve command line tool gained a new set of options
851 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
852 --set-nta= and --revert to configure per-interface DNS configuration
853 dynamically during runtime. It's useful for pushing DNS information
854 into elogind-resolved from DNS hook scripts that various interface
855 managing software supports (such as pppd).
857 * elogind-nspawn gained a new --network-namespace-path= command line
858 option, which may be used to make a container join an existing
859 network namespace, by specifying a path to a "netns" file.
861 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
862 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
863 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
864 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
865 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
866 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
867 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
868 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
869 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
870 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
871 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
872 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
873 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
874 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
875 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
876 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
877 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
878 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
879 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
880 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
881 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
882 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
883 Jędrzejewski-Szmek, Zeal Jagannatha
889 * INCOMPATIBILITY: systemd-logind.service and other long-running
890 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
891 communication with the outside. This generally improves security of
892 the system, and is in almost all cases a safe and good choice, as
893 these services do not and should not provide any network-facing
894 functionality. However, systemd-logind uses the glibc NSS API to
895 query the user database. This creates problems on systems where NSS
896 is set up to directly consult network services for user database
897 lookups. In particular, this creates incompatibilities with the
898 "nss-nis" module, which attempts to directly contact the NIS/YP
899 network servers it is configured for, and will now consistently
900 fail. In such cases, it is possible to turn off IP sandboxing for
901 systemd-logind.service (set IPAddressDeny= in its [Service] section
902 to the empty string, via a .d/ unit file drop-in). Downstream
903 distributions might want to update their nss-nis packaging to include
904 such a drop-in snippet, accordingly, to hide this incompatibility
905 from the user. Another option is to make use of glibc's nscd service
906 to proxy such network requests through a privilege-separated, minimal
907 local caching daemon, or to switch to more modern technologies such
908 sssd, whose NSS hook-ups generally do not involve direct network
909 access. In general, we think it's definitely time to question the
910 implementation choices of nss-nis, i.e. whether it's a good idea
911 today to embed a network-facing loadable module into all local
912 processes that need to query the user database, including the most
913 trivial and benign ones, such as "ls". For more details about
914 IPAddressDeny= see below.
916 * A new modprobe.d drop-in is now shipped by default that sets the
917 bonding module option max_bonds=0. This overrides the kernel default,
918 to avoid conflicts and ambiguity as to whether or not bond0 should be
919 managed by systemd-networkd or not. This resolves multiple issues
920 with bond0 properties not being applied, when bond0 is configured
921 with systemd-networkd. Distributors may choose to not package this,
922 however in that case users will be prevented from correctly managing
923 bond0 interface using systemd-networkd.
925 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
926 which print the logging level and target of the system manager. They
927 complement the existing "set-log-level" and "set-log-target" verbs
928 used to change those values.
930 * journald.conf gained a new boolean setting ReadKMsg= which defaults
931 to on. If turned off kernel log messages will not be read by
932 systemd-journald or included in the logs. It also gained a new
933 setting LineMax= for configuring the maximum line length in
934 STDOUT/STDERR log streams. The new default for this value is 48K, up
935 from the previous hardcoded 2048.
937 * A new unit setting RuntimeDirectoryPreserve= has been added, which
938 allows more detailed control of what to do with a runtime directory
939 configured with RuntimeDirectory= (i.e. a directory below /run or
940 $XDG_RUNTIME_DIR) after a unit is stopped.
942 * The RuntimeDirectory= setting for units gained support for creating
943 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
944 one top-level directory.
946 * Units gained new options StateDirectory=, CacheDirectory=,
947 LogsDirectory= and ConfigurationDirectory= which are closely related
948 to RuntimeDirectory= but manage per-service directories below
949 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
950 possible to write unit files which when activated automatically gain
951 properly owned service specific directories in these locations, thus
952 making unit files self-contained and increasing compatibility with
953 stateless systems and factory reset where /etc or /var are
954 unpopulated at boot. Matching these new settings there's also
955 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
956 ConfigurationDirectoryMode= for configuring the access mode of these
957 directories. These settings are particularly useful in combination
958 with DynamicUser=yes as they provide secure, properly-owned,
959 writable, and stateful locations for storage, excluded from the
960 sandbox that such services live in otherwise.
962 * Automake support has been removed from this release. systemd is now
965 * systemd-journald will now aggressively cache client metadata during
966 runtime, speeding up log write performance under pressure. This comes
967 at a small price though: as much of the metadata is read
968 asynchronously from /proc/ (and isn't implicitly attached to log
969 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
970 metadata stored alongside a log entry might be slightly
971 out-of-date. Previously it could only be slightly newer than the log
972 message. The time window is small however, and given that the kernel
973 is unlikely to be improved anytime soon in this regard, this appears
976 * nss-myhostname/systemd-resolved will now by default synthesize an
977 A/AAAA resource record for the "_gateway" hostname, pointing to the
978 current default IP gateway. Previously it did that for the "gateway"
979 name, hampering adoption, as some distributions wanted to leave that
980 host name open for local use. The old behaviour may still be
981 requested at build time.
983 * systemd-networkd's [Address] section in .network files gained a new
984 Scope= setting for configuring the IP address scope. The [Network]
985 section gained a new boolean setting ConfigureWithoutCarrier= that
986 tells systemd-networkd to ignore link sensing when configuring the
987 device. The [DHCP] section gained a new Anonymize= boolean option for
988 turning on a number of options suggested in RFC 7844. A new
989 [RoutingPolicyRule] section has been added for configuring the IP
990 routing policy. The [Route] section has gained support for a new
991 Type= setting which permits configuring
992 blackhole/unreachable/prohibit routes.
994 * The [VRF] section in .netdev files gained a new Table= setting for
995 configuring the routing table to use. The [Tunnel] section gained a
996 new Independent= boolean field for configuring tunnels independent of
997 an underlying network interface. The [Bridge] section gained a new
998 GroupForwardMask= option for configuration of propagation of link
999 local frames between bridge ports.
1001 * The WakeOnLan= setting in .link files gained support for a number of
1002 new modes. A new TCP6SegmentationOffload= setting has been added for
1003 configuring TCP/IPv6 hardware segmentation offload.
1005 * The IPv6 RA sender implementation may now optionally send out RDNSS
1006 and RDNSSL records to supply DNS configuration to peers.
1008 * systemd-nspawn gained support for a new --system-call-filter= command
1009 line option for adding and removing entries in the default system
1010 call filter it applies. Moreover systemd-nspawn has been changed to
1011 implement a system call whitelist instead of a blacklist.
1013 * systemd-run gained support for a new --pipe command line option. If
1014 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
1015 are directly passed on to the activated transient service
1016 executable. This allows invoking arbitrary processes as systemd
1017 services (for example to take benefit of dependency management,
1018 accounting management, resource management or log management that is
1019 done automatically for services) — while still allowing them to be
1020 integrated in a classic UNIX shell pipeline.
1022 * When a service sends RELOAD=1 via sd_notify() and reload propagation
1023 using ReloadPropagationTo= is configured, a reload is now propagated
1024 to configured units. (Previously this was only done on explicitly
1025 requested reloads, using "systemctl reload" or an equivalent
1028 * For each service unit a restart counter is now kept: it is increased
1029 each time the service is restarted due to Restart=, and may be
1030 queried using "systemctl show -p NRestarts …".
1032 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
1033 @signal and @timer have been added, for usage with SystemCallFilter=
1034 in unit files and the new --system-call-filter= command line option
1035 of systemd-nspawn (see above).
1037 * ExecStart= lines in unit files gained two new modifiers: when a
1038 command line is prefixed with "!" the command will be executed as
1039 configured, except for the credentials applied by
1040 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
1041 "+", but does still apply namespacing options unlike "+". There's
1042 also "!!" now, which is mostly identical, but becomes a NOP on
1043 systems that support ambient capabilities. This is useful to write
1044 unit files that work with ambient capabilities where possible but
1045 automatically fall back to traditional privilege dropping mechanisms
1046 on systems where this is not supported.
1048 * ListenNetlink= settings in socket units now support RDMA netlink
1051 * A new unit file setting LockPersonality= has been added which permits
1052 locking down the chosen execution domain ("personality") of a service
1055 * A new special target "getty-pre.target" has been added, which is
1056 ordered before all text logins, and may be used to order services
1057 before textual logins acquire access to the console.
1059 * systemd will now attempt to load the virtio-rng.ko kernel module very
1060 early on if a VM environment supporting this is detected. This should
1061 improve entropy during early boot in virtualized environments.
1063 * A _netdev option is now supported in /etc/crypttab that operates in a
1064 similar way as the same option in /etc/fstab: it permits configuring
1065 encrypted devices that need to be ordered after the network is up.
1066 Following this logic, two new special targets
1067 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
1068 added that are to cryptsetup.target what remote-fs.target and
1069 remote-fs-pre.target are to local-fs.target.
1071 * Service units gained a new UnsetEnvironment= setting which permits
1072 unsetting specific environment variables for services that are
1073 normally passed to it (for example in order to mask out locale
1074 settings for specific services that can't deal with it).
1076 * Units acquired a new boolean option IPAccounting=. When turned on, IP
1077 traffic accounting (packet count as well as byte count) is done for
1078 the service, and shown as part of "systemctl status" or "systemd-run
1081 * Service units acquired two new options IPAddressAllow= and
1082 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
1083 for configuring a simple IP access control list for all sockets of
1084 the unit. These options are available also on .slice and .socket
1085 units, permitting flexible access list configuration for individual
1086 services as well as groups of services (as defined by a slice unit),
1087 including system-wide. Note that IP ACLs configured this way are
1088 enforced on every single IPv4 and IPv6 socket created by any process
1089 of the service unit, and apply to ingress as well as egress traffic.
1091 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
1092 structured log message is generated each time the unit is stopped,
1093 containing information about the consumed resources of this
1096 * A new setting KeyringMode= has been added to unit files, which may be
1097 used to control how the kernel keyring is set up for executed
1100 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
1101 "systemctl kexec" and "systemctl exit" are now always asynchronous in
1102 behaviour (that is: these commands return immediately after the
1103 operation was enqueued instead of waiting for the operation to
1104 complete). Previously, "systemctl poweroff" and "systemctl reboot"
1105 were asynchronous on systems using systemd-logind (i.e. almost
1106 always, and like they were on sysvinit), and the other three commands
1107 were unconditionally synchronous. With this release this is cleaned
1108 up, and callers will see the same asynchronous behaviour on all
1109 systems for all five operations.
1111 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
1114 * .timer units now accept calendar specifications in other timezones
1115 than UTC or the local timezone.
1117 * The tmpfiles snippet var.conf has been changed to create
1118 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
1119 the "utmp" group already, and it appears to be generally understood
1120 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
1121 databases. Previously this was implemented correctly for all these
1122 databases excepts btmp, which has been opened up like this now
1123 too. Note that while the other databases are world-readable
1124 (i.e. 0644), btmp is not and remains more restrictive.
1126 * The systemd-resolve tool gained a new --reset-server-features
1127 switch. When invoked like this systemd-resolved will forget
1128 everything it learnt about the features supported by the configured
1129 upstream DNS servers, and restarts the feature probing logic on the
1130 next resolver look-up for them at the highest feature level
1133 * The status dump systemd-resolved sends to the logs upon receiving
1134 SIGUSR1 now also includes information about all DNS servers it is
1135 configured to use, and the features levels it probed for them.
1137 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
1138 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
1139 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
1140 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
1141 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
1142 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
1143 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
1144 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
1145 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
1146 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
1147 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
1148 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
1149 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
1150 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
1151 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
1152 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
1153 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
1154 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
1155 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
1156 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1158 — Berlin, 2017-10-06
1162 * Meson is now supported as build system in addition to Automake. It is
1163 our plan to remove Automake in one of our next releases, so that
1164 Meson becomes our exclusive build system. Hence, please start using
1165 the Meson build system in your downstream packaging. There's plenty
1166 of documentation around how to use Meson, the extremely brief
1169 ./autogen.sh && ./configure && make && sudo make install
1173 meson build && ninja -C build && sudo ninja -C build install
1175 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
1176 which permits configuring a timeout on the time a job is
1177 running. This is particularly useful for setting timeouts on jobs for
1180 * Unit files gained two new options ConditionUser= and ConditionGroup=
1181 for conditionalizing units based on the identity of the user/group
1182 running a systemd user instance.
1184 * systemd-networkd now understands a new FlowLabel= setting in the
1185 [VXLAN] section of .network files, as well as a Priority= in
1186 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
1187 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
1188 gained support for configuration of GENEVE links, and IPv6 address
1189 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
1191 * .link files now understand a new Port= setting.
1193 * systemd-networkd's DHCP support gained support for DHCP option 119
1194 (domain search list).
1196 * systemd-networkd gained support for serving IPv6 address ranges using
1197 the Router Advertisement protocol. The new .network configuration
1198 section [IPv6Prefix] may be used to configure the ranges to
1199 serve. This is implemented based on a new, minimal, native server
1200 implementation of RA.
1202 * journalctl's --output= switch gained support for a new parameter
1203 "short-iso-precise" for a mode where timestamps are shown as precise
1206 * systemd-udevd's "net_id" builtin may now generate stable network
1207 interface names from IBM PowerVM VIO devices as well as ACPI platform
1210 * MulticastDNS support in systemd-resolved may now be explicitly
1211 enabled/disabled using the new MulticastDNS= configuration file
1214 * systemd-resolved may now optionally use libidn2 instead of the libidn
1215 for processing internationalized domain names. Support for libidn2
1216 should be considered experimental and should not be enabled by
1219 * "machinectl pull-tar" and related call may now do verification of
1220 downloaded images using SUSE-style .sha256 checksum files in addition
1221 to the already existing support for validating using Ubuntu-style
1224 * sd-bus gained support for a new sd_bus_message_appendv() call which
1225 is va_list equivalent of sd_bus_message_append().
1227 * sd-boot gained support for validating images using SHIM/MOK.
1229 * The SMACK code learnt support for "onlycap".
1231 * systemd-mount --umount is now much smarter in figuring out how to
1232 properly unmount a device given its mount or device path.
1234 * The code to call libnss_dns as a fallback from libnss_resolve when
1235 the communication with systemd-resolved fails was removed. This
1236 fallback was redundant and interfered with the [!UNAVAIL=return]
1237 suffix. See nss-resolve(8) for the recommended configuration.
1239 * systemd-logind may now be restarted without losing state. It stores
1240 the file descriptors for devices it manages in the system manager
1241 using the FDSTORE= mechanism. Please note that further changes in
1242 other components may be required to make use of this (for example
1243 Xorg has code to listen for stops of systemd-logind and terminate
1244 itself when logind is stopped or restarted, in order to avoid using
1245 stale file descriptors for graphical devices, which is now
1246 counterproductive and must be reverted in order for restarts of
1247 systemd-logind to be safe. See
1248 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
1250 * All kernel install plugins are called with the environment variable
1251 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
1252 /etc/machine-id. If the file is missing or empty, the variable is
1253 empty and BOOT_DIR_ABS is the path of a temporary directory which is
1254 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
1255 is empty, all plugins should not put anything in BOOT_DIR_ABS.
1257 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
1258 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
1259 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
1260 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
1261 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
1262 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
1263 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
1264 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
1265 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
1266 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
1267 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
1268 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
1269 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
1270 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
1271 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
1272 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
1273 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
1274 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
1275 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
1276 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
1277 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
1278 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
1279 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
1280 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
1281 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
1282 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
1283 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
1284 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
1285 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
1288 — Berlin, 2017-07-12
1292 * The "hybrid" control group mode has been modified to improve
1293 compatibility with "legacy" cgroups-v1 setups. Specifically, the
1294 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
1295 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
1296 cgroups-v1 hierarchy), the only externally visible change being that
1297 the cgroups-v2 hierarchy is also mounted, to
1298 /sys/fs/cgroup/unified. This should provide a large degree of
1299 compatibility with "legacy" cgroups-v1, while taking benefit of the
1300 better management capabilities of cgroups-v2.
1302 * The default control group setup mode may be selected both a boot-time
1303 via a set of kernel command line parameters (specifically:
1304 systemd.unified_cgroup_hierarchy= and
1305 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
1306 default selected on the configure command line
1307 (--with-default-hierarchy=). The upstream default is "hybrid"
1308 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
1309 this will change in a future systemd version to be "unified" (pure
1310 cgroups-v2 mode). The third option for the compile time option is
1311 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
1312 distributions to default to "hybrid" mode for release distributions,
1313 starting with v233. We recommend "unified" for development
1314 distributions (specifically: distributions such as Fedora's rawhide)
1315 as that's where things are headed in the long run. Use "legacy" for
1316 greatest stability and compatibility only.
1318 * Note one current limitation of "unified" and "hybrid" control group
1319 setup modes: the kernel currently does not permit the systemd --user
1320 instance (i.e. unprivileged code) to migrate processes between two
1321 disconnected cgroup subtrees, even if both are managed and owned by
1322 the user. This effectively means "systemd-run --user --scope" doesn't
1323 work when invoked from outside of any "systemd --user" service or
1324 scope. Specifically, it is not supported from session scopes. We are
1325 working on fixing this in a future systemd version. (See #3388 for
1326 further details about this.)
1328 * DBus policy files are now installed into /usr rather than /etc. Make
1329 sure your system has dbus >= 1.9.18 running before upgrading to this
1330 version, or override the install path with --with-dbuspolicydir= .
1332 * All python scripts shipped with systemd (specifically: the various
1333 tests written in Python) now require Python 3.
1335 * systemd unit tests can now run standalone (without the source or
1336 build directories), and can be installed into /usr/lib/systemd/tests/
1337 with 'make install-tests'.
1339 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
1340 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
1343 * Support for the %c, %r, %R specifiers in unit files has been
1344 removed. Specifiers are not supposed to be dependent on configuration
1345 in the unit file itself (so that they resolve the same regardless
1346 where used in the unit files), but these specifiers were influenced
1347 by the Slice= option.
1349 * The shell invoked by debug-shell.service now defaults to /bin/sh in
1350 all cases. If distributions want to use a different shell for this
1351 purpose (for example Fedora's /sbin/sushell) they need to specify
1352 this explicitly at configure time using --with-debug-shell=.
1354 * The confirmation spawn prompt has been reworked to offer the
1357 (c)ontinue, proceed without asking anymore
1358 (D)ump, show the state of the unit
1359 (f)ail, don't execute the command and pretend it failed
1361 (i)nfo, show a short summary of the unit
1362 (j)obs, show jobs that are in progress
1363 (s)kip, don't execute the command and pretend it succeeded
1364 (y)es, execute the command
1366 The 'n' choice for the confirmation spawn prompt has been removed,
1367 because its meaning was confusing.
1369 The prompt may now also be redirected to an alternative console by
1370 specifying the console as parameter to systemd.confirm_spawn=.
1372 * Services of Type=notify require a READY=1 notification to be sent
1373 during startup. If no such message is sent, the service now fails,
1374 even if the main process exited with a successful exit code.
1376 * Services that fail to start up correctly now always have their
1377 ExecStopPost= commands executed. Previously, they'd enter "failed"
1378 state directly, without executing these commands.
1380 * The option MulticastDNS= of network configuration files has acquired
1381 an actual implementation. With MulticastDNS=yes a host can resolve
1382 names of remote hosts and reply to mDNS A and AAAA requests.
1384 * When units are about to be started an additional check is now done to
1385 ensure that all dependencies of type BindsTo= (when used in
1386 combination with After=) have been started.
1388 * systemd-analyze gained a new verb "syscall-filter" which shows which
1389 system call groups are defined for the SystemCallFilter= unit file
1390 setting, and which system calls they contain.
1392 * A new system call filter group "@filesystem" has been added,
1393 consisting of various file system related system calls. Group
1394 "@reboot" has been added, covering reboot, kexec and shutdown related
1395 calls. Finally, group "@swap" has been added covering swap
1396 configuration related calls.
1398 * A new unit file option RestrictNamespaces= has been added that may be
1399 used to restrict access to the various process namespace types the
1400 Linux kernel provides. Specifically, it may be used to take away the
1401 right for a service unit to create additional file system, network,
1402 user, and other namespaces. This sandboxing option is particularly
1403 relevant due to the high amount of recently discovered namespacing
1404 related vulnerabilities in the kernel.
1406 * systemd-udev's .link files gained support for a new AutoNegotiation=
1407 setting for configuring Ethernet auto-negotiation.
1409 * systemd-networkd's .network files gained support for a new
1410 ListenPort= setting in the [DHCP] section to explicitly configure the
1411 UDP client port the DHCP client shall listen on.
1413 * .network files gained a new Unmanaged= boolean setting for explicitly
1414 excluding one or more interfaces from management by systemd-networkd.
1416 * The systemd-networkd ProxyARP= option has been renamed to
1417 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
1418 renamed to ReduceARPProxy=. The old names continue to be available
1421 * systemd-networkd gained support for configuring IPv6 Proxy NDP
1422 addresses via the new IPv6ProxyNDPAddress= .network file setting.
1424 * systemd-networkd's bonding device support gained support for two new
1425 configuration options ActiveSlave= and PrimarySlave=.
1427 * The various options in the [Match] section of .network files gained
1428 support for negative matching.
1430 * New systemd-specific mount options are now understood in /etc/fstab:
1432 x-systemd.mount-timeout= may be used to configure the maximum
1433 permitted runtime of the mount command.
1435 x-systemd.device-bound may be set to bind a mount point to its
1436 backing device unit, in order to automatically remove a mount point
1437 if its backing device is unplugged. This option may also be
1438 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
1439 on the block device, which is now automatically set for all CDROM
1440 drives, so that mounted CDs are automatically unmounted when they are
1441 removed from the drive.
1443 x-systemd.after= and x-systemd.before= may be used to explicitly
1444 order a mount after or before another unit or mount point.
1446 * Enqueued start jobs for device units are now automatically garbage
1447 collected if there are no jobs waiting for them anymore.
1449 * systemctl list-jobs gained two new switches: with --after, for every
1450 queued job the jobs it's waiting for are shown; with --before the
1451 jobs which it's blocking are shown.
1453 * systemd-nspawn gained support for ephemeral boots from disk images
1454 (or in other words: --ephemeral and --image= may now be
1455 combined). Moreover, ephemeral boots are now supported for normal
1456 directories, even if the backing file system is not btrfs. Of course,
1457 if the file system does not support file system snapshots or
1458 reflinks, the initial copy operation will be relatively expensive, but
1459 this should still be suitable for many use cases.
1461 * Calendar time specifications in .timer units now support
1462 specifications relative to the end of a month by using "~" instead of
1463 "-" as separator between month and day. For example, "*-02~03" means
1464 "the third last day in February". In addition a new syntax for
1465 repeated events has been added using the "/" character. For example,
1466 "9..17/2:00" means "every two hours from 9am to 5pm".
1468 * systemd-socket-proxyd gained a new parameter --connections-max= for
1469 configuring the maximum number of concurrent connections.
1471 * sd-id128 gained a new API for generating unique IDs for the host in a
1472 way that does not leak the machine ID. Specifically,
1473 sd_id128_get_machine_app_specific() derives an ID based on the
1474 machine ID a in well-defined, non-reversible, stable way. This is
1475 useful whenever an identifier for the host is needed but where the
1476 identifier shall not be useful to identify the system beyond the
1477 scope of the application itself. (Internally this uses HMAC-SHA256 as
1478 keyed hash function using the machine ID as input.)
1480 * NotifyAccess= gained a new supported value "exec". When set
1481 notifications are accepted from all processes systemd itself invoked,
1482 including all control processes.
1484 * .nspawn files gained support for defining overlay mounts using the
1485 Overlay= and OverlayReadOnly= options. Previously this functionality
1486 was only available on the systemd-nspawn command line.
1488 * systemd-nspawn's --bind= and --overlay= options gained support for
1489 bind/overlay mounts whose source lies within the container tree by
1490 prefixing the source path with "+".
1492 * systemd-nspawn's --bind= and --overlay= options gained support for
1493 automatically allocating a temporary source directory in /var/tmp
1494 that is removed when the container dies. Specifically, if the source
1495 directory is specified as empty string this mechanism is selected. An
1496 example usage is --overlay=+/var::/var, which creates an overlay
1497 mount based on the original /var contained in the image, overlayed
1498 with a temporary directory in the host's /var/tmp. This way changes
1499 to /var are automatically flushed when the container shuts down.
1501 * systemd-nspawn --image= option does now permit raw file system block
1502 devices (in addition to images containing partition tables, as
1505 * The disk image dissection logic in systemd-nspawn gained support for
1506 automatically setting up LUKS encrypted as well as Verity protected
1507 partitions. When a container is booted from an encrypted image the
1508 passphrase is queried at start-up time. When a container with Verity
1509 data is started, the root hash is search in a ".roothash" file
1510 accompanying the disk image (alternatively, pass the root hash via
1511 the new --root-hash= command line option).
1513 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
1514 be used to dissect disk images the same way as systemd-nspawn does
1515 it, following the Bootable Partition Specification. It may even be
1516 used to mount disk images with complex partition setups (including
1517 LUKS and Verity partitions) to a local host directory, in order to
1518 inspect them. This tool is not considered public API (yet), and is
1519 thus not installed into /usr/bin. Please do not rely on its
1520 existence, since it might go away or be changed in later systemd
1523 * A new generator "systemd-verity-generator" has been added, similar in
1524 style to "systemd-cryptsetup-generator", permitting automatic setup of
1525 Verity root partitions when systemd boots up. In order to make use of
1526 this your partition setup should follow the Discoverable Partitions
1527 Specification, and the GPT partition ID of the root file system
1528 partition should be identical to the upper 128bit of the Verity root
1529 hash. The GPT partition ID of the Verity partition protecting it
1530 should be the lower 128bit of the Verity root hash. If the partition
1531 image follows this model it is sufficient to specify a single
1532 "roothash=" kernel command line argument to both configure which root
1533 image and verity partition to use as well as the root hash for
1534 it. Note that systemd-nspawn's Verity support follows the same
1535 semantics, meaning that disk images with proper Verity data in place
1536 may be booted in containers with systemd-nspawn as well as on
1537 physical systems via the verity generator. Also note that the "mkosi"
1538 tool available at https://github.com/systemd/mkosi has been updated
1539 to generate Verity protected disk images following this scheme. In
1540 fact, it has been updated to generate disk images that optionally
1541 implement a complete UEFI SecureBoot trust chain, involving a signed
1542 kernel and initrd image that incorporates such a root hash as well as
1543 a Verity-enabled root partition.
1545 * The hardware database (hwdb) udev supports has been updated to carry
1546 accelerometer quirks.
1548 * All system services are now run with a fresh kernel keyring set up
1549 for them. The invocation ID is stored by default in it, thus
1550 providing a safe, non-overridable way to determine the invocation
1553 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
1554 options for bind mounting arbitrary paths in a service-specific
1555 way. When these options are used, arbitrary host or service files and
1556 directories may be mounted to arbitrary locations in the service's
1559 * Documentation has been added that lists all of systemd's low-level
1560 environment variables:
1562 https://github.com/systemd/systemd/blob/master/doc/ENVIRONMENT.md
1564 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
1565 whether a specific socket file descriptor matches a specified socket
1568 * systemd-firstboot has been updated to check for the
1569 systemd.firstboot= kernel command line option. It accepts a boolean
1570 and when set to false the first boot questions are skipped.
1572 * systemd-fstab-generator has been updated to check for the
1573 systemd.volatile= kernel command line option, which either takes an
1574 optional boolean parameter or the special value "state". If used the
1575 system may be booted in a "volatile" boot mode. Specifically,
1576 "systemd.volatile" is used, the root directory will be mounted as
1577 tmpfs, and only /usr is mounted from the actual root file system. If
1578 "systemd.volatile=state" is used, the root directory will be mounted
1579 as usual, but /var is mounted as tmpfs. This concept provides similar
1580 functionality as systemd-nspawn's --volatile= option, but provides it
1581 on physical boots. Use this option for implementing stateless
1582 systems, or testing systems with all state and/or configuration reset
1583 to the defaults. (Note though that many distributions are not
1584 prepared to boot up without a populated /etc or /var, though.)
1586 * systemd-gpt-auto-generator gained support for LUKS encrypted root
1587 partitions. Previously it only supported LUKS encrypted partitions
1588 for all other uses, except for the root partition itself.
1590 * Socket units gained support for listening on AF_VSOCK sockets for
1591 communication in virtualized QEMU environments.
1593 * The "configure" script gained a new option --with-fallback-hostname=
1594 for specifying the fallback hostname to use if none is configured in
1595 /etc/hostname. For example, by specifying
1596 --with-fallback-hostname=fedora it is possible to default to a
1597 hostname of "fedora" on pristine installations.
1599 * systemd-cgls gained support for a new --unit= switch for listing only
1600 the control groups of a specific unit. Similar --user-unit= has been
1601 added for listing only the control groups of a specific user unit.
1603 * systemd-mount gained a new --umount switch for unmounting a mount or
1604 automount point (and all mount/automount points below it).
1606 * systemd will now refuse full configuration reloads (via systemctl
1607 daemon-reload and related calls) unless at least 16MiB of free space
1608 are available in /run. This is a safety precaution in order to ensure
1609 that generators can safely operate after the reload completed.
1611 * A new unit file option RootImage= has been added, which has a similar
1612 effect as RootDirectory= but mounts the service's root directory from
1613 a disk image instead of plain directory. This logic reuses the same
1614 image dissection and mount logic that systemd-nspawn already uses,
1615 and hence supports any disk images systemd-nspawn supports, including
1616 those following the Discoverable Partition Specification, as well as
1617 Verity enabled images. This option enables systemd to run system
1618 services directly off disk images acting as resource bundles,
1619 possibly even including full integrity data.
1621 * A new MountAPIVFS= unit file option has been added, taking a boolean
1622 argument. If enabled /proc, /sys and /dev (collectively called the
1623 "API VFS") will be mounted for the service. This is only relevant if
1624 RootDirectory= or RootImage= is used for the service, as these mounts
1625 are of course in place in the host mount namespace anyway.
1627 * systemd-nspawn gained support for a new --pivot-root= switch. If
1628 specified the root directory within the container image is pivoted to
1629 the specified mount point, while the original root disk is moved to a
1630 different place. This option enables booting of ostree images
1631 directly with systemd-nspawn.
1633 * The systemd build scripts will no longer complain if the NTP server
1634 addresses are not changed from the defaults. Google now supports
1635 these NTP servers officially. We still recommend downstreams to
1636 properly register an NTP pool with the NTP pool project though.
1638 * coredumpctl gained a new "--reverse" option for printing the list
1639 of coredumps in reverse order.
1641 * coredumpctl will now show additional information about truncated and
1642 inaccessible coredumps, as well as coredumps that are still being
1643 processed. It also gained a new --quiet switch for suppressing
1644 additional informational message in its output.
1646 * coredumpctl gained support for only showing coredumps newer and/or
1647 older than specific timestamps, using the new --since= and --until=
1648 options, reminiscent of journalctl's options by the same name.
1650 * The systemd-coredump logic has been improved so that it may be reused
1651 to collect backtraces in non-compiled languages, for example in
1652 scripting languages such as Python.
1654 * machinectl will now show the UID shift of local containers, if user
1655 namespacing is enabled for them.
1657 * systemd will now optionally run "environment generator" binaries at
1658 configuration load time. They may be used to add environment
1659 variables to the environment block passed to services invoked. One
1660 user environment generator is shipped by default that sets up
1661 environment variables based on files dropped into /etc/environment.d
1662 and ~/.config/environment.d/.
1664 * systemd-resolved now includes the new, recently published 2017 DNSSEC
1667 * hostnamed has been updated to report a new chassis type of
1668 "convertible" to cover "foldable" laptops that can both act as a
1669 tablet and as a laptop, such as various Lenovo Yoga devices.
1671 Contributions from: Adrián López, Alexander Galanin, Alexander
1672 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
1673 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
1674 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
1675 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
1676 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
1677 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
1678 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
1679 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
1680 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
1681 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
1682 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
1683 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
1684 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
1685 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
1686 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
1687 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
1688 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
1689 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
1690 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
1691 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
1692 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
1693 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
1694 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
1695 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
1696 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
1697 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
1698 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
1701 — Berlin, 2017-03-01
1705 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
1706 RestrictAddressFamilies= enabled. These sandboxing options should
1707 generally be compatible with the various external udev call-out
1708 binaries we are aware of, however there may be exceptions, in
1709 particular when exotic languages for these call-outs are used. In
1710 this case, consider turning off these settings locally.
1712 * The new RemoveIPC= option can be used to remove IPC objects owned by
1713 the user or group of a service when that service exits.
1715 * The new ProtectKernelModules= option can be used to disable explicit
1716 load and unload operations of kernel modules by a service. In
1717 addition access to /usr/lib/modules is removed if this option is set.
1719 * ProtectSystem= option gained a new value "strict", which causes the
1720 whole file system tree with the exception of /dev, /proc, and /sys,
1721 to be remounted read-only for a service.
1723 * The new ProtectKernelTunables= option can be used to disable
1724 modification of configuration files in /sys and /proc by a service.
1725 Various directories and files are remounted read-only, so access is
1726 restricted even if the file permissions would allow it.
1728 * The new ProtectControlGroups= option can be used to disable write
1729 access by a service to /sys/fs/cgroup.
1731 * Various systemd services have been hardened with
1732 ProtectKernelTunables=yes, ProtectControlGroups=yes,
1733 RestrictAddressFamilies=.
1735 * Support for dynamically creating users for the lifetime of a service
1736 has been added. If DynamicUser=yes is specified, user and group IDs
1737 will be allocated from the range 61184..65519 for the lifetime of the
1738 service. They can be resolved using the new nss-systemd.so NSS
1739 module. The module must be enabled in /etc/nsswitch.conf. Services
1740 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
1741 any resources allocated by the service will be cleaned up when the
1742 service exits. They also have ProtectHome=read-only and
1743 ProtectSystem=strict enabled, so they are not able to make any
1744 permanent modifications to the system.
1746 * The nss-systemd module also always resolves root and nobody, making
1747 it possible to have no /etc/passwd or /etc/group files in minimal
1748 container or chroot environments.
