1 systemd System and Service Manager
3 CHANGES WITH 239 in spe:
5 * NETWORK INTERFACE DEVICE NAMING CHANGES: elogind-udevd's "net_id"
6 builtin may name network interfaces differently than in previous
7 versions. SR-IOV virtual functions and NPAR partitions with PCI
8 function numbers of 8 and above will be named more predictably, and
9 udev may generate names based on the PCI slot number in some cases
10 where it previously did not.
12 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
13 elogind.service. Since v235, IPAddressDeny=any has been set to
14 the unit. So, it is expected that the default behavior of
15 elogind is not changed. However, if distribution packagers or
16 administrators disabled or modified IPAddressDeny= setting by a
17 drop-in config file, then it may be necessary to update the file to
18 re-enable AF_INET and AF_INET6 to support network user name services,
21 * When the RestrictNamespaces= unit property is specified multiple
22 times, then the specified types are merged now. Previously, only the
23 last assignment was used. So, if distribution packagers or
24 administrators modified the setting by a drop-in config file, then it
25 may be necessary to update the file.
27 * When OnFailure= is used in combination with Restart= on a service
28 unit, then the specified units will no longer be triggered on
29 failures that result in restarting. Previously, the specified units
30 would be activated each time the unit failed, even when the unit was
31 going to be restarted automatically. This behaviour contradicted the
32 documentation. With this release the code is adjusted to match the
35 * elogind-tmpfiles will now print a notice whenever it encounters
36 tmpfiles.d/ lines referencing the /var/run/ directory. It will
37 recommend reworking them to use the /run/ directory instead (for
38 which /var/run/ is simply a symlinked compatibility alias). This way
39 elogind-tmpfiles can properly detect line conflicts and merge lines
40 referencing the same file by two paths, without having to access
43 * systemctl disable/unmask/preset/preset-all cannot be used with
44 --runtime. Previously this was allowed, but resulted in unintuitive
45 behaviour that wasn't useful. systemctl disable/unmask will now undo
46 both runtime and persistent enablement/masking, i.e. it will remove
47 any relevant symlinks both in /run and /etc.
49 * Note that all long-running system services shipped with elogind will
50 now default to a system call whitelist (rather than a blacklist, as
51 before). In particular, elogind-udevd will now enforce one too. For
52 most cases this should be safe, however downstream distributions
53 which disabled sandboxing of elogind-udevd (specifically the
54 MountFlags= setting), might want to disable this security feature
55 too, as the default whitelisting will prohibit all mount, swap,
56 reboot and clock changing operations from udev rules.
58 * sd-boot acquired new loader configuration settings to optionally turn
59 off Windows and MacOS boot partition discovery as well as
60 reboot-into-firmware menu items. It is also able to pick a better
61 screen resolution for HiDPI systems, and now provides loader
62 configuration settings to change the resolution explicitly.
64 * The elogind-resolve tool has been renamed to resolvectl (it also
65 * elogind-resolved now supports DNS-over-TLS. It's still
66 turned off by default, use DNSOverTLS=opportunistic to turn it on in
67 resolved.conf. We intend to make this the default as soon as couple
68 of additional techniques for optimizing the initial latency caused by
69 establishing a TLS/TCP connection are implemented.
71 * elogind-resolved.service and elogind-networkd.service now set
72 DynamicUser=yes. The users elogind-resolve and elogind-network are
73 not created by elogind-sysusers.
75 remains available under the old name, for compatibility), and its
76 interface is now verb-based, similar in style to the other <xyz>ctl
77 tools, such as systemctl or loginctl.
79 * The resolvectl/elogind-resolve tool also provides 'resolvconf'
80 compatibility. It may be symlinked under the 'resolvconf' name, in
81 which case it will take arguments and input compatible with the
82 Debian and FreeBSD resolvconf tool.
84 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
85 where the system initially suspends, and after a time-out resumes and
88 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
89 set the client will only send a DUID as client identifier.
91 * The nss-elogind glibc NSS module will now enumerate dynamic users and
92 groups in effect. Previously, it could resolve UIDs/GIDs to user
93 names/groups and vice versa, but did not support enumeration.
95 * journald's Compress= configuration setting now optionally accepts a
96 byte threshold value. All journal objects larger than this threshold
97 will be compressed, smaller ones will not. Previously this threshold
98 was not configurable and set to 512.
100 * A new system.conf setting NoNewPrivileges= is now available which may
101 be used to turn off acquisition of new privileges system-wide
102 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
103 for all its children). Note that turning this option on means setuid
104 binaries and file system capabilities lose their special powers.
105 While turning on this option is a big step towards a more secure
106 system, doing so is likely to break numerous pre-existing UNIX tools,
107 in particular su and sudo.
109 * A new service elogind-time-sync-wait.service has been added. If
110 enabled it will delay the time-sync.target unit at boot until time
111 synchronization has been received from the network. This
112 functionality is useful on systems lacking a local RTC or where it is
113 acceptable that the boot process shall be delayed by external network
116 * When hibernating, elogind will now inform the kernel of the image
117 write offset, on kernels new enough to support this. This means swap
118 files should work for hibernation now.
120 * When loading unit files, elogind will now look for drop-in unit files
121 extensions in additional places. Previously, for a unit file name
122 "foo-bar-baz.service" it would look for dropin files in
123 "foo-bar-baz.service.d/*.conf". Now, it will also look in
124 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
125 service name truncated after all inner dashes. This scheme allows
126 writing drop-ins easily that apply to a whole set of unit files at
127 once. It's particularly useful for mount and slice units (as their
128 naming is prefix based), but is also useful for service and other
129 units, for packages that install multiple unit files at once,
130 following a strict naming regime of beginning the unit file name with
131 the package's name. Two new specifiers are now supported in unit
132 files to match this: %j and %J are replaced by the part of the unit
133 name following the last dash.
135 * Unit files and other configuration files that support specifier
136 expansion now understand another two new specifiers: %T and %V will
137 resolve to /tmp and /var/tmp respectively, or whatever temporary
138 directory has been set for the calling user.
140 * The ExecStart= lines of unit files are no longer required to
141 reference absolute paths. If non-absolute paths are specified the
142 specified binary name is searched within the service manager's
143 built-in $PATH, which may be queried with 'elogind-path
144 search-binaries-default'. It's generally recommended to continue to
145 use absolute paths for all binaries specified in unit files.
147 * Units gained a new load state "bad-setting", which is used when a
148 unit file was loaded, but contained fatal errors which prevent it
149 from being started (for example, a service unit has been defined
150 lacking both ExecStart= and ExecStop= lines).
152 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
153 support alternative debuggers, for example lldb. The old name
154 continues to be available however, for compatibility reasons. Use the
155 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
156 to pick an alternative debugger instead of the default gdb.
158 * systemctl and the other tools will now output escape sequences that
159 generate proper clickable hyperlinks in various terminal emulators
160 where useful (for example, in the "systemctl status" output you can
161 now click on the unit file name to quickly open it in the
162 editor/viewer of your choice). Note that not all terminal emulators
163 support this functionality yet, but many do. Unfortunately, the
164 "less" pager doesn't support this yet, hence this functionality is
165 currently automatically turned off when a pager is started (which
166 happens quite often due to auto-paging). We hope to remove this
167 limitation as soon as "less" learns these escape sequences. This new
168 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
169 environment variable. For details on these escape sequences see:
170 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
172 * networkd's .network files now support a new IPv6MTUBytes= option for
173 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
174 option in the [Route] section to configure the MTU to use for
175 specific routes. It also gained support for configuration of the DHCP
176 "UserClass" option through the new UserClass= setting. It gained
177 three new options in the new [CAN] section for configuring CAN
178 networks. The MULTICAST and ALLMULTI interface flags may now be
179 controlled explicitly with the new Multicast= and AllMulticast=
182 * networkd will now automatically make use of the kernel's route
183 expiration feature, if it is available.
185 * udevd's .link files now support setting the number of receive and
186 transmit channels, using the RxChannels=, TxChannels=,
187 OtherChannels=, CombinedChannels= settings.
189 * Support for UDPSegmentationOffload= has been removed, given its
190 limited support in hardware, and waning software support.
192 * networkd's .netdev files now support creating "netdevsim" interfaces.
194 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
195 to query the unit belonging to a specific kernel control group.
197 * elogind-analyze gained a new verb "cat-config", which may be used to
198 dump the contents of any configuration file, with all its matching
199 drop-in files added in, and honouring the usual search and masking
200 logic applied to elogind configuration files. For example use
201 "elogind-analyze cat-config elogind/system.conf" to get the complete
202 system configuration file of elogind how it would be loaded by PID 1
203 itself. Similar to this, various tools such as elogind-tmpfiles or
204 elogind-sysusers, gained a new option "--cat-config", which does the
205 corresponding operation for their own configuration settings. For
206 example, "elogind-tmpfiles --cat-config" will now output the full
207 list of tmpfiles.d/ lines in place.
209 * timedatectl gained two new verbs "timesync-status" (to show the
210 current NTP synchronization state of elogind-timesyncd) and
211 "show-timesync" (to show bus properties of elogind-timesyncd).
213 * elogind-timesyncd gained a bus interface on which it exposes details
216 * elogind-nspawn gained a new --rlimit= switch for setting initial
217 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
218 understood by elogind-timedated. It takes a colon-separated list of
219 unit names of NTP client services. The list is used by
220 "timedatectl set-ntp".
222 resource limits for the container payload. There's a new switch
223 --hostname= to explicitly override the container's hostname. A new
224 --no-new-privileges= switch may be used to control the
225 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
226 --oom-score-adjust= switch controls the OOM scoring adjustment value
227 for the payload. The new --cpu-affinity= switch controls the CPU
228 affinity of the container payload. The new --resolv-conf= switch
229 allows more detailed control of /etc/resolv.conf handling of the
230 container. Similarly, the new --timezone= switch allows more detailed
231 control of /etc/localtime handling of the container.
233 * elogind-detect-virt gained a new --list switch, which will print a
234 list of all currently known VM and container environments.
236 * Support for "Portable Services" has been added, see
237 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
238 experimental, but this is expected to change soon. Reflecting this
239 experimental state, the "portablectl" binary is not installed into
240 /usr/bin yet. The binary has to be called with the full path
241 /usr/lib/elogind/portablectl instead.
243 * journalctl's and systemctl's -o switch now knows a new log output
244 mode "with-unit". The output it generates is very similar to the
245 regular "short" mode, but displays the unit name instead of the
246 syslog tag for each log line. Also, the date is shown with timezone
247 information. This mode is probably more useful than the classic
248 "short" output mode for most purposes, except where pixel-perfect
249 compatibility with classic /var/log/messages formatting is required.
251 * A new --dump-bus-properties switch has been added to the elogind
252 binary, which may be used to dump all supported D-Bus properties.
253 (Options which are still supported, but are deprecated, are *not*
256 * sd-bus gained a set of new calls:
257 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
258 enable/disable the "floating" state of a bus slot object,
259 i.e. whether the slot object pins the bus it is allocated for into
260 memory or if the bus slot object gets disconnected when the bus goes
261 away. sd_bus_open_with_description(),
262 sd_bus_open_user_with_description(),
263 sd_bus_open_system_with_description() may be used to allocate bus
264 objects and set their description string already during allocation.
266 * sd-event gained support for watching inotify events from the event
267 loop, in an efficient way, sharing inotify handles between multiple
268 users. For this a new function sd_event_add_inotify() has been added.
270 * sd-event and sd-bus gained support for calling special user-supplied
271 destructor functions for userdata pointers associated with
272 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
273 functions sd_bus_slot_set_destroy_callback,
274 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
275 sd_bus_track_get_destroy_callback,
276 sd_event_source_set_destroy_callback,
277 sd_event_source_get_destroy_callback have been added.
279 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
281 * PID 1 will now automatically reschedule .timer units whenever the
282 local timezone changes. (They previously got rescheduled
283 automatically when the system clock changed.)
285 * New documentation has been added to document cgroups delegation,
286 portable services and the various code quality tools we have set up:
288 https://github.com/systemd/systemd/blob/master/doc/CGROUP_DELEGATION.md
289 https://github.com/systemd/systemd/blob/master/doc/PORTABLE_SERVICES.md
290 https://github.com/systemd/systemd/blob/master/doc/CODE_QUALITY.md
292 * pam_elogind will now look for PAM userdata fields elogind.memory_max,
293 elogind.tasks_max, elogind.cpu_weight, elogind.io_weight set by
294 earlier PAM modules. The data in these fields is used to initialize
295 the session scope's resource properties. Thus external PAM modules
296 may now configure per-session limits, for example sourced from
297 external user databases.
299 * socket units with Accept=yes will now maintain a "refused" counter in
300 addition to the existing "accepted" counter, counting connections
301 refused due to the enforced limits.
303 * The "elogind-path search-binaries-default" command may now be use to
304 query the default, built-in $PATH PID 1 will pass to the services it
307 * A new unit file setting PrivateMounts= has been added. It's a boolean
308 option. If enabled the unit's processes are invoked in their own file
309 system namespace. Note that this behaviour is also implied if any
310 other file system namespacing options (such as PrivateTmp=,
311 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
312 primarily useful for services that do not use any of the other file
313 system namespacing options. One such service is elogind-udevd.service
314 wher this is now used by default.
316 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
317 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
318 J. Murrell, Bruno Vernay, Chris Lesiak, Christian Brauner, Christian
319 Hesse, Daniel Dao, Daniel Lin, Danylo Korostil, Davide Cavalca, David
320 Tardon, Dimitri John Ledkov, Dmitriy Geels, Douglas Christman, Elia
321 Geretto, emelenas, Evegeny Vereshchagin, Evgeny Vereshchagin, Felipe
322 Sateler, Feng Sun, Filipe Brandenburger, Franck Bui, futpib, Giuseppe
323 Scrivano, Guillem Jover, guixxx, Hans de Goede, Henrique Dante de
324 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
325 James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir, João Paulo
326 Rechi Vita, Joost Heitbrink, Jui-Chi Ricky Liang, Jürg Billeter,
327 Kai-Heng Feng, Karol Augustin, Krzysztof Nowicki, Lauri Tirkkonen,
328 Lennart Poettering, Leonard König, Long Li, Luca Boccassi, Lucas
329 Werkmeister, Marcel Hoppe, Marc Kleine-Budde, Mario Limonciello, Martin
330 Jansa, Martin Wilck, Mathieu Malaterre, Matteo F. Vescovi, Matthew
331 McGinn, Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný,
332 Michal Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan
333 Pässler, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride Legovini,
334 Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot, Peter
335 Hutterer, Peter Jones, Philip Sequeira, Philip Withnall, Piotr Drąg,
336 Radostin Stoyanov, Ricardo Salveti de Araujo, Rosen Penev, Rubén Suárez
337 Alvarez, Ryan Gonzalez, Salvo Tomaselli, Sebastian Reichel, Sergio
338 Lindo Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes,
339 Susant Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias
340 Jungel, Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van
341 Mourik, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
347 * The MemoryAccounting= unit property now defaults to on. After
348 discussions with the upstream control group maintainers we learnt
349 that the negative impact of cgroup memory accounting on current
350 kernels is finally relatively minimal, so that it should be safe to
351 enable this by default without affecting system performance. Besides
352 memory accounting only task accounting is turned on by default, all
353 other forms of resource accounting (CPU, IO, IP) remain off for now,
354 because it's not clear yet that their impact is small enough to move
355 from opt-in to opt-out. We recommend downstreams to leave memory
356 accounting on by default if kernel 4.14 or higher is primarily
357 used. On very resource constrained systems or when support for old
358 kernels is a necessity, -Dmemory-accounting-default=false can be used
359 to revert this change.
361 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
362 %udev_rules_update) and the journal catalog (%journal_catalog_update)
363 from the upgrade scriptlets of individual packages now do nothing.
364 Transfiletriggers have been added which will perform those updates
365 once at the end of the transaction.
367 Similar transfiletriggers have been added to execute any sysctl.d
368 and binfmt.d rules. Thus, it should be unnecessary to provide any
369 scriptlets to execute this configuration from package installation
372 * elogind-sysusers gained a mode where the configuration to execute is
373 specified on the command line, but this configuration is not executed
374 directly, but instead it is merged with the configuration on disk,
375 and the result is executed. This is useful for package installation
376 scripts which want to create the user before installing any files on
377 disk (in case some of those files are owned by that user), while
378 still allowing local admin overrides.
380 This functionality is exposed to rpm scriptlets through a new
381 %sysusers_create_package macro. Old %sysusers_create and
382 %sysusers_create_inline macros are deprecated.
384 A transfiletrigger for sysusers.d configuration is now installed,
385 which means that it should be unnecessary to call elogind-sysusers from
386 package installation scripts, unless the package installs any files
387 owned by those newly-created users, in which case
388 %sysusers_create_package should be used.
390 * Analogous change has been done for elogind-tmpfiles: it gained a mode
391 where the command-line configuration is merged with the configuration
392 on disk. This is exposed as the new %tmpfiles_create_package macro,
393 and %tmpfiles_create is deprecated. A transfiletrigger is installed
394 for tmpfiles.d, hence it should be unnecessary to call elogind-tmpfiles
395 from package installation scripts.
397 * sysusers.d configuration for a user may now also specify the group
398 number, in addition to the user number ("u username 123:456"), or
399 without the user number ("u username -:456").
401 * Configution items for elogind-sysusers can now be specified as
402 positional arguments when the new --inline switch is used.
404 * The login shell of users created through sysusers.d may now be
405 specified (previously, it was always /bin/sh for root and
406 /sbin/nologin for other users).
408 * elogind-analyze gained a new --global switch to look at global user
409 configuration. It also gained a unit-paths verb to list the unit load
410 paths that are compiled into elogind (which can be used with
411 --elogind, --user, or --global).
413 * udevadm trigger gained a new --settle/-w option to wait for any
414 triggered events to finish (but just those, and not any other events
415 which are triggered meanwhile).
417 * The action that elogind takes when the lid is closed and the
418 machine is connected to external power can now be configured using
419 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
420 was determined by HandleLidSwitch=, and, for backwards compatibility,
421 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
423 * journalctl will periodically call sd_journal_process() to make it
424 resilient against inotify queue overruns when journal files are
425 rotated very quickly.
427 * Two new functions in libelogind — sd_bus_get_n_queued_read and
428 sd_bus_get_n_queued_write — may be used to check the number of
429 pending bus messages.