1750 * Services may be started with their own user namespace using the new
1751 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
1752 under which the service is running are mapped. All other users are
1755 * Support for the cgroup namespace has been added to systemd-nspawn. If
1756 supported by kernel, the container system started by systemd-nspawn
1757 will have its own view of the cgroup hierarchy. This new behaviour
1758 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
1760 * The new MemorySwapMax= option can be used to limit the maximum swap
1761 usage under the unified cgroup hierarchy.
1763 * Support for the CPU controller in the unified cgroup hierarchy has
1764 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
1765 options. This controller requires out-of-tree patches for the kernel
1766 and the support is provisional.
1768 * Mount and automount units may now be created transiently
1769 (i.e. dynamically at runtime via the bus API, instead of requiring
1770 unit files in the file system).
1772 * systemd-mount is a new tool which may mount file systems – much like
1773 mount(8), optionally pulling in additional dependencies through
1774 transient .mount and .automount units. For example, this tool
1775 automatically runs fsck on a backing block device before mounting,
1776 and allows the automount logic to be used dynamically from the
1777 command line for establishing mount points. This tool is particularly
1778 useful when dealing with removable media, as it will ensure fsck is
1779 run – if necessary – before the first access and that the file system
1780 is quickly unmounted after each access by utilizing the automount
1781 logic. This maximizes the chance that the file system on the
1782 removable media stays in a clean state, and if it isn't in a clean
1783 state is fixed automatically.
1785 * LazyUnmount=yes option for mount units has been added to expose the
1786 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
1789 * /efi will be used as the mount point of the EFI boot partition, if
1790 the directory is present, and the mount point was not configured
1791 through other means (e.g. fstab). If /efi directory does not exist,
1792 /boot will be used as before. This makes it easier to automatically
1793 mount the EFI partition on systems where /boot is used for something
1796 * When operating on GPT disk images for containers, systemd-nspawn will
1797 now mount the ESP to /boot or /efi according to the same rules as PID
1798 1 running on a host. This allows tools like "bootctl" to operate
1799 correctly within such containers, in order to make container images
1800 bootable on physical systems.
1802 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
1804 * Two new user session targets have been added to support running
1805 graphical sessions under the systemd --user instance:
1806 graphical-session.target and graphical-session-pre.target. See
1807 systemd.special(7) for a description of how those targets should be
1810 * The vconsole initialization code has been significantly reworked to
1811 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
1812 support unicode keymaps. Font and keymap configuration will now be
1813 copied to all allocated virtual consoles.
1815 * FreeBSD's bhyve virtualization is now detected.
1817 * Information recorded in the journal for core dumps now includes the
1818 contents of /proc/mountinfo and the command line of the process at
1819 the top of the process hierarchy (which is usually the init process
1822 * systemd-journal-gatewayd learned the --directory= option to serve
1823 files from the specified location.
1825 * journalctl --root=… can be used to peruse the journal in the
1826 /var/log/ directories inside of a container tree. This is similar to
1827 the existing --machine= option, but does not require the container to
1830 * The hardware database has been extended to support
1831 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
1834 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
1835 specify the click rate for mice which include a horizontal wheel with
1836 a click rate that is different than the one for the vertical wheel.
1838 * systemd-run gained a new --wait option that makes service execution
1839 synchronous. (Specifically, the command will not return until the
1840 specified service binary exited.)
1842 * systemctl gained a new --wait option that causes the start command to
1843 wait until the units being started have terminated again.
1845 * A new journal output mode "short-full" has been added which displays
1846 timestamps with abbreviated English day names and adds a timezone
1847 suffix. Those timestamps include more information than the default
1848 "short" output mode, and can be passed directly to journalctl's
1849 --since= and --until= options.
1851 * /etc/resolv.conf will be bind-mounted into containers started by
1852 systemd-nspawn, if possible, so any changes to resolv.conf contents
1853 are automatically propagated to the container.
1855 * The number of instances for socket-activated services originating
1856 from a single IP address can be limited with
1857 MaxConnectionsPerSource=, extending the existing setting of
1860 * systemd-networkd gained support for vcan ("Virtual CAN") interface
1863 * .netdev and .network configuration can now be extended through
1866 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
1867 Segmentation Offload, Generic Receive Offload, Large Receive Offload
1868 can be enabled and disabled using the new UDPSegmentationOffload=,
1869 TCPSegmentationOffload=, GenericSegmentationOffload=,
1870 GenericReceiveOffload=, LargeReceiveOffload= options in the
1871 [Link] section of .link files.
1873 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
1874 Port VLAN ID can be configured for bridge devices using the new STP=,
1875 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
1876 section of .netdev files.
1878 * The route table to which routes received over DHCP or RA should be
1879 added can be configured with the new RouteTable= option in the [DHCP]
1880 and [IPv6AcceptRA] sections of .network files.
1882 * The Address Resolution Protocol can be disabled on links managed by
1883 systemd-networkd using the ARP=no setting in the [Link] section of
1886 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
1887 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
1888 encode information about the result and exit codes of the current
1889 service runtime cycle.
1891 * systemd-sysctl will now configure kernel parameters in the order
1892 they occur in the configuration files. This matches what sysctl
1893 has been traditionally doing.
1895 * kernel-install "plugins" that are executed to perform various
1896 tasks after a new kernel is added and before an old one is removed
1897 can now return a special value to terminate the procedure and
1898 prevent any later plugins from running.
1900 * Journald's SplitMode=login setting has been deprecated. It has been
1901 removed from documentation, and its use is discouraged. In a future
1902 release it will be completely removed, and made equivalent to current
1903 default of SplitMode=uid.
1905 * Storage=both option setting in /etc/systemd/coredump.conf has been
1906 removed. With fast LZ4 compression storing the core dump twice is not
1909 * The --share-system systemd-nspawn option has been replaced with an
1910 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
1911 this functionality is discouraged. In addition the variables
1912 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
1913 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
1914 individual namespaces.
1916 * "machinectl list" now shows the IP address of running containers in
1917 the output, as well as OS release information.
1919 * "loginctl list" now shows the TTY of each session in the output.
1921 * sd-bus gained new API calls sd_bus_track_set_recursive(),
1922 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
1923 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
1924 tracking objects in a "recursive" mode, where a single client can be
1925 counted multiple times, if it takes multiple references.
1927 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
1928 sd_bus_get_exit_on_disconnect(). They may be used to make a
1929 process using sd-bus automatically exit if the bus connection is
1932 * Bus clients of the service manager may now "pin" loaded units into
1933 memory, by taking an explicit reference on them. This is useful to
1934 ensure the client can retrieve runtime data about the service even
1935 after the service completed execution. Taking such a reference is
1936 available only for privileged clients and should be helpful to watch
1937 running services in a race-free manner, and in particular collect
1938 information about exit statuses and results.
1940 * The nss-resolve module has been changed to strictly return UNAVAIL
1941 when communication via D-Bus with resolved failed, and NOTFOUND when
1942 a lookup completed but was negative. This means it is now possible to
1943 neatly configure fallbacks using nsswitch.conf result checking
1944 expressions. Taking benefit of this, the new recommended
1945 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
1947 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
1949 * A new setting CtrlAltDelBurstAction= has been added to
1950 /etc/systemd/system.conf which may be used to configure the precise
1951 behaviour if the user on the console presses Ctrl-Alt-Del more often
1952 than 7 times in 2s. Previously this would unconditionally result in
1953 an expedited, immediate reboot. With this new setting the precise
1954 operation may be configured in more detail, and also turned off
1957 * In .netdev files two new settings RemoteChecksumTx= and
1958 RemoteChecksumRx= are now understood that permit configuring the
1959 remote checksumming logic for VXLAN networks.
1961 * The service manager learnt a new "invocation ID" concept for invoked
1962 services. Each runtime cycle of a service will get a new invocation
1963 ID (a 128bit random UUID) assigned that identifies the current
1964 run of the service uniquely and globally. A new invocation ID
1965 is generated each time a service starts up. The journal will store
1966 the invocation ID of a service along with any logged messages, thus
1967 making the invocation ID useful for matching the online runtime of a
1968 service with the offline log data it generated in a safe way without
1969 relying on synchronized timestamps. In many ways this new service
1970 invocation ID concept is similar to the kernel's boot ID concept that
1971 uniquely and globally identifies the runtime of each boot. The
1972 invocation ID of a service is passed to the service itself via an
1973 environment variable ($INVOCATION_ID). A new bus call
1974 GetUnitByInvocationID() has been added that is similar to GetUnit()
1975 but instead of retrieving the bus path for a unit by its name
1976 retrieves it by its invocation ID. The returned path is valid only as
1977 long as the passed invocation ID is current.
1979 * systemd-resolved gained a new "DNSStubListener" setting in
1980 resolved.conf. It either takes a boolean value or the special values
1981 "udp" and "tcp", and configures whether to enable the stub DNS
1982 listener on 127.0.0.53:53.
1984 * IP addresses configured via networkd may now carry additional
1985 configuration settings supported by the kernel. New options include:
1986 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
1987 PrefixRoute=, AutoJoin=.
1989 * The PAM configuration fragment file for "user@.service" shipped with
1990 systemd (i.e. the --user instance of systemd) has been stripped to
1991 the minimum necessary to make the system boot. Previously, it
1992 contained Fedora-specific stanzas that did not apply to other
1993 distributions. It is expected that downstream distributions add
1994 additional configuration lines, matching their needs to this file,
1995 using it only as rough template of what systemd itself needs. Note
1996 that this reduced fragment does not even include an invocation of
1997 pam_limits which most distributions probably want to add, even though
1998 systemd itself does not need it. (There's also the new build time
1999 option --with-pamconfdir=no to disable installation of the PAM
2002 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
2003 capability is now also dropped from its set (in addition to
2004 CAP_SYS_MKNOD as before).
2006 * In service unit files it is now possible to connect a specific named
2007 file descriptor with stdin/stdout/stdout of an executed service. The
2008 name may be specified in matching .socket units using the
2009 FileDescriptorName= setting.
2011 * A number of journal settings may now be configured on the kernel
2012 command line. Specifically, the following options are now understood:
2013 systemd.journald.max_level_console=,
2014 systemd.journald.max_level_store=,
2015 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
2016 systemd.journald.max_level_wall=.
2018 * "systemctl is-enabled --full" will now show by which symlinks a unit
2019 file is enabled in the unit dependency tree.
2021 * Support for VeraCrypt encrypted partitions has been added to the
2022 "cryptsetup" logic and /etc/crypttab.
2024 * systemd-detect-virt gained support for a new --private-users switch
2025 that checks whether the invoking processes are running inside a user
2026 namespace. Similar, a new special value "private-users" for the
2027 existing ConditionVirtualization= setting has been added, permitting
2028 skipping of specific units in user namespace environments.
2030 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
2031 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
2032 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
2033 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
2034 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
2035 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
2036 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
2037 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
2038 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
2039 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
2040 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
2041 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
2042 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
2043 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
2044 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
2045 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
2046 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
2047 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
2048 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
2049 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
2050 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
2051 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
2052 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
2053 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
2054 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
2055 Jędrzejewski-Szmek, Zeal Jagannatha
2057 — Santa Fe, 2016-11-03
2061 * In service units the various ExecXYZ= settings have been extended
2062 with an additional special character as first argument of the
2063 assigned value: if the character '+' is used the specified command
2064 line it will be run with full privileges, regardless of User=,
2065 Group=, CapabilityBoundingSet= and similar options. The effect is
2066 similar to the existing PermissionsStartOnly= option, but allows
2067 configuration of this concept for each executed command line
2070 * Services may now alter the service watchdog timeout at runtime by
2071 sending a WATCHDOG_USEC= message via sd_notify().
2073 * MemoryLimit= and related unit settings now optionally take percentage
2074 specifications. The percentage is taken relative to the amount of
2075 physical memory in the system (or in case of containers, the assigned
2076 amount of memory). This allows scaling service resources neatly with
2077 the amount of RAM available on the system. Similarly, systemd-logind's
2078 RuntimeDirectorySize= option now also optionally takes percentage
2081 * In similar fashion TasksMax= takes percentage values now, too. The
2082 value is taken relative to the configured maximum number of processes
2083 on the system. The per-service task maximum has been changed to 15%
2084 using this functionality. (Effectively this is an increase of 512 →
2085 4915 for service units, given the kernel's default pid_max setting.)
2087 * Calendar time specifications in .timer units now understand a ".."
2088 syntax for time ranges. Example: "4..7:10" may now be used for
2089 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
2092 * The InaccessableDirectories=, ReadOnlyDirectories= and
2093 ReadWriteDirectories= unit file settings have been renamed to
2094 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
2095 applied to all kinds of file nodes, and not just directories, with
2096 the exception of symlinks. Specifically these settings may now be
2097 used on block and character device nodes, UNIX sockets and FIFOS as
2098 well as regular files. The old names of these settings remain
2099 available for compatibility.
2101 * systemd will now log about all service processes it kills forcibly
2102 (using SIGKILL) because they remained after the clean shutdown phase
2103 of the service completed. This should help identifying services that
2104 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
2105 systemd-logind's configuration a similar log message is generated for
2106 processes killed at the end of each session due to this setting.
2108 * systemd will now set the $JOURNAL_STREAM environment variable for all
2109 services whose stdout/stderr are connected to the Journal (which
2110 effectively means by default: all services). The variable contains
2111 the device and inode number of the file descriptor used for
2112 stdout/stderr. This may be used by invoked programs to detect whether
2113 their stdout/stderr is connected to the Journal, in which case they
2114 can switch over to direct Journal communication, thus being able to
2115 pass extended, structured metadata along with their log messages. As
2116 one example, this is now used by glib's logging primitives.
2118 * When using systemd's default tmp.mount unit for /tmp, the mount point
2119 will now be established with the "nosuid" and "nodev" options. This
2120 avoids privilege escalation attacks that put traps and exploits into
2121 /tmp. However, this might cause problems if you e. g. put container
2122 images or overlays into /tmp; if you need this, override tmp.mount's
2123 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
2126 * systemd now supports the "memory" cgroup controller also on
2129 * The systemd-cgtop tool now optionally takes a control group path as
2130 command line argument. If specified, the control group list shown is
2131 limited to subgroups of that group.
2133 * The SystemCallFilter= unit file setting gained support for
2134 pre-defined, named system call filter sets. For example
2135 SystemCallFilter=@clock is now an effective way to make all clock
2136 changing-related system calls unavailable to a service. A number of
2137 similar pre-defined groups are defined. Writing system call filters
2138 for system services is simplified substantially with this new
2139 concept. Accordingly, all of systemd's own, long-running services now
2140 enable system call filtering based on this, by default.
2142 * A new service setting MemoryDenyWriteExecute= has been added, taking
2143 a boolean value. If turned on, a service may no longer create memory
2144 mappings that are writable and executable at the same time. This
2145 enhances security for services where this is enabled as it becomes
2146 harder to dynamically write and then execute memory in exploited
2147 service processes. This option has been enabled for all of systemd's
2148 own long-running services.
2150 * A new RestrictRealtime= service setting has been added, taking a
2151 boolean argument. If set the service's processes may no longer
2152 acquire realtime scheduling. This improves security as realtime
2153 scheduling may otherwise be used to easily freeze the system.
2155 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
2156 value. This may be used for requesting that the system manager inside
2157 of the container reports start-up completion to nspawn which then
2158 propagates this notification further to the service manager
2159 supervising nspawn itself. A related option NotifyReady= in .nspawn
2160 files has been added too. This functionality allows ordering of the
2161 start-up of multiple containers using the usual systemd ordering
2164 * machinectl gained a new command "stop" that is an alias for
2167 * systemd-resolved gained support for contacting DNS servers on
2168 link-local IPv6 addresses.
2170 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
2171 its caches. A method call for requesting the same operation has been
2172 added to the bus API too, and is made available via "systemd-resolve
2175 * systemd-resolve gained a new --status switch. If passed a brief
2176 summary of the used DNS configuration with per-interface information
2179 * resolved.conf gained a new Cache= boolean option, defaulting to
2180 on. If turned off local DNS caching is disabled. This comes with a
2181 performance penalty in particular when DNSSEC is enabled. Note that
2182 resolved disables its internal caching implicitly anyway, when the
2183 configured DNS server is on a host-local IP address such as ::1 or
2184 127.0.0.1, thus automatically avoiding double local caching.
2186 * systemd-resolved now listens on the local IP address 127.0.0.53:53
2187 for DNS requests. This improves compatibility with local programs
2188 that do not use the libc NSS or systemd-resolved's bus APIs for name
2189 resolution. This minimal DNS service is only available to local
2190 programs and does not implement the full DNS protocol, but enough to
2191 cover local DNS clients. A new, static resolv.conf file, listing just
2192 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
2193 now recommended to make /etc/resolv.conf a symlink to this file in
2194 order to route all DNS lookups to systemd-resolved, regardless if
2195 done via NSS, the bus API or raw DNS packets. Note that this local
2196 DNS service is not as fully featured as the libc NSS or
2197 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
2198 used to deliver link-local address information (as this implies
2199 sending a local interface index along), LLMNR/mDNS support via this
2200 interface is severely restricted. It is thus strongly recommended for
2201 all applications to use the libc NSS API or native systemd-resolved
2204 * systemd-networkd's bridge support learned a new setting
2205 VLANFiltering= for controlling VLAN filtering. Moreover a new section
2206 in .network files has been added for configuring VLAN bridging in
2207 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
2209 * systemd-networkd's IPv6 Router Advertisement code now makes use of
2210 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
2211 now be acquired without relying on DHCPv6. Two new options
2212 UseDomains= and UseDNS= have been added to configure this behaviour.
2214 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
2215 renamed IPv6AcceptRA=, without altering its behaviour. The old
2216 setting name remains available for compatibility reasons.
2218 * The systemd-networkd VTI/VTI6 tunneling support gained new options
2219 Key=, InputKey= and OutputKey=.
2221 * systemd-networkd gained support for VRF ("Virtual Routing Function")
2222 interface configuration.
2224 * "systemctl edit" may now be used to create new unit files by
2225 specifying the --force switch.
2227 * sd-event gained a new function sd_event_get_iteration() for
2228 requesting the current iteration counter of the event loop. It starts
2229 at zero and is increased by one with each event loop iteration.
2231 * A new rpm macro %systemd_ordering is provided by the macros.systemd
2232 file. It can be used in lieu of %systemd_requires in packages which
2233 don't use any systemd functionality and are intended to be installed
2234 in minimal containers without systemd present. This macro provides
2235 ordering dependencies to ensure that if the package is installed in
2236 the same rpm transaction as systemd, systemd will be installed before
2237 the scriptlets for the package are executed, allowing unit presets
2240 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
2241 been added to simplify packaging of generators.
2243 * The os-release file gained VERSION_CODENAME field for the
2244 distribution nickname (e.g. VERSION_CODENAME=woody).
2246 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
2247 can be set to disable parsing of metadata and the creation
2248 of persistent symlinks for that device.
2250 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
2251 to make them available to logged-in users has been reverted.
2253 * Much of the common code of the various systemd components is now
2254 built into an internal shared library libsystemd-shared-231.so
2255 (incorporating the systemd version number in the name, to be updated
2256 with future releases) that the components link to. This should
2257 decrease systemd footprint both in memory during runtime and on
2258 disk. Note that the shared library is not for public use, and is
2259 neither API nor ABI stable, but is likely to change with every new
2260 released update. Packagers need to make sure that binaries
2261 linking to libsystemd-shared.so are updated in step with the
2264 * Configuration for "mkosi" is now part of the systemd
2265 repository. mkosi is a tool to easily build legacy-free OS images,
2266 and is available on github: https://github.com/systemd/mkosi. If
2267 "mkosi" is invoked in the build tree a new raw OS image is generated
2268 incorporating the systemd sources currently being worked on and a
2269 clean, fresh distribution installation. The generated OS image may be
2270 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
2271 UEFI PC. This functionality is particularly useful to easily test
2272 local changes made to systemd in a pristine, defined environment. See
2273 doc/HACKING for details.
2275 * configure learned the --with-support-url= option to specify the
2276 distribution's bugtracker.
2278 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
2279 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
2280 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
2281 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
2282 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
2283 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
2284 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
2285 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
2286 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
2287 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
2288 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
2289 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
2290 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
2291 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
2292 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
2293 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
2294 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
2295 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
2296 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
2298 — Berlin, 2016-07-25
2302 * DNSSEC is now turned on by default in systemd-resolved (in
2303 "allow-downgrade" mode), but may be turned off during compile time by
2304 passing "--with-default-dnssec=no" to "configure" (and of course,
2305 during runtime with DNSSEC= in resolved.conf). We recommend
2306 downstreams to leave this on at least during development cycles and
2307 report any issues with the DNSSEC logic upstream. We are very
2308 interested in collecting feedback about the DNSSEC validator and its
2309 limitations in the wild. Note however, that DNSSEC support is
2310 probably nothing downstreams should turn on in stable distros just
2311 yet, as it might create incompatibilities with a few DNS servers and
2312 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
2313 automatically whenever we detect such incompatible setups, but there
2314 might be systems we do not cover yet. Hence: please help us testing
2315 the DNSSEC code, leave this on where you can, report back, but then
2316 again don't consider turning this on in your stable, LTS or
2317 production release just yet. (Note that you have to enable
2318 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
2319 and its DNSSEC mode for host name resolution from local
2322 * systemd-resolve conveniently resolves DANE records with the --tlsa
2323 option and OPENPGPKEY records with the --openpgp option. It also
2324 supports dumping raw DNS record data via the new --raw= switch.
2326 * systemd-logind will now by default terminate user processes that are
2327 part of the user session scope unit (session-XX.scope) when the user
2328 logs out. This behavior is controlled by the KillUserProcesses=
2329 setting in logind.conf, and the previous default of "no" is now
2330 changed to "yes". This means that user sessions will be properly
2331 cleaned up after, but additional steps are necessary to allow
2332 intentionally long-running processes to survive logout.
2334 While the user is logged in at least once, user@.service is running,
2335 and any service that should survive the end of any individual login
2336 session can be started at a user service or scope using systemd-run.
2337 systemd-run(1) man page has been extended with an example which shows
2338 how to run screen in a scope unit underneath user@.service. The same
2339 command works for tmux.
2341 After the user logs out of all sessions, user@.service will be
2342 terminated too, by default, unless the user has "lingering" enabled.
2343 To effectively allow users to run long-term tasks even if they are
2344 logged out, lingering must be enabled for them. See loginctl(1) for
2345 details. The default polkit policy was modified to allow users to
2346 set lingering for themselves without authentication.
2348 Previous defaults can be restored at compile time by the
2349 --without-kill-user-processes option to "configure".
2351 * systemd-logind gained new configuration settings SessionsMax= and
2352 InhibitorsMax=, both with a default of 8192. It will not register new
2353 user sessions or inhibitors above this limit.
2355 * systemd-logind will now reload configuration on SIGHUP.
2357 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
2358 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
2359 enable. Also, support for the "io" cgroup controller in the unified
2360 hierarchy has been added, so that the "memory", "pids" and "io" are
2361 now the controllers that are supported on the unified hierarchy.
2363 WARNING: it is not possible to use previous systemd versions with
2364 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
2365 is necessary to also update systemd in the initramfs if using the
2366 unified hierarchy. An updated SELinux policy is also required.
2368 * LLDP support has been extended, and both passive (receive-only) and
2369 active (sender) modes are supported. Passive mode ("routers-only") is
2370 enabled by default in systemd-networkd. Active LLDP mode is enabled
2371 by default for containers on the internal network. The "networkctl
2372 lldp" command may be used to list information gathered. "networkctl
2373 status" will also show basic LLDP information on connected peers now.
2375 * The IAID and DUID unique identifier sent in DHCP requests may now be
2376 configured for the system and each .network file managed by
2377 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
2379 * systemd-networkd gained support for configuring proxy ARP support for
2380 each interface, via the ProxyArp= setting in .network files. It also
2381 gained support for configuring the multicast querier feature of
2382 bridge devices, via the new MulticastQuerier= setting in .netdev
2383 files. Similarly, snooping on the IGMP traffic can be controlled
2384 via the new setting MulticastSnooping=.
2386 A new setting PreferredLifetime= has been added for addresses
2387 configured in .network file to configure the lifetime intended for an
2390 The systemd-networkd DHCP server gained the option EmitRouter=, which
2391 defaults to yes, to configure whether the DHCP Option 3 (Router)
2394 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
2395 systemd-socket-activate and installed into /usr/bin. It is now fully
2398 * systemd-journald now uses separate threads to flush changes to disk
2399 when closing journal files, thus reducing impact of slow disk I/O on
2400 logging performance.
2402 * The sd-journal API gained two new calls
2403 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
2404 can be used to open journal files using file descriptors instead of
2405 file or directory paths. sd_journal_open_container() has been
2406 deprecated, sd_journal_open_directory_fd() should be used instead
2407 with the flag SD_JOURNAL_OS_ROOT.
2409 * journalctl learned a new output mode "-o short-unix" that outputs log
2410 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
2411 UTC). It also gained support for a new --no-hostname setting to
2412 suppress the hostname column in the family of "short" output modes.
2414 * systemd-ask-password now optionally skips printing of the password to
2415 stdout with --no-output which can be useful in scripts.
2417 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
2418 (devices tagged with ID_MAKER_TOOL) are now tagged with
2419 "uaccess" and are available to logged in users.
2421 * The DeviceAllow= unit setting now supports specifiers (with "%").
2423 * "systemctl show" gained a new --value switch, which allows print a
2424 only the contents of a specific unit property, without also printing
2425 the property's name. Similar support was added to "show*" verbs
2426 of loginctl and machinectl that output "key=value" lists.
2428 * A new unit type "generated" was added for files dynamically generated
2429 by generator tools. Similarly, a new unit type "transient" is used
2430 for unit files created using the runtime API. "systemctl enable" will
2431 refuse to operate on such files.
2433 * A new command "systemctl revert" has been added that may be used to
2434 revert to the vendor version of a unit file, in case local changes
2435 have been made by adding drop-ins or overriding the unit file.
2437 * "machinectl clean" gained a new verb to automatically remove all or
2438 just hidden container images.
2440 * systemd-tmpfiles gained support for a new line type "e" for emptying
2441 directories, if they exist, without creating them if they don't.
2443 * systemd-nspawn gained support for automatically patching the UID/GIDs
2444 of the owners and the ACLs of all files and directories in a
2445 container tree to match the UID/GID user namespacing range selected
2446 for the container invocation. This mode is enabled via the new
2447 --private-users-chown switch. It also gained support for
2448 automatically choosing a free, previously unused UID/GID range when
2449 starting a container, via the new --private-users=pick setting (which
2450 implies --private-users-chown). Together, these options for the first
2451 time make user namespacing for nspawn containers fully automatic and
2452 thus deployable. The systemd-nspawn@.service template unit file has
2453 been changed to use this functionality by default.
2455 * systemd-nspawn gained a new --network-zone= switch, that allows
2456 creating ad-hoc virtual Ethernet links between multiple containers,
2457 that only exist as long as at least one container referencing them is
2458 running. This allows easy connecting of multiple containers with a
2459 common link that implements an Ethernet broadcast domain. Each of
2460 these network "zones" may be named relatively freely by the user, and
2461 may be referenced by any number of containers, but each container may
2462 only reference one of these "zones". On the lower level, this is
2463 implemented by an automatically managed bridge network interface for
2464 each zone, that is created when the first container referencing its
2465 zone is created and removed when the last one referencing its zone
2468 * The default start timeout may now be configured on the kernel command
2469 line via systemd.default_timeout_start_sec=. It was already
2470 configurable via the DefaultTimeoutStartSec= option in
2471 /etc/systemd/system.conf.
2473 * Socket units gained a new TriggerLimitIntervalSec= and
2474 TriggerLimitBurst= setting to configure a limit on the activation
2475 rate of the socket unit.
2477 * The LimitNICE= setting now optionally takes normal UNIX nice values
2478 in addition to the raw integer limit value. If the specified
2479 parameter is prefixed with "+" or "-" and is in the range -20..19 the
2480 value is understood as UNIX nice value. If not prefixed like this it
2481 is understood as raw RLIMIT_NICE limit.
2483 * Note that the effect of the PrivateDevices= unit file setting changed
2484 slightly with this release: the per-device /dev file system will be
2485 mounted read-only from this version on, and will have "noexec"
2486 set. This (minor) change of behavior might cause some (exceptional)
2487 legacy software to break, when PrivateDevices=yes is set for its
2488 service. Please leave PrivateDevices= off if you run into problems
2491 * systemd-bootchart has been split out to a separate repository:
2492 https://github.com/systemd/systemd-bootchart
2494 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
2495 merged into the kernel in its current form.
2497 * The compatibility libraries libsystemd-daemon.so,
2498 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
2499 which have been deprecated since systemd-209 have been removed along
2500 with the corresponding pkg-config files. All symbols provided by
2501 those libraries are provided by libsystemd.so.
2503 * The Capabilities= unit file setting has been removed (it is ignored
2504 for backwards compatibility). AmbientCapabilities= and
2505 CapabilityBoundingSet= should be used instead.
2507 * A new special target has been added, initrd-root-device.target,
2508 which creates a synchronization point for dependencies of the root
2509 device in early userspace. Initramfs builders must ensure that this
2510 target is now included in early userspace.
2512 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
2513 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
2514 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
2515 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
2516 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
2517 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
2518 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
2519 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
2520 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
2521 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
2522 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
2523 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
2524 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
2525 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
2526 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
2527 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
2528 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
2529 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
2530 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
2531 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
2532 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
2533 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
2534 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
2535 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
2536 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
2539 — Fairfax, 2016-05-21
2543 * The systemd-resolved DNS resolver service has gained a substantial
2544 set of new features, most prominently it may now act as a DNSSEC
2545 validating stub resolver. DNSSEC mode is currently turned off by
2546 default, but is expected to be turned on by default in one of the
2547 next releases. For now, we invite everybody to test the DNSSEC logic
2548 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
2549 service also gained a full set of D-Bus interfaces, including calls
2550 to configure DNS and DNSSEC settings per link (for use by external
2551 network management software). systemd-resolved and systemd-networkd
2552 now distinguish between "search" and "routing" domains. The former
2553 are used to qualify single-label names, the latter are used purely
2554 for routing lookups within certain domains to specific links.
2555 resolved now also synthesizes RRs for all entries from /etc/hosts.
2557 * The systemd-resolve tool (which is a client utility for
2558 systemd-resolved) has been improved considerably and is now fully
2559 supported and documented. Hence it has moved from /usr/lib/systemd to
2562 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
2565 * The coredump collection logic has been reworked: when a coredump is
2566 collected it is now written to disk, compressed and processed
2567 (including stacktrace extraction) from a new instantiated service
2568 systemd-coredump@.service, instead of directly from the
2569 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
2570 processing large coredumps can take up a substantial amount of
2571 resources and time, and this previously happened entirely outside of
2572 systemd's service supervision. With the new logic the core_pattern
2573 hook only does minimal metadata collection before passing off control
2574 to the new instantiated service, which is configured with a time
2575 limit, a nice level and other settings to minimize negative impact on
2576 the rest of the system. Also note that the new logic will honour the
2577 RLIMIT_CORE setting of the crashed process, which now allows users
2578 and processes to turn off coredumping for their processes by setting
2581 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
2582 and all forked processes by default. Previously, PID 1 would leave
2583 the setting at "0" for all processes, as set by the kernel. Note that
2584 the resource limit traditionally has no effect on the generated
2585 coredumps on the system if the /proc/sys/kernel/core_pattern hook
2586 logic is used. Since the limit is now honoured (see above) its
2587 default has been changed so that the coredumping logic is enabled by
2588 default for all processes, while allowing specific opt-out.
2590 * When the stacktrace is extracted from processes of system users, this
2591 is now done as "systemd-coredump" user, in order to sandbox this
2592 potentially security sensitive parsing operation. (Note that when
2593 processing coredumps of normal users this is done under the user ID
2594 of process that crashed, as before.) Packagers should take notice
2595 that it is now necessary to create the "systemd-coredump" system user
2596 and group at package installation time.
2598 * The systemd-activate socket activation testing tool gained support
2599 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
2600 and --seqpacket switches. It also has been extended to support both
2601 new-style and inetd-style file descriptor passing. Use the new
2602 --inetd switch to request inetd-style file descriptor passing.
2604 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
2605 variable, which takes a boolean value. If set to false, ANSI color
2606 output is disabled in the tools even when run on a terminal that
2609 * The VXLAN support in networkd now supports two new settings
2610 DestinationPort= and PortRange=.
2612 * A new systemd.machine_id= kernel command line switch has been added,
2613 that may be used to set the machine ID in /etc/machine-id if it is
2614 not initialized yet. This command line option has no effect if the
2615 file is already initialized.
2617 * systemd-nspawn gained a new --as-pid2 switch that invokes any
2618 specified command line as PID 2 rather than PID 1 in the
2619 container. In this mode PID 1 is a minimal stub init process that
2620 implements the special POSIX and Linux semantics of PID 1 regarding
2621 signal and child process management. Note that this stub init process
2622 is implemented in nspawn itself and requires no support from the
2623 container image. This new logic is useful to support running
2624 arbitrary commands in the container, as normal processes are
2625 generally not prepared to run as PID 1.
2627 * systemd-nspawn gained a new --chdir= switch for setting the current
2628 working directory for the process started in the container.
2630 * "journalctl /dev/sda" will now output all kernel log messages for
2631 specified device from the current boot, in addition to all devices
2632 that are parents of it. This should make log output about devices
2633 pretty useful, as long as kernel drivers attach enough metadata to
2634 the log messages. (The usual SATA drivers do.)
2636 * The sd-journal API gained two new calls
2637 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
2638 that report whether log data from /run or /var has been found.
2640 * journalctl gained a new switch "--fields" that prints all journal
2641 record field names currently in use in the journal. This is backed
2642 by two new sd-journal API calls sd_journal_enumerate_fields() and
2643 sd_journal_restart_fields().
2645 * Most configurable timeouts in systemd now expect an argument of
2646 "infinity" to turn them off, instead of "0" as before. The semantics
2647 from now on is that a timeout of "0" means "now", and "infinity"
2648 means "never". To maintain backwards compatibility, "0" continues to
2649 turn off previously existing timeout settings.