431 * elogind gained a new
432 org.freedesktop.elogind1.Manager.AttachProcessesToUnit dbus call
433 which can be used to migrate foreign processes to scope and service
434 units. The primary user for this new API is elogind itself: the
435 elogind --user instance uses this call of the elogind --system
436 instance to migrate processes if it itself gets the request to
437 migrate processes and the kernel refuses this due to access
438 restrictions. Thanks to this "elogind-run --scope --user …" works
439 again in pure cgroups v2 environments when invoked from the user
442 * A new TemporaryFileSystem= setting can be used to mask out part of
443 the real file system tree with tmpfs mounts. This may be combined
444 with BindPaths= and BindReadOnlyPaths= to hide files or directories
445 not relevant to the unit, while still allowing some paths lower in
446 the tree to be accessed.
448 ProtectHome=tmpfs may now be used to hide user home and runtime
449 directories from units, in a way that is mostly equivalent to
450 "TemporaryFileSystem=/home /run/user /root".
452 * Non-service units are now started with KeyringMode=shared by default.
453 This means that mount and swapon and other mount tools have access
454 to keys in the main keyring.
456 * /sys/fs/bpf is now mounted automatically.
458 * QNX virtualization is now detected by elogind-detect-virt and may
459 be used in ConditionVirtualization=.
461 * IPAccounting= may now be enabled also for slice units.
463 * A new -Dsplit-bin= build configuration switch may be used to specify
464 whether bin and sbin directories are merged, or if they should be
465 included separately in $PATH and various listings of executable
466 directories. The build configuration scripts will try to autodetect
467 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
468 system, but distributions are encouraged to configure this
471 * A new -Dok-color= build configuration switch may be used to change
472 the colour of "OK" status messages.
474 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
475 PrivateNetwork=yes was buggy in previous versions of elogind. This
476 means that after the upgrade and daemon-reexec, any such units must
479 * INCOMPATIBILITY: as announced in the NEWS for 237, elogind-tmpfiles
480 will not exclude read-only files owned by root from cleanup.
482 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
483 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
484 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
485 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
486 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
487 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
488 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
489 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
490 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
491 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
492 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
493 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
494 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
495 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
496 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
497 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
503 * Some keyboards come with a zoom see-saw or rocker which until now got
504 mapped to the Linux "zoomin/out" keys in hwdb. However, these
505 keycodes are not recognized by any major desktop. They now produce
506 Up/Down key events so that they can be used for scrolling.
508 * INCOMPATIBILITY: elogind-tmpfiles' "f" lines changed behaviour
509 slightly: previously, if an argument was specified for lines of this
510 type (i.e. the right-most column was set) this string was appended to
511 existing files each time elogind-tmpfiles was run. This behaviour was
512 different from what the documentation said, and not particularly
513 useful, as repeated elogind-tmpfiles invocations would not be
514 idempotent and grow such files without bounds. With this release
515 behaviour has been altered slightly, to match what the documentation
516 says: lines of this type only have an effect if the indicated files
517 don't exist yet, and only then the argument string is written to the
520 * FUTURE INCOMPATIBILITY: In elogind v238 we intend to slightly change
521 elogind-tmpfiles behaviour: previously, read-only files owned by root
522 were always excluded from the file "aging" algorithm (i.e. the
523 automatic clean-up of directories like /tmp based on
524 atime/mtime/ctime). We intend to drop this restriction, and age files
525 by default even when owned by root and read-only. This behaviour was
526 inherited from older tools, but there have been requests to remove
527 it, and it's not obvious why this restriction was made in the first
528 place. Please speak up now, if you are aware of software that reqires
529 this behaviour, otherwise we'll remove the restriction in v238.
531 * A new environment variable $SYSTEMD_OFFLINE is now understood by
532 systemctl. It takes a boolean argument. If on, systemctl assumes it
533 operates on an "offline" OS tree, and will not attempt to talk to the
534 service manager. Previously, this mode was implicitly enabled if a
535 chroot() environment was detected, and this new environment variable
536 now provides explicit control.
538 * .path and .socket units may now be created transiently, too.
539 Previously only service, mount, automount and timer units were
540 supported as transient units. The elogind-run tool has been updated
541 to expose this new functionality, you may hence use it now to bind
542 arbitrary commands to path or socket activation on-the-fly from the
543 command line. Moreover, almost all properties are now exposed for the
544 unit types that already supported transient operation.
546 * The elogind-mount command gained support for a new --owner= parameter
547 which takes a user name, which is then resolved and included in uid=
548 and gid= mount options string of the file system to mount.
550 * A new unit condition ConditionControlGroupController= has been added
551 that checks whether a specific cgroup controller is available.
553 * Unit files, udev's .link files, and elogind-networkd's .netdev and
554 .network files all gained support for a new condition
555 ConditionKernelVersion= for checking against specific kernel
558 * In elogind-networkd, the [IPVLAN] section in .netdev files gained
559 support for configuring device flags in the Flags= setting. In the
560 same files, the [Tunnel] section gained support for configuring
561 AllowLocalRemote=. The [Route] section in .network files gained
562 support for configuring InitialCongestionWindow=,
563 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
564 understands RapidCommit=.
566 * elogind-networkd's DHCPv6 support gained support for Prefix
569 * sd-bus gained support for a new "watch-bind" feature. When this
570 feature is enabled, an sd_bus connection may be set up to connect to
571 an AF_UNIX socket in the file system as soon as it is created. This
572 functionality is useful for writing early-boot services that
573 automatically connect to the system bus as soon as it is started,
574 without ugly time-based polling. elogind-networkd and
575 elogind-resolved have been updated to make use of this
576 functionality. busctl exposes this functionality in a new
577 --watch-bind= command line switch.
579 * sd-bus will now optionally synthesize a local "Connected" signal as
580 soon as a D-Bus connection is set up fully. This message mirrors the
581 already existing "Disconnected" signal which is synthesized when the
582 connection is terminated. This signal is generally useful but
583 particularly handy in combination with the "watch-bind" feature
584 described above. Synthesizing of this message has to be requested
585 explicitly through the new API call sd_bus_set_connected_signal(). In
586 addition a new call sd_bus_is_ready() has been added that checks
587 whether a connection is fully set up (i.e. between the "Connected" and
588 "Disconnected" signals).
590 * sd-bus gained two new calls sd_bus_request_name_async() and
591 sd_bus_release_name_async() for asynchronously registering bus
592 names. Similar, there is now sd_bus_add_match_async() for installing
593 a signal match asynchronously. All of elogind's own services have
594 been updated to make use of these calls. Doing these operations
595 asynchronously has two benefits: it reduces the risk of deadlocks in
596 case of cyclic dependencies between bus services, and it speeds up
597 service initialization since synchronization points for bus
598 round-trips are removed.
600 * sd-bus gained two new calls sd_bus_match_signal() and
601 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
602 and sd_bus_add_match_async() but instead of taking a D-Bus match
603 string take match fields as normal function parameters.
605 * sd-bus gained two new calls sd_bus_set_sender() and
606 sd_bus_message_set_sender() for setting the sender name of outgoing
607 messages (either for all outgoing messages or for just one specific
608 one). These calls are only useful in direct connections as on
609 brokered connections the broker fills in the sender anyway,
610 overwriting whatever the client filled in.
612 * sd-event gained a new pseudo-handle that may be specified on all API
613 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
614 used this refers to the default event loop object of the calling
615 thread. Note however that this does not implicitly allocate one —
616 which has to be done prior by using sd_event_default(). Similarly
617 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
618 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
619 to the default bus of the specified type of the calling thread. Here
620 too this does not implicitly allocate bus connection objects, this
621 has to be done prior with sd_bus_default() and friends.
623 * sd-event gained a new call pair
624 sd_event_source_{get|set}_io_fd_own(). This may be used to request
625 automatic closure of the file descriptor an IO event source watches
626 when the event source is destroyed.
628 * elogind-networkd gained support for natively configuring WireGuard
631 * In previous versions elogind synthesized user records both for the
632 "nobody" (UID 65534) and "root" (UID 0) users in nss-elogind and
633 internally. In order to simplify distribution-wide renames of the
634 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
635 new transitional flag file has been added: if
636 /etc/elogind/dont-synthesize-nobody exists synthesizing of the 65534
637 user and group record within the elogind codebase is disabled.
639 * elogind-notify gained a new --uid= option for selecting the source
640 user/UID to use for notification messages sent to the service
643 * journalctl gained a new --grep= option to list only entries in which
644 the message matches a certain pattern. By default matching is case
645 insensitive if the pattern is lowercase, and case sensitive
646 otherwise. Option --case-sensitive=yes|no can be used to override
647 this an specify case sensitivity or case insensitivity.
649 * There's now a "elogind-analyze service-watchdogs" command for printing
650 the current state of the service runtime watchdog, and optionally
651 enabling or disabling the per-service watchdogs system-wide if given a
652 boolean argument (i.e. the concept you configure in WatchdogSec=), for
653 debugging purposes. There's also a kernel command line option
654 elogind.service_watchdogs= for controlling the same.
656 * Two new "log-level" and "log-target" options for elogind-analyze were
657 added that merge the now deprecated get-log-level, set-log-level and
658 get-log-target, set-log-target pairs. The deprecated options are still
659 understood for backwards compatibility. The two new options print the
660 current value when no arguments are given, and set them when a
661 level/target is given as an argument.
663 * sysusers.d's "u" lines now optionally accept both a UID and a GID
664 specification, separated by a ":" character, in order to create users
665 where UID and GID do not match.
667 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
668 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
669 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
670 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
671 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
672 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
673 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
674 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
675 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
676 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
677 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
678 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
679 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
680 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
681 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
682 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
683 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
684 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
685 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
686 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
693 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
694 in v235 has been extended to also set the dummy.ko module option
695 numdummies=0, preventing the kernel from automatically creating
696 dummy0. All dummy interfaces must now be explicitly created.
698 * Unknown '%' specifiers in configuration files are now rejected. This
699 applies to units and tmpfiles.d configuration. Any percent characters
700 that are followed by a letter or digit that are not supposed to be
701 interpreted as the beginning of a specifier should be escaped by
702 doubling ("%%"). (So "size=5%" is still accepted, as well as
703 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
704 valid specifiers today.)
706 * systemd-resolved now maintains a new dynamic
707 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
708 recommended to make /etc/resolv.conf a symlink to it. This file
709 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
710 includes dynamically acquired search domains, achieving more correct
711 DNS resolution by software that bypasses local DNS APIs such as NSS.
713 * The "uaccess" udev tag has been dropped from /dev/kvm and
714 /dev/dri/renderD*. These devices now have the 0666 permissions by
715 default (but this may be changed at build-time). /dev/dri/renderD*
716 will now be owned by the "render" group along with /dev/kfd.
718 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
719 systemd-journal-gatewayd.service and
720 systemd-journal-upload.service. This means "nss-systemd" must be
721 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
722 services are resolved properly.
724 * In /etc/fstab two new mount options are now understood:
725 x-systemd.makefs and x-systemd.growfs. The former has the effect that
726 the configured file system is formatted before it is mounted, the
727 latter that the file system is resized to the full block device size
728 after it is mounted (i.e. if the file system is smaller than the
729 partition it resides on, it's grown). This is similar to the fsck
730 logic in /etc/fstab, and pulls in systemd-makefs@.service and
731 systemd-growfs@.service as necessary, similar to
732 systemd-fsck@.service. Resizing is currently only supported on ext4
735 * In systemd-networkd, the IPv6 RA logic now optionally may announce
736 DNS server and domain information.
738 * Support for the LUKS2 on-disk format for encrypted partitions has
739 been added. This requires libcryptsetup2 during compilation and
742 * The systemd --user instance will now signal "readiness" when its
743 basic.target unit has been reached, instead of when the run queue ran
744 empty for the first time.
746 * Tmpfiles.d with user configuration are now also supported.
747 systemd-tmpfiles gained a new --user switch, and snippets placed in
748 ~/.config/user-tmpfiles.d/ and corresponding directories will be
749 executed by systemd-tmpfiles --user running in the new
750 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
751 running in the user session.
753 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
754 %S resolves to the top-level state directory (/var/lib for the system
755 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
756 top-level cache directory (/var/cache for the system instance,
757 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
758 logs directory (/var/log for the system instance,
759 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
760 existing %t specifier, that resolves to the top-level runtime
761 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
764 * journalctl learnt a new parameter --output-fields= for limiting the
765 set of journal fields to output in verbose and JSON output modes.
767 * systemd-timesyncd's configuration file gained a new option
768 RootDistanceMaxSec= for setting the maximum root distance of servers
769 it'll use, as well as the new options PollIntervalMinSec= and
770 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
772 * bootctl gained a new command "list" for listing all available boot
773 menu items on systems that follow the boot loader specification.
775 * systemctl gained a new --dry-run switch that shows what would be done
776 instead of doing it, and is currently supported by the shutdown and
779 * ConditionSecurity= can now detect the TOMOYO security module.
781 * Unit file [Install] sections are now also respected in unit drop-in
782 files. This is intended to be used by drop-ins under /usr/lib/.
784 * systemd-firstboot may now also set the initial keyboard mapping.
786 * Udev "changed" events for devices which are exposed as systemd
787 .device units are now propagated to units specified in
788 ReloadPropagatedFrom= as reload requests.
790 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
791 unit template name (i.e. a name in the form of 'foobar@.service',
792 without the instance component between the '@' and - the '.'), then
793 the escaped sysfs path of the device is automatically used as the
796 * SystemCallFilter= in unit files has been extended so that an "errno"
797 can be specified individually for each system call. Example:
798 SystemCallFilter=~uname:EILSEQ.
800 * The cgroup delegation logic has been substantially updated. Delegate=
801 now optionally takes a list of controllers (instead of a boolean, as
802 before), which lists the controllers to delegate at least.
804 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
806 * A new LogLevelMax= setting configures the maximum log level any
807 process of the service may log at (i.e. anything with a lesser
808 priority than what is specified is automatically dropped). A new
809 LogExtraFields= setting allows configuration of additional journal
810 fields to attach to all log records generated by any of the unit's
813 * New StandardInputData= and StandardInputText= settings along with the
814 new option StandardInput=data may be used to configure textual or
815 binary data that shall be passed to the executed service process via
816 standard input, encoded in-line in the unit file.
818 * StandardInput=, StandardOutput= and StandardError= may now be used to
819 connect stdin/stdout/stderr of executed processes directly with a
820 file or AF_UNIX socket in the file system, using the new "file:" option.
822 * A new unit file option CollectMode= has been added, that allows
823 tweaking the garbage collection logic for units. It may be used to
824 tell systemd to garbage collect units that have failed automatically
825 (normally it only GCs units that exited successfully). systemd-run
826 and systemd-mount expose this new functionality with a new -G option.
828 * "machinectl bind" may now be used to bind mount non-directories
829 (i.e. regularfiles, devices, fifos, sockets).
831 * systemd-analyze gained a new verb "calendar" for validating and
832 testing calendar time specifications to use for OnCalendar= in timer
833 units. Besides validating the expression it will calculate the next
834 time the specified expression would elapse.
836 * In addition to the pre-existing FailureAction= unit file setting
837 there's now SuccessAction=, for configuring a shutdown action to
838 execute when a unit completes successfully. This is useful in
839 particular inside containers that shall terminate after some workload
840 has been completed. Also, both options are now supported for all unit
841 types, not just services.
843 * networkds's IP rule support gained two new options
844 IncomingInterface= and OutgoingInterface= for configuring the incoming
845 and outgoing interfaces of configured rules. systemd-networkd also
846 gained support for "vxcan" network devices.
848 * networkd gained a new setting RequiredForOnline=, taking a
849 boolean. If set, systemd-wait-online will take it into consideration
850 when determining that the system is up, otherwise it will ignore the
851 interface for this purpose.
853 * The sd_notify() protocol gained support for a new operation: with
854 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
855 store again, ahead of POLLHUP or POLLERR when they are removed
858 * A new document doc/UIDS-GIDS.md has been added to the source tree,
859 that documents the UID/GID range and assignment assumptions and
860 requirements of systemd.
862 * The watchdog device PID 1 will ping may now be configured through the
863 WatchdogDevice= configuration file setting, or by setting the
864 systemd.watchdog_service= kernel commandline option.
866 * systemd-resolved's gained support for registering DNS-SD services on
867 the local network using MulticastDNS. Services may either be
868 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
869 the same dir below /run, /usr/lib), or through its D-Bus API.
871 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
872 extend the effective start, runtime, and stop time. The service must
873 continue to send EXTEND_TIMEOUT_USEC within the period specified to
874 prevent the service manager from making the service as timedout.
876 * elogind-resolved's DNSSEC support gained support for RFC 8080
877 (Ed25519 keys and signatures).
879 * The elogind-resolve command line tool gained a new set of options
880 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
881 --set-nta= and --revert to configure per-interface DNS configuration
882 dynamically during runtime. It's useful for pushing DNS information
883 into elogind-resolved from DNS hook scripts that various interface
884 managing software supports (such as pppd).
886 * elogind-nspawn gained a new --network-namespace-path= command line
887 option, which may be used to make a container join an existing
888 network namespace, by specifying a path to a "netns" file.
890 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
891 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
892 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
893 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
894 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
895 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
896 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
897 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
898 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
899 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
900 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
901 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
902 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
903 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
904 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
905 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
906 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
907 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
908 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
909 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
910 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
911 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
912 Jędrzejewski-Szmek, Zeal Jagannatha
918 * INCOMPATIBILITY: systemd-logind.service and other long-running
919 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
920 communication with the outside. This generally improves security of
921 the system, and is in almost all cases a safe and good choice, as
922 these services do not and should not provide any network-facing
923 functionality. However, systemd-logind uses the glibc NSS API to
924 query the user database. This creates problems on systems where NSS
925 is set up to directly consult network services for user database
926 lookups. In particular, this creates incompatibilities with the
927 "nss-nis" module, which attempts to directly contact the NIS/YP
928 network servers it is configured for, and will now consistently
929 fail. In such cases, it is possible to turn off IP sandboxing for
930 systemd-logind.service (set IPAddressDeny= in its [Service] section
931 to the empty string, via a .d/ unit file drop-in). Downstream
932 distributions might want to update their nss-nis packaging to include
933 such a drop-in snippet, accordingly, to hide this incompatibility
934 from the user. Another option is to make use of glibc's nscd service
935 to proxy such network requests through a privilege-separated, minimal
936 local caching daemon, or to switch to more modern technologies such
937 sssd, whose NSS hook-ups generally do not involve direct network
938 access. In general, we think it's definitely time to question the
939 implementation choices of nss-nis, i.e. whether it's a good idea
940 today to embed a network-facing loadable module into all local
941 processes that need to query the user database, including the most
942 trivial and benign ones, such as "ls". For more details about
943 IPAddressDeny= see below.
945 * A new modprobe.d drop-in is now shipped by default that sets the
946 bonding module option max_bonds=0. This overrides the kernel default,
947 to avoid conflicts and ambiguity as to whether or not bond0 should be
948 managed by systemd-networkd or not. This resolves multiple issues
949 with bond0 properties not being applied, when bond0 is configured
950 with systemd-networkd. Distributors may choose to not package this,
951 however in that case users will be prevented from correctly managing
952 bond0 interface using systemd-networkd.