2651 * "systemctl reload-or-try-restart" has been renamed to "systemctl
2652 try-reload-or-restart" to clarify what it actually does: the "try"
2653 logic applies to both reloading and restarting, not just restarting.
2654 The old name continues to be accepted for compatibility.
2656 * On boot-up, when PID 1 detects that the system clock is behind the
2657 release date of the systemd version in use, the clock is now set
2658 to the latter. Previously, this was already done in timesyncd, in order
2659 to avoid running with clocks set to the various clock epochs such as
2660 1902, 1938 or 1970. With this change the logic is now done in PID 1
2661 in addition to timesyncd during early boot-up, so that it is enforced
2662 before the first process is spawned by systemd. Note that the logic
2663 in timesyncd remains, as it is more comprehensive and ensures
2664 clock monotonicity by maintaining a persistent timestamp file in
2665 /var. Since /var is generally not available in earliest boot or the
2666 initrd, this part of the logic remains in timesyncd, and is not done
2669 * Support for tweaking details in net_cls.class_id through the
2670 NetClass= configuration directive has been removed, as the kernel
2671 people have decided to deprecate that controller in cgroup v2.
2672 Userspace tools such as nftables are moving over to setting rules
2673 that are specific to the full cgroup path of a task, which obsoletes
2674 these controllers anyway. The NetClass= directive is kept around for
2675 legacy compatibility reasons. For a more in-depth description of the
2676 kernel change, please refer to the respective upstream commit:
2678 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
2680 * A new service setting RuntimeMaxSec= has been added that may be used
2681 to specify a maximum runtime for a service. If the timeout is hit, the
2682 service is terminated and put into a failure state.
2684 * A new service setting AmbientCapabilities= has been added. It allows
2685 configuration of additional Linux process capabilities that are
2686 passed to the activated processes. This is only available on very
2689 * The process resource limit settings in service units may now be used
2690 to configure hard and soft limits individually.
2692 * The various libsystemd APIs such as sd-bus or sd-event now publicly
2693 expose support for gcc's __attribute__((cleanup())) C extension.
2694 Specifically, for many object destructor functions alternative
2695 versions have been added that have names suffixed with "p" and take a
2696 pointer to a pointer to the object to destroy, instead of just a
2697 pointer to the object itself. This is useful because these destructor
2698 functions may be used directly as parameters to the cleanup
2699 construct. Internally, systemd has been a heavy user of this GCC
2700 extension for a long time, and with this change similar support is
2701 now available to consumers of the library outside of systemd. Note
2702 that by using this extension in your sources compatibility with old
2703 and strictly ANSI compatible C compilers is lost. However, all gcc or
2704 LLVM versions of recent years support this extension.
2706 * Timer units gained support for a new setting RandomizedDelaySec= that
2707 allows configuring some additional randomized delay to the configured
2708 time. This is useful to spread out timer events to avoid load peaks in
2709 clusters or larger setups.
2711 * Calendar time specifications now support sub-second accuracy.
2713 * Socket units now support listening on SCTP and UDP-lite protocol
2716 * The sd-event API now comes with a full set of man pages.
2718 * Older versions of systemd contained experimental support for
2719 compressing journal files and coredumps with the LZ4 compressor that
2720 was not compatible with the lz4 binary (due to API limitations of the
2721 lz4 library). This support has been removed; only support for files
2722 compatible with the lz4 binary remains. This LZ4 logic is now
2723 officially supported and no longer considered experimental.
2725 * The dkr image import logic has been removed again from importd. dkr's
2726 micro-services focus doesn't fit into the machine image focus of
2727 importd, and quickly got out of date with the upstream dkr API.
2729 * Creation of the /run/lock/lockdev/ directory was dropped from
2730 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
2731 been available for many years. If you still need this, you need to
2732 create your own tmpfiles.d config file with:
2734 d /run/lock/lockdev 0775 root lock -
2736 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
2737 and RebootArgument= have been moved from the [Service] section of
2738 unit files to [Unit], and they are now supported on all unit types,
2739 not just service units. Of course, elogind will continue to
2740 understand these settings also at the old location, in order to
2741 maintain compatibility.
2743 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
2744 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
2745 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
2746 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
2747 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
2748 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
2749 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
2750 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
2751 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
2752 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
2753 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
2754 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
2755 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
2756 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
2757 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
2758 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
2759 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
2760 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
2761 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2763 — Berlin, 2016-02-11
2767 * A number of properties previously only settable in unit
2768 files are now also available as properties to set when
2769 creating transient units programmatically via the bus, as it
2770 is exposed with systemd-run's --property=
2771 setting. Specifically, these are: SyslogIdentifier=,
2772 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
2773 EnvironmentFile=, ReadWriteDirectories=,
2774 ReadOnlyDirectories=, InaccessibleDirectories=,
2775 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
2777 * When creating transient services via the bus API it is now
2778 possible to pass in a set of file descriptors to use as
2779 STDIN/STDOUT/STDERR for the invoked process.
2781 * Slice units may now be created transiently via the bus APIs,
2782 similar to the way service and scope units may already be
2783 created transiently.
2785 * Wherever systemd expects a calendar timestamp specification
2786 (like in journalctl's --since= and --until= switches) UTC
2787 timestamps are now supported. Timestamps suffixed with "UTC"
2788 are now considered to be in Universal Time Coordinated
2789 instead of the local timezone. Also, timestamps may now
2790 optionally be specified with sub-second accuracy. Both of
2791 these additions also apply to recurring calendar event
2792 specification, such as OnCalendar= in timer units.
2794 * journalctl gained a new "--sync" switch that asks the
2795 journal daemon to write all so far unwritten log messages to
2796 disk and sync the files, before returning.
2798 * systemd-tmpfiles learned two new line types "q" and "Q" that
2799 operate like "v", but also set up a basic btrfs quota
2800 hierarchy when used on a btrfs file system with quota
2803 * tmpfiles' "v", "q" and "Q" will now create a plain directory
2804 instead of a subvolume (even on a btrfs file system) if the
2805 root directory is a plain directory, and not a
2806 subvolume. This should simplify things with certain chroot()
2807 environments which are not aware of the concept of btrfs
2810 * systemd-detect-virt gained a new --chroot switch to detect
2811 whether execution takes place in a chroot() environment.
2813 * CPUAffinity= now takes CPU index ranges in addition to
2816 * The various memory-related resource limit settings (such as
2817 LimitAS=) now understand the usual K, M, G, ... suffixes to
2818 the base of 1024 (IEC). Similar, the time-related resource
2819 limit settings understand the usual min, h, day, ...
2822 * There's a new system.conf setting DefaultTasksMax= to
2823 control the default TasksMax= setting for services and
2824 scopes running on the system. (TasksMax= is the primary
2825 setting that exposes the "pids" cgroup controller on systemd
2826 and was introduced in the previous systemd release.) The
2827 setting now defaults to 512, which means services that are
2828 not explicitly configured otherwise will only be able to
2829 create 512 processes or threads at maximum, from this
2830 version on. Note that this means that thread- or
2831 process-heavy services might need to be reconfigured to set
2832 TasksMax= to a higher value. It is sufficient to set
2833 TasksMax= in these specific unit files to a higher value, or
2834 even "infinity". Similar, there's now a logind.conf setting
2835 UserTasksMax= that defaults to 4096 and limits the total
2836 number of processes or tasks each user may own
2837 concurrently. nspawn containers also have the TasksMax=
2838 value set by default now, to 8192. Note that all of this
2839 only has an effect if the "pids" cgroup controller is
2840 enabled in the kernel. The general benefit of these changes
2841 should be a more robust and safer system, that provides a
2842 certain amount of per-service fork() bomb protection.
2844 * systemd-nspawn gained the new --network-veth-extra= switch
2845 to define additional and arbitrarily-named virtual Ethernet
2846 links between the host and the container.
2848 * A new service execution setting PassEnvironment= has been
2849 added that allows importing select environment variables
2850 from PID1's environment block into the environment block of
2853 * Timer units gained support for a new RemainAfterElapse=
2854 setting which takes a boolean argument. It defaults to on,
2855 exposing behaviour unchanged to previous releases. If set to
2856 off, timer units are unloaded after they elapsed if they
2857 cannot elapse again. This is particularly useful for
2858 transient timer units, which shall not stay around longer
2859 than until they first elapse.
2861 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
2862 default now (the kernel default is 16). This is beneficial
2863 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
2864 allows substantially larger numbers of queued
2865 datagrams. This should increase the capability of systemd to
2866 parallelize boot-up, as logging and sd_notify() are unlikely
2867 to stall execution anymore. If you need to change the value
2868 from the new defaults, use the usual sysctl.d/ snippets.
2870 * The compression framing format used by the journal or
2871 coredump processing has changed to be in line with what the
2872 official LZ4 tools generate. LZ4 compression support in
2873 systemd was considered unsupported previously, as the format
2874 was not compatible with the normal tools. With this release
2875 this has changed now, and it is hence safe for downstream
2876 distributions to turn it on. While not compressing as well
2877 as the XZ, LZ4 is substantially faster, which makes
2878 it a good default choice for the compression logic in the
2879 journal and in coredump handling.
2881 * Any reference to /etc/mtab has been dropped from
2882 systemd. The file has been obsolete since a while, but
2883 systemd refused to work on systems where it was incorrectly
2884 set up (it should be a symlink or non-existent). Please make
2885 sure to update to util-linux 2.27.1 or newer in conjunction
2886 with this systemd release, which also drops any reference to
2887 /etc/mtab. If you maintain a distribution make sure that no
2888 software you package still references it, as this is a
2889 likely source of bugs. There's also a glibc bug pending,
2890 asking for removal of any reference to this obsolete file:
2892 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
2894 Note that only util-linux versions built with
2895 --enable-libmount-force-mountinfo are supported.
2897 * Support for the ".snapshot" unit type has been removed. This
2898 feature turned out to be little useful and little used, and
2899 has now been removed from the core and from systemctl.
2901 * The dependency types RequiresOverridable= and
2902 RequisiteOverridable= have been removed from systemd. They
2903 have been used only very sparingly to our knowledge and
2904 other options that provide a similar effect (such as
2905 systemctl --mode=ignore-dependencies) are much more useful
2906 and commonly used. Moreover, they were only half-way
2907 implemented as the option to control behaviour regarding
2908 these dependencies was never added to systemctl. By removing
2909 these dependency types the execution engine becomes a bit
2910 simpler. Unit files that use these dependencies should be
2911 changed to use the non-Overridable dependency types
2912 instead. In fact, when parsing unit files with these
2913 options, that's what systemd will automatically convert them
2914 too, but it will also warn, asking users to fix the unit
2915 files accordingly. Removal of these dependency types should
2916 only affect a negligible number of unit files in the wild.
2918 * Behaviour of networkd's IPForward= option changed
2919 (again). It will no longer maintain a per-interface setting,
2920 but propagate one way from interfaces where this is enabled
2921 to the global kernel setting. The global setting will be
2922 enabled when requested by a network that is set up, but
2923 never be disabled again. This change was made to make sure
2924 IPv4 and IPv6 behaviour regarding packet forwarding is
2925 similar (as the Linux IPv6 stack does not support
2926 per-interface control of this setting) and to minimize
2929 * In unit files the behaviour of %u, %U, %h, %s has
2930 changed. These specifiers will now unconditionally resolve
2931 to the various user database fields of the user that the
2932 systemd instance is running as, instead of the user
2933 configured in the specific unit via User=. Note that this
2934 effectively doesn't change much, as resolving of these
2935 specifiers was already turned off in the --system instance
2936 of systemd, as we cannot do NSS lookups from PID 1. In the
2937 --user instance of systemd these specifiers where correctly
2938 resolved, but hardly made any sense, since the user instance
2939 lacks privileges to do user switches anyway, and User= is
2940 hence useless. Moreover, even in the --user instance of
2941 systemd behaviour was awkward as it would only take settings
2942 from User= assignment placed before the specifier into
2943 account. In order to unify and simplify the logic around
2944 this the specifiers will now always resolve to the
2945 credentials of the user invoking the manager (which in case
2946 of PID 1 is the root user).
2948 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
2949 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
2950 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
2951 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
2952 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
2953 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
2954 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
2955 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
2956 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
2957 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
2958 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
2959 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
2960 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
2961 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
2964 — Berlin, 2015-11-18
2968 * systemd now depends on util-linux v2.27. More specifically,
2969 the newly added mount monitor feature in libmount now
2970 replaces systemd's former own implementation.
2972 * libmount mandates /etc/mtab not to be regular file, and
2973 systemd now enforces this condition at early boot.
2974 /etc/mtab has been deprecated and warned about for a very
2975 long time, so systems running systemd should already have
2976 stopped having this file around as anything else than a
2977 symlink to /proc/self/mounts.
2979 * Support for the "pids" cgroup controller has been added. It
2980 allows accounting the number of tasks in a cgroup and
2981 enforcing limits on it. This adds two new setting
2982 TasksAccounting= and TasksMax= to each unit, as well as a
2983 global option DefaultTasksAccounting=.
2985 * Support for the "net_cls" cgroup controller has been added.
2986 It allows assigning a net class ID to each task in the
2987 cgroup, which can then be used in firewall rules and traffic
2988 shaping configurations. Note that the kernel netfilter net
2989 class code does not currently work reliably for ingress
2990 packets on unestablished sockets.
2992 This adds a new config directive called NetClass= to CGroup
2993 enabled units. Allowed values are positive numbers for fixed
2994 assignments and "auto" for picking a free value
2997 * 'systemctl is-system-running' now returns 'offline' if the
2998 system is not booted with systemd. This command can now be
2999 used as a substitute for 'systemd-notify --booted'.
3001 * Watchdog timeouts have been increased to 3 minutes for all
3002 in-tree service files. Apparently, disk IO issues are more
3003 frequent than we hoped, and user reported >1 minute waiting
3006 * 'machine-id-commit' functionality has been merged into
3007 'machine-id-setup --commit'. The separate binary has been
3010 * The WorkingDirectory= directive in unit files may now be set
3011 to the special value '~'. In this case, the working
3012 directory is set to the home directory of the user
3013 configured in User=.
3015 * "machinectl shell" will now open the shell in the home
3016 directory of the selected user by default.
3018 * The CrashChVT= configuration file setting is renamed to
3019 CrashChangeVT=, following our usual logic of not
3020 abbreviating unnecessarily. The old directive is still
3021 supported for compat reasons. Also, this directive now takes
3022 an integer value between 1 and 63, or a boolean value. The
3023 formerly supported '-1' value for disabling stays around for
3026 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
3027 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
3028 RootDirectory= properties can now be set for transient
3031 * The systemd-analyze tool gained a new "set-log-target" verb
3032 to change the logging target the system manager logs to
3033 dynamically during runtime. This is similar to how
3034 "systemd-analyze set-log-level" already changes the log
3037 * In nspawn /sys is now mounted as tmpfs, with only a selected
3038 set of subdirectories mounted in from the real sysfs. This
3039 enhances security slightly, and is useful for ensuring user
3040 namespaces work correctly.
3042 * Support for USB FunctionFS activation has been added. This
3043 allows implementation of USB gadget services that are
3044 activated as soon as they are requested, so that they don't
3045 have to run continuously, similar to classic socket
3048 * The "systemctl exit" command now optionally takes an
3049 additional parameter that sets the exit code to return from
3050 the systemd manager when exiting. This is only relevant when
3051 running the systemd user instance, or when running the
3052 system instance in a container.
3054 * sd-bus gained the new API calls sd_bus_path_encode_many()
3055 and sd_bus_path_decode_many() that allow easy encoding and
3056 decoding of multiple identifier strings inside a D-Bus
3057 object path. Another new call sd_bus_default_flush_close()
3058 has been added to flush and close per-thread default
3061 * systemd-cgtop gained support for a -M/--machine= switch to
3062 show the control groups within a certain container only.
3064 * "systemctl kill" gained support for an optional --fail
3065 switch. If specified the requested operation will fail of no
3066 processes have been killed, because the unit had no
3067 processes attached, or similar.
3069 * A new systemd.crash_reboot=1 kernel command line option has
3070 been added that triggers a reboot after crashing. This can
3071 also be set through CrashReboot= in systemd.conf.
3073 * The RuntimeDirectory= setting now understands unit
3074 specifiers like %i or %f.
3076 * A new (still internal) library API sd-ipv4acd has been added,
3077 that implements address conflict detection for IPv4. It's
3078 based on code from sd-ipv4ll, and will be useful for
3079 detecting DHCP address conflicts.
3081 * File descriptors passed during socket activation may now be
3082 named. A new API sd_listen_fds_with_names() is added to
3083 access the names. The default names may be overridden,
3084 either in the .socket file using the FileDescriptorName=
3085 parameter, or by passing FDNAME= when storing the file
3086 descriptors using sd_notify().
3088 * systemd-networkd gained support for:
3090 - Setting the IPv6 Router Advertisement settings via
3091 IPv6AcceptRouterAdvertisements= in .network files.
3093 - Configuring the HelloTimeSec=, MaxAgeSec= and
3094 ForwardDelaySec= bridge parameters in .netdev files.
3096 - Configuring PreferredSource= for static routes in
3099 * The "ask-password" framework used to query for LUKS harddisk
3100 passwords or SSL passwords during boot gained support for
3101 caching passwords in the kernel keyring, if it is
3102 available. This makes sure that the user only has to type in
3103 a passphrase once if there are multiple objects to unlock
3104 with the same one. Previously, such password caching was
3105 available only when Plymouth was used; this moves the
3106 caching logic into the systemd codebase itself. The
3107 "systemd-ask-password" utility gained a new --keyname=
3108 switch to control which kernel keyring key to use for
3109 caching a password in. This functionality is also useful for
3110 enabling display managers such as gdm to automatically
3111 unlock the user's GNOME keyring if its passphrase, the
3112 user's password and the harddisk password are the same, if
3113 gdm-autologin is used.
3115 * When downloading tar or raw images using "machinectl
3116 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
3117 file is now also downloaded, if it is available and stored
3118 next to the image file.
3120 * Units of type ".socket" gained a new boolean setting
3121 Writable= which is only useful in conjunction with
3122 ListenSpecial=. If true, enables opening the specified
3123 special file in O_RDWR mode rather than O_RDONLY mode.
3125 * systemd-rfkill has been reworked to become a singleton
3126 service that is activated through /dev/rfkill on each rfkill
3127 state change and saves the settings to disk. This way,
3128 systemd-rfkill is now compatible with devices that exist
3129 only intermittendly, and even restores state if the previous
3130 system shutdown was abrupt rather than clean.
3132 * The journal daemon gained support for vacuuming old journal
3133 files controlled by the number of files that shall remain,
3134 in addition to the already existing control by size and by
3135 date. This is useful as journal interleaving performance
3136 degrades with too many separate journal files, and allows
3137 putting an effective limit on them. The new setting defaults
3138 to 100, but this may be changed by setting SystemMaxFiles=
3139 and RuntimeMaxFiles= in journald.conf. Also, the
3140 "journalctl" tool gained the new --vacuum-files= switch to
3141 manually vacuum journal files to leave only the specified
3142 number of files in place.
3144 * udev will now create /dev/disk/by-path links for ATA devices
3145 on kernels where that is supported.
3147 * Galician, Serbian, Turkish and Korean translations were added.
3149 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
3150 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
3151 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
3152 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
3153 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
3154 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
3155 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
3156 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
3157 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
3158 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
3159 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
3160 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
3161 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
3162 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
3163 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
3164 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
3165 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
3166 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
3168 — Berlin, 2015-10-07
3172 * The DHCP implementation of systemd-networkd gained a set of
3175 - The DHCP server now supports emitting DNS and NTP
3176 information. It may be enabled and configured via
3177 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
3178 and NTP information is enabled, but no servers are
3179 configured, the corresponding uplink information (if there
3180 is any) is propagated.
3182 - Server and client now support transmission and reception
3183 of timezone information. It can be configured via the
3184 newly introduced network options UseTimezone=,
3185 EmitTimezone=, and Timezone=. Transmission of timezone
3186 information is enabled between host and containers by
3187 default now: the container will change its local timezone
3188 to what the host has set.
3190 - Lease timeouts can now be configured via
3191 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
3193 - The DHCP server improved on the stability of
3194 leases. Clients are more likely to get the same lease
3195 information back, even if the server loses state.
3197 - The DHCP server supports two new configuration options to
3198 control the lease address pool metrics, PoolOffset= and
3201 * The encapsulation limit of tunnels in systemd-networkd may
3202 now be configured via 'EncapsulationLimit='. It allows
3203 modifying the maximum additional levels of encapsulation
3204 that are permitted to be prepended to a packet.
3206 * systemd now supports the concept of user buses replacing
3207 session buses, if used with dbus-1.10 (and enabled via dbus
3208 --enable-user-session). It previously only supported this on
3209 kdbus-enabled systems, and this release expands this to
3210 'dbus-daemon' systems.
3212 * systemd-networkd now supports predictable interface names
3215 * systemd now optionally supports the new Linux kernel
3216 "unified" control group hierarchy. If enabled via the kernel
3217 command-line option 'systemd.unified_cgroup_hierarchy=1',
3218 systemd will try to mount the unified cgroup hierarchy
3219 directly on /sys/fs/cgroup. If not enabled, or not
3220 available, systemd will fall back to the legacy cgroup
3221 hierarchy setup, as before. Host system and containers can
3222 mix and match legacy and unified hierarchies as they
3223 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
3224 environment variable to individually select the hierarchy to
3225 use for executed containers. By default, nspawn will use the
3226 unified hierarchy for the containers if the host uses the
3227 unified hierarchy, and the legacy hierarchy otherwise.
3228 Please note that at this point the unified hierarchy is an
3229 experimental kernel feature and is likely to change in one
3230 of the next kernel releases. Therefore, it should not be
3231 enabled by default in downstream distributions yet. The
3232 minimum required kernel version for the unified hierarchy to
3233 work is 4.2. Note that when the unified hierarchy is used
3234 for the first time delegated access to controllers is
3235 safe. Because of this systemd-nspawn containers will get
3236 access to controllers now, as will systemd user
3237 sessions. This means containers and user sessions may now
3238 manage their own resources, partitioning up what the system
3241 * A new special scope unit "init.scope" has been introduced
3242 that encapsulates PID 1 of the system. It may be used to
3243 determine resource usage and enforce resource limits on PID
3244 1 itself. PID 1 hence moved out of the root of the control
3247 * The cgtop tool gained support for filtering out kernel
3248 threads when counting tasks in a control group. Also, the
3249 count of processes is now recursively summed up by
3250 default. Two options -k and --recursive= have been added to
3251 revert to old behaviour. The tool has also been updated to
3252 work correctly in containers now.
3254 * systemd-nspawn's --bind= and --bind-ro= options have been
3255 extended to allow creation of non-recursive bind mounts.
3257 * libsystemd gained two new calls sd_pid_get_cgroup() and
3258 sd_peer_get_cgroup() which return the control group path of
3259 a process or peer of a connected AF_UNIX socket. This
3260 function call is particularly useful when implementing
3261 delegated subtrees support in the control group hierarchy.
3263 * The "sd-event" event loop API of libsystemd now supports
3264 correct dequeuing of real-time signals, without losing
3267 * When systemd requests a PolicyKit decision when managing
3268 units it will now add additional fields to the request,
3269 including unit name and desired operation. This enables more
3270 powerful PolicyKit policies, that make decisions depending
3271 on these parameters.
3273 * nspawn learnt support for .nspawn settings files, that may
3274 accompany the image files or directories of containers, and
3275 may contain additional settings for the container. This is
3276 an alternative to configuring container parameters via the
3277 nspawn command line.
3279 Contributions from: Cristian Rodríguez, Daniel Mack, David
3280 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
3281 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
3282 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
3283 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
3284 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
3285 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
3286 Andersen, Tom Gundersen, Torstein Husebø
3288 — Berlin, 2015-09-08
3292 * machinectl gained a new verb 'shell' which opens a fresh
3293 shell on the target container or the host. It is similar to
3294 the existing 'login' command of machinectl, but spawns the
3295 shell directly without prompting for username or
3296 password. The pseudo machine '.host' now refers to the local
3297 host and is used by default. Hence, 'machinectl shell' can
3298 be used as replacement for 'su -' which spawns a session as
3299 a fresh systemd unit in a way that is fully isolated from
3300 the originating session.
3302 * systemd-networkd learned to cope with private-zone DHCP
3303 options and allows other programs to query the values.
3305 * SELinux access control when enabling/disabling units is no
3306 longer enforced with this release. The previous
3307 implementation was incorrect, and a new corrected
3308 implementation is not yet available. As unit file operations
3309 are still protected via PolicyKit and D-Bus policy this is
3310 not a security problem. Yet, distributions which care about
3311 optimal SELinux support should probably not stabilize on
3314 * sd-bus gained support for matches of type "arg0has=", that
3315 test for membership of strings in string arrays sent in bus
3318 * systemd-resolved now dumps the contents of its DNS and LLMNR
3319 caches to the logs on reception of the SIGUSR1 signal. This
3320 is useful to debug DNS behaviour.
3322 * The coredumpctl tool gained a new --directory= option to
3323 operate on journal files in a specific directory.
3325 * "systemctl reboot" and related commands gained a new
3326 "--message=" option which may be used to set a free-text
3327 wall message when shutting down or rebooting the
3328 system. This message is also logged, which is useful for
3329 figuring out the reason for a reboot or shutdown a
3332 * The "systemd-resolve-host" tool's -i switch now takes
3333 network interface numbers as alternative to interface names.
3335 * A new unit file setting for services has been introduced:
3336 UtmpMode= allows configuration of how precisely systemd
3337 handles utmp and wtmp entries for the service if this is
3338 enabled. This allows writing services that appear similar to
3339 user sessions in the output of the "w", "who", "last" and
3342 * systemd-resolved will now locally synthesize DNS resource
3343 records for the "localhost" and "gateway" domains as well as
3344 the local hostname. This should ensure that clients querying
3345 RRs via resolved will get similar results as those going via
3346 NSS, if nss-myhostname is enabled.
3348 Contributions from: Alastair Hughes, Alex Crawford, Daniel
3349 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
3350 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
3351 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
3352 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
3353 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
3354 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
3355 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
3356 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
3357 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
3358 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
3359 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3361 — Berlin, 2015-08-27
3365 * The systemd-efi-boot-generator functionality was merged into
3366 systemd-gpt-auto-generator.
3368 * systemd-networkd now supports Group Policy for vxlan
3369 devices. It can be enabled via the new boolean configuration
3370 option called 'GroupPolicyExtension='.
3372 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
3373 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
3374 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
3376 — Berlin, 2015-07-31
3380 * The python-systemd code has been removed from the systemd repository.
3381 A new repository has been created which accommodates the code from
3382 now on, and we kindly ask distributions to create a separate package
3383 for this: https://github.com/systemd/python-systemd
3385 * The systemd daemon will now reload its main configuration
3386 (/etc/systemd/system.conf) on daemon-reload.
3388 * sd-dhcp now exposes vendor specific extensions via
3389 sd_dhcp_lease_get_vendor_specific().
3391 * systemd-networkd gained a number of new configuration options.
3393 - A new boolean configuration option for TAP devices called
3394 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
3395 device, thus allowing to send and receive GSO packets.
3397 - A new tunnel configuration option called 'CopyDSCP='.
3398 If enabled, the DSCP field of ip6 tunnels is copied into the
3399 decapsulated packet.
3401 - A set of boolean bridge configuration options were added.
3402 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
3403 and 'UnicastFlood=' are now parsed by networkd and applied to the
3404 respective bridge link device via the respective IFLA_BRPORT_*
3407 - A new string configuration option to override the hostname sent
3408 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
3409 is true, networkd will use the configured hostname instead of the
3410 system hostname when sending DHCP requests.
3412 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
3413 networkd will configure the IPv6 flow-label of the tunnel device
3414 according to RFC2460.
3416 - The 'macvtap' virtual network devices are now supported, similar to
3417 the already supported 'macvlan' devices.
3419 * systemd-resolved now implements RFC5452 to improve resilience against
3420 cache poisoning. Additionally, source port randomization is enabled
3421 by default to further protect against DNS spoofing attacks.
3423 * nss-mymachines now supports translating UIDs and GIDs of running
3424 containers with user-namespaces enabled. If a container 'foo'
3425 translates a host uid 'UID' to the container uid 'TUID', then
3426 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
3427 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
3428 mapped as 'vg-foo-TGID'.
3430 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
3431 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
3432 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
3433 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
3434 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
3435 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
3436 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
3437 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
3438 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
3439 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
3441 — Berlin, 2015-07-29
3445 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
3446 There are no known issues with current sysfs, and udev does not need
3447 or should be used to work around such bugs.
3449 * udev does no longer enable USB HID power management. Several reports
3450 indicate, that some devices cannot handle that setting.
3452 * The udev accelerometer helper was removed. The functionality
3453 is now fully included in iio-sensor-proxy. But this means,
3454 older iio-sensor-proxy versions will no longer provide
3455 accelerometer/orientation data with this systemd version.
3456 Please upgrade iio-sensor-proxy to version 1.0.
3458 * networkd gained a new configuration option IPv6PrivacyExtensions=
3459 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
3460 for Stateless Address") on selected networks.
3462 * For the sake of fewer build-time dependencies and less code in the
3463 main repository, the python bindings are about to be removed in the
3464 next release. A new repository has been created which accommodates
3465 the code from now on, and we kindly ask distributions to create a
3466 separate package for this. The removal will take place in v223.
3468 https://github.com/systemd/python-systemd
3470 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
3471 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
3472 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
3473 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
3474 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
3475 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
3476 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
3477 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
3478 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3479 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3481 — Berlin, 2015-07-07
3485 * The sd-bus.h and sd-event.h APIs have now been declared
3486 stable and have been added to the official interface of
3487 libsystemd.so. sd-bus implements an alternative D-Bus client
3488 library, that is relatively easy to use, very efficient and
3489 supports both classic D-Bus as well as kdbus as transport
3490 backend. sd-event is a generic event loop abstraction that
3491 is built around Linux epoll, but adds features such as event
3492 prioritization or efficient timer handling. Both APIs are good
3493 choices for C programs looking for a bus and/or event loop
3494 implementation that is minimal and does not have to be
3495 portable to other kernels.
3497 * kdbus support is no longer compile-time optional. It is now
3498 always built-in. However, it can still be disabled at
3499 runtime using the kdbus=0 kernel command line setting, and
3500 that setting may be changed to default to off, by specifying
3501 --disable-kdbus at build-time. Note though that the kernel
3502 command line setting has no effect if the kdbus.ko kernel
3503 module is not installed, in which case kdbus is (obviously)
3504 also disabled. We encourage all downstream distributions to
3505 begin testing kdbus by adding it to the kernel images in the
3506 development distributions, and leaving kdbus support in
3509 * The minimal required util-linux version has been bumped to
3512 * Support for chkconfig (--enable-chkconfig) was removed in
3513 favor of calling an abstraction tool
3514 /lib/systemd/systemd-sysv-install. This needs to be
3515 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
3516 in README for details.
3518 * If there's a systemd unit and a SysV init script for the
3519 same service name, and the user executes "systemctl enable"
3520 for it (or a related call), then this will now enable both
3521 (or execute the related operation on both), not just the
3524 * The libudev API documentation has been converted from gtkdoc
3527 * gudev has been removed from the systemd tree, it is now an
3530 * The systemd-cgtop tool learnt a new --raw switch to generate
3531 "raw" (machine parsable) output.
3533 * networkd's IPForwarding= .network file setting learnt the
3534 new setting "kernel", which ensures that networkd does not
3535 change the IP forwarding sysctl from the default kernel
3538 * The systemd-logind bus API now exposes a new boolean
3539 property "Docked" that reports whether logind considers the
3540 system "docked", i.e. connected to a docking station or not.
3542 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
3543 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
3544 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
3545 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
3546 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
3547 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
3548 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
3549 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
3550 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
3551 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
3552 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
3553 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
3554 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
3555 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3556 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
3557 Fink, Zbigniew Jędrzejewski-Szmek
3559 — Berlin, 2015-06-19
3563 * The gudev library has been extracted into a separate repository
3564 available at: https://git.gnome.org/browse/libgudev/
3565 It is now managed as part of the Gnome project. Distributions
3566 are recommended to pass --disable-gudev to systemd and use
3567 gudev from the Gnome project instead. gudev is still included
3568 in systemd, for now. It will be removed soon, though. Please
3569 also see the announcement-thread on systemd-devel:
3570 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
3572 * systemd now exposes a CPUUsageNSec= property for each
3573 service unit on the bus, that contains the overall consumed
3574 CPU time of a service (the sum of what each process of the
3575 service consumed). This value is only available if
3576 CPUAccounting= is turned on for a service, and is then shown
3577 in the "systemctl status" output.
3579 * Support for configuring alternative mappings of the old SysV
3580 runlevels to systemd targets has been removed. They are now
3581 hardcoded in a way that runlevels 2, 3, 4 all map to
3582 multi-user.target and 5 to graphical.target (which
3583 previously was already the default behaviour).
3585 * The auto-mounter logic gained support for mount point
3586 expiry, using a new TimeoutIdleSec= setting in .automount
3587 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
3589 * The EFI System Partition (ESP) as mounted to /boot by
3590 systemd-efi-boot-generator will now be unmounted
3591 automatically after 2 minutes of not being used. This should
3592 minimize the risk of ESP corruptions.
3594 * New /etc/fstab options x-systemd.requires= and
3595 x-systemd.requires-mounts-for= are now supported to express
3596 additional dependencies for mounts. This is useful for
3597 journalling file systems that support external journal
3598 devices or overlay file systems that require underlying file
3599 systems to be mounted.
3601 * systemd does not support direct live-upgrades (via systemctl
3602 daemon-reexec) from versions older than v44 anymore. As no
3603 distribution we are aware of shipped such old versions in a
3604 stable release this should not be problematic.
3606 * When systemd forks off a new per-connection service instance
3607 it will now set the $REMOTE_ADDR environment variable to the
3608 remote IP address, and $REMOTE_PORT environment variable to
3609 the remote IP port. This behaviour is similar to the
3610 corresponding environment variables defined by CGI.
3612 * systemd-networkd gained support for uplink failure
3613 detection. The BindCarrier= option allows binding interface
3614 configuration dynamically to the link sense of other
3615 interfaces. This is useful to achieve behaviour like in
3618 * systemd-networkd gained support for configuring the DHCP
3619 client identifier to use when requesting leases.
3621 * systemd-networkd now has a per-network UseNTP= option to
3622 configure whether NTP server information acquired via DHCP
3623 is passed on to services like systemd-timesyncd.
3625 * systemd-networkd gained support for vti6 tunnels.
3627 * Note that systemd-networkd manages the sysctl variable
3628 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
3629 it is configured for since v219. The variable controls IP
3630 forwarding, and is a per-interface alternative to the global
3631 /proc/sys/net/ipv[46]/ip_forward. This setting is
3632 configurable in the IPForward= option, which defaults to
3633 "no". This means if networkd is used for an interface it is
3634 no longer sufficient to set the global sysctl option to turn
3635 on IP forwarding! Instead, the .network file option
3636 IPForward= needs to be turned on! Note that the
3637 implementation of this behaviour was broken in v219 and has
3640 * Many bonding and vxlan options are now configurable in
3643 * systemd-nspawn gained a new --property= setting to set unit
3644 properties for the container scope. This is useful for
3645 setting resource parameters (e.g. "CPUShares=500") on
3646 containers started from the command line.
3648 * systemd-nspawn gained a new --private-users= switch to make
3649 use of user namespacing available on recent Linux kernels.
3651 * systemd-nspawn may now be called as part of a shell pipeline
3652 in which case the pipes used for stdin and stdout are passed
3653 directly to the process invoked in the container, without
3654 indirection via a pseudo tty.
3656 * systemd-nspawn gained a new switch to control the UNIX
3657 signal to use when killing the init process of the container
3660 * systemd-nspawn gained a new --overlay= switch for mounting
3661 overlay file systems into the container using the new kernel
3664 * When a container image is imported via systemd-importd and
3665 the host file system is not btrfs, a loopback block device
3666 file is created in /var/lib/machines.raw with a btrfs file
3667 system inside. It is then mounted to /var/lib/machines to
3668 enable btrfs features for container management. The loopback
3669 file and btrfs file system is grown as needed when container
3670 images are imported via systemd-importd.
3672 * systemd-machined/systemd-importd gained support for btrfs
3673 quota, to enforce container disk space limits on disk. This
3674 is exposed in "machinectl set-limit".
3676 * systemd-importd now can import containers from local .tar,
3677 .raw and .qcow2 images, and export them to .tar and .raw. It
3678 can also import dkr v2 images now from the network (on top
3681 * systemd-importd gained support for verifying downloaded
3682 images with gpg2 (previously only gpg1 was supported).
3684 * systemd-machined, systemd-logind, systemd: most bus calls
3685 are now accessible to unprivileged processes via
3686 PolicyKit. Also, systemd-logind will now allow users to kill
3687 their own sessions without further privileges or
3690 * systemd-shutdownd has been removed. This service was
3691 previously responsible for implementing scheduled shutdowns
3692 as exposed in /usr/bin/shutdown's time parameter. This
3693 functionality has now been moved into systemd-logind and is
3694 accessible via a bus interface.
3696 * "systemctl reboot" gained a new switch --firmware-setup that
3697 can be used to reboot into the EFI firmware setup, if that
3698 is available. systemd-logind now exposes an API on the bus
3699 to trigger such reboots, in case graphical desktop UIs want
3700 to cover this functionality.
3702 * "systemctl enable", "systemctl disable" and "systemctl mask"
3703 now support a new "--now" switch. If specified the units
3704 that are enabled will also be started, and the ones
3705 disabled/masked also stopped.
3707 * The Gummiboot EFI boot loader tool has been merged into
3708 systemd, and renamed to "systemd-boot". The bootctl tool has been
3709 updated to support systemd-boot.
3711 * An EFI kernel stub has been added that may be used to create
3712 kernel EFI binaries that contain not only the actual kernel,
3713 but also an initrd, boot splash, command line and OS release
3714 information. This combined binary can then be signed as a
3715 single image, so that the firmware can verify it all in one
3716 step. systemd-boot has special support for EFI binaries created
3717 like this and can extract OS release information from them
3718 and show them in the boot menu. This functionality is useful
3719 to implement cryptographically verified boot schemes.
3721 * Optional support has been added to systemd-fsck to pass
3722 fsck's progress report to an AF_UNIX socket in the file
3725 * udev will no longer create device symlinks for all block
3726 devices by default. A blacklist for excluding special block
3727 devices from this logic has been turned into a whitelist
3728 that requires picking block devices explicitly that require
3731 * A new (currently still internal) API sd-device.h has been
3732 added to libsystemd. This modernized API is supposed to
3733 replace libudev eventually. In fact, already much of libudev
3734 is now just a wrapper around sd-device.h.
3736 * A new hwdb database for storing metadata about pointing
3737 stick devices has been added.
3739 * systemd-tmpfiles gained support for setting file attributes
3740 similar to the "chattr" tool with new 'h' and 'H' lines.
3742 * systemd-journald will no longer unconditionally set the
3743 btrfs NOCOW flag on new journal files. This is instead done
3744 with tmpfiles snippet using the new 'h' line type. This
3745 allows easy disabling of this logic, by masking the
3746 journal-nocow.conf tmpfiles file.
3748 * systemd-journald will now translate audit message types to
3749 human readable identifiers when writing them to the
3750 journal. This should improve readability of audit messages.
3752 * The LUKS logic gained support for the offset= and skip=
3753 options in /etc/crypttab, as previously implemented by
3756 * /usr/lib/os-release gained a new optional field VARIANT= for
3757 distributions that support multiple variants (such as a
3758 desktop edition, a server edition, ...)
3760 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
3761 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
3762 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
3763 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
3764 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
3765 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
3766 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
3767 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
3768 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
3769 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
3770 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
3771 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
3772 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
3773 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
3774 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
3775 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
3776 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
3777 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
3778 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
3779 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
3780 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
3781 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
3782 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
3783 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
3784 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
3785 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
3786 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
3788 — Berlin, 2015-05-22
3792 * Introduce a new API "sd-hwdb.h" for querying the hardware
3793 metadata database. With this minimal interface one can query
3794 and enumerate the udev hwdb, decoupled from the old libudev
3795 library. libudev's interface for this is now only a wrapper
3796 around sd-hwdb. A new tool systemd-hwdb has been added to
3797 interface with and update the database.
3799 * When any of systemd's tools copies files (for example due to
3800 tmpfiles' C lines) a btrfs reflink will attempted first,
3801 before bytewise copying is done.
3803 * systemd-nspawn gained a new --ephemeral switch. When
3804 specified a btrfs snapshot is taken of the container's root
3805 directory, and immediately removed when the container
3806 terminates again. Thus, a container can be started whose
3807 changes never alter the container's root directory, and are
3808 lost on container termination. This switch can also be used
3809 for starting a container off the root file system of the
3810 host without affecting the host OS. This switch is only
3811 available on btrfs file systems.
3813 * systemd-nspawn gained a new --template= switch. It takes the
3814 path to a container tree to use as template for the tree
3815 specified via --directory=, should that directory be
3816 missing. This allows instantiating containers dynamically,
3817 on first run. This switch is only available on btrfs file
3820 * When a .mount unit refers to a mount point on which multiple
3821 mounts are stacked, and the .mount unit is stopped all of
3822 the stacked mount points will now be unmounted until no
3823 mount point remains.
3825 * systemd now has an explicit notion of supported and
3826 unsupported unit types. Jobs enqueued for unsupported unit
3827 types will now fail with an "unsupported" error code. More
3828 specifically .swap, .automount and .device units are not
3829 supported in containers, .busname units are not supported on
3830 non-kdbus systems. .swap and .automount are also not
3831 supported if their respective kernel compile time options
3834 * machinectl gained support for two new "copy-from" and
3835 "copy-to" commands for copying files from a running
3836 container to the host or vice versa.
3838 * machinectl gained support for a new "bind" command to bind
3839 mount host directories into local containers. This is
3840 currently only supported for nspawn containers.
3842 * networkd gained support for configuring bridge forwarding
3843 database entries (fdb) from .network files.
3845 * A new tiny daemon "systemd-importd" has been added that can
3846 download container images in tar, raw, qcow2 or dkr formats,
3847 and make them available locally in /var/lib/machines, so
3848 that they can run as nspawn containers. The daemon can GPG
3849 verify the downloads (not supported for dkr, since it has no
3850 provisions for verifying downloads). It will transparently
3851 decompress bz2, xz, gzip compressed downloads if necessary,
3852 and restore sparse files on disk. The daemon uses privilege
3853 separation to ensure the actual download logic runs with
3854 fewer privileges than the daemon itself. machinectl has
3855 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
3856 make the functionality of importd available to the
3857 user. With this in place the Fedora and Ubuntu "Cloud"
3858 images can be downloaded and booted as containers unmodified
3859 (the Fedora images lack the appropriate GPG signature files
3860 currently, so they cannot be verified, but this will change
3861 soon, hopefully). Note that downloading images is currently
3862 only fully supported on btrfs.
3864 * machinectl is now able to list container images found in
3865 /var/lib/machines, along with some metadata about sizes of
3866 disk and similar. If the directory is located on btrfs and
3867 quota is enabled, this includes quota display. A new command
3868 "image-status" has been added that shows additional
3869 information about images.
3871 * machinectl is now able to clone container images
3872 efficiently, if the underlying file system (btrfs) supports
3873 it, with the new "machinectl clone" command. It also
3874 gained commands for renaming and removing images, as well as
3875 marking them read-only or read-write (supported also on
3876 legacy file systems).
3878 * networkd gained support for collecting LLDP network
3879 announcements, from hardware that supports this. This is
3880 shown in networkctl output.
3882 * systemd-run gained support for a new -t (--pty) switch for
3883 invoking a binary on a pty whose input and output is
3884 connected to the invoking terminal. This allows executing
3885 processes as system services while interactively
3886 communicating with them via the terminal. Most interestingly
3887 this is supported across container boundaries. Invoking
3888 "systemd-run -t /bin/bash" is an alternative to running a
3889 full login session, the difference being that the former
3890 will not register a session, nor go through the PAM session
3893 * tmpfiles gained support for a new "v" line type for creating
3894 btrfs subvolumes. If the underlying file system is a legacy
3895 file system, this automatically degrades to creating a
3896 normal directory. Among others /var/lib/machines is now
3897 created like this at boot, should it be missing.
3899 * The directory /var/lib/containers/ has been deprecated and
3900 been replaced by /var/lib/machines. The term "machines" has
3901 been used in the systemd context as generic term for both
3902 VMs and containers, and hence appears more appropriate for
3903 this, as the directory can also contain raw images bootable
3906 * systemd-nspawn when invoked with -M but without --directory=
3907 or --image= is now capable of searching for the container
3908 root directory, subvolume or disk image automatically, in
3909 /var/lib/machines. systemd-nspawn@.service has been updated
3910 to make use of this, thus allowing it to be used for raw
3913 * A new machines.target unit has been introduced that is
3914 supposed to group all containers/VMs invoked as services on
3915 the system. systemd-nspawn@.service has been updated to
3916 integrate with that.
3918 * machinectl gained a new "start" command, for invoking a
3919 container as a service. "machinectl start foo" is mostly
3920 equivalent to "systemctl start systemd-nspawn@foo.service",
3921 but handles escaping in a nicer way.
3923 * systemd-nspawn will now mount most of the cgroupfs tree
3924 read-only into each container, with the exception of the
3925 container's own subtree in the name=systemd hierarchy.
3927 * journald now sets the special FS_NOCOW file flag for its
3928 journal files. This should improve performance on btrfs, by
3929 avoiding heavy fragmentation when journald's write-pattern
3930 is used on COW file systems. It degrades btrfs' data
3931 integrity guarantees for the files to the same levels as for
3932 ext3/ext4 however. This should be OK though as journald does
3933 its own data integrity checks and all its objects are
3934 checksummed on disk. Also, journald should handle btrfs disk
3935 full events a lot more gracefully now, by processing SIGBUS
3936 errors, and not relying on fallocate() anymore.
3938 * When journald detects that journal files it is writing to
3939 have been deleted it will immediately start new journal
3942 * systemd now provides a way to store file descriptors
3943 per-service in PID 1. This is useful for daemons to ensure
3944 that fds they require are not lost during a daemon
3945 restart. The fds are passed to the daemon on the next
3946 invocation in the same way socket activation fds are
3947 passed. This is now used by journald to ensure that the
3948 various sockets connected to all the system's stdout/stderr
3949 are not lost when journald is restarted. File descriptors
3950 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
3951 an extension to sd_notify(). Note that a limit is enforced
3952 on the number of fds a service can store in PID 1, and it
3953 defaults to 0, so that no fds may be stored, unless this is
3954 explicitly turned on.
3956 * The default TERM variable to use for units connected to a
3957 terminal, when no other value is explicitly is set is now
3958 vt220 rather than vt102. This should be fairly safe still,
3959 but allows PgUp/PgDn work.
3961 * The /etc/crypttab option header= as known from Debian is now
3964 * "loginctl user-status" and "loginctl session-status" will
3965 now show the last 10 lines of log messages of the
3966 user/session following the status output. Similar,
3967 "machinectl status" will show the last 10 log lines
3968 associated with a virtual machine or container
3969 service. (Note that this is usually not the log messages
3970 done in the VM/container itself, but simply what the
3971 container manager logs. For nspawn this includes all console
3974 * "loginctl session-status" without further argument will now
3975 show the status of the session of the caller. Similar,
3976 "lock-session", "unlock-session", "activate",
3977 "enable-linger", "disable-linger" may now be called without
3978 session/user parameter in which case they apply to the
3979 caller's session/user.
3981 * An X11 session scriptlet is now shipped that uploads
3982 $DISPLAY and $XAUTHORITY into the environment of the systemd
3983 --user daemon if a session begins. This should improve
3984 compatibility with X11 enabled applications run as systemd
3987 * Generators are now subject to masking via /etc and /run, the
3988 same way as unit files.
3990 * networkd .network files gained support for configuring
3991 per-link IPv4/IPv6 packet forwarding as well as IPv4
3992 masquerading. This is by default turned on for veth links to
3993 containers, as registered by systemd-nspawn. This means that
3994 nspawn containers run with --network-veth will now get
3995 automatic routed access to the host's networks without any
3996 further configuration or setup, as long as networkd runs on
3999 * systemd-nspawn gained the --port= (-p) switch to expose TCP
4000 or UDP posts of a container on the host. With this in place
4001 it is possible to run containers with private veth links
4002 (--network-veth), and have their functionality exposed on
4003 the host as if their services were running directly on the
4006 * systemd-nspawn's --network-veth switch now gained a short
4007 version "-n", since with the changes above it is now truly
4008 useful out-of-the-box. The systemd-nspawn@.service has been
4009 updated to make use of it too by default.
4011 * systemd-nspawn will now maintain a per-image R/W lock, to
4012 ensure that the same image is not started more than once
4013 writable. (It's OK to run an image multiple times
4014 simultaneously in read-only mode.)
4016 * systemd-nspawn's --image= option is now capable of
4017 dissecting and booting MBR and GPT disk images that contain
4018 only a single active Linux partition. Previously it
4019 supported only GPT disk images with proper GPT type
4020 IDs. This allows running cloud images from major
4021 distributions directly with systemd-nspawn, without
4024 * In addition to collecting mouse dpi data in the udev
4025 hardware database, there's now support for collecting angle
4026 information for mouse scroll wheels. The database is
4027 supposed to guarantee similar scrolling behavior on mice
4028 that it knows about. There's also support for collecting
4029 information about Touchpad types.
4031 * udev's input_id built-in will now also collect touch screen
4032 dimension data and attach it to probed devices.
4034 * /etc/os-release gained support for a Distribution Privacy
4037 * networkd gained support for creating "ipvlan", "gretap",
4038 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
4040 * systemd-tmpfiles gained support for "a" lines for setting
4043 * systemd-nspawn will now mount /tmp in the container to
4044 tmpfs, automatically.
4046 * systemd now exposes the memory.usage_in_bytes cgroup
4047 attribute and shows it for each service in the "systemctl
4048 status" output, if available.
4050 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
4051 immediate reboot is triggered. This useful if shutdown is
4052 hung and is unable to complete, to expedite the
4053 operation. Note that this kind of reboot will still unmount
4054 all file systems, and hence should not result in fsck being
4057 * A .device unit for an optical block device will now be
4058 considered active only when a medium is in the drive. Also,
4059 mount units are now bound to their backing devices thus
4060 triggering automatic unmounting when devices become
4061 unavailable. With this in place systemd will now
4062 automatically unmount left-over mounts when a CD-ROM is
4063 ejected or an USB stick is yanked from the system.
4065 * networkd-wait-online now has support for waiting for
4066 specific interfaces only (with globbing), and for giving up
4067 after a configurable timeout.
4069 * networkd now exits when idle. It will be automatically
4070 restarted as soon as interfaces show up, are removed or
4071 change state. networkd will stay around as long as there is
4072 at least one DHCP state machine or similar around, that keep
4075 * networkd may now configure IPv6 link-local addressing in
4076 addition to IPv4 link-local addressing.
4078 * The IPv6 "token" for use in SLAAC may now be configured for
4079 each .network interface in networkd.
4081 * Routes configured with networkd may now be assigned a scope
4084 * networkd's [Match] sections now support globbing and lists
4085 of multiple space-separated matches per item.
4087 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
4088 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
4089 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
4090 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
4091 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
4092 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
4093 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
4094 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
4095 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
4096 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
4097 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
4098 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
4099 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
4100 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
4101 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
4102 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
4103 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
4104 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
4105 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
4106 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
4107 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
4108 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
4109 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
4110 Hoffmann, Zbigniew Jędrzejewski-Szmek
4112 — Berlin, 2015-02-16
4116 * When querying unit file enablement status (for example via
4117 "systemctl is-enabled"), a new state "indirect" is now known
4118 which indicates that a unit might not be enabled itself, but
4119 another unit listed in its Also= setting might be.
4121 * Similar to the various existing ConditionXYZ= settings for
4122 units, there are now matching AssertXYZ= settings. While
4123 failing conditions cause a unit to be skipped, but its job
4124 to succeed, failing assertions declared like this will cause
4125 a unit start operation and its job to fail.
4127 * hostnamed now knows a new chassis type "embedded".
4129 * systemctl gained a new "edit" command. When used on a unit
4130 file, this allows extending unit files with .d/ drop-in
4131 configuration snippets or editing the full file (after
4132 copying it from /usr/lib to /etc). This will invoke the
4133 user's editor (as configured with $EDITOR), and reload the
4134 modified configuration after editing.
4136 * "systemctl status" now shows the suggested enablement state
4137 for a unit, as declared in the (usually vendor-supplied)
4138 system preset files.
4140 * nss-myhostname will now resolve the single-label host name
4141 "gateway" to the locally configured default IP routing
4142 gateways, ordered by their metrics. This assigns a stable
4143 name to the used gateways, regardless which ones are
4144 currently configured. Note that the name will only be
4145 resolved after all other name sources (if nss-myhostname is
4146 configured properly) and should hence not negatively impact
4147 systems that use the single-label host name "gateway" in
4150 * systemd-inhibit now allows filtering by mode when listing
4153 * Scope and service units gained a new "Delegate" boolean
4154 property, which, when set, allows processes running inside the
4155 unit to further partition resources. This is primarily
4156 useful for systemd user instances as well as container
4159 * journald will now pick up audit messages directly from
4160 the kernel, and log them like any other log message. The
4161 audit fields are split up and fully indexed. This means that
4162 journalctl in many ways is now a (nicer!) alternative to
4163 ausearch, the traditional audit client. Note that this
4164 implements only a minimal audit client. If you want the
4165 special audit modes like reboot-on-log-overflow, please use
4166 the traditional auditd instead, which can be used in
4167 parallel to journald.
4169 * The ConditionSecurity= unit file option now understands the
4170 special string "audit" to check whether auditing is
4173 * journalctl gained two new commands --vacuum-size= and
4174 --vacuum-time= to delete old journal files until the
4175 remaining ones take up no more than the specified size on disk,
4176 or are not older than the specified time.
4178 * A new, native PPPoE library has been added to sd-network,
4179 systemd's library of light-weight networking protocols. This
4180 library will be used in a future version of networkd to
4181 enable PPPoE communication without an external pppd daemon.
4183 * The busctl tool now understands a new "capture" verb that
4184 works similar to "monitor", but writes a packet capture
4185 trace to STDOUT that can be redirected to a file which is
4186 compatible with libcap's capture file format. This can then
4187 be loaded in Wireshark and similar tools to inspect bus
4190 * The busctl tool now understands a new "tree" verb that shows
4191 the object trees of a specific service on the bus, or of all
4194 * The busctl tool now understands a new "introspect" verb that
4195 shows all interfaces and members of objects on the bus,
4196 including their signature and values. This is particularly
4197 useful to get more information about bus objects shown by
4198 the new "busctl tree" command.
4200 * The busctl tool now understands new verbs "call",
4201 "set-property" and "get-property" for invoking bus method
4202 calls, setting and getting bus object properties in a
4205 * busctl gained a new --augment-creds= argument that controls
4206 whether the tool shall augment credential information it
4207 gets from the bus with data from /proc, in a possibly
4210 * nspawn's --link-journal= switch gained two new values
4211 "try-guest" and "try-host" that work like "guest" and
4212 "host", but do not fail if the host has no persistent
4213 journalling enabled. -j is now equivalent to
4214 --link-journal=try-guest.
4216 * macvlan network devices created by nspawn will now have
4217 stable MAC addresses.
4219 * A new SmackProcessLabel= unit setting has been added, which
4220 controls the SMACK security label processes forked off by
4221 the respective unit shall use.
4223 * If compiled with --enable-xkbcommon, systemd-localed will
4224 verify x11 keymap settings by compiling the given keymap. It
4225 will spew out warnings if the compilation fails. This
4226 requires libxkbcommon to be installed.
4228 * When a coredump is collected, a larger number of metadata
4229 fields is now collected and included in the journal records
4230 created for it. More specifically, control group membership,
4231 environment variables, memory maps, working directory,
4232 chroot directory, /proc/$PID/status, and a list of open file
4233 descriptors is now stored in the log entry.
4235 * The udev hwdb now contains DPI information for mice. For
4238 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
4240 * All systemd programs that read standalone configuration
4241 files in /etc now also support a corresponding series of
4242 .conf.d configuration directories in /etc/, /run/,
4243 /usr/local/lib/, /usr/lib/, and (if configured with
4244 --enable-split-usr) /lib/. In particular, the following
4245 configuration files now have corresponding configuration
4246 directories: system.conf user.conf, logind.conf,
4247 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
4248 resolved.conf, timesyncd.conf, journal-remote.conf, and
4249 journal-upload.conf. Note that distributions should use the
4250 configuration directories in /usr/lib/; the directories in
4251 /etc/ are reserved for the system administrator.
4253 * systemd-rfkill will no longer take the rfkill device name
4254 into account when storing rfkill state on disk, as the name
4255 might be dynamically assigned and not stable. Instead, the
4256 ID_PATH udev variable combined with the rfkill type (wlan,
4257 bluetooth, ...) is used.
4259 * A new service systemd-machine-id-commit.service has been
4260 added. When used on systems where /etc is read-only during
4261 boot, and /etc/machine-id is not initialized (but an empty
4262 file), this service will copy the temporary machine ID
4263 created as replacement into /etc after the system is fully
4264 booted up. This is useful for systems that are freshly
4265 installed with a non-initialized machine ID, but should get
4266 a fixed machine ID for subsequent boots.
4268 * networkd's .netdev files now provide a large set of
4269 configuration parameters for VXLAN devices. Similarly, the
4270 bridge port cost parameter is now configurable in .network
4271 files. There's also new support for configuring IP source
4272 routing. networkd .link files gained support for a new
4273 OriginalName= match that is useful to match against the
4274 original interface name the kernel assigned. .network files
4275 may include MTU= and MACAddress= fields for altering the MTU
4276 and MAC address while being connected to a specific network
4279 * The LUKS logic gained supported for configuring
4280 UUID-specific key files. There's also new support for naming
4281 LUKS device from the kernel command line, using the new
4282 luks.name= argument.
4284 * Timer units may now be transiently created via the bus API
4285 (this was previously already available for scope and service
4286 units). In addition it is now possible to create multiple
4287 transient units at the same time with a single bus call. The
4288 "systemd-run" tool has been updated to make use of this for
4289 running commands on a specified time, in at(1)-style.
4291 * tmpfiles gained support for "t" lines, for assigning
4292 extended attributes to files. Among other uses this may be
4293 used to assign SMACK labels to files.
4295 Contributions from: Alin Rauta, Alison Chaiken, Andrej
4296 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
4297 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
4298 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
4299 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
4300 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
4301 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
4302 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
4303 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
4304 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
4305 Michael Biebl, Michael Chapman, Michael Marineau, Michal
4306 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
4307 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
4308 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
4309 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
4310 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4311 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
4312 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
4314 — Berlin, 2014-12-10
4318 * journalctl gained the new options -t/--identifier= to match
4319 on the syslog identifier (aka "tag"), as well as --utc to
4320 show log timestamps in the UTC timezone. journalctl now also
4321 accepts -n/--lines=all to disable line capping in a pager.
4323 * journalctl gained a new switch, --flush, that synchronously
4324 flushes logs from /run/log/journal to /var/log/journal if
4325 persistent storage is enabled. systemd-journal-flush.service
4326 now waits until the operation is complete.
4328 * Services can notify the manager before they start a reload
4329 (by sending RELOADING=1) or shutdown (by sending
4330 STOPPING=1). This allows the manager to track and show the
4331 internal state of daemons and closes a race condition when
4332 the process is still running but has closed its D-Bus
4335 * Services with Type=oneshot do not have to have any ExecStart
4338 * User units are now loaded also from
4339 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
4340 /run/systemd/user directory that was already previously
4341 supported, but is under the control of the user.
4343 * Job timeouts (i.e. time-outs on the time a job that is
4344 queued stays in the run queue) can now optionally result in
4345 immediate reboot or power-off actions (JobTimeoutAction= and
4346 JobTimeoutRebootArgument=). This is useful on ".target"
4347 units, to limit the maximum time a target remains
4348 undispatched in the run queue, and to trigger an emergency
4349 operation in such a case. This is now used by default to
4350 turn off the system if boot-up (as defined by everything in
4351 basic.target) hangs and does not complete for at least
4352 15min. Also, if power-off or reboot hang for at least 30min
4353 an immediate power-off/reboot operation is triggered. This
4354 functionality is particularly useful to increase reliability
4355 on embedded devices, but also on laptops which might
4356 accidentally get powered on when carried in a backpack and
4357 whose boot stays stuck in a hard disk encryption passphrase
4360 * systemd-logind can be configured to also handle lid switch
4361 events even when the machine is docked or multiple displays
4362 are attached (HandleLidSwitchDocked= option).
4364 * A helper binary and a service have been added which can be
4365 used to resume from hibernation in the initramfs. A
4366 generator will parse the resume= option on the kernel
4367 command line to trigger resume.
4369 * A user console daemon systemd-consoled has been
4370 added. Currently, it is a preview, and will so far open a
4371 single terminal on each session of the user marked as
4372 Desktop=systemd-console.
4374 * Route metrics can be specified for DHCP routes added by
4377 * The SELinux context of socket-activated services can be set
4378 from the information provided by the networking stack
4379 (SELinuxContextFromNet= option).
4381 * Userspace firmware loading support has been removed and
4382 the minimum supported kernel version is thus bumped to 3.7.
4384 * Timeout for udev workers has been increased from 1 to 3
4385 minutes, but a warning will be printed after 1 minute to
4386 help diagnose kernel modules that take a long time to load.
4388 * Udev rules can now remove tags on devices with TAG-="foobar".
4390 * systemd's readahead implementation has been removed. In many
4391 circumstances it didn't give expected benefits even for
4392 rotational disk drives and was becoming less relevant in the
4393 age of SSDs. As none of the developers has been using
4394 rotating media anymore, and nobody stepped up to actively
4395 maintain this component of systemd it has now been removed.
4397 * Swap units can use Options= to specify discard options.
4398 Discard options specified for swaps in /etc/fstab are now
4401 * Docker containers are now detected as a separate type of
4404 * The Password Agent protocol gained support for queries where
4405 the user input is shown, useful e.g. for user names.
4406 systemd-ask-password gained a new --echo option to turn that
4409 * The default sysctl.d/ snippets will now set:
4411 net.core.default_qdisc = fq_codel
4413 This selects Fair Queuing Controlled Delay as the default
4414 queuing discipline for network interfaces. fq_codel helps
4415 fight the network bufferbloat problem. It is believed to be
4416 a good default with no tuning required for most workloads.
4417 Downstream distributions may override this choice. On 10Gbit
4418 servers that do not do forwarding, "fq" may perform better.
4419 Systems without a good clocksource should use "pfifo_fast".
4421 * If kdbus is enabled during build a new option BusPolicy= is
4422 available for service units, that allows locking all service
4423 processes into a stricter bus policy, in order to limit
4424 access to various bus services, or even hide most of them
4425 from the service's view entirely.
4427 * networkctl will now show the .network and .link file
4428 networkd has applied to a specific interface.
4430 * sd-login gained a new API call sd_session_get_desktop() to
4431 query which desktop environment has been selected for a
4434 * UNIX utmp support is now compile-time optional to support
4435 legacy-free systems.
4437 * systemctl gained two new commands "add-wants" and
4438 "add-requires" for pulling in units from specific targets
4441 * If the word "rescue" is specified on the kernel command line
4442 the system will now boot into rescue mode (aka
4443 rescue.target), which was previously available only by
4444 specifying "1" or "systemd.unit=rescue.target" on the kernel
4445 command line. This new kernel command line option nicely
4446 mirrors the already existing "emergency" kernel command line
4449 * New kernel command line options mount.usr=, mount.usrflags=,
4450 mount.usrfstype= have been added that match root=, rootflags=,
4451 rootfstype= but allow mounting a specific file system to
4454 * The $NOTIFY_SOCKET is now also passed to control processes of
4455 services, not only the main process.
4457 * This version reenables support for fsck's -l switch. This
4458 means at least version v2.25 of util-linux is required for
4459 operation, otherwise dead-locks on device nodes may
4460 occur. Again: you need to update util-linux to at least
4461 v2.25 when updating systemd to v217.
4463 * The "multi-seat-x" tool has been removed from systemd, as
4464 its functionality has been integrated into X servers 1.16,
4465 and the tool is hence redundant. It is recommended to update
4466 display managers invoking this tool to simply invoke X
4467 directly from now on, again.
4469 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
4470 message flag has been added for all of systemd's PolicyKit
4471 authenticated method calls has been added. In particular
4472 this now allows optional interactive authorization via
4473 PolicyKit for many of PID1's privileged operations such as
4474 unit file enabling and disabling.
4476 * "udevadm hwdb --update" learnt a new switch "--usr" for
4477 placing the rebuilt hardware database in /usr instead of
4478 /etc. When used only hardware database entries stored in
4479 /usr will be used, and any user database entries in /etc are
4480 ignored. This functionality is useful for vendors to ship a
4481 pre-built database on systems where local configuration is
4482 unnecessary or unlikely.
4484 * Calendar time specifications in .timer units now also
4485 understand the strings "semi-annually", "quarterly" and
4486 "minutely" as shortcuts (in addition to the preexisting
4487 "anually", "hourly", ...).
4489 * systemd-tmpfiles will now correctly create files in /dev
4490 at boot which are marked for creation only at boot. It is
4491 recommended to always create static device nodes with 'c!'
4492 and 'b!', so that they are created only at boot and not
4493 overwritten at runtime.
4495 * When the watchdog logic is used for a service (WatchdogSec=)
4496 and the watchdog timeout is hit the service will now be
4497 terminated with SIGABRT (instead of just SIGTERM), in order
4498 to make sure a proper coredump and backtrace is
4499 generated. This ensures that hanging services will result in
4500 similar coredump/backtrace behaviour as services that hit a
4503 Contributions from: Andreas Henriksson, Andrei Borzenkov,
4504 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
4505 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
4506 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
4507 Herrmann, David Sommerseth, David Strauss, Emil Renner
4508 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
4509 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
4510 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
4511 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
4512 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
4513 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
4514 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
4515 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
4516 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
4517 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
4518 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
4519 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
4520 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
4521 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
4522 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
4523 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
4526 — Berlin, 2014-10-28
4530 * timedated no longer reads NTP implementation unit names from
4531 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
4532 implementations should add a
4534 Conflicts=systemd-timesyncd.service
4536 to their unit files to take over and replace systemd's NTP
4537 default functionality.
4539 * systemd-sysusers gained a new line type "r" for configuring
4540 which UID/GID ranges to allocate system users/groups
4541 from. Lines of type "u" may now add an additional column
4542 that specifies the home directory for the system user to be
4543 created. Also, systemd-sysusers may now optionally read user
4544 information from STDIN instead of a file. This is useful for
4545 invoking it from RPM preinst scriptlets that need to create
4546 users before the first RPM file is installed since these
4547 files might need to be owned by them. A new
4548 %sysusers_create_inline RPM macro has been introduced to do
4549 just that. systemd-sysusers now updates the shadow files as
4550 well as the user/group databases, which should enhance
4551 compatibility with certain tools like grpck.
4553 * A number of bus APIs of PID 1 now optionally consult
4554 PolicyKit to permit access for otherwise unprivileged
4555 clients under certain conditions. Note that this currently
4556 doesn't support interactive authentication yet, but this is
4557 expected to be added eventually, too.
4559 * /etc/machine-info now has new fields for configuring the
4560 deployment environment of the machine, as well as the
4561 location of the machine. hostnamectl has been updated with
4562 new command to update these fields.
4564 * systemd-timesyncd has been updated to automatically acquire
4565 NTP server information from systemd-networkd, which might
4566 have been discovered via DHCP.