954 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
955 which print the logging level and target of the system manager. They
956 complement the existing "set-log-level" and "set-log-target" verbs
957 used to change those values.
959 * journald.conf gained a new boolean setting ReadKMsg= which defaults
960 to on. If turned off kernel log messages will not be read by
961 systemd-journald or included in the logs. It also gained a new
962 setting LineMax= for configuring the maximum line length in
963 STDOUT/STDERR log streams. The new default for this value is 48K, up
964 from the previous hardcoded 2048.
966 * A new unit setting RuntimeDirectoryPreserve= has been added, which
967 allows more detailed control of what to do with a runtime directory
968 configured with RuntimeDirectory= (i.e. a directory below /run or
969 $XDG_RUNTIME_DIR) after a unit is stopped.
971 * The RuntimeDirectory= setting for units gained support for creating
972 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
973 one top-level directory.
975 * Units gained new options StateDirectory=, CacheDirectory=,
976 LogsDirectory= and ConfigurationDirectory= which are closely related
977 to RuntimeDirectory= but manage per-service directories below
978 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
979 possible to write unit files which when activated automatically gain
980 properly owned service specific directories in these locations, thus
981 making unit files self-contained and increasing compatibility with
982 stateless systems and factory reset where /etc or /var are
983 unpopulated at boot. Matching these new settings there's also
984 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
985 ConfigurationDirectoryMode= for configuring the access mode of these
986 directories. These settings are particularly useful in combination
987 with DynamicUser=yes as they provide secure, properly-owned,
988 writable, and stateful locations for storage, excluded from the
989 sandbox that such services live in otherwise.
991 * Automake support has been removed from this release. systemd is now
994 * systemd-journald will now aggressively cache client metadata during
995 runtime, speeding up log write performance under pressure. This comes
996 at a small price though: as much of the metadata is read
997 asynchronously from /proc/ (and isn't implicitly attached to log
998 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
999 metadata stored alongside a log entry might be slightly
1000 out-of-date. Previously it could only be slightly newer than the log
1001 message. The time window is small however, and given that the kernel
1002 is unlikely to be improved anytime soon in this regard, this appears
1005 * nss-myhostname/systemd-resolved will now by default synthesize an
1006 A/AAAA resource record for the "_gateway" hostname, pointing to the
1007 current default IP gateway. Previously it did that for the "gateway"
1008 name, hampering adoption, as some distributions wanted to leave that
1009 host name open for local use. The old behaviour may still be
1010 requested at build time.
1012 * systemd-networkd's [Address] section in .network files gained a new
1013 Scope= setting for configuring the IP address scope. The [Network]
1014 section gained a new boolean setting ConfigureWithoutCarrier= that
1015 tells systemd-networkd to ignore link sensing when configuring the
1016 device. The [DHCP] section gained a new Anonymize= boolean option for
1017 turning on a number of options suggested in RFC 7844. A new
1018 [RoutingPolicyRule] section has been added for configuring the IP
1019 routing policy. The [Route] section has gained support for a new
1020 Type= setting which permits configuring
1021 blackhole/unreachable/prohibit routes.
1023 * The [VRF] section in .netdev files gained a new Table= setting for
1024 configuring the routing table to use. The [Tunnel] section gained a
1025 new Independent= boolean field for configuring tunnels independent of
1026 an underlying network interface. The [Bridge] section gained a new
1027 GroupForwardMask= option for configuration of propagation of link
1028 local frames between bridge ports.
1030 * The WakeOnLan= setting in .link files gained support for a number of
1031 new modes. A new TCP6SegmentationOffload= setting has been added for
1032 configuring TCP/IPv6 hardware segmentation offload.
1034 * The IPv6 RA sender implementation may now optionally send out RDNSS
1035 and RDNSSL records to supply DNS configuration to peers.
1037 * systemd-nspawn gained support for a new --system-call-filter= command
1038 line option for adding and removing entries in the default system
1039 call filter it applies. Moreover systemd-nspawn has been changed to
1040 implement a system call whitelist instead of a blacklist.
1042 * systemd-run gained support for a new --pipe command line option. If
1043 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
1044 are directly passed on to the activated transient service
1045 executable. This allows invoking arbitrary processes as systemd
1046 services (for example to take benefit of dependency management,
1047 accounting management, resource management or log management that is
1048 done automatically for services) — while still allowing them to be
1049 integrated in a classic UNIX shell pipeline.
1051 * When a service sends RELOAD=1 via sd_notify() and reload propagation
1052 using ReloadPropagationTo= is configured, a reload is now propagated
1053 to configured units. (Previously this was only done on explicitly
1054 requested reloads, using "systemctl reload" or an equivalent
1057 * For each service unit a restart counter is now kept: it is increased
1058 each time the service is restarted due to Restart=, and may be
1059 queried using "systemctl show -p NRestarts …".
1061 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
1062 @signal and @timer have been added, for usage with SystemCallFilter=
1063 in unit files and the new --system-call-filter= command line option
1064 of systemd-nspawn (see above).
1066 * ExecStart= lines in unit files gained two new modifiers: when a
1067 command line is prefixed with "!" the command will be executed as
1068 configured, except for the credentials applied by
1069 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
1070 "+", but does still apply namespacing options unlike "+". There's
1071 also "!!" now, which is mostly identical, but becomes a NOP on
1072 systems that support ambient capabilities. This is useful to write
1073 unit files that work with ambient capabilities where possible but
1074 automatically fall back to traditional privilege dropping mechanisms
1075 on systems where this is not supported.
1077 * ListenNetlink= settings in socket units now support RDMA netlink
1080 * A new unit file setting LockPersonality= has been added which permits
1081 locking down the chosen execution domain ("personality") of a service
1084 * A new special target "getty-pre.target" has been added, which is
1085 ordered before all text logins, and may be used to order services
1086 before textual logins acquire access to the console.
1088 * systemd will now attempt to load the virtio-rng.ko kernel module very
1089 early on if a VM environment supporting this is detected. This should
1090 improve entropy during early boot in virtualized environments.
1092 * A _netdev option is now supported in /etc/crypttab that operates in a
1093 similar way as the same option in /etc/fstab: it permits configuring
1094 encrypted devices that need to be ordered after the network is up.
1095 Following this logic, two new special targets
1096 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
1097 added that are to cryptsetup.target what remote-fs.target and
1098 remote-fs-pre.target are to local-fs.target.
1100 * Service units gained a new UnsetEnvironment= setting which permits
1101 unsetting specific environment variables for services that are
1102 normally passed to it (for example in order to mask out locale
1103 settings for specific services that can't deal with it).
1105 * Units acquired a new boolean option IPAccounting=. When turned on, IP
1106 traffic accounting (packet count as well as byte count) is done for
1107 the service, and shown as part of "systemctl status" or "systemd-run
1110 * Service units acquired two new options IPAddressAllow= and
1111 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
1112 for configuring a simple IP access control list for all sockets of
1113 the unit. These options are available also on .slice and .socket
1114 units, permitting flexible access list configuration for individual
1115 services as well as groups of services (as defined by a slice unit),
1116 including system-wide. Note that IP ACLs configured this way are
1117 enforced on every single IPv4 and IPv6 socket created by any process
1118 of the service unit, and apply to ingress as well as egress traffic.
1120 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
1121 structured log message is generated each time the unit is stopped,
1122 containing information about the consumed resources of this
1125 * A new setting KeyringMode= has been added to unit files, which may be
1126 used to control how the kernel keyring is set up for executed
1129 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
1130 "systemctl kexec" and "systemctl exit" are now always asynchronous in
1131 behaviour (that is: these commands return immediately after the
1132 operation was enqueued instead of waiting for the operation to
1133 complete). Previously, "systemctl poweroff" and "systemctl reboot"
1134 were asynchronous on systems using systemd-logind (i.e. almost
1135 always, and like they were on sysvinit), and the other three commands
1136 were unconditionally synchronous. With this release this is cleaned
1137 up, and callers will see the same asynchronous behaviour on all
1138 systems for all five operations.
1140 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
1143 * .timer units now accept calendar specifications in other timezones
1144 than UTC or the local timezone.
1146 * The tmpfiles snippet var.conf has been changed to create
1147 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
1148 the "utmp" group already, and it appears to be generally understood
1149 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
1150 databases. Previously this was implemented correctly for all these
1151 databases excepts btmp, which has been opened up like this now
1152 too. Note that while the other databases are world-readable
1153 (i.e. 0644), btmp is not and remains more restrictive.