4568 * systemd-resolved now includes a caching DNS stub resolver
4569 and a complete LLMNR name resolution implementation. A new
4570 NSS module "nss-resolve" has been added which can be used
4571 instead of glibc's own "nss-dns" to resolve hostnames via
4572 systemd-resolved. Hostnames, addresses and arbitrary RRs may
4573 be resolved via systemd-resolved D-Bus APIs. In contrast to
4574 the glibc internal resolver systemd-resolved is aware of
4575 multi-homed system, and keeps DNS server and caches separate
4576 and per-interface. Queries are sent simultaneously on all
4577 interfaces that have DNS servers configured, in order to
4578 properly handle VPNs and local LANs which might resolve
4579 separate sets of domain names. systemd-resolved may acquire
4580 DNS server information from systemd-networkd automatically,
4581 which in turn might have discovered them via DHCP. A tool
4582 "systemd-resolve-host" has been added that may be used to
4583 query the DNS logic in resolved. systemd-resolved implements
4584 IDNA and automatically uses IDNA or UTF-8 encoding depending
4585 on whether classic DNS or LLMNR is used as transport. In the
4586 next releases we intend to add a DNSSEC and mDNS/DNS-SD
4587 implementation to systemd-resolved.
4589 * A new NSS module nss-mymachines has been added, that
4590 automatically resolves the names of all local registered
4591 containers to their respective IP addresses.
4593 * A new client tool "networkctl" for systemd-networkd has been
4594 added. It currently is entirely passive and will query
4595 networking configuration from udev, rtnetlink and networkd,
4596 and present it to the user in a very friendly
4597 way. Eventually, we hope to extend it to become a full
4598 control utility for networkd.
4600 * .socket units gained a new DeferAcceptSec= setting that
4601 controls the kernels' TCP_DEFER_ACCEPT sockopt for
4602 TCP. Similarly, support for controlling TCP keep-alive
4603 settings has been added (KeepAliveTimeSec=,
4604 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
4605 turning off Nagle's algorithm on TCP has been added
4608 * logind learned a new session type "web", for use in projects
4609 like Cockpit which register web clients as PAM sessions.
4611 * timer units with at least one OnCalendar= setting will now
4612 be started only after time-sync.target has been
4613 reached. This way they will not elapse before the system
4614 clock has been corrected by a local NTP client or
4615 similar. This is particular useful on RTC-less embedded
4616 machines, that come up with an invalid system clock.
4618 * systemd-nspawn's --network-veth= switch should now result in
4619 stable MAC addresses for both the outer and the inner side
4622 * systemd-nspawn gained a new --volatile= switch for running
4623 container instances with /etc or /var unpopulated.
4625 * The kdbus client code has been updated to use the new Linux
4626 3.17 memfd subsystem instead of the old kdbus-specific one.
4628 * systemd-networkd's DHCP client and server now support
4629 FORCERENEW. There are also new configuration options to
4630 configure the vendor client identifier and broadcast mode
4633 * systemd will no longer inform the kernel about the current
4634 timezone, as this is necessarily incorrect and racy as the
4635 kernel has no understanding of DST and similar
4636 concepts. This hence means FAT timestamps will be always
4637 considered UTC, similar to what Android is already
4638 doing. Also, when the RTC is configured to the local time
4639 (rather than UTC) systemd will never synchronize back to it,
4640 as this might confuse Windows at a later boot.
4642 * systemd-analyze gained a new command "verify" for offline
4643 validation of unit files.
4645 * systemd-networkd gained support for a couple of additional
4646 settings for bonding networking setups. Also, the metric for
4647 statically configured routes may now be configured. For
4648 network interfaces where this is appropriate the peer IP
4649 address may now be configured.
4651 * systemd-networkd's DHCP client will no longer request
4652 broadcasting by default, as this tripped up some networks.
4653 For hardware where broadcast is required the feature should
4654 be switched back on using RequestBroadcast=yes.
4656 * systemd-networkd will now set up IPv4LL addresses (when
4657 enabled) even if DHCP is configured successfully.
4659 * udev will now default to respect network device names given
4660 by the kernel when the kernel indicates that these are
4661 predictable. This behavior can be tweaked by changing
4662 NamePolicy= in the relevant .link file.
4664 * A new library systemd-terminal has been added that
4665 implements full TTY stream parsing and rendering. This
4666 library is supposed to be used later on for implementing a
4667 full userspace VT subsystem, replacing the current kernel
4670 * A new tool systemd-journal-upload has been added to push
4671 journal data to a remote system running
4672 systemd-journal-remote.
4674 * journald will no longer forward all local data to another
4675 running syslog daemon. This change has been made because
4676 rsyslog (which appears to be the most commonly used syslog
4677 implementation these days) no longer makes use of this, and
4678 instead pulls the data out of the journal on its own. Since
4679 forwarding the messages to a non-existent syslog server is
4680 more expensive than we assumed we have now turned this
4681 off. If you run a syslog server that is not a recent rsyslog
4682 version, you have to turn this option on again
4683 (ForwardToSyslog= in journald.conf).
4685 * journald now optionally supports the LZ4 compressor for
4686 larger journal fields. This compressor should perform much
4687 better than XZ which was the previous default.
4689 * machinectl now shows the IP addresses of local containers,
4690 if it knows them, plus the interface name of the container.
4692 * A new tool "systemd-escape" has been added that makes it
4693 easy to escape strings to build unit names and similar.
4695 * sd_notify() messages may now include a new ERRNO= field
4696 which is parsed and collected by systemd and shown among the
4697 "systemctl status" output for a service.
4699 * A new component "systemd-firstboot" has been added that
4700 queries the most basic systemd information (timezone,
4701 hostname, root password) interactively on first
4702 boot. Alternatively it may also be used to provision these
4703 things offline on OS images installed into directories.
4705 * The default sysctl.d/ snippets will now set
4707 net.ipv4.conf.default.promote_secondaries=1
4709 This has the benefit of no flushing secondary IP addresses
4710 when primary addresses are removed.
4712 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
4713 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
4714 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
4715 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
4716 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
4717 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
4718 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
4719 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
4720 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
4721 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
4722 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
4723 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
4724 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
4725 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
4726 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
4728 — Berlin, 2014-08-19
4732 * A new tool systemd-sysusers has been added. This tool
4733 creates system users and groups in /etc/passwd and
4734 /etc/group, based on static declarative system user/group
4735 definitions in /usr/lib/sysusers.d/. This is useful to
4736 enable factory resets and volatile systems that boot up with
4737 an empty /etc directory, and thus need system users and
4738 groups created during early boot. systemd now also ships
4739 with two default sysusers.d/ files for the most basic
4740 users and groups systemd and the core operating system
4743 * A new tmpfiles snippet has been added that rebuilds the
4744 essential files in /etc on boot, should they be missing.
4746 * A directive for ensuring automatic clean-up of
4747 /var/cache/man/ has been removed from the default
4748 configuration. This line should now be shipped by the man
4749 implementation. The necessary change has been made to the
4750 man-db implementation. Note that you need to update your man
4751 implementation to one that ships this line, otherwise no
4752 automatic clean-up of /var/cache/man will take place.
4754 * A new condition ConditionNeedsUpdate= has been added that
4755 may conditionalize services to only run when /etc or /var
4756 are "older" than the vendor operating system resources in
4757 /usr. This is useful for reconstructing or updating /etc
4758 after an offline update of /usr or a factory reset, on the
4759 next reboot. Services that want to run once after such an
4760 update or reset should use this condition and order
4761 themselves before the new systemd-update-done.service, which
4762 will mark the two directories as fully updated. A number of
4763 service files have been added making use of this, to rebuild
4764 the udev hardware database, the journald message catalog and
4765 dynamic loader cache (ldconfig). The systemd-sysusers tool
4766 described above also makes use of this now. With this in
4767 place it is now possible to start up a minimal operating
4768 system with /etc empty cleanly. For more information on the
4769 concepts involved see this recent blog story:
4771 http://0pointer.de/blog/projects/stateless.html
4773 * A new system group "input" has been introduced, and all
4774 input device nodes get this group assigned. This is useful
4775 for system-level software to get access to input devices. It
4776 complements what is already done for "audio" and "video".
4778 * systemd-networkd learnt minimal DHCPv4 server support in
4779 addition to the existing DHCPv4 client support. It also
4780 learnt DHCPv6 client and IPv6 Router Solicitation client
4781 support. The DHCPv4 client gained support for static routes
4782 passed in from the server. Note that the [DHCPv4] section
4783 known in older systemd-networkd versions has been renamed to
4784 [DHCP] and is now also used by the DHCPv6 client. Existing
4785 .network files using settings of this section should be
4786 updated, though compatibility is maintained. Optionally, the
4787 client hostname may now be sent to the DHCP server.
4789 * networkd gained support for vxlan virtual networks as well
4790 as tun/tap and dummy devices.
4792 * networkd gained support for automatic allocation of address
4793 ranges for interfaces from a system-wide pool of
4794 addresses. This is useful for dynamically managing a large
4795 number of interfaces with a single network configuration
4796 file. In particular this is useful to easily assign
4797 appropriate IP addresses to the veth links of a large number
4798 of nspawn instances.
4800 * RPM macros for processing sysusers, sysctl and binfmt
4801 drop-in snippets at package installation time have been
4804 * The /etc/os-release file should now be placed in
4805 /usr/lib/os-release. The old location is automatically
4806 created as symlink. /usr/lib is the more appropriate
4807 location of this file, since it shall actually describe the
4808 vendor operating system shipped in /usr, and not the
4809 configuration stored in /etc.
4811 * .mount units gained a new boolean SloppyOptions= setting
4812 that maps to mount(8)'s -s option which enables permissive
4813 parsing of unknown mount options.
4815 * tmpfiles learnt a new "L+" directive which creates a symlink
4816 but (unlike "L") deletes a pre-existing file first, should
4817 it already exist and not already be the correct
4818 symlink. Similarly, "b+", "c+" and "p+" directives have been
4819 added as well, which create block and character devices, as
4820 well as fifos in the filesystem, possibly removing any
4821 pre-existing files of different types.
4823 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
4824 'argument' field (which so far specified the source to
4825 symlink/copy the files from) is now optional. If omitted the
4826 same file os copied from /usr/share/factory/ suffixed by the
4827 full destination path. This is useful for populating /etc
4828 with essential files, by copying them from vendor defaults
4829 shipped in /usr/share/factory/etc.
4831 * A new command "systemctl preset-all" has been added that
4832 applies the service preset settings to all installed unit
4833 files. A new switch --preset-mode= has been added that
4834 controls whether only enable or only disable operations
4837 * A new command "systemctl is-system-running" has been added
4838 that allows checking the overall state of the system, for
4839 example whether it is fully up and running.
4841 * When the system boots up with an empty /etc, the equivalent
4842 to "systemctl preset-all" is executed during early boot, to
4843 make sure all default services are enabled after a factory
4846 * systemd now contains a minimal preset file that enables the
4847 most basic services systemd ships by default.
4849 * Unit files' [Install] section gained a new DefaultInstance=
4850 field for defining the default instance to create if a
4851 template unit is enabled with no instance specified.
4853 * A new passive target cryptsetup-pre.target has been added
4854 that may be used by services that need to make they run and
4855 finish before the first LUKS cryptographic device is set up.
4857 * The /dev/loop-control and /dev/btrfs-control device nodes
4858 are now owned by the "disk" group by default, opening up
4859 access to this group.
4861 * systemd-coredump will now automatically generate a
4862 stack trace of all core dumps taking place on the system,
4863 based on elfutils' libdw library. This stack trace is logged
4866 * systemd-coredump may now optionally store coredumps directly
4867 on disk (in /var/lib/systemd/coredump, possibly compressed),
4868 instead of storing them unconditionally in the journal. This
4869 mode is the new default. A new configuration file
4870 /etc/systemd/coredump.conf has been added to configure this
4871 and other parameters of systemd-coredump.
4873 * coredumpctl gained a new "info" verb to show details about a
4874 specific coredump. A new switch "-1" has also been added
4875 that makes sure to only show information about the most
4876 recent entry instead of all entries. Also, as the tool is
4877 generally useful now the "systemd-" prefix of the binary
4878 name has been removed. Distributions that want to maintain
4879 compatibility with the old name should add a symlink from
4880 the old name to the new name.
4882 * journald's SplitMode= now defaults to "uid". This makes sure
4883 that unprivileged users can access their own coredumps with
4884 coredumpctl without restrictions.
4886 * New kernel command line options "systemd.wants=" (for
4887 pulling an additional unit during boot), "systemd.mask="
4888 (for masking a specific unit for the boot), and
4889 "systemd.debug-shell" (for enabling the debug shell on tty9)
4890 have been added. This is implemented in the new generator
4891 "systemd-debug-generator".
4893 * systemd-nspawn will now by default filter a couple of
4894 syscalls for containers, among them those required for
4895 kernel module loading, direct x86 IO port access, swap
4896 management, and kexec. Most importantly though
4897 open_by_handle_at() is now prohibited for containers,
4898 closing a hole similar to a recently discussed vulnerability
4899 in docker regarding access to files on file hierarchies the
4900 container should normally not have access to. Note that, for
4901 nspawn, we generally make no security claims anyway (and
4902 this is explicitly documented in the man page), so this is
4903 just a fix for one of the most obvious problems.
4905 * A new man page file-hierarchy(7) has been added that
4906 contains a minimized, modernized version of the file system
4907 layout systemd expects, similar in style to the FHS
4908 specification or hier(5). A new tool systemd-path(1) has
4909 been added to query many of these paths for the local
4912 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
4913 longer done. Since the directory now has a per-user size
4914 limit, and is cleaned on logout this appears unnecessary,
4915 in particular since this now brings the lifecycle of this
4916 directory closer in line with how IPC objects are handled.
4918 * systemd.pc now exports a number of additional directories,
4919 including $libdir (which is useful to identify the library
4920 path for the primary architecture of the system), and a
4921 couple of drop-in directories.
4923 * udev's predictable network interface names now use the dev_port
4924 sysfs attribute, introduced in linux 3.15 instead of dev_id to
4925 distinguish between ports of the same PCI function. dev_id should
4926 only be used for ports using the same HW address, hence the need
4929 * machined has been updated to export the OS version of a
4930 container (read from /etc/os-release and
4931 /usr/lib/os-release) on the bus. This is now shown in
4932 "machinectl status" for a machine.
4934 * A new service setting RestartForceExitStatus= has been
4935 added. If configured to a set of exit signals or process
4936 return values, the service will be restarted when the main
4937 daemon process exits with any of them, regardless of the
4940 * systemctl's -H switch for connecting to remote systemd
4941 machines has been extended so that it may be used to
4942 directly connect to a specific container on the
4943 host. "systemctl -H root@foobar:waldi" will now connect as
4944 user "root" to host "foobar", and then proceed directly to
4945 the container named "waldi". Note that currently you have to
4946 authenticate as user "root" for this to work, as entering
4947 containers is a privileged operation.
4949 Contributions from: Andreas Henriksson, Benjamin Steinwender,
4950 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
4951 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
4952 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
4953 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
4954 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
4955 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
4956 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
4957 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
4958 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
4959 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
4960 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
4962 — Berlin, 2014-07-03
4966 * As an experimental feature, udev now tries to lock the
4967 disk device node (flock(LOCK_SH|LOCK_NB)) while it
4968 executes events for the disk or any of its partitions.
4969 Applications like partitioning programs can lock the
4970 disk device node (flock(LOCK_EX)) and claim temporary
4971 device ownership that way; udev will entirely skip all event
4972 handling for this disk and its partitions. If the disk
4973 was opened for writing, the close will trigger a partition
4974 table rescan in udev's "watch" facility, and if needed
4975 synthesize "change" events for the disk and all its partitions.
4976 This is now unconditionally enabled, and if it turns out to
4977 cause major problems, we might turn it on only for specific
4978 devices, or might need to disable it entirely. Device Mapper
4979 devices are excluded from this logic.
4981 * We temporarily dropped the "-l" switch for fsck invocations,
4982 since they collide with the flock() logic above. util-linux
4983 upstream has been changed already to avoid this conflict,
4984 and we will readd "-l" as soon as util-linux with this
4985 change has been released.
4987 * The dependency on libattr has been removed. Since a long
4988 time, the extended attribute calls have moved to glibc, and
4989 libattr is thus unnecessary.
4991 * Virtualization detection works without privileges now. This
4992 means the systemd-detect-virt binary no longer requires
4993 CAP_SYS_PTRACE file capabilities, and our daemons can run
4994 with fewer privileges.
4996 * systemd-networkd now runs under its own "systemd-network"
4997 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
4998 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
4999 loses the ability to write to files owned by root this way.
5001 * Similarly, systemd-resolved now runs under its own
5002 "systemd-resolve" user with no capabilities remaining.
5004 * Similarly, systemd-bus-proxyd now runs under its own
5005 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
5007 * systemd-networkd gained support for setting up "veth"
5008 virtual Ethernet devices for container connectivity, as well
5009 as GRE and VTI tunnels.
5011 * systemd-networkd will no longer automatically attempt to
5012 manually load kernel modules necessary for certain tunnel
5013 transports. Instead, it is assumed the kernel loads them
5014 automatically when required. This only works correctly on
5015 very new kernels. On older kernels, please consider adding
5016 the kernel modules to /etc/modules-load.d/ as a work-around.
5018 * The resolv.conf file systemd-resolved generates has been
5019 moved to /run/systemd/resolve/. If you have a symlink from
5020 /etc/resolv.conf, it might be necessary to correct it.
5022 * Two new service settings, ProtectHome= and ProtectSystem=,
5023 have been added. When enabled, they will make the user data
5024 (such as /home) inaccessible or read-only and the system
5025 (such as /usr) read-only, for specific services. This allows
5026 very light-weight per-service sandboxing to avoid
5027 modifications of user data or system files from
5028 services. These two new switches have been enabled for all
5029 of systemd's long-running services, where appropriate.
5031 * Socket units gained new SocketUser= and SocketGroup=
5032 settings to set the owner user and group of AF_UNIX sockets
5033 and FIFOs in the file system.
5035 * Socket units gained a new RemoveOnStop= setting. If enabled,
5036 all FIFOS and sockets in the file system will be removed
5037 when the specific socket unit is stopped.
5039 * Socket units gained a new Symlinks= setting. It takes a list
5040 of symlinks to create to file system sockets or FIFOs
5041 created by the specific Unix sockets. This is useful to
5042 manage symlinks to socket nodes with the same life-cycle as
5045 * The /dev/log socket and /dev/initctl FIFO have been moved to
5046 /run, and have been replaced by symlinks. This allows
5047 connecting to these facilities even if PrivateDevices=yes is
5048 used for a service (which makes /dev/log itself unavailable,
5049 but /run is left). This also has the benefit of ensuring
5050 that /dev only contains device nodes, directories and
5051 symlinks, and nothing else.
5053 * sd-daemon gained two new calls sd_pid_notify() and
5054 sd_pid_notifyf(). They are similar to sd_notify() and
5055 sd_notifyf(), but allow overriding of the source PID of
5056 notification messages if permissions permit this. This is
5057 useful to send notify messages on behalf of a different
5058 process (for example, the parent process). The
5059 systemd-notify tool has been updated to make use of this
5060 when sending messages (so that notification messages now
5061 originate from the shell script invoking systemd-notify and
5062 not the systemd-notify process itself. This should minimize
5063 a race where systemd fails to associate notification
5064 messages to services when the originating process already
5067 * A new "on-abnormal" setting for Restart= has been added. If
5068 set, it will result in automatic restarts on all "abnormal"
5069 reasons for a process to exit, which includes unclean
5070 signals, core dumps, timeouts and watchdog timeouts, but
5071 does not include clean and unclean exit codes or clean
5072 signals. Restart=on-abnormal is an alternative for
5073 Restart=on-failure for services that shall be able to
5074 terminate and avoid restarts on certain errors, by
5075 indicating so with an unclean exit code. Restart=on-failure
5076 or Restart=on-abnormal is now the recommended setting for
5077 all long-running services.
5079 * If the InaccessibleDirectories= service setting points to a
5080 mount point (or if there are any submounts contained within
5081 it), it is now attempted to completely unmount it, to make
5082 the file systems truly unavailable for the respective
5085 * The ReadOnlyDirectories= service setting and
5086 systemd-nspawn's --read-only parameter are now recursively
5087 applied to all submounts, too.
5089 * Mount units may now be created transiently via the bus APIs.
5091 * The support for SysV and LSB init scripts has been removed
5092 from the systemd daemon itself. Instead, it is now
5093 implemented as a generator that creates native systemd units
5094 from these scripts when needed. This enables us to remove a
5095 substantial amount of legacy code from PID 1, following the
5096 fact that many distributions only ship a very small number
5097 of LSB/SysV init scripts nowadays.
5099 * Privileged Xen (dom0) domains are not considered
5100 virtualization anymore by the virtualization detection
5101 logic. After all, they generally have unrestricted access to
5102 the hardware and usually are used to manage the unprivileged
5105 * systemd-tmpfiles gained a new "C" line type, for copying
5106 files or entire directories.
5108 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
5109 lines. So far, they have been non-globbing versions of the
5110 latter, and have thus been redundant. In future, it is
5111 recommended to only use "z". "m" has hence been removed
5112 from the documentation, even though it stays supported.
5114 * A tmpfiles snippet to recreate the most basic structure in
5115 /var has been added. This is enough to create the /var/run →
5116 /run symlink and create a couple of structural
5117 directories. This allows systems to boot up with an empty or
5118 volatile /var. Of course, while with this change, the core OS
5119 now is capable with dealing with a volatile /var, not all
5120 user services are ready for it. However, we hope that sooner
5121 or later, many service daemons will be changed upstream so
5122 that they are able to automatically create their necessary
5123 directories in /var at boot, should they be missing. This is
5124 the first step to allow state-less systems that only require
5125 the vendor image for /usr to boot.
5127 * systemd-nspawn has gained a new --tmpfs= switch to mount an
5128 empty tmpfs instance to a specific directory. This is
5129 particularly useful for making use of the automatic
5130 reconstruction of /var (see above), by passing --tmpfs=/var.
5132 * Access modes specified in tmpfiles snippets may now be
5133 prefixed with "~", which indicates that they shall be masked
5134 by whether the existing file or directory is currently
5135 writable, readable or executable at all. Also, if specified,
5136 the sgid/suid/sticky bits will be masked for all
5139 * A new passive target unit "network-pre.target" has been
5140 added which is useful for services that shall run before any
5141 network is configured, for example firewall scripts.
5143 * The "floppy" group that previously owned the /dev/fd*
5144 devices is no longer used. The "disk" group is now used
5145 instead. Distributions should probably deprecate usage of
5148 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
5149 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
5150 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
5151 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
5152 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
5153 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
5156 — Berlin, 2014-06-11
5160 * A new "systemd-timesyncd" daemon has been added for
5161 synchronizing the system clock across the network. It
5162 implements an SNTP client. In contrast to NTP
5163 implementations such as chrony or the NTP reference server,
5164 this only implements a client side, and does not bother with
5165 the full NTP complexity, focusing only on querying time from
5166 one remote server and synchronizing the local clock to
5167 it. Unless you intend to serve NTP to networked clients or
5168 want to connect to local hardware clocks, this simple NTP
5169 client should be more than appropriate for most
5170 installations. The daemon runs with minimal privileges, and
5171 has been hooked up with networkd to only operate when
5172 network connectivity is available. The daemon saves the
5173 current clock to disk every time a new NTP sync has been
5174 acquired, and uses this to possibly correct the system clock
5175 early at bootup, in order to accommodate for systems that
5176 lack an RTC such as the Raspberry Pi and embedded devices,
5177 and to make sure that time monotonically progresses on these
5178 systems, even if it is not always correct. To make use of
5179 this daemon, a new system user and group "systemd-timesync"
5180 needs to be created on installation of systemd.
5182 * The queue "seqnum" interface of libudev has been disabled, as
5183 it was generally incompatible with device namespacing as
5184 sequence numbers of devices go "missing" if the devices are
5185 part of a different namespace.
5187 * "systemctl list-timers" and "systemctl list-sockets" gained
5188 a --recursive switch for showing units of these types also
5189 for all local containers, similar in style to the already
5190 supported --recursive switch for "systemctl list-units".
5192 * A new RebootArgument= setting has been added for service
5193 units, which may be used to specify a kernel reboot argument
5194 to use when triggering reboots with StartLimitAction=.
5196 * A new FailureAction= setting has been added for service
5197 units which may be used to specify an operation to trigger
5198 when a service fails. This works similarly to
5199 StartLimitAction=, but unlike it, controls what is done
5200 immediately rather than only after several attempts to
5201 restart the service in question.
5203 * hostnamed got updated to also expose the kernel name,
5204 release, and version on the bus. This is useful for
5205 executing commands like hostnamectl with the -H switch.
5206 systemd-analyze makes use of this to properly display
5207 details when running non-locally.
5209 * The bootchart tool can now show cgroup information in the
5210 graphs it generates.
5212 * The CFS CPU quota cgroup attribute is now exposed for
5213 services. The new CPUQuota= switch has been added for this
5214 which takes a percentage value. Setting this will have the
5215 result that a service may never get more CPU time than the
5216 specified percentage, even if the machine is otherwise idle.
5218 * systemd-networkd learned IPIP and SIT tunnel support.
5220 * LSB init scripts exposing a dependency on $network will now
5221 get a dependency on network-online.target rather than simply
5222 network.target. This should bring LSB handling closer to
5223 what it was on SysV systems.
5225 * A new fsck.repair= kernel option has been added to control
5226 how fsck shall deal with unclean file systems at boot.
5228 * The (.ini) configuration file parser will now silently
5229 ignore sections whose name begins with "X-". This may be
5230 used to maintain application-specific extension sections in unit
5233 * machined gained a new API to query the IP addresses of
5234 registered containers. "machinectl status" has been updated
5235 to show these addresses in its output.
5237 * A new call sd_uid_get_display() has been added to the
5238 sd-login APIs for querying the "primary" session of a
5239 user. The "primary" session of the user is elected from the
5240 user's sessions and generally a graphical session is
5241 preferred over a text one.
5243 * A minimal systemd-resolved daemon has been added. It
5244 currently simply acts as a companion to systemd-networkd and
5245 manages resolv.conf based on per-interface DNS
5246 configuration, possibly supplied via DHCP. In the long run
5247 we hope to extend this into a local DNSSEC enabled DNS and
5250 * The systemd-networkd-wait-online tool is now enabled by
5251 default. It will delay network-online.target until a network
5252 connection has been configured. The tool primarily integrates
5253 with networkd, but will also make a best effort to make sense
5254 of network configuration performed in some other way.
5256 * Two new service options StartupCPUShares= and
5257 StartupBlockIOWeight= have been added that work similarly to
5258 CPUShares= and BlockIOWeight= however only apply during
5259 system startup. This is useful to prioritize certain services
5260 differently during bootup than during normal runtime.
5262 * hostnamed has been changed to prefer the statically
5263 configured hostname in /etc/hostname (unless set to
5264 'localhost' or empty) over any dynamic one supplied by
5265 dhcp. With this change, the rules for picking the hostname
5266 match more closely the rules of other configuration settings
5267 where the local administrator's configuration in /etc always
5268 overrides any other settings.
5270 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
5271 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
5272 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
5273 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
5274 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
5275 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
5276 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
5277 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
5278 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
5279 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
5280 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
5281 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
5282 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
5283 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
5284 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
5285 Lindskog, WaLyong Cho, Will Woods, Zbigniew
5288 — Beijing, 2014-05-28
5292 * When restoring the screen brightness at boot, stay away from
5293 the darkest setting or from the lowest 5% of the available
5294 range, depending on which is the larger value of both. This
5295 should effectively protect the user from rebooting into a
5296 black screen, should the brightness have been set to minimum
5299 * sd-login gained a new sd_machine_get_class() call to
5300 determine the class ("vm" or "container") of a machine
5301 registered with machined.
5303 * sd-login gained new calls
5304 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
5305 to query the identity of the peer of a local AF_UNIX
5306 connection. They operate similarly to their sd_pid_get_xyz()
5309 * PID 1 will now maintain a system-wide system state engine
5310 with the states "starting", "running", "degraded",
5311 "maintenance", "stopping". These states are bound to system
5312 startup, normal runtime, runtime with at least one failed
5313 service, rescue/emergency mode and system shutdown. This
5314 state is shown in the "systemctl status" output when no unit
5315 name is passed. It is useful to determine system state, in
5316 particularly when doing so for many systems or containers at
5319 * A new command "list-machines" has been added to "systemctl"
5320 that lists all local OS containers and shows their system
5321 state (see above), if systemd runs inside of them.
5323 * systemctl gained a new "-r" switch to recursively enumerate
5324 units on all local containers, when used with the
5325 "list-unit" command (which is the default one that is
5326 executed when no parameters are specified).
5328 * The GPT automatic partition discovery logic will now honour
5329 two GPT partition flags: one may be set on a partition to
5330 cause it to be mounted read-only, and the other may be set
5331 on a partition to ignore it during automatic discovery.
5333 * Two new GPT type UUIDs have been added for automatic root
5334 partition discovery, for 32-bit and 64-bit ARM. This is not
5335 particularly useful for discovering the root directory on
5336 these architectures during bare-metal boots (since UEFI is
5337 not common there), but still very useful to allow booting of
5338 ARM disk images in nspawn with the -i option.
5340 * MAC addresses of interfaces created with nspawn's
5341 --network-interface= switch will now be generated from the
5342 machine name, and thus be stable between multiple invocations
5345 * logind will now automatically remove all IPC objects owned
5346 by a user if she or he fully logs out. This makes sure that
5347 users who are logged out cannot continue to consume IPC
5348 resources. This covers SysV memory, semaphores and message
5349 queues as well as POSIX shared memory and message
5350 queues. Traditionally, SysV and POSIX IPC had no life-cycle
5351 limits. With this functionality, that is corrected. This may
5352 be turned off by using the RemoveIPC= switch of logind.conf.
5354 * The systemd-machine-id-setup and tmpfiles tools gained a
5355 --root= switch to operate on a specific root directory,
5358 * journald can now forward logged messages to the TTYs of all
5359 logged in users ("wall"). This is the default for all
5360 emergency messages now.
5362 * A new tool systemd-journal-remote has been added to stream
5363 journal log messages across the network.
5365 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
5366 controller trees are mounted into it. Note that the
5367 directories mounted beneath it are not read-only. This is a
5368 security measure and is particularly useful because glibc
5369 actually includes a search logic to pick any tmpfs it can
5370 find to implement shm_open() if /dev/shm is not available
5371 (which it might very well be in namespaced setups).
5373 * machinectl gained a new "poweroff" command to cleanly power
5374 down a local OS container.
5376 * The PrivateDevices= unit file setting will now also drop the
5377 CAP_MKNOD capability from the capability bound set, and
5378 imply DevicePolicy=closed.
5380 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
5381 comprehensively on all long-running systemd services where
5382 this is appropriate.
5384 * systemd-udevd will now run in a disassociated mount
5385 namespace. To mount directories from udev rules, make sure to
5386 pull in mount units via SYSTEMD_WANTS properties.
5388 * The kdbus support gained support for uploading policy into
5389 the kernel. sd-bus gained support for creating "monitoring"
5390 connections that can eavesdrop into all bus communication
5391 for debugging purposes.
5393 * Timestamps may now be specified in seconds since the UNIX
5394 epoch Jan 1st, 1970 by specifying "@" followed by the value
5397 * Native tcpwrap support in systemd has been removed. tcpwrap
5398 is old code, not really maintained anymore and has serious
5399 shortcomings, and better options such as firewalls
5400 exist. For setups that require tcpwrap usage, please
5401 consider invoking your socket-activated service via tcpd,
5402 like on traditional inetd.
5404 * A new system.conf configuration option
5405 DefaultTimerAccuracySec= has been added that controls the
5406 default AccuracySec= setting of .timer units.
5408 * Timer units gained a new WakeSystem= switch. If enabled,
5409 timers configured this way will cause the system to resume
5410 from system suspend (if the system supports that, which most
5413 * Timer units gained a new Persistent= switch. If enabled,
5414 timers configured this way will save to disk when they have
5415 been last triggered. This information is then used on next
5416 reboot to possible execute overdue timer events, that
5417 could not take place because the system was powered off.
5418 This enables simple anacron-like behaviour for timer units.
5420 * systemctl's "list-timers" will now also list the time a
5421 timer unit was last triggered in addition to the next time
5422 it will be triggered.
5424 * systemd-networkd will now assign predictable IPv4LL
5425 addresses to its local interfaces.
5427 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
5428 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
5429 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
5430 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
5431 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
5432 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
5433 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
5434 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
5437 — Berlin, 2014-03-25
5441 * A new unit file setting RestrictAddressFamilies= has been
5442 added to restrict which socket address families unit
5443 processes gain access to. This takes address family names
5444 like "AF_INET" or "AF_UNIX", and is useful to minimize the
5445 attack surface of services via exotic protocol stacks. This
5446 is built on seccomp system call filters.
5448 * Two new unit file settings RuntimeDirectory= and
5449 RuntimeDirectoryMode= have been added that may be used to
5450 manage a per-daemon runtime directories below /run. This is
5451 an alternative for setting up directory permissions with
5452 tmpfiles snippets, and has the advantage that the runtime
5453 directory's lifetime is bound to the daemon runtime and that
5454 the daemon starts up with an empty directory each time. This
5455 is particularly useful when writing services that drop
5456 privileges using the User= or Group= setting.
5458 * The DeviceAllow= unit setting now supports globbing for
5459 matching against device group names.
5461 * The systemd configuration file system.conf gained new
5462 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
5463 DefaultMemoryAccounting= to globally turn on/off accounting
5464 for specific resources (cgroups) for all units. These
5465 settings may still be overridden individually in each unit
5468 * systemd-gpt-auto-generator is now able to discover /srv and
5469 root partitions in addition to /home and swap partitions. It
5470 also supports LUKS-encrypted partitions now. With this in
5471 place, automatic discovery of partitions to mount following
5472 the Discoverable Partitions Specification
5473 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
5474 is now a lot more complete. This allows booting without
5475 /etc/fstab and without root= on the kernel command line on
5476 systems prepared appropriately.
5478 * systemd-nspawn gained a new --image= switch which allows
5479 booting up disk images and Linux installations on any block
5480 device that follow the Discoverable Partitions Specification
5481 (see above). This means that installations made with
5482 appropriately updated installers may now be started and
5483 deployed using container managers, completely
5484 unmodified. (We hope that libvirt-lxc will add support for
5485 this feature soon, too.)
5487 * systemd-nspawn gained a new --network-macvlan= setting to
5488 set up a private macvlan interface for the
5489 container. Similarly, systemd-networkd gained a new
5490 Kind=macvlan setting in .netdev files.
5492 * systemd-networkd now supports configuring local addresses
5495 * A new tool systemd-network-wait-online has been added to
5496 synchronously wait for network connectivity using
5499 * The sd-bus.h bus API gained a new sd_bus_track object for
5500 tracking the life-cycle of bus peers. Note that sd-bus.h is
5501 still not a public API though (unless you specify
5502 --enable-kdbus on the configure command line, which however
5503 voids your warranty and you get no API stability guarantee).
5505 * The $XDG_RUNTIME_DIR runtime directories for each user are
5506 now individual tmpfs instances, which has the benefit of
5507 introducing separate pools for each user, with individual
5508 size limits, and thus making sure that unprivileged clients
5509 can no longer negatively impact the system or other users by
5510 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
5511 RuntimeDirectorySize= has been introduced that allows
5512 controlling the default size limit for all users. It
5513 defaults to 10% of the available physical memory. This is no
5514 replacement for quotas on tmpfs though (which the kernel
5515 still does not support), as /dev/shm and /tmp are still
5516 shared resources used by both the system and unprivileged
5519 * logind will now automatically turn off automatic suspending
5520 on laptop lid close when more than one display is
5521 connected. This was previously expected to be implemented
5522 individually in desktop environments (such as GNOME),
5523 however has been added to logind now, in order to fix a
5524 boot-time race where a desktop environment might not have
5525 been started yet and thus not been able to take an inhibitor
5526 lock at the time where logind already suspends the system
5527 due to a closed lid.
5529 * logind will now wait at least 30s after each system
5530 suspend/resume cycle, and 3min after system boot before
5531 suspending the system due to a closed laptop lid. This
5532 should give USB docking stations and similar enough time to
5533 be probed and configured after system resume and boot in
5534 order to then act as suspend blocker.
5536 * systemd-run gained a new --property= setting which allows
5537 initialization of resource control properties (and others)
5538 for the created scope or service unit. Example: "systemd-run
5539 --property=BlockIOWeight=10 updatedb" may be used to run
5540 updatedb at a low block IO scheduling weight.
5542 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
5543 now also work in --scope mode.
5545 * When systemd is compiled with kdbus support, basic support
5546 for enforced policies is now in place. (Note that enabling
5547 kdbus still voids your warranty and no API compatibility
5550 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
5551 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5552 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
5553 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
5554 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
5555 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
5556 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
5557 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
5558 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
5559 Zbigniew Jędrzejewski-Szmek
5561 — Berlin, 2014-03-12
5565 * systemd will now relabel /dev after loading the SMACK policy
5566 according to SMACK rules.
5568 * A new unit file option AppArmorProfile= has been added to
5569 set the AppArmor profile for the processes of a unit.
5571 * A new condition check ConditionArchitecture= has been added
5572 to conditionalize units based on the system architecture, as
5573 reported by uname()'s "machine" field.
5575 * systemd-networkd now supports matching on the system
5576 virtualization, architecture, kernel command line, host name
5579 * logind is now a lot more aggressive when suspending the
5580 machine due to a closed laptop lid. Instead of acting only
5581 on the lid close action, it will continuously watch the lid
5582 status and act on it. This is useful for laptops where the
5583 power button is on the outside of the chassis so that it can
5584 be reached without opening the lid (such as the Lenovo
5585 Yoga). On those machines, logind will now immediately
5586 re-suspend the machine if the power button has been
5587 accidentally pressed while the laptop was suspended and in a
5588 backpack or similar.
5590 * logind will now watch SW_DOCK switches and inhibit reaction
5591 to the lid switch if it is pressed. This means that logind
5592 will not suspend the machine anymore if the lid is closed
5593 and the system is docked, if the laptop supports SW_DOCK
5594 notifications via the input layer. Note that ACPI docking
5595 stations do not generate this currently. Also note that this
5596 logic is usually not fully sufficient and Desktop
5597 Environments should take a lid switch inhibitor lock when an
5598 external display is connected, as systemd will not watch
5601 * nspawn will now make use of the devices cgroup controller by
5602 default, and only permit creation of and access to the usual
5603 API device nodes like /dev/null or /dev/random, as well as
5604 access to (but not creation of) the pty devices.
5606 * We will now ship a default .network file for
5607 systemd-networkd that automatically configures DHCP for
5608 network interfaces created by nspawn's --network-veth or
5609 --network-bridge= switches.
5611 * systemd will now understand the usual M, K, G, T suffixes
5612 according to SI conventions (i.e. to the base 1000) when
5613 referring to throughput and hardware metrics. It will stay
5614 with IEC conventions (i.e. to the base 1024) for software
5615 metrics, according to what is customary according to
5616 Wikipedia. We explicitly document which base applies for
5617 each configuration option.
5619 * The DeviceAllow= setting in unit files now supports a syntax
5620 to whitelist an entire group of devices node majors at once,
5621 based on the /proc/devices listing. For example, with the
5622 string "char-pts", it is now possible to whitelist all
5623 current and future pseudo-TTYs at once.
5625 * sd-event learned a new "post" event source. Event sources of
5626 this type are triggered by the dispatching of any event
5627 source of a type that is not "post". This is useful for
5628 implementing clean-up and check event sources that are
5629 triggered by other work being done in the program.
5631 * systemd-networkd is no longer statically enabled, but uses
5632 the usual [Install] sections so that it can be
5633 enabled/disabled using systemctl. It still is enabled by
5636 * When creating a veth interface pair with systemd-nspawn, the
5637 host side will now be prefixed with "vb-" if
5638 --network-bridge= is used, and with "ve-" if --network-veth
5639 is used. This way, it is easy to distinguish these cases on
5640 the host, for example to apply different configuration to
5641 them with systemd-networkd.
5643 * The compatibility libraries for libsystemd-journal.so,
5644 libsystem-id128.so, libsystemd-login.so and
5645 libsystemd-daemon.so do not make use of IFUNC
5646 anymore. Instead, we now build libsystemd.so multiple times
5647 under these alternative names. This means that the footprint
5648 is drastically increased, but given that these are
5649 transitional compatibility libraries, this should not matter
5650 much. This change has been made necessary to support the ARM
5651 platform for these compatibility libraries, as the ARM
5652 toolchain is not really at the same level as the toolchain
5653 for other architectures like x86 and does not support
5654 IFUNC. Please make sure to use --enable-compat-libs only
5655 during a transitional period!
5657 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
5658 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5659 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
5660 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
5661 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
5662 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
5663 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
5664 Zbigniew Jędrzejewski-Szmek
5666 — Berlin, 2014-02-24
5670 * A new component "systemd-networkd" has been added that can
5671 be used to configure local network interfaces statically or
5672 via DHCP. It is capable of bringing up bridges, VLANs, and
5673 bonding. Currently, no hook-ups for interactive network
5674 configuration are provided. Use this for your initrd,
5675 container, embedded, or server setup if you need a simple,
5676 yet powerful, network configuration solution. This
5677 configuration subsystem is quite nifty, as it allows wildcard
5678 hotplug matching in interfaces. For example, with a single
5679 configuration snippet, you can configure that all Ethernet
5680 interfaces showing up are automatically added to a bridge,
5681 or similar. It supports link-sensing and more.
5683 * A new tool "systemd-socket-proxyd" has been added which can
5684 act as a bidirectional proxy for TCP sockets. This is
5685 useful for adding socket activation support to services that
5686 do not actually support socket activation, including virtual
5687 machines and the like.
5689 * Add a new tool to save/restore rfkill state on
5692 * Save/restore state of keyboard backlights in addition to
5693 display backlights on shutdown/boot.
5695 * udev learned a new SECLABEL{} construct to label device
5696 nodes with a specific security label when they appear. For
5697 now, only SECLABEL{selinux} is supported, but the syntax is
5698 prepared for additional security frameworks.
5700 * udev gained a new scheme to configure link-level attributes
5701 from files in /etc/systemd/network/*.link. These files can
5702 match against MAC address, device path, driver name and type,
5703 and will apply attributes like the naming policy, link speed,
5704 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
5705 address assignment policy (randomized, ...).
5707 * The configuration of network interface naming rules for
5708 "permanent interface names" has changed: a new NamePolicy=
5709 setting in the [Link] section of .link files determines the
5710 priority of possible naming schemes (onboard, slot, MAC,
5711 path). The default value of this setting is determined by
5712 /usr/lib/net/links/99-default.link. Old
5713 80-net-name-slot.rules udev configuration file has been
5714 removed, so local configuration overriding this file should
5715 be adapted to override 99-default.link instead.
5717 * When the User= switch is used in a unit file, also
5718 initialize $SHELL= based on the user database entry.
5720 * systemd no longer depends on libdbus. All communication is
5721 now done with sd-bus, systemd's low-level bus library
5724 * kdbus support has been added to PID 1 itself. When kdbus is
5725 enabled, this causes PID 1 to set up the system bus and
5726 enable support for a new ".busname" unit type that
5727 encapsulates bus name activation on kdbus. It works a little
5728 bit like ".socket" units, except for bus names. A new
5729 generator has been added that converts classic dbus1 service
5730 activation files automatically into native systemd .busname
5733 * sd-bus: add a light-weight vtable implementation that allows
5734 defining objects on the bus with a simple static const
5735 vtable array of its methods, signals and properties.
5737 * systemd will not generate or install static dbus
5738 introspection data anymore to /usr/share/dbus-1/interfaces,
5739 as the precise format of these files is unclear, and
5740 nothing makes use of it.
5742 * A proxy daemon is now provided to proxy clients connecting
5743 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
5744 compatibility with classic D-Bus.
5746 * A bus driver implementation has been added that supports the
5747 classic D-Bus bus driver calls on kdbus, also for
5748 compatibility purposes.
5750 * A new API "sd-event.h" has been added that implements a
5751 minimal event loop API built around epoll. It provides a
5752 couple of features that direct epoll usage is lacking:
5753 prioritization of events, scales to large numbers of timer
5754 events, per-event timer slack (accuracy), system-wide
5755 coalescing of timer events, exit handlers, watchdog
5756 supervision support using systemd's sd_notify() API, child
5759 * A new API "sd-rntl.h" has been added that provides an API
5760 around the route netlink interface of the kernel, similar in
5761 style to "sd-bus.h".
5763 * A new API "sd-dhcp-client.h" has been added that provides a
5764 small DHCPv4 client-side implementation. This is used by
5767 * There is a new kernel command line option
5768 "systemd.restore_state=0|1". When set to "0", none of the
5769 systemd tools will restore saved runtime state to hardware
5770 devices. More specifically, the rfkill and backlight states
5773 * The FsckPassNo= compatibility option in mount/service units
5774 has been removed. The fstab generator will now add the
5775 necessary dependencies automatically, and does not require
5776 PID1's support for that anymore.
5778 * journalctl gained a new switch, --list-boots, that lists
5779 recent boots with their times and boot IDs.
5781 * The various tools like systemctl, loginctl, timedatectl,
5782 busctl, systemd-run, ... have gained a new switch "-M" to
5783 connect to a specific, local OS container (as direct
5784 connection, without requiring SSH). This works on any
5785 container that is registered with machined, such as those
5786 created by libvirt-lxc or nspawn.
5788 * systemd-run and systemd-analyze also gained support for "-H"
5789 to connect to remote hosts via SSH. This is particularly
5790 useful for systemd-run because it enables queuing of jobs
5791 onto remote systems.
5793 * machinectl gained a new command "login" to open a getty
5794 login in any local container. This works with any container
5795 that is registered with machined (such as those created by
5796 libvirt-lxc or nspawn), and which runs systemd inside.
5798 * machinectl gained a new "reboot" command that may be used to
5799 trigger a reboot on a specific container that is registered
5800 with machined. This works on any container that runs an init
5801 system of some kind.
5803 * systemctl gained a new "list-timers" command to print a nice
5804 listing of installed timer units with the times they elapse
5807 * Alternative reboot() parameters may now be specified on the
5808 "systemctl reboot" command line and are passed to the
5809 reboot() system call.
5811 * systemctl gained a new --job-mode= switch to configure the
5812 mode to queue a job with. This is a more generic version of
5813 --fail, --irreversible, and --ignore-dependencies, which are
5814 still available but not advertised anymore.
5816 * /etc/systemd/system.conf gained new settings to configure
5817 various default timeouts of units, as well as the default
5818 start limit interval and burst. These may still be overridden
5821 * PID1 will now export on the bus profile data of the security
5822 policy upload process (such as the SELinux policy upload to
5825 * journald: when forwarding logs to the console, include
5826 timestamps (following the setting in
5827 /sys/module/printk/parameters/time).
5829 * OnCalendar= in timer units now understands the special
5830 strings "yearly" and "annually". (Both are equivalent)
5832 * The accuracy of timer units is now configurable with the new
5833 AccuracySec= setting. It defaults to 1min.
5835 * A new dependency type JoinsNamespaceOf= has been added that
5836 allows running two services within the same /tmp and network
5837 namespace, if PrivateNetwork= or PrivateTmp= are used.
5839 * A new command "cat" has been added to systemctl. It outputs
5840 the original unit file of a unit, and concatenates the
5841 contents of additional "drop-in" unit file snippets, so that
5842 the full configuration is shown.
5844 * systemctl now supports globbing on the various "list-xyz"
5845 commands, like "list-units" or "list-sockets", as well as on
5846 those commands which take multiple unit names.
5848 * journalctl's --unit= switch gained support for globbing.
5850 * All systemd daemons now make use of the watchdog logic so
5851 that systemd automatically notices when they hang.
5853 * If the $container_ttys environment variable is set,
5854 getty-generator will automatically spawn a getty for each
5855 listed tty. This is useful for container managers to request
5856 login gettys to be spawned on as many ttys as needed.
5858 * %h, %s, %U specifier support is not available anymore when
5859 used in unit files for PID 1. This is because NSS calls are
5860 not safe from PID 1. They stay available for --user
5861 instances of systemd, and as special case for the root user.
5863 * loginctl gained a new "--no-legend" switch to turn off output
5866 * The "sd-login.h" API gained three new calls:
5867 sd_session_is_remote(), sd_session_get_remote_user(),
5868 sd_session_get_remote_host() to query information about
5871 * The udev hardware database now also carries vendor/product
5872 information of SDIO devices.
5874 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
5875 determine whether watchdog notifications are requested by
5878 * Socket-activated per-connection services now include a
5879 short description of the connection parameters in the
5882 * tmpfiles gained a new "--boot" option. When this is not used,
5883 only lines where the command character is not suffixed with
5884 "!" are executed. When this option is specified, those
5885 options are executed too. This partitions tmpfiles
5886 directives into those that can be safely executed at any
5887 time, and those which should be run only at boot (for
5888 example, a line that creates /run/nologin).
5890 * A new API "sd-resolve.h" has been added which provides a simple
5891 asynchronous wrapper around glibc NSS host name resolution
5892 calls, such as getaddrinfo(). In contrast to glibc's
5893 getaddrinfo_a(), it does not use signals. In contrast to most
5894 other asynchronous name resolution libraries, this one does
5895 not reimplement DNS, but reuses NSS, so that alternate
5896 host name resolution systems continue to work, such as mDNS,
5897 LDAP, etc. This API is based on libasyncns, but it has been
5898 cleaned up for inclusion in systemd.
5900 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
5901 "sd-daemon.h" are no longer found in individual libraries
5902 libsystemd-journal.so, libsystemd-login.so,
5903 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
5904 merged them into a single library, libsystemd.so, which
5905 provides all symbols. The reason for this is cyclic
5906 dependencies, as these libraries tend to use each other's
5907 symbols. So far, we have managed to workaround that by linking
5908 a copy of a good part of our code into each of these
5909 libraries again and again, which, however, makes certain
5910 things hard to do, like sharing static variables. Also, it
5911 substantially increases footprint. With this change, there
5912 is only one library for the basic APIs systemd
5913 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
5914 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
5915 library as well, however are subject to the --enable-kdbus
5916 switch (see below). Note that "sd-dhcp-client.h" is not part
5917 of this library (this is because it only consumes, never
5918 provides, services of/to other APIs). To make the transition
5919 easy from the separate libraries to the unified one, we
5920 provide the --enable-compat-libs compile-time switch which
5921 will generate stub libraries that are compatible with the
5922 old ones but redirect all calls to the new one.
5924 * All of the kdbus logic and the new APIs "sd-bus.h",
5925 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
5926 and "sd-utf8.h" are compile-time optional via the
5927 "--enable-kdbus" switch, and they are not compiled in by
5928 default. To make use of kdbus, you have to explicitly enable
5929 the switch. Note however, that neither the kernel nor the
5930 userspace API for all of this is considered stable yet. We
5931 want to maintain the freedom to still change the APIs for
5932 now. By specifying this build-time switch, you acknowledge
5933 that you are aware of the instability of the current
5936 * Also, note that while kdbus is pretty much complete,
5937 it lacks one thing: proper policy support. This means you
5938 can build a fully working system with all features; however,
5939 it will be highly insecure. Policy support will be added in
5940 one of the next releases, at the same time that we will
5941 declare the APIs stable.
5943 * When the kernel command line argument "kdbus" is specified,
5944 systemd will automatically load the kdbus.ko kernel module. At
5945 this stage of development, it is only useful for testing kdbus
5946 and should not be used in production. Note: if "--enable-kdbus"
5947 is specified, and the kdbus.ko kernel module is available, and
5948 "kdbus" is added to the kernel command line, the entire system
5949 runs with kdbus instead of dbus-daemon, with the above mentioned
5950 problem of missing the system policy enforcement. Also a future
5951 version of kdbus.ko or a newer systemd will not be compatible with
5952 each other, and will unlikely be able to boot the machine if only
5953 one of them is updated.
5955 * systemctl gained a new "import-environment" command which
5956 uploads the caller's environment (or parts thereof) into the
5957 service manager so that it is inherited by services started
5958 by the manager. This is useful to upload variables like
5959 $DISPLAY into the user service manager.
5961 * A new PrivateDevices= switch has been added to service units
5962 which allows running a service with a namespaced /dev
5963 directory that does not contain any device nodes for
5964 physical devices. More specifically, it only includes devices
5965 such as /dev/null, /dev/urandom, and /dev/zero which are API
5968 * logind has been extended to support behaviour like VT
5969 switching on seats that do not support a VT. This makes
5970 multi-session available on seats that are not the first seat
5971 (seat0), and on systems where kernel support for VTs has
5972 been disabled at compile-time.
5974 * If a process holds a delay lock for system sleep or shutdown
5975 and fails to release it in time, we will now log its
5976 identity. This makes it easier to identify processes that
5977 cause slow suspends or power-offs.
5979 * When parsing /etc/crypttab, support for a new key-slot=
5980 option as supported by Debian is added. It allows indicating
5981 which LUKS slot to use on disk, speeding up key loading.
5983 * The sd_journal_sendv() API call has been checked and
5984 officially declared to be async-signal-safe so that it may
5985 be invoked from signal handlers for logging purposes.
5987 * Boot-time status output is now enabled automatically after a
5988 short timeout if boot does not progress, in order to give
5989 the user an indication what she or he is waiting for.
5991 * The boot-time output has been improved to show how much time
5992 remains until jobs expire.
5994 * The KillMode= switch in service units gained a new possible
5995 value "mixed". If set, and the unit is shut down, then the
5996 initial SIGTERM signal is sent only to the main daemon
5997 process, while the following SIGKILL signal is sent to
5998 all remaining processes of the service.
6000 * When a scope unit is registered, a new property "Controller"
6001 may be set. If set to a valid bus name, systemd will send a
6002 RequestStop() signal to this name when it would like to shut
6003 down the scope. This may be used to hook manager logic into
6004 the shutdown logic of scope units. Also, scope units may now
6005 be put in a special "abandoned" state, in which case the
6006 manager process which created them takes no further
6007 responsibilities for it.
6009 * When reading unit files, systemd will now verify
6010 the access mode of these files, and warn about certain
6011 suspicious combinations. This has been added to make it
6012 easier to track down packaging bugs where unit files are
6013 marked executable or world-writable.
6015 * systemd-nspawn gained a new "--setenv=" switch to set
6016 container-wide environment variables. The similar option in
6017 systemd-activate was renamed from "--environment=" to
6018 "--setenv=" for consistency.
6020 * systemd-nspawn has been updated to create a new kdbus domain
6021 for each container that is invoked, thus allowing each
6022 container to have its own set of system and user buses,
6023 independent of the host.
6025 * systemd-nspawn gained a new --drop-capability= switch to run
6026 the container with less capabilities than the default. Both
6027 --drop-capability= and --capability= now take the special
6028 string "all" for dropping or keeping all capabilities.
6030 * systemd-nspawn gained new switches for executing containers
6031 with specific SELinux labels set.
6033 * systemd-nspawn gained a new --quiet switch to not generate
6034 any additional output but the container's own console
6037 * systemd-nspawn gained a new --share-system switch to run a
6038 container without PID namespacing enabled.
6040 * systemd-nspawn gained a new --register= switch to control
6041 whether the container is registered with systemd-machined or
6042 not. This is useful for containers that do not run full
6043 OS images, but only specific apps.
6045 * systemd-nspawn gained a new --keep-unit which may be used
6046 when invoked as the only program from a service unit, and
6047 results in registration of the unit service itself in
6048 systemd-machined, instead of a newly opened scope unit.
6050 * systemd-nspawn gained a new --network-interface= switch for
6051 moving arbitrary interfaces to the container. The new
6052 --network-veth switch creates a virtual Ethernet connection
6053 between host and container. The new --network-bridge=
6054 switch then allows assigning the host side of this virtual
6055 Ethernet connection to a bridge device.
6057 * systemd-nspawn gained a new --personality= switch for
6058 setting the kernel personality for the container. This is
6059 useful when running a 32-bit container on a 64-bit host. A
6060 similar option Personality= is now also available for service
6063 * logind will now also track a "Desktop" identifier for each
6064 session which encodes the desktop environment of it. This is
6065 useful for desktop environments that want to identify
6066 multiple running sessions of itself easily.
6068 * A new SELinuxContext= setting for service units has been
6069 added that allows setting a specific SELinux execution
6070 context for a service.
6072 * Most systemd client tools will now honour $SYSTEMD_LESS for
6073 settings of the "less" pager. By default, these tools will
6074 override $LESS to allow certain operations to work, such as
6075 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
6076 influence this logic.
6078 * systemd's "seccomp" hook-up has been changed to make use of
6079 the libseccomp library instead of using its own
6080 implementation. This has benefits for portability among
6083 * For usage together with SystemCallFilter=, a new
6084 SystemCallErrorNumber= setting has been introduced that
6085 allows configuration of a system error number to be returned
6086 on filtered system calls, instead of immediately killing the
6087 process. Also, SystemCallArchitectures= has been added to
6088 limit access to system calls of a particular architecture
6089 (in order to turn off support for unused secondary
6090 architectures). There is also a global
6091 SystemCallArchitectures= setting in system.conf now to turn
6092 off support for non-native system calls system-wide.
6094 * systemd requires a kernel with a working name_to_handle_at(),
6095 please see the kernel config requirements in the README file.
6097 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
6098 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
6099 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
6100 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
6101 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
6102 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
6103 Elia Pinto, Florian Weimer, George McCollister, Goffredo
6104 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
6105 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
6106 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
6107 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
6108 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
6109 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
6110 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
6111 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
6112 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
6113 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
6114 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
6115 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
6116 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
6117 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
6118 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
6119 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
6120 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
6122 — Berlin, 2014-02-20
6126 * logind has gained support for facilitating privileged input
6127 and drm device access for unprivileged clients. This work is
6128 useful to allow Wayland display servers (and similar
6129 programs, such as kmscon) to run under the user's ID and
6130 access input and drm devices which are normally
6131 protected. When this is used (and the kernel is new enough)
6132 logind will "mute" IO on the file descriptors passed to
6133 Wayland as long as it is in the background and "unmute" it
6134 if it returns into the foreground. This allows secure
6135 session switching without allowing background sessions to
6136 eavesdrop on input and display data. This also introduces
6137 session switching support if VT support is turned off in the
6138 kernel, and on seats that are not seat0.
6140 * A new kernel command line option luks.options= is understood
6141 now which allows specifying LUKS options for usage for LUKS
6142 encrypted partitions specified with luks.uuid=.
6144 * tmpfiles.d(5) snippets may now use specifier expansion in
6145 path names. More specifically %m, %b, %H, %v, are now
6146 replaced by the local machine id, boot id, hostname, and
6147 kernel version number.
6149 * A new tmpfiles.d(5) command "m" has been introduced which
6150 may be used to change the owner/group/access mode of a file
6151 or directory if it exists, but do nothing if it does not.
6153 * This release removes high-level support for the
6154 MemorySoftLimit= cgroup setting. The underlying kernel
6155 cgroup attribute memory.soft_limit= is currently badly
6156 designed and likely to be removed from the kernel API in its
6157 current form, hence we should not expose it for now.
6159 * The memory.use_hierarchy cgroup attribute is now enabled for
6160 all cgroups systemd creates in the memory cgroup
6161 hierarchy. This option is likely to be come the built-in
6162 default in the kernel anyway, and the non-hierarchical mode
6163 never made much sense in the intrinsically hierarchical
6166 * A new field _SYSTEMD_SLICE= is logged along with all journal
6167 messages containing the slice a message was generated
6168 from. This is useful to allow easy per-customer filtering of
6169 logs among other things.
6171 * systemd-journald will no longer adjust the group of journal
6172 files it creates to the "systemd-journal" group. Instead we
6173 rely on the journal directory to be owned by the
6174 "systemd-journal" group, and its setgid bit set, so that the
6175 kernel file system layer will automatically enforce that
6176 journal files inherit this group assignment. The reason for
6177 this change is that we cannot allow NSS look-ups from
6178 journald which would be necessary to resolve
6179 "systemd-journal" to a numeric GID, because this might
6180 create deadlocks if NSS involves synchronous queries to
6181 other daemons (such as nscd, or sssd) which in turn are
6182 logging clients of journald and might block on it, which
6183 would then dead lock. A tmpfiles.d(5) snippet included in
6184 systemd will make sure the setgid bit and group are
6185 properly set on the journal directory if it exists on every
6186 boot. However, we recommend adjusting it manually after
6187 upgrades too (or from RPM scriptlets), so that the change is
6188 not delayed until next reboot.
6190 * Backlight and random seed files in /var/lib/ have moved into
6191 the /var/lib/systemd/ directory, in order to centralize all
6192 systemd generated files in one directory.
6194 * Boot time performance measurements (as displayed by
6195 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
6196 performance information if that's available to determine how
6197 much time BIOS and boot loader initialization required. With
6198 a sufficiently new BIOS you hence no longer need to boot
6199 with Gummiboot to get access to such information.
6201 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
6202 Cristian Rodríguez, Dave Reisner, David Herrmann, David
6203 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
6204 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
6205 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
6206 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
6207 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6209 — Berlin, 2013-10-02
6213 * The Restart= option for services now understands a new
6214 on-watchdog setting, which will restart the service
6215 automatically if the service stops sending out watchdog keep
6216 alive messages (as configured with WatchdogSec=).
6218 * The getty generator (which is responsible for bringing up a
6219 getty on configured serial consoles) will no longer only
6220 start a getty on the primary kernel console but on all
6221 others, too. This makes the order in which console= is
6222 specified on the kernel command line less important.
6224 * libsystemd-logind gained a new sd_session_get_vt() call to
6225 retrieve the VT number of a session.
6227 * If the option "tries=0" is set for an entry of /etc/crypttab
6228 its passphrase is queried indefinitely instead of any
6229 maximum number of tries.
6231 * If a service with a configure PID file terminates its PID
6232 file will now be removed automatically if it still exists
6233 afterwards. This should put an end to stale PID files.
6235 * systemd-run will now also take relative binary path names
6236 for execution and no longer insists on absolute paths.
6238 * InaccessibleDirectories= and ReadOnlyDirectories= now take
6239 paths that are optionally prefixed with "-" to indicate that
6240 it should not be considered a failure if they do not exist.
6242 * journalctl -o (and similar commands) now understands a new
6243 output mode "short-precise", it is similar to "short" but
6244 shows timestamps with usec accuracy.
6246 * The option "discard" (as known from Debian) is now
6247 synonymous to "allow-discards" in /etc/crypttab. In fact,
6248 "discard" is preferred now (since it is easier to remember
6251 * Some licensing clean-ups were made, so that more code is now
6252 LGPL-2.1 licensed than before.
6254 * A minimal tool to save/restore the display backlight
6255 brightness across reboots has been added. It will store the
6256 backlight setting as late as possible at shutdown, and
6257 restore it as early as possible during reboot.
6259 * A logic to automatically discover and enable home and swap
6260 partitions on GPT disks has been added. With this in place
6261 /etc/fstab becomes optional for many setups as systemd can
6262 discover certain partitions located on the root disk
6263 automatically. Home partitions are recognized under their
6264 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
6265 partitions are recognized under their GPT type ID
6266 0657fd6da4ab43c484e50933c84b4f4f.
6268 * systemd will no longer pass any environment from the kernel
6269 or initrd to system services. If you want to set an
6270 environment for all services, do so via the kernel command
6271 line systemd.setenv= assignment.
6273 * The systemd-sysctl tool no longer natively reads the file
6274 /etc/sysctl.conf. If desired, the file should be symlinked
6275 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
6276 legacy support by a symlink rather than built-in code, it
6277 also makes the otherwise hidden order of application of the
6278 different files visible. (Note that this partly reverts to a
6279 pre-198 application order of sysctl knobs!)
6281 * The "systemctl set-log-level" and "systemctl dump" commands
6282 have been moved to systemd-analyze.
6284 * systemd-run learned the new --remain-after-exit switch,
6285 which causes the scope unit not to be cleaned up
6286 automatically after the process terminated.
6288 * tmpfiles learned a new --exclude-prefix= switch to exclude
6289 certain paths from operation.
6291 * journald will now automatically flush all messages to disk
6292 as soon as a message at the log level CRIT, ALERT or EMERG
6295 Contributions from: Andrew Cook, Brandon Philips, Christian
6296 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
6297 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
6298 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
6299 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
6300 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
6301 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
6302 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
6303 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
6304 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
6305 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
6306 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
6307 William Giokas, Zbigniew Jędrzejewski-Szmek
6309 — Berlin, 2013-09-13
6313 * The documentation has been updated to cover the various new
6314 concepts introduced with 205.
6316 * Unit files now understand the new %v specifier which
6317 resolves to the kernel version string as returned by "uname
6320 * systemctl now supports filtering the unit list output by
6321 load state, active state and sub state, using the new
6324 * "systemctl status" will now show the results of the
6325 condition checks (like ConditionPathExists= and similar) of
6326 the last start attempts of the unit. They are also logged to
6329 * "journalctl -b" may now be used to look for boot output of a
6330 specific boot. Try "journalctl -b -1" for the previous boot,
6331 but the syntax is substantially more powerful.
6333 * "journalctl --show-cursor" has been added which prints the
6334 cursor string the last shown log line. This may then be used
6335 with the new "journalctl --after-cursor=" switch to continue
6336 browsing logs from that point on.
6338 * "journalctl --force" may now be used to force regeneration
6341 * Creation of "dead" device nodes has been moved from udev
6342 into kmod and tmpfiles. Previously, udev would read the kmod
6343 databases to pre-generate dead device nodes based on meta
6344 information contained in kernel modules, so that these would
6345 be auto-loaded on access rather then at boot. As this
6346 does not really have much to do with the exposing actual
6347 kernel devices to userspace this has always been slightly
6348 alien in the udev codebase. Following the new scheme kmod
6349 will now generate a runtime snippet for tmpfiles from the
6350 module meta information and it now is tmpfiles' job to the
6351 create the nodes. This also allows overriding access and
6352 other parameters for the nodes using the usual tmpfiles
6353 facilities. As side effect this allows us to remove the
6354 CAP_SYS_MKNOD capability bit from udevd entirely.
6356 * logind's device ACLs may now be applied to these "dead"
6357 devices nodes too, thus finally allowing managed access to
6358 devices such as /dev/snd/sequencer without loading the
6359 backing module right-away.
6361 * A new RPM macro has been added that may be used to apply
6362 tmpfiles configuration during package installation.
6364 * systemd-detect-virt and ConditionVirtualization= now can
6365 detect User-Mode-Linux machines (UML).
6367 * journald will now implicitly log the effective capabilities
6368 set of processes in the message metadata.
6370 * systemd-cryptsetup has gained support for TrueCrypt volumes.
6372 * The initrd interface has been simplified (more specifically,
6373 support for passing performance data via environment
6374 variables and fsck results via files in /run has been
6375 removed). These features were non-essential, and are
6376 nowadays available in a much nicer way by having systemd in
6377 the initrd serialize its state and have the hosts systemd
6378 deserialize it again.
6380 * The udev "keymap" data files and tools to apply keyboard
6381 specific mappings of scan to key codes, and force-release
6382 scan code lists have been entirely replaced by a udev
6383 "keyboard" builtin and a hwdb data file.
6385 * systemd will now honour the kernel's "quiet" command line
6386 argument also during late shutdown, resulting in a
6387 completely silent shutdown when used.
6389 * There's now an option to control the SO_REUSEPORT socket
6390 option in .socket units.
6392 * Instance units will now automatically get a per-template
6393 subslice of system.slice unless something else is explicitly
6394 configured. For example, instances of sshd@.service will now
6395 implicitly be placed in system-sshd.slice rather than
6396 system.slice as before.
6398 * Test coverage support may now be enabled at build time.
6400 Contributions from: Dave Reisner, Frederic Crozat, Harald
6401 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
6402 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
6403 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
6404 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
6405 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
6406 Giokas, Zbigniew Jędrzejewski-Szmek
6408 — Berlin, 2013-07-23
6412 * Two new unit types have been introduced:
6414 Scope units are very similar to service units, however, are
6415 created out of pre-existing processes — instead of PID 1
6416 forking off the processes. By using scope units it is
6417 possible for system services and applications to group their
6418 own child processes (worker processes) in a powerful way
6419 which then maybe used to organize them, or kill them
6420 together, or apply resource limits on them.