1155 * The systemd-resolve tool gained a new --reset-server-features
1156 switch. When invoked like this systemd-resolved will forget
1157 everything it learnt about the features supported by the configured
1158 upstream DNS servers, and restarts the feature probing logic on the
1159 next resolver look-up for them at the highest feature level
1162 * The status dump systemd-resolved sends to the logs upon receiving
1163 SIGUSR1 now also includes information about all DNS servers it is
1164 configured to use, and the features levels it probed for them.
1166 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
1167 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
1168 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
1169 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
1170 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
1171 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
1172 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
1173 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
1174 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
1175 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
1176 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
1177 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
1178 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
1179 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
1180 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
1181 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
1182 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
1183 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
1184 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
1185 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1187 — Berlin, 2017-10-06
1191 * Meson is now supported as build system in addition to Automake. It is
1192 our plan to remove Automake in one of our next releases, so that
1193 Meson becomes our exclusive build system. Hence, please start using
1194 the Meson build system in your downstream packaging. There's plenty
1195 of documentation around how to use Meson, the extremely brief
1198 ./autogen.sh && ./configure && make && sudo make install
1202 meson build && ninja -C build && sudo ninja -C build install
1204 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
1205 which permits configuring a timeout on the time a job is
1206 running. This is particularly useful for setting timeouts on jobs for
1209 * Unit files gained two new options ConditionUser= and ConditionGroup=
1210 for conditionalizing units based on the identity of the user/group
1211 running a systemd user instance.
1213 * systemd-networkd now understands a new FlowLabel= setting in the
1214 [VXLAN] section of .network files, as well as a Priority= in
1215 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
1216 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
1217 gained support for configuration of GENEVE links, and IPv6 address
1218 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
1220 * .link files now understand a new Port= setting.
1222 * systemd-networkd's DHCP support gained support for DHCP option 119
1223 (domain search list).
1225 * systemd-networkd gained support for serving IPv6 address ranges using
1226 the Router Advertisement protocol. The new .network configuration
1227 section [IPv6Prefix] may be used to configure the ranges to
1228 serve. This is implemented based on a new, minimal, native server
1229 implementation of RA.
1231 * journalctl's --output= switch gained support for a new parameter
1232 "short-iso-precise" for a mode where timestamps are shown as precise
1235 * systemd-udevd's "net_id" builtin may now generate stable network
1236 interface names from IBM PowerVM VIO devices as well as ACPI platform
1239 * MulticastDNS support in systemd-resolved may now be explicitly
1240 enabled/disabled using the new MulticastDNS= configuration file
1243 * systemd-resolved may now optionally use libidn2 instead of the libidn
1244 for processing internationalized domain names. Support for libidn2
1245 should be considered experimental and should not be enabled by
1248 * "machinectl pull-tar" and related call may now do verification of
1249 downloaded images using SUSE-style .sha256 checksum files in addition
1250 to the already existing support for validating using Ubuntu-style
1253 * sd-bus gained support for a new sd_bus_message_appendv() call which
1254 is va_list equivalent of sd_bus_message_append().
1256 * sd-boot gained support for validating images using SHIM/MOK.
1258 * The SMACK code learnt support for "onlycap".
1260 * systemd-mount --umount is now much smarter in figuring out how to
1261 properly unmount a device given its mount or device path.
1263 * The code to call libnss_dns as a fallback from libnss_resolve when
1264 the communication with systemd-resolved fails was removed. This
1265 fallback was redundant and interfered with the [!UNAVAIL=return]
1266 suffix. See nss-resolve(8) for the recommended configuration.
1268 * systemd-logind may now be restarted without losing state. It stores
1269 the file descriptors for devices it manages in the system manager
1270 using the FDSTORE= mechanism. Please note that further changes in
1271 other components may be required to make use of this (for example
1272 Xorg has code to listen for stops of systemd-logind and terminate
1273 itself when logind is stopped or restarted, in order to avoid using
1274 stale file descriptors for graphical devices, which is now
1275 counterproductive and must be reverted in order for restarts of
1276 systemd-logind to be safe. See
1277 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
1279 * All kernel install plugins are called with the environment variable
1280 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
1281 /etc/machine-id. If the file is missing or empty, the variable is
1282 empty and BOOT_DIR_ABS is the path of a temporary directory which is
1283 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
1284 is empty, all plugins should not put anything in BOOT_DIR_ABS.
1286 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
1287 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
1288 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
1289 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
1290 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
1291 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
1292 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
1293 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
1294 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
1295 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
1296 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
1297 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
1298 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
1299 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
1300 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
1301 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
1302 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
1303 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
1304 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
1305 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
1306 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
1307 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
1308 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
1309 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
1310 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
1311 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
1312 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
1313 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
1314 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
1317 — Berlin, 2017-07-12
1321 * The "hybrid" control group mode has been modified to improve
1322 compatibility with "legacy" cgroups-v1 setups. Specifically, the
1323 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
1324 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
1325 cgroups-v1 hierarchy), the only externally visible change being that
1326 the cgroups-v2 hierarchy is also mounted, to
1327 /sys/fs/cgroup/unified. This should provide a large degree of
1328 compatibility with "legacy" cgroups-v1, while taking benefit of the
1329 better management capabilities of cgroups-v2.
1331 * The default control group setup mode may be selected both a boot-time
1332 via a set of kernel command line parameters (specifically:
1333 systemd.unified_cgroup_hierarchy= and
1334 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
1335 default selected on the configure command line
1336 (--with-default-hierarchy=). The upstream default is "hybrid"
1337 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
1338 this will change in a future systemd version to be "unified" (pure
1339 cgroups-v2 mode). The third option for the compile time option is
1340 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
1341 distributions to default to "hybrid" mode for release distributions,
1342 starting with v233. We recommend "unified" for development
1343 distributions (specifically: distributions such as Fedora's rawhide)
1344 as that's where things are headed in the long run. Use "legacy" for
1345 greatest stability and compatibility only.
1347 * Note one current limitation of "unified" and "hybrid" control group
1348 setup modes: the kernel currently does not permit the systemd --user
1349 instance (i.e. unprivileged code) to migrate processes between two
1350 disconnected cgroup subtrees, even if both are managed and owned by
1351 the user. This effectively means "systemd-run --user --scope" doesn't
1352 work when invoked from outside of any "systemd --user" service or
1353 scope. Specifically, it is not supported from session scopes. We are
1354 working on fixing this in a future systemd version. (See #3388 for
1355 further details about this.)
1357 * DBus policy files are now installed into /usr rather than /etc. Make
1358 sure your system has dbus >= 1.9.18 running before upgrading to this
1359 version, or override the install path with --with-dbuspolicydir= .
1361 * All python scripts shipped with systemd (specifically: the various
1362 tests written in Python) now require Python 3.
1364 * systemd unit tests can now run standalone (without the source or
1365 build directories), and can be installed into /usr/lib/systemd/tests/
1366 with 'make install-tests'.
1368 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
1369 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
1372 * Support for the %c, %r, %R specifiers in unit files has been
1373 removed. Specifiers are not supposed to be dependent on configuration
1374 in the unit file itself (so that they resolve the same regardless
1375 where used in the unit files), but these specifiers were influenced
1376 by the Slice= option.
1378 * The shell invoked by debug-shell.service now defaults to /bin/sh in
1379 all cases. If distributions want to use a different shell for this
1380 purpose (for example Fedora's /sbin/sushell) they need to specify
1381 this explicitly at configure time using --with-debug-shell=.
1383 * The confirmation spawn prompt has been reworked to offer the
1386 (c)ontinue, proceed without asking anymore
1387 (D)ump, show the state of the unit
1388 (f)ail, don't execute the command and pretend it failed
1390 (i)nfo, show a short summary of the unit
1391 (j)obs, show jobs that are in progress
1392 (s)kip, don't execute the command and pretend it succeeded
1393 (y)es, execute the command
1395 The 'n' choice for the confirmation spawn prompt has been removed,
1396 because its meaning was confusing.
1398 The prompt may now also be redirected to an alternative console by
1399 specifying the console as parameter to systemd.confirm_spawn=.
1401 * Services of Type=notify require a READY=1 notification to be sent
1402 during startup. If no such message is sent, the service now fails,
1403 even if the main process exited with a successful exit code.
1405 * Services that fail to start up correctly now always have their
1406 ExecStopPost= commands executed. Previously, they'd enter "failed"
1407 state directly, without executing these commands.
1409 * The option MulticastDNS= of network configuration files has acquired
1410 an actual implementation. With MulticastDNS=yes a host can resolve
1411 names of remote hosts and reply to mDNS A and AAAA requests.
1413 * When units are about to be started an additional check is now done to
1414 ensure that all dependencies of type BindsTo= (when used in
1415 combination with After=) have been started.
1417 * systemd-analyze gained a new verb "syscall-filter" which shows which
1418 system call groups are defined for the SystemCallFilter= unit file
1419 setting, and which system calls they contain.
1421 * A new system call filter group "@filesystem" has been added,
1422 consisting of various file system related system calls. Group
1423 "@reboot" has been added, covering reboot, kexec and shutdown related
1424 calls. Finally, group "@swap" has been added covering swap
1425 configuration related calls.
1427 * A new unit file option RestrictNamespaces= has been added that may be
1428 used to restrict access to the various process namespace types the
1429 Linux kernel provides. Specifically, it may be used to take away the
1430 right for a service unit to create additional file system, network,
1431 user, and other namespaces. This sandboxing option is particularly
1432 relevant due to the high amount of recently discovered namespacing
1433 related vulnerabilities in the kernel.