6422 Slice units may be used to partition system resources in an
6423 hierarchical fashion and then assign other units to them. By
6424 default there are now three slices: system.slice (for all
6425 system services), user.slice (for all user sessions),
6426 machine.slice (for VMs and containers).
6428 Slices and scopes have been introduced primarily in
6429 context of the work to move cgroup handling to a
6430 single-writer scheme, where only PID 1
6431 creates/removes/manages cgroups.
6433 * There's a new concept of "transient" units. In contrast to
6434 normal units these units are created via an API at runtime,
6435 not from configuration from disk. More specifically this
6436 means it is now possible to run arbitrary programs as
6437 independent services, with all execution parameters passed
6438 in via bus APIs rather than read from disk. Transient units
6439 make systemd substantially more dynamic then it ever was,
6440 and useful as a general batch manager.
6442 * logind has been updated to make use of scope and slice units
6443 for managing user sessions. As a user logs in he will get
6444 his own private slice unit, to which all sessions are added
6445 as scope units. We also added support for automatically
6446 adding an instance of user@.service for the user into the
6447 slice. Effectively logind will no longer create cgroup
6448 hierarchies on its own now, it will defer entirely to PID 1
6449 for this by means of scope, service and slice units. Since
6450 user sessions this way become entities managed by PID 1
6451 the output of "systemctl" is now a lot more comprehensive.
6453 * A new mini-daemon "systemd-machined" has been added which
6454 may be used by virtualization managers to register local
6455 VMs/containers. nspawn has been updated accordingly, and
6456 libvirt will be updated shortly. machined will collect a bit
6457 of meta information about the VMs/containers, and assign
6458 them their own scope unit (see above). The collected
6459 meta-data is then made available via the "machinectl" tool,
6460 and exposed in "ps" and similar tools. machined/machinectl
6461 is compile-time optional.
6463 * As discussed earlier, the low-level cgroup configuration
6464 options ControlGroup=, ControlGroupModify=,
6465 ControlGroupPersistent=, ControlGroupAttribute= have been
6466 removed. Please use high-level attribute settings instead as
6467 well as slice units.
6469 * A new bus call SetUnitProperties() has been added to alter
6470 various runtime parameters of a unit. This is primarily
6471 useful to alter cgroup parameters dynamically in a nice way,
6472 but will be extended later on to make more properties
6473 modifiable at runtime. systemctl gained a new set-properties
6474 command that wraps this call.
6476 * A new tool "systemd-run" has been added which can be used to
6477 run arbitrary command lines as transient services or scopes,
6478 while configuring a number of settings via the command
6479 line. This tool is currently very basic, however already
6480 very useful. We plan to extend this tool to even allow
6481 queuing of execution jobs with time triggers from the
6482 command line, similar in fashion to "at".
6484 * nspawn will now inform the user explicitly that kernels with
6485 audit enabled break containers, and suggest the user to turn
6488 * Support for detecting the IMA and AppArmor security
6489 frameworks with ConditionSecurity= has been added.
6491 * journalctl gained a new "-k" switch for showing only kernel
6492 messages, mimicking dmesg output; in addition to "--user"
6493 and "--system" switches for showing only user's own logs
6496 * systemd-delta can now show information about drop-in
6497 snippets extending unit files.
6499 * libsystemd-bus has been substantially updated but is still
6500 not available as public API.
6502 * systemd will now look for the "debug" argument on the kernel
6503 command line and enable debug logging, similar to what
6504 "systemd.log_level=debug" already did before.
6506 * "systemctl set-default", "systemctl get-default" has been
6507 added to configure the default.target symlink, which
6508 controls what to boot into by default.
6510 * "systemctl set-log-level" has been added as a convenient
6511 way to raise and lower systemd logging threshold.
6513 * "systemd-analyze plot" will now show the time the various
6514 generators needed for execution, as well as information
6515 about the unit file loading.
6517 * libsystemd-journal gained a new sd_journal_open_files() call
6518 for opening specific journal files. journactl also gained a
6519 new switch to expose this new functionality. Previously we
6520 only supported opening all files from a directory, or all
6521 files from the system, as opening individual files only is
6522 racy due to journal file rotation.
6524 * systemd gained the new DefaultEnvironment= setting in
6525 /etc/systemd/system.conf to set environment variables for
6528 * If a privileged process logs a journal message with the
6529 OBJECT_PID= field set, then journald will automatically
6530 augment this with additional OBJECT_UID=, OBJECT_GID=,
6531 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
6532 system services want to log events about specific client
6533 processes. journactl/systemctl has been updated to make use
6534 of this information if all log messages regarding a specific
6537 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
6538 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
6539 Reisner, David Coppa, David King, David Strauss, Eelco
6540 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
6541 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
6542 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
6543 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
6544 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
6545 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
6546 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
6547 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
6548 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
6549 Łukasz Stelmach, 장동준
6553 * The Python bindings gained some minimal support for the APIs
6554 exposed by libsystemd-logind.
6556 * ConditionSecurity= gained support for detecting SMACK. Since
6557 this condition already supports SELinux and AppArmor we only
6558 miss IMA for this. Patches welcome!
6560 Contributions from: Karol Lewandowski, Lennart Poettering,
6561 Zbigniew Jędrzejewski-Szmek
6565 * systemd-nspawn will now create /etc/resolv.conf if
6566 necessary, before bind-mounting the host's file onto it.
6568 * systemd-nspawn will now store meta information about a
6569 container on the container's cgroup as extended attribute
6570 fields, including the root directory.
6572 * The cgroup hierarchy has been reworked in many ways. All
6573 objects any of the components systemd creates in the cgroup
6574 tree are now suffixed. More specifically, user sessions are
6575 now placed in cgroups suffixed with ".session", users in
6576 cgroups suffixed with ".user", and nspawn containers in
6577 cgroups suffixed with ".nspawn". Furthermore, all cgroup
6578 names are now escaped in a simple scheme to avoid collision
6579 of userspace object names with kernel filenames. This work
6580 is preparation for making these objects relocatable in the
6581 cgroup tree, in order to allow easy resource partitioning of
6582 these objects without causing naming conflicts.
6584 * systemctl list-dependencies gained the new switches
6585 --plain, --reverse, --after and --before.
6587 * systemd-inhibit now shows the process name of processes that
6588 have taken an inhibitor lock.
6590 * nss-myhostname will now also resolve "localhost"
6591 implicitly. This makes /etc/hosts an optional file and
6592 nicely handles that on IPv6 ::1 maps to both "localhost" and
6595 * libsystemd-logind.so gained a new call
6596 sd_get_machine_names() to enumerate running containers and
6597 VMs (currently only supported by very new libvirt and
6598 nspawn). sd_login_monitor can now be used to watch
6599 VMs/containers coming and going.
6601 * .include is not allowed recursively anymore, and only in
6602 unit files. Usually it is better to use drop-in snippets in
6603 .d/*.conf anyway, as introduced with systemd 198.
6605 * systemd-analyze gained a new "critical-chain" command that
6606 determines the slowest chain of units run during system
6607 boot-up. It is very useful for tracking down where
6608 optimizing boot time is the most beneficial.
6610 * systemd will no longer allow manipulating service paths in
6611 the name=systemd:/system cgroup tree using ControlGroup= in
6612 units. (But is still fine with it in all other dirs.)
6614 * There's a new systemd-nspawn@.service service file that may
6615 be used to easily run nspawn containers as system
6616 services. With the container's root directory in
6617 /var/lib/container/foobar it is now sufficient to run
6618 "systemctl start systemd-nspawn@foobar.service" to boot it.
6620 * systemd-cgls gained a new parameter "--machine" to list only
6621 the processes within a certain container.
6623 * ConditionSecurity= now can check for "apparmor". We still
6624 are lacking checks for SMACK and IMA for this condition
6625 check though. Patches welcome!
6627 * A new configuration file /etc/systemd/sleep.conf has been
6628 added that may be used to configure which kernel operation
6629 systemd is supposed to execute when "suspend", "hibernate"
6630 or "hybrid-sleep" is requested. This makes the new kernel
6631 "freeze" state accessible to the user.
6633 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
6634 the passed argument if applicable.
6636 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
6637 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
6638 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
6639 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
6640 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
6641 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
6642 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
6647 * The output of 'systemctl list-jobs' got some polishing. The
6648 '--type=' argument may now be passed more than once. A new
6649 command 'systemctl list-sockets' has been added which shows
6650 a list of kernel sockets systemd is listening on with the
6651 socket units they belong to, plus the units these socket
6654 * The experimental libsystemd-bus library got substantial
6655 updates to work in conjunction with the (also experimental)
6656 kdbus kernel project. It works well enough to exchange
6657 messages with some sophistication. Note that kdbus is not
6658 ready yet, and the library is mostly an elaborate test case
6659 for now, and not installable.
6661 * systemd gained a new unit 'systemd-static-nodes.service'
6662 that generates static device nodes earlier during boot, and
6663 can run in conjunction with udev.
6665 * libsystemd-login gained a new call sd_pid_get_user_unit()
6666 to retrieve the user systemd unit a process is running
6667 in. This is useful for systems where systemd is used as
6670 * systemd-nspawn now places all containers in the new /machine
6671 top-level cgroup directory in the name=systemd
6672 hierarchy. libvirt will soon do the same, so that we get a
6673 uniform separation of /system, /user and /machine for system
6674 services, user processes and containers/virtual
6675 machines. This new cgroup hierarchy is also useful to stick
6676 stable names to specific container instances, which can be
6677 recognized later this way (this name may be controlled
6678 via systemd-nspawn's new -M switch). libsystemd-login also
6679 gained a new call sd_pid_get_machine_name() to retrieve the
6680 name of the container/VM a specific process belongs to.
6682 * bootchart can now store its data in the journal.
6684 * libsystemd-journal gained a new call
6685 sd_journal_add_conjunction() for AND expressions to the
6686 matching logic. This can be used to express more complex
6687 logical expressions.
6689 * journactl can now take multiple --unit= and --user-unit=
6692 * The cryptsetup logic now understands the "luks.key=" kernel
6693 command line switch for specifying a file to read the
6694 decryption key from. Also, if a configured key file is not
6695 found the tool will now automatically fall back to prompting
6698 * Python systemd.journal module was updated to wrap recently
6699 added functions from libsystemd-journal. The interface was
6700 changed to bring the low level interface in s.j._Reader
6701 closer to the C API, and the high level interface in
6702 s.j.Reader was updated to wrap and convert all data about
6705 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
6706 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
6707 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
6708 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
6709 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
6710 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6714 * journalctl --update-catalog now understands a new --root=
6715 option to operate on catalogs found in a different root
6718 * During shutdown after systemd has terminated all running
6719 services a final killing loop kills all remaining left-over
6720 processes. We will now print the name of these processes
6721 when we send SIGKILL to them, since this usually indicates a
6724 * If /etc/crypttab refers to password files stored on
6725 configured mount points automatic dependencies will now be
6726 generated to ensure the specific mount is established first
6727 before the key file is attempted to be read.
6729 * 'systemctl status' will now show information about the
6730 network sockets a socket unit is listening on.
6732 * 'systemctl status' will also shown information about any
6733 drop-in configuration file for units. (Drop-In configuration
6734 files in this context are files such as
6735 /etc/systemd/systemd/foobar.service.d/*.conf)
6737 * systemd-cgtop now optionally shows summed up CPU times of
6738 cgroups. Press '%' while running cgtop to switch between
6739 percentage and absolute mode. This is useful to determine
6740 which cgroups use up the most CPU time over the entire
6741 runtime of the system. systemd-cgtop has also been updated
6742 to be 'pipeable' for processing with further shell tools.
6744 * 'hostnamectl set-hostname' will now allow setting of FQDN
6747 * The formatting and parsing of time span values has been
6748 changed. The parser now understands fractional expressions
6749 such as "5.5h". The formatter will now output fractional
6750 expressions for all time spans under 1min, i.e. "5.123456s"
6751 rather than "5s 123ms 456us". For time spans under 1s
6752 millisecond values are shown, for those under 1ms
6753 microsecond values are shown. This should greatly improve
6754 all time-related output of systemd.
6756 * libsystemd-login and libsystemd-journal gained new
6757 functions for querying the poll() events mask and poll()
6758 timeout value for integration into arbitrary event
6761 * localectl gained the ability to list available X11 keymaps
6762 (models, layouts, variants, options).
6764 * 'systemd-analyze dot' gained the ability to filter for
6765 specific units via shell-style globs, to create smaller,
6766 more useful graphs. I.e. it is now possible to create simple
6767 graphs of all the dependencies between only target units, or
6768 of all units that Avahi has dependencies with.
6770 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
6771 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
6772 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
6773 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
6774 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
6775 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
6776 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
6780 * The boot-time readahead implementation for rotating media
6781 will now read the read-ahead data in multiple passes which
6782 consist of all read requests made in equidistant time
6783 intervals. This means instead of strictly reading read-ahead
6784 data in its physical order on disk we now try to find a
6785 middle ground between physical and access time order.
6787 * /etc/os-release files gained a new BUILD_ID= field for usage
6788 on operating systems that provide continuous builds of OS
6791 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
6792 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
6793 William Douglas, Zbigniew Jędrzejewski-Szmek
6797 * systemd-python gained an API exposing libsystemd-daemon.
6799 * The SMACK setup logic gained support for uploading CIPSO
6802 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
6803 ReadOnlyDirectories= and InaccessibleDirectories= has
6804 changed. The private /tmp and /var/tmp directories are now
6805 shared by all processes of a service (which means
6806 ExecStartPre= may now leave data in /tmp that ExecStart= of
6807 the same service can still access). When a service is
6808 stopped its temporary directories are immediately deleted
6809 (normal clean-up with tmpfiles is still done in addition to
6812 * By default, systemd will now set a couple of sysctl
6813 variables in the kernel: the safe sysrq options are turned
6814 on, IP route verification is turned on, and source routing
6815 disabled. The recently added hardlink and softlink
6816 protection of the kernel is turned on. These settings should
6817 be reasonably safe, and good defaults for all new systems.
6819 * The predictable network naming logic may now be turned off
6820 with a new kernel command line switch: net.ifnames=0.
6822 * A new libsystemd-bus module has been added that implements a
6823 pretty complete D-Bus client library. For details see:
6825 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
6827 * journald will now explicitly flush the journal files to disk
6828 at the latest 5min after each write. The file will then also
6829 be marked offline until the next write. This should increase
6830 reliability in case of a crash. The synchronization delay
6831 can be configured via SyncIntervalSec= in journald.conf.
6833 * There's a new remote-fs-setup.target unit that can be used
6834 to pull in specific services when at least one remote file
6835 system is to be mounted.
6837 * There are new targets timers.target and paths.target as
6838 canonical targets to pull user timer and path units in
6839 from. This complements sockets.target with a similar
6840 purpose for socket units.
6842 * libudev gained a new call udev_device_set_attribute_value()
6843 to set sysfs attributes of a device.
6845 * The udev daemon now sets the default number of worker
6846 processes executed in parallel based on the number of available
6847 CPUs instead of the amount of available RAM. This is supposed
6848 to provide a more reliable default and limit a too aggressive
6849 parallelism for setups with 1000s of devices connected.
6851 Contributions from: Auke Kok, Colin Walters, Cristian
6852 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
6853 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
6854 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
6855 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
6856 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
6857 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
6858 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
6859 Zbigniew Jędrzejewski-Szmek
6863 * Configuration of unit files may now be extended via drop-in
6864 files without having to edit/override the unit files
6865 themselves. More specifically, if the administrator wants to
6866 change one value for a service file foobar.service he can
6867 now do so by dropping in a configuration snippet into
6868 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
6869 will load all these snippets and apply them on top of the
6870 main unit configuration file, possibly extending or
6871 overriding its settings. Using these drop-in snippets is
6872 generally nicer than the two earlier options for changing
6873 unit files locally: copying the files from
6874 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
6875 them there; or creating a new file in /etc/systemd/system/
6876 that incorporates the original one via ".include". Drop-in
6877 snippets into these .d/ directories can be placed in any
6878 directory systemd looks for units in, and the usual
6879 overriding semantics between /usr/lib, /etc and /run apply
6882 * Most unit file settings which take lists of items can now be
6883 reset by assigning the empty string to them. For example,
6884 normally, settings such as Environment=FOO=BAR append a new
6885 environment variable assignment to the environment block,
6886 each time they are used. By assigning Environment= the empty
6887 string the environment block can be reset to empty. This is
6888 particularly useful with the .d/*.conf drop-in snippets
6889 mentioned above, since this adds the ability to reset list
6890 settings from vendor unit files via these drop-ins.
6892 * systemctl gained a new "list-dependencies" command for
6893 listing the dependencies of a unit recursively.
6895 * Inhibitors are now honored and listed by "systemctl
6896 suspend", "systemctl poweroff" (and similar) too, not only
6897 GNOME. These commands will also list active sessions by
6900 * Resource limits (as exposed by the various control group
6901 controllers) can now be controlled dynamically at runtime
6902 for all units. More specifically, you can now use a command
6903 like "systemctl set-cgroup-attr foobar.service cpu.shares
6904 2000" to alter the CPU shares a specific service gets. These
6905 settings are stored persistently on disk, and thus allow the
6906 administrator to easily adjust the resource usage of
6907 services with a few simple commands. This dynamic resource
6908 management logic is also available to other programs via the
6909 bus. Almost any kernel cgroup attribute and controller is
6912 * systemd-vconsole-setup will now copy all font settings to
6913 all allocated VTs, where it previously applied them only to
6916 * libsystemd-login gained the new sd_session_get_tty() API
6919 * This release drops support for a few legacy or
6920 distribution-specific LSB facility names when parsing init
6921 scripts: $x-display-manager, $mail-transfer-agent,
6922 $mail-transport-agent, $mail-transfer-agent, $smtp,
6923 $null. Also, the mail-transfer-agent.target unit backing
6924 this has been removed. Distributions which want to retain
6925 compatibility with this should carry the burden for
6926 supporting this themselves and patch support for these back
6927 in, if they really need to. Also, the facilities $syslog and
6928 $local_fs are now ignored, since systemd does not support
6929 early-boot LSB init scripts anymore, and these facilities
6930 are implied anyway for normal services. syslog.target has
6933 * There are new bus calls on PID1's Manager object for
6934 cancelling jobs, and removing snapshot units. Previously,
6935 both calls were only available on the Job and Snapshot
6938 * systemd-journal-gatewayd gained SSL support.
6940 * The various "environment" files, such as /etc/locale.conf
6941 now support continuation lines with a backslash ("\") as
6942 last character in the line, similarly in style (but different)
6943 to how this is supported in shells.
6945 * For normal user processes the _SYSTEMD_USER_UNIT= field is
6946 now implicitly appended to every log entry logged. systemctl
6947 has been updated to filter by this field when operating on a
6948 user systemd instance.
6950 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
6951 CAP_AUDIT_CONTROL capabilities to the capabilities set for
6952 the container. This makes it easier to boot unmodified
6953 Fedora systems in a container, which however still requires
6954 audit=0 to be passed on the kernel command line. Auditing in
6955 kernel and userspace is unfortunately still too broken in
6956 context of containers, hence we recommend compiling it out
6957 of the kernel or using audit=0. Hopefully this will be fixed
6958 one day for good in the kernel.
6960 * nspawn gained the new --bind= and --bind-ro= parameters to
6961 bind mount specific directories from the host into the
6964 * nspawn will now mount its own devpts file system instance
6965 into the container, in order not to leak pty devices from
6966 the host into the container.
6968 * systemd will now read the firmware boot time performance
6969 information from the EFI variables, if the used boot loader
6970 supports this, and takes it into account for boot performance
6971 analysis via "systemd-analyze". This is currently supported
6972 only in conjunction with Gummiboot, but could be supported
6973 by other boot loaders too. For details see:
6975 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
6977 * A new generator has been added that automatically mounts the
6978 EFI System Partition (ESP) to /boot, if that directory
6979 exists, is empty, and no other file system has been
6980 configured to be mounted there.
6982 * logind will now send out PrepareForSleep(false) out
6983 unconditionally, after coming back from suspend. This may be
6984 used by applications as asynchronous notification for
6985 system resume events.
6987 * "systemctl unlock-sessions" has been added, that allows
6988 unlocking the screens of all user sessions at once, similar
6989 to how "systemctl lock-sessions" already locked all users
6990 sessions. This is backed by a new D-Bus call UnlockSessions().
6992 * "loginctl seat-status" will now show the master device of a
6993 seat. (i.e. the device of a seat that needs to be around for
6994 the seat to be considered available, usually the graphics
6997 * tmpfiles gained a new "X" line type, that allows
6998 configuration of files and directories (with wildcards) that
6999 shall be excluded from automatic cleanup ("aging").
7001 * udev default rules set the device node permissions now only
7002 at "add" events, and do not change them any longer with a
7003 later "change" event.
7005 * The log messages for lid events and power/sleep keypresses
7006 now carry a message ID.
7008 * We now have a substantially larger unit test suite, but this
7009 continues to be work in progress.
7011 * udevadm hwdb gained a new --root= parameter to change the
7012 root directory to operate relative to.
7014 * logind will now issue a background sync() request to the kernel
7015 early at shutdown, so that dirty buffers are flushed to disk early
7016 instead of at the last moment, in order to optimize shutdown
7019 * A new bootctl tool has been added that is an interface for
7020 certain boot loader operations. This is currently a preview
7021 and is likely to be extended into a small mechanism daemon
7022 like timedated, localed, hostnamed, and can be used by
7023 graphical UIs to enumerate available boot options, and
7024 request boot into firmware operations.
7026 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
7027 the rest of the package. It also has been updated to work
7028 correctly in initrds.
7030 * Policykit previously has been runtime optional, and is now
7031 also compile time optional via a configure switch.
7033 * systemd-analyze has been reimplemented in C. Also "systemctl
7034 dot" has moved into systemd-analyze.
7036 * "systemctl status" with no further parameters will now print
7037 the status of all active or failed units.
7039 * Operations such as "systemctl start" can now be executed
7040 with a new mode "--irreversible" which may be used to queue
7041 operations that cannot accidentally be reversed by a later
7042 job queuing. This is by default used to make shutdown
7043 requests more robust.
7045 * The Python API of systemd now gained a new module for
7046 reading journal files.
7048 * A new tool kernel-install has been added that can install
7049 kernel images according to the Boot Loader Specification:
7051 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
7053 * Boot time console output has been improved to provide
7054 animated boot time output for hanging jobs.
7056 * A new tool systemd-activate has been added which can be used
7057 to test socket activation with, directly from the command
7058 line. This should make it much easier to test and debug
7059 socket activation in daemons.
7061 * journalctl gained a new "--reverse" (or -r) option to show
7062 journal output in reverse order (i.e. newest line first).
7064 * journalctl gained a new "--pager-end" (or -e) option to jump
7065 to immediately jump to the end of the journal in the
7066 pager. This is only supported in conjunction with "less".
7068 * journalctl gained a new "--user-unit=" option, that works
7069 similarly to "--unit=" but filters for user units rather than
7072 * A number of unit files to ease adoption of systemd in
7073 initrds has been added. This moves some minimal logic from
7074 the various initrd implementations into systemd proper.
7076 * The journal files are now owned by a new group
7077 "systemd-journal", which exists specifically to allow access
7078 to the journal, and nothing else. Previously, we used the
7079 "adm" group for that, which however possibly covers more
7080 than just journal/log file access. This new group is now
7081 already used by systemd-journal-gatewayd to ensure this
7082 daemon gets access to the journal files and as little else
7083 as possible. Note that "make install" will also set FS ACLs
7084 up for /var/log/journal to give "adm" and "wheel" read
7085 access to it, in addition to "systemd-journal" which owns
7086 the journal files. We recommend that packaging scripts also
7087 add read access to "adm" + "wheel" to /var/log/journal, and
7088 all existing/future journal files. To normal users and
7089 administrators little changes, however packagers need to
7090 ensure to create the "systemd-journal" system group at
7091 package installation time.
7093 * The systemd-journal-gatewayd now runs as unprivileged user
7094 systemd-journal-gateway:systemd-journal-gateway. Packaging
7095 scripts need to create these system user/group at
7098 * timedated now exposes a new boolean property CanNTP that
7099 indicates whether a local NTP service is available or not.
7101 * systemd-detect-virt will now also detect xen PVs
7103 * The pstore file system is now mounted by default, if it is
7106 * In addition to the SELinux and IMA policies we will now also
7107 load SMACK policies at early boot.
7109 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
7110 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
7111 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
7112 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
7113 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
7114 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
7115 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
7116 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
7117 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
7118 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
7119 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
7120 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
7121 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
7122 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
7126 * Timer units now support calendar time events in addition to
7127 monotonic time events. That means you can now trigger a unit
7128 based on a calendar time specification such as "Thu,Fri
7129 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
7130 or fifth day of any month of the year 2013, given that it is
7131 a thursday or friday. This brings timer event support
7132 considerably closer to cron's capabilities. For details on
7133 the supported calendar time specification language see
7136 * udev now supports a number of different naming policies for
7137 network interfaces for predictable names, and a combination
7138 of these policies is now the default. Please see this wiki
7139 document for details:
7141 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
7143 * Auke Kok's bootchart implementation has been added to the
7144 systemd tree. It is an optional component that can graph the
7145 boot in quite some detail. It is one of the best bootchart
7146 implementations around and minimal in its code and
7149 * nss-myhostname has been integrated into the systemd source
7150 tree. nss-myhostname guarantees that the local hostname
7151 always stays resolvable via NSS. It has been a weak
7152 requirement of systemd-hostnamed since a long time, and
7153 since its code is actually trivial we decided to just
7154 include it in systemd's source tree. It can be turned off
7155 with a configure switch.
7157 * The read-ahead logic is now capable of properly detecting
7158 whether a btrfs file system is on SSD or rotating media, in
7159 order to optimize the read-ahead scheme. Previously, it was
7160 only capable of detecting this on traditional file systems
7163 * In udev, additional device properties are now read from the
7164 IAB in addition to the OUI database. Also, Bluetooth company
7165 identities are attached to the devices as well.
7167 * In service files %U may be used as specifier that is
7168 replaced by the configured user name of the service.
7170 * nspawn may now be invoked without a controlling TTY. This
7171 makes it suitable for invocation as its own service. This
7172 may be used to set up a simple containerized server system
7173 using only core OS tools.
7175 * systemd and nspawn can now accept socket file descriptors
7176 when they are started for socket activation. This enables
7177 implementation of socket activated nspawn
7178 containers. i.e. think about autospawning an entire OS image
7179 when the first SSH or HTTP connection is received. We expect
7180 that similar functionality will also be added to libvirt-lxc
7183 * journalctl will now suppress ANSI color codes when
7184 presenting log data.
7186 * systemctl will no longer show control group information for
7187 a unit if the control group is empty anyway.
7189 * logind can now automatically suspend/hibernate/shutdown the
7192 * /etc/machine-info and hostnamed now also expose the chassis
7193 type of the system. This can be used to determine whether
7194 the local system is a laptop, desktop, handset or
7195 tablet. This information may either be configured by the
7196 user/vendor or is automatically determined from ACPI and DMI
7197 information if possible.
7199 * A number of PolicyKit actions are now bound together with
7200 "imply" rules. This should simplify creating UIs because
7201 many actions will now authenticate similar ones as well.
7203 * Unit files learnt a new condition ConditionACPower= which
7204 may be used to conditionalize a unit depending on whether an
7205 AC power source is connected or not, of whether the system
7206 is running on battery power.
7208 * systemctl gained a new "is-failed" verb that may be used in
7209 shell scripts and suchlike to check whether a specific unit
7210 is in the "failed" state.
7212 * The EnvironmentFile= setting in unit files now supports file
7213 globbing, and can hence be used to easily read a number of
7214 environment files at once.
7216 * systemd will no longer detect and recognize specific
7217 distributions. All distribution-specific #ifdeffery has been
7218 removed, systemd is now fully generic and
7219 distribution-agnostic. Effectively, not too much is lost as
7220 a lot of the code is still accessible via explicit configure
7221 switches. However, support for some distribution specific
7222 legacy configuration file formats has been dropped. We
7223 recommend distributions to simply adopt the configuration
7224 files everybody else uses now and convert the old
7225 configuration from packaging scripts. Most distributions
7226 already did that. If that's not possible or desirable,
7227 distributions are welcome to forward port the specific
7228 pieces of code locally from the git history.
7230 * When logging a message about a unit systemd will now always
7231 log the unit name in the message meta data.
7233 * localectl will now also discover system locale data that is
7234 not stored in locale archives, but directly unpacked.
7236 * logind will no longer unconditionally use framebuffer
7237 devices as seat masters, i.e. as devices that are required
7238 to be existing before a seat is considered preset. Instead,
7239 it will now look for all devices that are tagged as
7240 "seat-master" in udev. By default, framebuffer devices will
7241 be marked as such, but depending on local systems, other
7242 devices might be marked as well. This may be used to
7243 integrate graphics cards using closed source drivers (such
7244 as NVidia ones) more nicely into logind. Note however, that
7245 we recommend using the open source NVidia drivers instead,
7246 and no udev rules for the closed-source drivers will be
7247 shipped from us upstream.
7249 Contributions from: Adam Williamson, Alessandro Crismani, Auke
7250 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
7251 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
7252 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
7253 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
7254 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
7255 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
7256 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
7257 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
7258 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
7259 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
7260 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
7265 * udev gained support for loading additional device properties
7266 from an indexed database that is keyed by vendor/product IDs
7267 and similar device identifiers. For the beginning this
7268 "hwdb" is populated with data from the well-known PCI and
7269 USB database, but also includes PNP, ACPI and OID data. In
7270 the longer run this indexed database shall grow into
7271 becoming the one central database for non-essential
7272 userspace device metadata. Previously, data from the PCI/USB
7273 database was only attached to select devices, since the
7274 lookup was a relatively expensive operation due to O(n) time
7275 complexity (with n being the number of entries in the
7276 database). Since this is now O(1), we decided to add in this
7277 data for all devices where this is available, by
7278 default. Note that the indexed database needs to be rebuilt
7279 when new data files are installed. To achieve this you need
7280 to update your packaging scripts to invoke "udevadm hwdb
7281 --update" after installation of hwdb data files. For
7282 RPM-based distributions we introduced the new
7283 %udev_hwdb_update macro for this purpose.
7285 * The Journal gained support for the "Message Catalog", an
7286 indexed database to link up additional information with
7287 journal entries. For further details please check:
7289 https://www.freedesktop.org/wiki/Software/systemd/catalog
7291 The indexed message catalog database also needs to be
7292 rebuilt after installation of message catalog files. Use
7293 "journalctl --update-catalog" for this. For RPM-based
7294 distributions we introduced the %journal_catalog_update
7295 macro for this purpose.
7297 * The Python Journal bindings gained support for the standard
7298 Python logging framework.
7300 * The Journal API gained new functions for checking whether
7301 the underlying file system of a journal file is capable of
7302 properly reporting file change notifications, or whether
7303 applications that want to reflect journal changes "live"
7304 need to recheck journal files continuously in appropriate
7307 * It is now possible to set the "age" field for tmpfiles
7308 entries to 0, indicating that files matching this entry
7309 shall always be removed when the directories are cleaned up.
7311 * coredumpctl gained a new "gdb" verb which invokes gdb
7312 right-away on the selected coredump.
7314 * There's now support for "hybrid sleep" on kernels that
7315 support this, in addition to "suspend" and "hibernate". Use
7316 "systemctl hybrid-sleep" to make use of this.
7318 * logind's HandleSuspendKey= setting (and related settings)
7319 now gained support for a new "lock" setting to simply
7320 request the screen lock on all local sessions, instead of
7321 actually executing a suspend or hibernation.
7323 * systemd will now mount the EFI variables file system by
7326 * Socket units now gained support for configuration of the
7327 SMACK security label.
7329 * timedatectl will now output the time of the last and next
7330 daylight saving change.
7332 * We dropped support for various legacy and distro-specific
7333 concepts, such as insserv, early-boot SysV services
7334 (i.e. those for non-standard runlevels such as 'b' or 'S')
7335 or ArchLinux /etc/rc.conf support. We recommend the
7336 distributions who still need support this to either continue
7337 to maintain the necessary patches downstream, or find a
7338 different solution. (Talk to us if you have questions!)
7340 * Various systemd components will now bypass PolicyKit checks
7341 for root and otherwise handle properly if PolicyKit is not
7342 found to be around. This should fix most issues for
7343 PolicyKit-less systems. Quite frankly this should have been
7344 this way since day one. It is absolutely our intention to
7345 make systemd work fine on PolicyKit-less systems, and we
7346 consider it a bug if something does not work as it should if
7347 PolicyKit is not around.
7349 * For embedded systems it is now possible to build udev and
7350 systemd without blkid and/or kmod support.
7352 * "systemctl switch-root" is now capable of switching root
7353 more than once. I.e. in addition to transitions from the
7354 initrd to the host OS it is now possible to transition to
7355 further OS images from the host. This is useful to implement
7356 offline updating tools.
7358 * Various other additions have been made to the RPM macros
7359 shipped with systemd. Use %udev_rules_update() after
7360 installing new udev rules files. %_udevhwdbdir,
7361 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
7362 %_sysctldir are now available which resolve to the right
7363 directories for packages to place various data files in.
7365 * journalctl gained the new --full switch (in addition to
7366 --all, to disable ellipsation for long messages.
7368 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
7369 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
7370 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
7371 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
7372 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
7373 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
7374 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
7375 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
7376 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
7380 * journalctl gained new --since= and --until= switches to
7381 filter by time. It also now supports nice filtering for
7382 units via --unit=/-u.
7384 * Type=oneshot services may use ExecReload= and do the
7387 * The journal daemon now supports time-based rotation and
7388 vacuuming, in addition to the usual disk-space based
7391 * The journal will now index the available field values for
7392 each field name. This enables clients to show pretty drop
7393 downs of available match values when filtering. The bash
7394 completion of journalctl has been updated
7395 accordingly. journalctl gained a new switch -F to list all
7396 values a certain field takes in the journal database.
7398 * More service events are now written as structured messages
7399 to the journal, and made recognizable via message IDs.