1435 * systemd-udev's .link files gained support for a new AutoNegotiation=
1436 setting for configuring Ethernet auto-negotiation.
1438 * systemd-networkd's .network files gained support for a new
1439 ListenPort= setting in the [DHCP] section to explicitly configure the
1440 UDP client port the DHCP client shall listen on.
1442 * .network files gained a new Unmanaged= boolean setting for explicitly
1443 excluding one or more interfaces from management by systemd-networkd.
1445 * The systemd-networkd ProxyARP= option has been renamed to
1446 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
1447 renamed to ReduceARPProxy=. The old names continue to be available
1450 * systemd-networkd gained support for configuring IPv6 Proxy NDP
1451 addresses via the new IPv6ProxyNDPAddress= .network file setting.
1453 * systemd-networkd's bonding device support gained support for two new
1454 configuration options ActiveSlave= and PrimarySlave=.
1456 * The various options in the [Match] section of .network files gained
1457 support for negative matching.
1459 * New systemd-specific mount options are now understood in /etc/fstab:
1461 x-systemd.mount-timeout= may be used to configure the maximum
1462 permitted runtime of the mount command.
1464 x-systemd.device-bound may be set to bind a mount point to its
1465 backing device unit, in order to automatically remove a mount point
1466 if its backing device is unplugged. This option may also be
1467 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
1468 on the block device, which is now automatically set for all CDROM
1469 drives, so that mounted CDs are automatically unmounted when they are
1470 removed from the drive.
1472 x-systemd.after= and x-systemd.before= may be used to explicitly
1473 order a mount after or before another unit or mount point.
1475 * Enqueued start jobs for device units are now automatically garbage
1476 collected if there are no jobs waiting for them anymore.
1478 * systemctl list-jobs gained two new switches: with --after, for every
1479 queued job the jobs it's waiting for are shown; with --before the
1480 jobs which it's blocking are shown.
1482 * systemd-nspawn gained support for ephemeral boots from disk images
1483 (or in other words: --ephemeral and --image= may now be
1484 combined). Moreover, ephemeral boots are now supported for normal
1485 directories, even if the backing file system is not btrfs. Of course,
1486 if the file system does not support file system snapshots or
1487 reflinks, the initial copy operation will be relatively expensive, but
1488 this should still be suitable for many use cases.
1490 * Calendar time specifications in .timer units now support
1491 specifications relative to the end of a month by using "~" instead of
1492 "-" as separator between month and day. For example, "*-02~03" means
1493 "the third last day in February". In addition a new syntax for
1494 repeated events has been added using the "/" character. For example,
1495 "9..17/2:00" means "every two hours from 9am to 5pm".
1497 * systemd-socket-proxyd gained a new parameter --connections-max= for
1498 configuring the maximum number of concurrent connections.
1500 * sd-id128 gained a new API for generating unique IDs for the host in a
1501 way that does not leak the machine ID. Specifically,
1502 sd_id128_get_machine_app_specific() derives an ID based on the
1503 machine ID a in well-defined, non-reversible, stable way. This is
1504 useful whenever an identifier for the host is needed but where the
1505 identifier shall not be useful to identify the system beyond the
1506 scope of the application itself. (Internally this uses HMAC-SHA256 as
1507 keyed hash function using the machine ID as input.)
1509 * NotifyAccess= gained a new supported value "exec". When set
1510 notifications are accepted from all processes systemd itself invoked,
1511 including all control processes.
1513 * .nspawn files gained support for defining overlay mounts using the
1514 Overlay= and OverlayReadOnly= options. Previously this functionality
1515 was only available on the systemd-nspawn command line.
1517 * systemd-nspawn's --bind= and --overlay= options gained support for
1518 bind/overlay mounts whose source lies within the container tree by
1519 prefixing the source path with "+".
1521 * systemd-nspawn's --bind= and --overlay= options gained support for
1522 automatically allocating a temporary source directory in /var/tmp
1523 that is removed when the container dies. Specifically, if the source
1524 directory is specified as empty string this mechanism is selected. An
1525 example usage is --overlay=+/var::/var, which creates an overlay
1526 mount based on the original /var contained in the image, overlayed
1527 with a temporary directory in the host's /var/tmp. This way changes
1528 to /var are automatically flushed when the container shuts down.
1530 * systemd-nspawn --image= option does now permit raw file system block
1531 devices (in addition to images containing partition tables, as
1534 * The disk image dissection logic in systemd-nspawn gained support for
1535 automatically setting up LUKS encrypted as well as Verity protected
1536 partitions. When a container is booted from an encrypted image the
1537 passphrase is queried at start-up time. When a container with Verity
1538 data is started, the root hash is search in a ".roothash" file
1539 accompanying the disk image (alternatively, pass the root hash via
1540 the new --root-hash= command line option).
1542 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
1543 be used to dissect disk images the same way as systemd-nspawn does
1544 it, following the Bootable Partition Specification. It may even be
1545 used to mount disk images with complex partition setups (including
1546 LUKS and Verity partitions) to a local host directory, in order to
1547 inspect them. This tool is not considered public API (yet), and is
1548 thus not installed into /usr/bin. Please do not rely on its
1549 existence, since it might go away or be changed in later systemd
1552 * A new generator "systemd-verity-generator" has been added, similar in
1553 style to "systemd-cryptsetup-generator", permitting automatic setup of
1554 Verity root partitions when systemd boots up. In order to make use of
1555 this your partition setup should follow the Discoverable Partitions
1556 Specification, and the GPT partition ID of the root file system
1557 partition should be identical to the upper 128bit of the Verity root
1558 hash. The GPT partition ID of the Verity partition protecting it
1559 should be the lower 128bit of the Verity root hash. If the partition
1560 image follows this model it is sufficient to specify a single
1561 "roothash=" kernel command line argument to both configure which root
1562 image and verity partition to use as well as the root hash for
1563 it. Note that systemd-nspawn's Verity support follows the same
1564 semantics, meaning that disk images with proper Verity data in place
1565 may be booted in containers with systemd-nspawn as well as on
1566 physical systems via the verity generator. Also note that the "mkosi"
1567 tool available at https://github.com/systemd/mkosi has been updated
1568 to generate Verity protected disk images following this scheme. In
1569 fact, it has been updated to generate disk images that optionally
1570 implement a complete UEFI SecureBoot trust chain, involving a signed
1571 kernel and initrd image that incorporates such a root hash as well as
1572 a Verity-enabled root partition.
1574 * The hardware database (hwdb) udev supports has been updated to carry
1575 accelerometer quirks.
1577 * All system services are now run with a fresh kernel keyring set up
1578 for them. The invocation ID is stored by default in it, thus
1579 providing a safe, non-overridable way to determine the invocation
1582 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
1583 options for bind mounting arbitrary paths in a service-specific
1584 way. When these options are used, arbitrary host or service files and
1585 directories may be mounted to arbitrary locations in the service's
1588 * Documentation has been added that lists all of systemd's low-level
1589 environment variables:
1591 https://github.com/systemd/systemd/blob/master/doc/ENVIRONMENT.md
1593 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
1594 whether a specific socket file descriptor matches a specified socket
1597 * systemd-firstboot has been updated to check for the
1598 systemd.firstboot= kernel command line option. It accepts a boolean
1599 and when set to false the first boot questions are skipped.
1601 * systemd-fstab-generator has been updated to check for the
1602 systemd.volatile= kernel command line option, which either takes an
1603 optional boolean parameter or the special value "state". If used the
1604 system may be booted in a "volatile" boot mode. Specifically,
1605 "systemd.volatile" is used, the root directory will be mounted as
1606 tmpfs, and only /usr is mounted from the actual root file system. If
1607 "systemd.volatile=state" is used, the root directory will be mounted
1608 as usual, but /var is mounted as tmpfs. This concept provides similar
1609 functionality as systemd-nspawn's --volatile= option, but provides it
1610 on physical boots. Use this option for implementing stateless
1611 systems, or testing systems with all state and/or configuration reset
1612 to the defaults. (Note though that many distributions are not
1613 prepared to boot up without a populated /etc or /var, though.)
1615 * systemd-gpt-auto-generator gained support for LUKS encrypted root
1616 partitions. Previously it only supported LUKS encrypted partitions
1617 for all other uses, except for the root partition itself.
1619 * Socket units gained support for listening on AF_VSOCK sockets for
1620 communication in virtualized QEMU environments.
1622 * The "configure" script gained a new option --with-fallback-hostname=
1623 for specifying the fallback hostname to use if none is configured in
1624 /etc/hostname. For example, by specifying
1625 --with-fallback-hostname=fedora it is possible to default to a
1626 hostname of "fedora" on pristine installations.
1628 * systemd-cgls gained support for a new --unit= switch for listing only
1629 the control groups of a specific unit. Similar --user-unit= has been
1630 added for listing only the control groups of a specific user unit.
1632 * systemd-mount gained a new --umount switch for unmounting a mount or
1633 automount point (and all mount/automount points below it).
1635 * systemd will now refuse full configuration reloads (via systemctl
1636 daemon-reload and related calls) unless at least 16MiB of free space
1637 are available in /run. This is a safety precaution in order to ensure
1638 that generators can safely operate after the reload completed.