7401 * The timedated, localed and hostnamed mini-services which
7402 previously only provided support for changing time, locale
7403 and hostname settings from graphical DEs such as GNOME now
7404 also have a minimal (but very useful) text-based client
7405 utility each. This is probably the nicest way to changing
7406 these settings from the command line now, especially since
7407 it lists available options and is fully integrated with bash
7410 * There's now a new tool "systemd-coredumpctl" to list and
7411 extract coredumps from the journal.
7413 * We now install a README each in /var/log/ and
7414 /etc/rc.d/init.d explaining where the system logs and init
7415 scripts went. This hopefully should help folks who go to
7416 that dirs and look into the otherwise now empty void and
7417 scratch their heads.
7419 * When user-services are invoked (by systemd --user) the
7420 $MANAGERPID env var is set to the PID of systemd.
7422 * SIGRTMIN+24 when sent to a --user instance will now result
7423 in immediate termination of systemd.
7425 * gatewayd received numerous feature additions such as a
7426 "follow" mode, for live syncing and filtering.
7428 * browse.html now allows filtering and showing detailed
7429 information on specific entries. Keyboard navigation and
7430 mouse screen support has been added.
7432 * gatewayd/journalctl now supports HTML5/JSON
7433 Server-Sent-Events as output.
7435 * The SysV init script compatibility logic will now
7436 heuristically determine whether a script supports the
7437 "reload" verb, and only then make this available as
7440 * "systemctl status --follow" has been removed, use "journalctl
7443 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
7444 have been removed since they are hardly useful to be
7447 * And I'd like to take the opportunity to specifically mention
7448 Zbigniew for his great contributions. Zbigniew, you rock!
7450 Contributions from: Andrew Eikum, Christian Hesse, Colin
7451 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
7452 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
7453 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
7454 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
7455 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
7456 Jędrzejewski-Szmek, Сковорода Никита Андреевич
7460 * If /etc/vconsole.conf is non-existent or empty we will no
7461 longer load any console font or key map at boot by
7462 default. Instead the kernel defaults will be left
7463 intact. This is definitely the right thing to do, as no
7464 configuration should mean no configuration, and hard-coding
7465 font names that are different on all archs is probably a bad
7466 idea. Also, the kernel default key map and font should be
7467 good enough for most cases anyway, and mostly identical to
7468 the userspace fonts/key maps we previously overloaded them
7469 with. If distributions want to continue to default to a
7470 non-kernel font or key map they should ship a default
7471 /etc/vconsole.conf with the appropriate contents.
7473 Contributions from: Colin Walters, Daniel J Walsh, Dave
7474 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
7475 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
7479 * journalctl gained a new --cursor= switch to show entries
7480 starting from the specified location in the journal.
7482 * We now enforce a size limit on journal entry fields exported
7483 with "-o json" in journalctl. Fields larger than 4K will be
7484 assigned null. This can be turned off with --all.
7486 * An (optional) journal gateway daemon is now available as
7487 "systemd-journal-gatewayd.service". This service provides
7488 access to the journal via HTTP and JSON. This functionality
7489 will be used to implement live log synchronization in both
7490 pull and push modes, but has various other users too, such
7491 as easy log access for debugging of embedded devices. Right
7492 now it is already useful to retrieve the journal via HTTP:
7494 # systemctl start systemd-journal-gatewayd.service
7495 # wget http://localhost:19531/entries
7497 This will download the journal contents in a
7498 /var/log/messages compatible format. The same as JSON:
7500 # curl -H"Accept: application/json" http://localhost:19531/entries
7502 This service is also accessible via a web browser where a
7503 single static HTML5 app is served that uses the JSON logic
7504 to enable the user to do some basic browsing of the
7505 journal. This will be extended later on. Here's an example
7506 screenshot of this app in its current state:
7508 http://0pointer.de/public/journal-gatewayd
7510 Contributions from: Kay Sievers, Lennart Poettering, Robert
7511 Milasan, Tom Gundersen
7515 * The bash completion logic is now available for journalctl
7518 * We do not mount the "cpuset" controller anymore together with
7519 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
7520 started if no parameters are assigned to it. "cpuset" hence
7521 broke code that assumed it could create "cpu" groups and
7524 * journalctl -f will now subscribe to terminal size changes,
7525 and line break accordingly.
7527 Contributions from: Dave Reisner, Kay Sievers, Lennart
7528 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
7532 * nspawn will now create a symlink /etc/localtime in the
7533 container environment, copying the host's timezone
7534 setting. Previously this has been done via a bind mount, but
7535 since symlinks cannot be bind mounted this has now been
7536 changed to create/update the appropriate symlink.
7538 * journalctl -n's line number argument is now optional, and
7539 will default to 10 if omitted.
7541 * journald will now log the maximum size the journal files may
7542 take up on disk. This is particularly useful if the default
7543 built-in logic of determining this parameter from the file
7544 system size is used. Use "systemctl status
7545 systemd-journald.service" to see this information.
7547 * The multi-seat X wrapper tool has been stripped down. As X
7548 is now capable of enumerating graphics devices via udev in a
7549 seat-aware way the wrapper is not strictly necessary
7550 anymore. A stripped down temporary stop-gap is still shipped
7551 until the upstream display managers have been updated to
7552 fully support the new X logic. Expect this wrapper to be
7553 removed entirely in one of the next releases.
7555 * HandleSleepKey= in logind.conf has been split up into
7556 HandleSuspendKey= and HandleHibernateKey=. The old setting
7557 is not available anymore. X11 and the kernel are
7558 distinguishing between these keys and we should too. This
7559 also means the inhibition lock for these keys has been split
7562 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
7563 Poettering, Lukas Nykryn, Václav Pavlín
7567 * Whenever a unit changes state we will now log this to the
7568 journal and show along the unit's own log output in
7571 * ConditionPathIsMountPoint= can now properly detect bind
7572 mount points too. (Previously, a bind mount of one file
7573 system to another place in the same file system could not be
7574 detected as mount, since they shared struct stat's st_dev
7577 * We will now mount the cgroup controllers cpu, cpuacct,
7578 cpuset and the controllers net_cls, net_prio together by
7581 * nspawn containers will now have a virtualized boot
7582 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
7583 over with a randomized ID at container initialization). This
7584 has the effect of making "journalctl -b" do the right thing
7587 * The JSON output journal serialization has been updated not
7588 to generate "endless" list objects anymore, but rather one
7589 JSON object per line. This is more in line how most JSON
7590 parsers expect JSON objects. The new output mode
7591 "json-pretty" has been added to provide similar output, but
7592 neatly aligned for readability by humans.
7594 * We dropped all explicit sync() invocations in the shutdown
7595 code. The kernel does this implicitly anyway in the kernel
7596 reboot() syscall. halt(8)'s -n option is now a compatibility
7599 * We now support virtualized reboot() in containers, as
7600 supported by newer kernels. We will fall back to exit() if
7601 CAP_SYS_REBOOT is not available to the container. Also,
7602 nspawn makes use of this now and will actually reboot the
7603 container if the containerized OS asks for that.
7605 * journalctl will only show local log output by default
7606 now. Use --merge (-m) to show remote log output, too.
7608 * libsystemd-journal gained the new sd_journal_get_usage()
7609 call to determine the current disk usage of all journal
7610 files. This is exposed in the new "journalctl --disk-usage"
7613 * journald gained a new configuration setting SplitMode= in
7614 journald.conf which may be used to control how user journals
7615 are split off. See journald.conf(5) for details.
7617 * A new condition type ConditionFileNotEmpty= has been added.
7619 * tmpfiles' "w" lines now support file globbing, to write
7620 multiple files at once.
7622 * We added Python bindings for the journal submission
7623 APIs. More Python APIs for a number of selected APIs will
7624 likely follow. Note that we intend to add native bindings
7625 only for the Python language, as we consider it common
7626 enough to deserve bindings shipped within systemd. There are
7627 various projects outside of systemd that provide bindings
7628 for languages such as PHP or Lua.
7630 * Many conditions will now resolve specifiers such as %i. In
7631 addition, PathChanged= and related directives of .path units
7632 now support specifiers as well.
7634 * There's now a new RPM macro definition for the system preset
7637 * journald will now warn if it ca not forward a message to the
7638 syslog daemon because its socket is full.
7640 * timedated will no longer write or process /etc/timezone,
7641 except on Debian. As we do not support late mounted /usr
7642 anymore /etc/localtime always being a symlink is now safe,
7643 and hence the information in /etc/timezone is not necessary
7646 * logind will now always reserve one VT for a text getty (VT6
7647 by default). Previously if more than 6 X sessions where
7648 started they took up all the VTs with auto-spawned gettys,
7649 so that no text gettys were available anymore.
7651 * udev will now automatically inform the btrfs kernel logic
7652 about btrfs RAID components showing up. This should make
7653 simple hotplug based btrfs RAID assembly work.
7655 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
7656 (but not for its children which will stay at the kernel
7657 default). This should allow setups with a lot more listening
7660 * systemd will now always pass the configured timezone to the
7661 kernel at boot. timedated will do the same when the timezone
7664 * logind's inhibition logic has been updated. By default,
7665 logind will now handle the lid switch, the power and sleep
7666 keys all the time, even in graphical sessions. If DEs want
7667 to handle these events on their own they should take the new
7668 handle-power-key, handle-sleep-key and handle-lid-switch
7669 inhibitors during their runtime. A simple way to achieve
7670 that is to invoke the DE wrapped in an invocation of:
7672 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
7674 * Access to unit operations is now checked via SELinux taking
7675 the unit file label and client process label into account.
7677 * systemd will now notify the administrator in the journal
7678 when he over-mounts a non-empty directory.
7680 * There are new specifiers that are resolved in unit files,
7681 for the host name (%H), the machine ID (%m) and the boot ID
7684 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
7685 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
7686 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
7687 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
7688 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
7689 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
7690 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
7694 * Support for reading structured kernel messages from
7695 /dev/kmsg has now been added and is enabled by default.
7697 * Support for reading kernel messages from /proc/kmsg has now
7698 been removed. If you want kernel messages in the journal
7699 make sure to run a recent kernel (>= 3.5) that supports
7700 reading structured messages from /dev/kmsg (see
7701 above). /proc/kmsg is now exclusive property of classic
7702 syslog daemons again.
7704 * The libudev API gained the new
7705 udev_device_new_from_device_id() call.
7707 * The logic for file system namespace (ReadOnlyDirectory=,
7708 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
7709 require pivot_root() anymore. This means fewer temporary
7710 directories are created below /tmp for this feature.
7712 * nspawn containers will now see and receive all submounts
7713 made on the host OS below the root file system of the
7716 * Forward Secure Sealing is now supported for Journal files,
7717 which provide cryptographical sealing of journal files so
7718 that attackers cannot alter log history anymore without this
7719 being detectable. Lennart will soon post a blog story about
7720 this explaining it in more detail.
7722 * There are two new service settings RestartPreventExitStatus=
7723 and SuccessExitStatus= which allow configuration of exit
7724 status (exit code or signal) which will be excepted from the
7725 restart logic, resp. consider successful.
7727 * journalctl gained the new --verify switch that can be used
7728 to check the integrity of the structure of journal files and
7729 (if Forward Secure Sealing is enabled) the contents of
7732 * nspawn containers will now be run with /dev/stdin, /dev/fd/
7733 and similar symlinks pre-created. This makes running shells
7734 as container init process a lot more fun.
7736 * The fstab support can now handle PARTUUID= and PARTLABEL=
7739 * A new ConditionHost= condition has been added to match
7740 against the hostname (with globs) and machine ID. This is
7741 useful for clusters where a single OS image is used to
7742 provision a large number of hosts which shall run slightly
7743 different sets of services.
7745 * Services which hit the restart limit will now be placed in a
7748 Contributions from: Bertram Poettering, Dave Reisner, Huang
7749 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
7750 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
7754 * When running in --user mode systemd will now become a
7755 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
7756 tree a lot more organized.
7758 * A new PartOf= unit dependency type has been introduced that
7759 may be used to group services in a natural way.
7761 * "systemctl enable" may now be used to enable instances of
7764 * journalctl now prints error log levels in red, and
7765 warning/notice log levels in bright white. It also supports
7766 filtering by log level now.
7768 * cgtop gained a new -n switch (similar to top), to configure
7769 the maximum number of iterations to run for. It also gained
7770 -b, to run in batch mode (accepting no input).
7772 * The suffix ".service" may now be omitted on most systemctl
7773 command lines involving service unit names.
7775 * There's a new bus call in logind to lock all sessions, as
7776 well as a loginctl verb for it "lock-sessions".
7778 * libsystemd-logind.so gained a new call sd_journal_perror()
7779 that works similar to libc perror() but logs to the journal
7780 and encodes structured information about the error number.
7782 * /etc/crypttab entries now understand the new keyfile-size=
7785 * shutdown(8) now can send a (configurable) wall message when
7786 a shutdown is cancelled.
7788 * The mount propagation mode for the root file system will now
7789 default to "shared", which is useful to make containers work
7790 nicely out-of-the-box so that they receive new mounts from
7791 the host. This can be undone locally by running "mount
7792 --make-rprivate /" if needed.
7794 * The prefdm.service file has been removed. Distributions
7795 should maintain this unit downstream if they intend to keep
7796 it around. However, we recommend writing normal unit files
7797 for display managers instead.
7799 * Since systemd is a crucial part of the OS we will now
7800 default to a number of compiler switches that improve
7801 security (hardening) such as read-only relocations, stack
7802 protection, and suchlike.
7804 * The TimeoutSec= setting for services is now split into
7805 TimeoutStartSec= and TimeoutStopSec= to allow configuration
7806 of individual time outs for the start and the stop phase of
7809 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
7810 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
7811 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
7812 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
7813 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
7814 Gundersen, Zbigniew Jędrzejewski-Szmek
7818 * The journal and id128 C APIs are now fully documented as man
7821 * Extra safety checks have been added when transitioning from
7822 the initial RAM disk to the main system to avoid accidental
7825 * /etc/crypttab entries now understand the new keyfile-offset=
7828 * systemctl -t can now be used to filter by unit load state.
7830 * The journal C API gained the new sd_journal_wait() call to
7831 make writing synchronous journal clients easier.
7833 * journalctl gained the new -D switch to show journals from a
7836 * journalctl now displays a special marker between log
7837 messages of two different boots.
7839 * The journal is now explicitly flushed to /var via a service
7840 systemd-journal-flush.service, rather than implicitly simply
7841 by seeing /var/log/journal to be writable.
7843 * journalctl (and the journal C APIs) can now match for much
7844 more complex expressions, with alternatives and
7847 * When transitioning from the initial RAM disk to the main
7848 system we will now kill all processes in a killing spree to
7849 ensure no processes stay around by accident.
7851 * Three new specifiers may be used in unit files: %u, %h, %s
7852 resolve to the user name, user home directory resp. user
7853 shell. This is useful for running systemd user instances.
7855 * We now automatically rotate journal files if their data
7856 object hash table gets a fill level > 75%. We also size the
7857 hash table based on the configured maximum file size. This
7858 together should lower hash collisions drastically and thus
7859 speed things up a bit.
7861 * journalctl gained the new "--header" switch to introspect
7862 header data of journal files.
7864 * A new setting SystemCallFilters= has been added to services
7865 which may be used to apply blacklists or whitelists to
7866 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
7868 * nspawn gained a new --link-journal= switch (and quicker: -j)
7869 to link the container journal with the host. This makes it
7870 very easy to centralize log viewing on the host for all
7871 guests while still keeping the journal files separated.
7873 * Many bugfixes and optimizations
7875 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
7876 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
7877 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
7882 * Several tools now understand kernel command line arguments,
7883 which are only read when run in an initial RAM disk. They
7884 usually follow closely their normal counterparts, but are
7887 * There's a new tool to analyze the readahead files that are
7888 automatically generated at boot. Use:
7890 /usr/lib/systemd/systemd-readahead analyze /.readahead
7892 * We now provide an early debug shell on tty9 if this enabled. Use:
7894 systemctl enable debug-shell.service
7896 * All plymouth related units have been moved into the Plymouth
7897 package. Please make sure to upgrade your Plymouth version
7900 * systemd-tmpfiles now supports getting passed the basename of
7901 a configuration file only, in which case it will look for it
7902 in all appropriate directories automatically.
7904 * udevadm info now takes a /dev or /sys path as argument, and
7905 does the right thing. Example:
7907 udevadm info /dev/sda
7908 udevadm info /sys/class/block/sda
7910 * systemctl now prints a warning if a unit is stopped but a
7911 unit that might trigger it continues to run. Example: a
7912 service is stopped but the socket that activates it is left
7915 * "systemctl status" will now mention if the log output was
7916 shortened due to rotation since a service has been started.
7918 * The journal API now exposes functions to determine the
7919 "cutoff" times due to rotation.
7921 * journald now understands SIGUSR1 and SIGUSR2 for triggering
7922 immediately flushing of runtime logs to /var if possible,
7923 resp. for triggering immediate rotation of the journal
7926 * It is now considered an error if a service is attempted to
7927 be stopped that is not loaded.
7929 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
7931 * systemd-analyze now supports Python 3
7933 * tmpfiles now supports cleaning up directories via aging
7934 where the first level dirs are always kept around but
7935 directories beneath it automatically aged. This is enabled
7936 by prefixing the age field with '~'.
7938 * Seat objects now expose CanGraphical, CanTTY properties
7939 which is required to deal with very fast bootups where the
7940 display manager might be running before the graphics drivers
7941 completed initialization.
7943 * Seat objects now expose a State property.
7945 * We now include RPM macros for service enabling/disabling
7946 based on the preset logic. We recommend RPM based
7947 distributions to make use of these macros if possible. This
7948 makes it simpler to reuse RPM spec files across
7951 * We now make sure that the collected systemd unit name is
7952 always valid when services log to the journal via
7955 * There's a new man page kernel-command-line(7) detailing all
7956 command line options we understand.
7958 * The fstab generator may now be disabled at boot by passing
7959 fstab=0 on the kernel command line.
7961 * A new kernel command line option modules-load= is now understood
7962 to load a specific kernel module statically, early at boot.
7964 * Unit names specified on the systemctl command line are now
7965 automatically escaped as needed. Also, if file system or
7966 device paths are specified they are automatically turned
7967 into the appropriate mount or device unit names. Example:
7969 systemctl status /home
7970 systemctl status /dev/sda
7972 * The SysVConsole= configuration option has been removed from
7973 system.conf parsing.
7975 * The SysV search path is no longer exported on the D-Bus
7978 * The Names= option has been removed from unit file parsing.
7980 * There's a new man page bootup(7) detailing the boot process.
7982 * Every unit and every generator we ship with systemd now
7983 comes with full documentation. The self-explanatory boot is
7986 * A couple of services gained "systemd-" prefixes in their
7987 name if they wrap systemd code, rather than only external
7988 code. Among them fsck@.service which is now
7989 systemd-fsck@.service.
7991 * The HaveWatchdog property has been removed from the D-Bus
7994 * systemd.confirm_spawn= on the kernel command line should now
7997 * There's a new man page crypttab(5) which details all options
7998 we actually understand.
8000 * systemd-nspawn gained a new --capability= switch to pass
8001 additional capabilities to the container.
8003 * timedated will now read known NTP implementation unit names
8004 from /usr/lib/systemd/ntp-units.d/*.list,
8005 systemd-timedated-ntp.target has been removed.
8007 * journalctl gained a new switch "-b" that lists log data of
8008 the current boot only.
8010 * The notify socket is in the abstract namespace again, in
8011 order to support daemons which chroot() at start-up.
8013 * There is a new Storage= configuration option for journald
8014 which allows configuration of where log data should go. This
8015 also provides a way to disable journal logging entirely, so
8016 that data collected is only forwarded to the console, the
8017 kernel log buffer or another syslog implementation.
8019 * Many bugfixes and optimizations
8021 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
8022 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
8023 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
8024 Shawn Landden, Tom Gundersen
8028 * "systemctl help <unit>" now shows the man page if one is
8031 * Several new man pages have been added.
8033 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
8034 MaxLevelConsole= can now be specified in
8035 journald.conf. These options allow reducing the amount of
8036 data stored on disk or forwarded by the log level.
8038 * TimerSlackNSec= can now be specified in system.conf for
8039 PID1. This allows system-wide power savings.
8041 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
8042 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
8047 * logind is now capable of (optionally) handling power and
8048 sleep keys as well as the lid switch.
8050 * journalctl now understands the syntax "journalctl
8051 /usr/bin/avahi-daemon" to get all log output of a specific
8054 * CapabilityBoundingSet= in system.conf now also influences
8055 the capability bound set of usermode helpers of the kernel.
8057 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
8058 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
8059 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
8060 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
8064 * Note that we skipped 139 releases here in order to set the
8065 new version to something that is greater than both udev's
8066 and systemd's most recent version number.
8068 * udev: all udev sources are merged into the systemd source tree now.
8069 All future udev development will happen in the systemd tree. It
8070 is still fully supported to use the udev daemon and tools without
8071 systemd running, like in initramfs or other init systems. Building
8072 udev though, will require the *build* of the systemd tree, but
8073 udev can be properly *run* without systemd.
8075 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
8076 should be used to create dead device nodes as workarounds for broken
8079 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
8080 no longer supported. udev_monitor_new_from_netlink() needs to be
8081 used to subscribe to events.
8083 * udev: when udevd is started by systemd, processes which are left
8084 behind by forking them off of udev rules, are unconditionally cleaned
8085 up and killed now after the event handling has finished. Services or
8086 daemons must be started as systemd services. Services can be
8087 pulled-in by udev to get started, but they can no longer be directly
8088 forked by udev rules.
8090 * udev: the daemon binary is called systemd-udevd now and installed
8091 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
8092 to adapt to that, create symlink, or rename the binary after building
8095 * libudev no longer provides these symbols:
8096 udev_monitor_from_socket()
8097 udev_queue_get_failed_list_entry()
8098 udev_get_{dev,sys,run}_path()
8099 The versions number was bumped and symbol versioning introduced.
8101 * systemd-loginctl and systemd-journalctl have been renamed
8102 to loginctl and journalctl to match systemctl.
8104 * The config files: /etc/systemd/systemd-logind.conf and
8105 /etc/systemd/systemd-journald.conf have been renamed to
8106 logind.conf and journald.conf. Package updates should rename
8107 the files to the new names on upgrade.
8109 * For almost all files the license is now LGPL2.1+, changed
8110 from the previous GPL2.0+. Exceptions are some minor stuff
8111 of udev (which will be changed to LGPL2.1 eventually, too),
8112 and the MIT licensed sd-daemon.[ch] library that is suitable
8113 to be used as drop-in files.
8115 * systemd and logind now handle system sleep states, in
8116 particular suspending and hibernating.
8118 * logind now implements a sleep/shutdown/idle inhibiting logic
8119 suitable for a variety of uses. Soonishly Lennart will blog
8120 about this in more detail.
8122 * var-run.mount and var-lock.mount are no longer provided
8123 (which previously bind mounted these directories to their new
8124 places). Distributions which have not converted these
8125 directories to symlinks should consider stealing these files
8126 from git history and add them downstream.
8128 * We introduced the Documentation= field for units and added
8129 this to all our shipped units. This is useful to make it
8130 easier to explore the boot and the purpose of the various
8133 * All smaller setup units (such as
8134 systemd-vconsole-setup.service) now detect properly if they
8135 are run in a container and are skipped when
8136 appropriate. This guarantees an entirely noise-free boot in
8137 Linux container environments such as systemd-nspawn.
8139 * A framework for implementing offline system updates is now
8140 integrated, for details see:
8141 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
8143 * A new service type Type=idle is available now which helps us
8144 avoiding ugly interleaving of getty output and boot status
8147 * There's now a system-wide CapabilityBoundingSet= option to
8148 globally reduce the set of capabilities for the
8149 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
8150 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
8151 even CAP_NET_ADMIN system-wide for secure systems.
8153 * There are now system-wide DefaultLimitXXX= options to
8154 globally change the defaults of the various resource limits
8155 for all units started by PID 1.
8157 * Harald Hoyer's systemd test suite has been integrated into
8158 systemd which allows easy testing of systemd builds in qemu
8159 and nspawn. (This is really awesome! Ask us for details!)
8161 * The fstab parser is now implemented as generator, not inside
8164 * systemctl will now warn you if .mount units generated from
8165 /etc/fstab are out of date due to changes in fstab that
8166 have not been read by systemd yet.
8168 * systemd is now suitable for usage in initrds. Dracut has
8169 already been updated to make use of this. With this in place
8170 initrds get a slight bit faster but primarily are much
8171 easier to introspect and debug since "systemctl status" in
8172 the host system can be used to introspect initrd services,
8173 and the journal from the initrd is kept around too.
8175 * systemd-delta has been added, a tool to explore differences
8176 between user/admin configuration and vendor defaults.
8178 * PrivateTmp= now affects both /tmp and /var/tmp.
8180 * Boot time status messages are now much prettier and feature
8181 proper english language. Booting up systemd has never been
8184 * Read-ahead pack files now include the inode number of all
8185 files to pre-cache. When the inode changes the pre-caching
8186 is not attempted. This should be nicer to deal with updated
8187 packages which might result in changes of read-ahead
8190 * We now temporaritly lower the kernel's read_ahead_kb variable
8191 when collecting read-ahead data to ensure the kernel's
8192 built-in read-ahead does not add noise to our measurements
8193 of necessary blocks to pre-cache.
8195 * There's now RequiresMountsFor= to add automatic dependencies
8196 for all mounts necessary for a specific file system path.
8198 * MountAuto= and SwapAuto= have been removed from
8199 system.conf. Mounting file systems at boot has to take place
8202 * nspawn now learned a new switch --uuid= to set the machine
8203 ID on the command line.
8205 * nspawn now learned the -b switch to automatically search
8208 * vt102 is now the default TERM for serial TTYs, upgraded from
8211 * systemd-logind now works on VT-less systems.
8213 * The build tree has been reorganized. The individual
8214 components now have directories of their own.
8216 * A new condition type ConditionPathIsReadWrite= is now available.
8218 * nspawn learned the new -C switch to create cgroups for the
8219 container in other hierarchies.
8221 * We now have support for hardware watchdogs, configurable in
8224 * The scheduled shutdown logic now has a public API.
8226 * We now mount /tmp as tmpfs by default, but this can be
8227 masked and /etc/fstab can override it.
8229 * Since udisks does not make use of /media anymore we are not
8230 mounting a tmpfs on it anymore.
8232 * journalctl gained a new --local switch to only interleave
8233 locally generated journal files.
8235 * We can now load the IMA policy at boot automatically.
8237 * The GTK tools have been split off into a systemd-ui.
8239 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
8240 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
8241 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
8242 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
8243 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
8244 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
8245 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
8246 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
8247 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
8252 * This is mostly a bugfix release
8254 * Support optional initialization of the machine ID from the
8255 KVM or container configured UUID.
8257 * Support immediate reboots with "systemctl reboot -ff"
8259 * Show /etc/os-release data in systemd-analyze output
8261 * Many bugfixes for the journal, including endianness fixes and
8262 ensuring that disk space enforcement works
8264 * sd-login.h is C++ compatible again
8266 * Extend the /etc/os-release format on request of the Debian
8269 * We now refuse non-UTF8 strings used in various configuration
8270 and unit files. This is done to ensure we do not pass invalid
8271 data over D-Bus or expose it elsewhere.
8273 * Register Mimo USB Screens as suitable for automatic seat
8276 * Read SELinux client context from journal clients in a race
8279 * Reorder configuration file lookup order. /etc now always
8280 overrides /run in order to allow the administrator to always
8281 and unconditionally override vendor-supplied or
8282 automatically generated data.
8284 * The various user visible bits of the journal now have man
8285 pages. We still lack man pages for the journal API calls
8288 * We now ship all man pages in HTML format again in the
8291 Contributions from: Dave Reisner, Dirk Eibach, Frederic
8292 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
8293 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
8298 * This is mostly a bugfix release
8300 * systems lacking /etc/os-release are no longer supported.
8302 * Various functionality updates to libsystemd-login.so
8304 * Track class of PAM logins to distinguish greeters from
8307 Contributions from: Kay Sievers, Lennart Poettering, Michael
8312 * This is an important bugfix release for v41.
8314 * Building man pages is now optional which should be useful
8315 for those building systemd from git but unwilling to install
8318 * Watchdog support for supervising services is now usable. In
8319 a future release support for hardware watchdogs
8320 (i.e. /dev/watchdog) will be added building on this.
8322 * Service start rate limiting is now configurable and can be
8323 turned off per service. When a start rate limit is hit a
8324 reboot can automatically be triggered.
8326 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
8328 Contributions from: Benjamin Franzke, Bill Nottingham,
8329 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
8330 Schmidt, Michał Górny, Piotr Drąg
8334 * The systemd binary is installed /usr/lib/systemd/systemd now;
8335 An existing /sbin/init symlink needs to be adapted with the
8338 * The code that loads kernel modules has been ported to invoke
8339 libkmod directly, instead of modprobe. This means we do not
8340 support systems with module-init-tools anymore.
8342 * Watchdog support is now already useful, but still not
8345 * A new kernel command line option systemd.setenv= is
8346 understood to set system wide environment variables
8347 dynamically at boot.
8349 * We now limit the set of capabilities of systemd-journald.
8351 * We now set SIGPIPE to ignore by default, since it only is
8352 useful in shell pipelines, and has little use in general
8353 code. This can be disabled with IgnoreSIPIPE=no in unit
8356 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
8357 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
8362 * This is mostly a bugfix release
8364 * We now expose the reason why a service failed in the
8365 "Result" D-Bus property.
8367 * Rudimentary service watchdog support (will be completed over
8368 the next few releases.)
8370 * When systemd forks off in order execute some service we will
8371 now immediately changes its argv[0] to reflect which process
8372 it will execute. This is useful to minimize the time window
8373 with a generic argv[0], which makes bootcharts more useful
8375 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
8376 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
8377 Mike Kazantsev, Ray Strode
8381 * This is mostly a test release, but incorporates many
8384 * New systemd-cgtop tool to show control groups by their
8387 * Linking against libacl for ACLs is optional again. If
8388 disabled, support tracking device access for active logins
8389 goes becomes unavailable, and so does access to the user
8390 journals by the respective users.
8392 * If a group "adm" exists, journal files are automatically
8393 owned by them, thus allow members of this group full access
8394 to the system journal as well as all user journals.
8396 * The journal now stores the SELinux context of the logging
8397 client for all entries.
8399 * Add C++ inclusion guards to all public headers
8401 * New output mode "cat" in the journal to print only text
8402 messages, without any meta data like date or time.
8404 * Include tiny X server wrapper as a temporary stop-gap to
8405 teach XOrg udev display enumeration. This is used by display
8406 managers such as gdm, and will go away as soon as XOrg
8407 learned native udev hotplugging for display devices.
8409 * Add new systemd-cat tool for executing arbitrary programs
8410 with STDERR/STDOUT connected to the journal. Can also act as
8411 BSD logger replacement, and does so by default.
8413 * Optionally store all locally generated coredumps in the
8414 journal along with meta data.
8416 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
8417 writing short strings to files (for usage for /sys), and for
8418 creating symlinks, character and block device nodes.
8420 * New unit file option ControlGroupPersistent= to make cgroups
8421 persistent, following the mechanisms outlined in
8422 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
8424 * Support multiple local RTCs in a sane way
8426 * No longer monopolize IO when replaying readahead data on
8427 rotating disks, since we might starve non-file-system IO to
8428 death, since fanotify() will not see accesses done by blkid,
8431 * Do not show kernel threads in systemd-cgls anymore, unless
8432 requested with new -k switch.
8434 Contributions from: Dan Horák, Kay Sievers, Lennart
8435 Poettering, Michal Schmidt
8439 * This is mostly a test release, but incorporates many
8442 * The git repository moved to:
8443 git://anongit.freedesktop.org/systemd/systemd
8444 ssh://git.freedesktop.org/git/systemd/systemd
8446 * First release with the journal
8447 http://0pointer.de/blog/projects/the-journal.html
8449 * The journal replaces both systemd-kmsg-syslogd and
8450 systemd-stdout-bridge.
8452 * New sd_pid_get_unit() API call in libsystemd-logind
8454 * Many systemadm clean-ups
8456 * Introduce remote-fs-pre.target which is ordered before all
8457 remote mounts and may be used to start services before all
8460 * Added Mageia support
8462 * Add bash completion for systemd-loginctl
8464 * Actively monitor PID file creation for daemons which exit in
8465 the parent process before having finished writing the PID
8466 file in the daemon process. Daemons which do this need to be
8467 fixed (i.e. PID file creation must have finished before the
8468 parent exits), but we now react a bit more gracefully to them.
8470 * Add colourful boot output, mimicking the well-known output
8471 of existing distributions.
8473 * New option PassCredentials= for socket units, for
8474 compatibility with a recent kernel ABI breakage.
8476 * /etc/rc.local is now hooked in via a generator binary, and
8477 thus will no longer act as synchronization point during
8480 * systemctl list-unit-files now supports --root=.
8482 * systemd-tmpfiles now understands two new commands: z, Z for
8483 relabelling files according to the SELinux database. This is
8484 useful to apply SELinux labels to specific files in /sys,
8487 * Output of SysV services is now forwarded to both the console
8488 and the journal by default, not only just the console.
8490 * New man pages for all APIs from libsystemd-login.
8492 * The build tree got reorganized and the build system is a
8493 lot more modular allowing embedded setups to specifically
8494 select the components of systemd they are interested in.
8496 * Support for Linux systems lacking the kernel VT subsystem is
8499 * configure's --with-rootdir= got renamed to
8500 --with-rootprefix= to follow the naming used by udev and
8503 * Unless specified otherwise we will now install to /usr instead
8504 of /usr/local by default.
8506 * Processes with '@' in argv[0][0] are now excluded from the
8507 final shut-down killing spree, following the logic explained
8509 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
8511 * All processes remaining in a service cgroup when we enter
8512 the START or START_PRE states are now killed with
8513 SIGKILL. That means it is no longer possible to spawn
8514 background processes from ExecStart= lines (which was never
8515 supported anyway, and bad style).
8517 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
8518 reloading of units together.
8520 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
8521 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
8522 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
8523 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
8524 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
~ianmdlvl / elogind.git / blob