1640 * A new unit file option RootImage= has been added, which has a similar
1641 effect as RootDirectory= but mounts the service's root directory from
1642 a disk image instead of plain directory. This logic reuses the same
1643 image dissection and mount logic that systemd-nspawn already uses,
1644 and hence supports any disk images systemd-nspawn supports, including
1645 those following the Discoverable Partition Specification, as well as
1646 Verity enabled images. This option enables systemd to run system
1647 services directly off disk images acting as resource bundles,
1648 possibly even including full integrity data.
1650 * A new MountAPIVFS= unit file option has been added, taking a boolean
1651 argument. If enabled /proc, /sys and /dev (collectively called the
1652 "API VFS") will be mounted for the service. This is only relevant if
1653 RootDirectory= or RootImage= is used for the service, as these mounts
1654 are of course in place in the host mount namespace anyway.
1656 * systemd-nspawn gained support for a new --pivot-root= switch. If
1657 specified the root directory within the container image is pivoted to
1658 the specified mount point, while the original root disk is moved to a
1659 different place. This option enables booting of ostree images
1660 directly with systemd-nspawn.
1662 * The systemd build scripts will no longer complain if the NTP server
1663 addresses are not changed from the defaults. Google now supports
1664 these NTP servers officially. We still recommend downstreams to
1665 properly register an NTP pool with the NTP pool project though.
1667 * coredumpctl gained a new "--reverse" option for printing the list
1668 of coredumps in reverse order.
1670 * coredumpctl will now show additional information about truncated and
1671 inaccessible coredumps, as well as coredumps that are still being
1672 processed. It also gained a new --quiet switch for suppressing
1673 additional informational message in its output.
1675 * coredumpctl gained support for only showing coredumps newer and/or
1676 older than specific timestamps, using the new --since= and --until=
1677 options, reminiscent of journalctl's options by the same name.
1679 * The systemd-coredump logic has been improved so that it may be reused
1680 to collect backtraces in non-compiled languages, for example in
1681 scripting languages such as Python.
1683 * machinectl will now show the UID shift of local containers, if user
1684 namespacing is enabled for them.
1686 * systemd will now optionally run "environment generator" binaries at
1687 configuration load time. They may be used to add environment
1688 variables to the environment block passed to services invoked. One
1689 user environment generator is shipped by default that sets up
1690 environment variables based on files dropped into /etc/environment.d
1691 and ~/.config/environment.d/.
1693 * systemd-resolved now includes the new, recently published 2017 DNSSEC
1696 * hostnamed has been updated to report a new chassis type of
1697 "convertible" to cover "foldable" laptops that can both act as a
1698 tablet and as a laptop, such as various Lenovo Yoga devices.
1700 Contributions from: Adrián López, Alexander Galanin, Alexander
1701 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
1702 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
1703 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
1704 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
1705 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
1706 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
1707 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
1708 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
1709 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
1710 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
1711 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
1712 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
1713 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
1714 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
1715 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
1716 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
1717 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
1718 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
1719 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
1720 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
1721 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
1722 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
1723 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
1724 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
1725 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
1726 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
1727 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
1730 — Berlin, 2017-03-01
1734 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
1735 RestrictAddressFamilies= enabled. These sandboxing options should
1736 generally be compatible with the various external udev call-out
1737 binaries we are aware of, however there may be exceptions, in
1738 particular when exotic languages for these call-outs are used. In
1739 this case, consider turning off these settings locally.
1741 * The new RemoveIPC= option can be used to remove IPC objects owned by
1742 the user or group of a service when that service exits.
1744 * The new ProtectKernelModules= option can be used to disable explicit
1745 load and unload operations of kernel modules by a service. In
1746 addition access to /usr/lib/modules is removed if this option is set.
1748 * ProtectSystem= option gained a new value "strict", which causes the
1749 whole file system tree with the exception of /dev, /proc, and /sys,
1750 to be remounted read-only for a service.
1752 * The new ProtectKernelTunables= option can be used to disable
1753 modification of configuration files in /sys and /proc by a service.
1754 Various directories and files are remounted read-only, so access is
1755 restricted even if the file permissions would allow it.
1757 * The new ProtectControlGroups= option can be used to disable write
1758 access by a service to /sys/fs/cgroup.
1760 * Various systemd services have been hardened with
1761 ProtectKernelTunables=yes, ProtectControlGroups=yes,
1762 RestrictAddressFamilies=.
1764 * Support for dynamically creating users for the lifetime of a service
1765 has been added. If DynamicUser=yes is specified, user and group IDs
1766 will be allocated from the range 61184..65519 for the lifetime of the
1767 service. They can be resolved using the new nss-systemd.so NSS
1768 module. The module must be enabled in /etc/nsswitch.conf. Services
1769 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
1770 any resources allocated by the service will be cleaned up when the
1771 service exits. They also have ProtectHome=read-only and
1772 ProtectSystem=strict enabled, so they are not able to make any
1773 permanent modifications to the system.
1775 * The nss-systemd module also always resolves root and nobody, making
1776 it possible to have no /etc/passwd or /etc/group files in minimal
1777 container or chroot environments.
1779 * Services may be started with their own user namespace using the new
1780 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
1781 under which the service is running are mapped. All other users are
1784 * Support for the cgroup namespace has been added to systemd-nspawn. If
1785 supported by kernel, the container system started by systemd-nspawn
1786 will have its own view of the cgroup hierarchy. This new behaviour
1787 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
1789 * The new MemorySwapMax= option can be used to limit the maximum swap
1790 usage under the unified cgroup hierarchy.
1792 * Support for the CPU controller in the unified cgroup hierarchy has
1793 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
1794 options. This controller requires out-of-tree patches for the kernel
1795 and the support is provisional.
1797 * Mount and automount units may now be created transiently
1798 (i.e. dynamically at runtime via the bus API, instead of requiring
1799 unit files in the file system).
1801 * systemd-mount is a new tool which may mount file systems – much like
1802 mount(8), optionally pulling in additional dependencies through
1803 transient .mount and .automount units. For example, this tool
1804 automatically runs fsck on a backing block device before mounting,
1805 and allows the automount logic to be used dynamically from the
1806 command line for establishing mount points. This tool is particularly
1807 useful when dealing with removable media, as it will ensure fsck is
1808 run – if necessary – before the first access and that the file system
1809 is quickly unmounted after each access by utilizing the automount
1810 logic. This maximizes the chance that the file system on the
1811 removable media stays in a clean state, and if it isn't in a clean
1812 state is fixed automatically.
1814 * LazyUnmount=yes option for mount units has been added to expose the
1815 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
1818 * /efi will be used as the mount point of the EFI boot partition, if
1819 the directory is present, and the mount point was not configured
1820 through other means (e.g. fstab). If /efi directory does not exist,
1821 /boot will be used as before. This makes it easier to automatically
1822 mount the EFI partition on systems where /boot is used for something
1825 * When operating on GPT disk images for containers, systemd-nspawn will
1826 now mount the ESP to /boot or /efi according to the same rules as PID
1827 1 running on a host. This allows tools like "bootctl" to operate
1828 correctly within such containers, in order to make container images
1829 bootable on physical systems.
1831 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
1833 * Two new user session targets have been added to support running
1834 graphical sessions under the systemd --user instance:
1835 graphical-session.target and graphical-session-pre.target. See
1836 systemd.special(7) for a description of how those targets should be
1839 * The vconsole initialization code has been significantly reworked to
1840 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
1841 support unicode keymaps. Font and keymap configuration will now be
1842 copied to all allocated virtual consoles.
1844 * FreeBSD's bhyve virtualization is now detected.
1846 * Information recorded in the journal for core dumps now includes the
1847 contents of /proc/mountinfo and the command line of the process at
1848 the top of the process hierarchy (which is usually the init process
1851 * systemd-journal-gatewayd learned the --directory= option to serve
1852 files from the specified location.
1854 * journalctl --root=… can be used to peruse the journal in the
1855 /var/log/ directories inside of a container tree. This is similar to
1856 the existing --machine= option, but does not require the container to
1859 * The hardware database has been extended to support
1860 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
1863 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
1864 specify the click rate for mice which include a horizontal wheel with
1865 a click rate that is different than the one for the vertical wheel.
1867 * systemd-run gained a new --wait option that makes service execution
1868 synchronous. (Specifically, the command will not return until the
1869 specified service binary exited.)
1871 * systemctl gained a new --wait option that causes the start command to
1872 wait until the units being started have terminated again.
1874 * A new journal output mode "short-full" has been added which displays
1875 timestamps with abbreviated English day names and adds a timezone
1876 suffix. Those timestamps include more information than the default
1877 "short" output mode, and can be passed directly to journalctl's
1878 --since= and --until= options.
1880 * /etc/resolv.conf will be bind-mounted into containers started by
1881 systemd-nspawn, if possible, so any changes to resolv.conf contents
1882 are automatically propagated to the container.
1884 * The number of instances for socket-activated services originating
1885 from a single IP address can be limited with
1886 MaxConnectionsPerSource=, extending the existing setting of
1889 * systemd-networkd gained support for vcan ("Virtual CAN") interface
1892 * .netdev and .network configuration can now be extended through
1895 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
1896 Segmentation Offload, Generic Receive Offload, Large Receive Offload
1897 can be enabled and disabled using the new UDPSegmentationOffload=,
1898 TCPSegmentationOffload=, GenericSegmentationOffload=,
1899 GenericReceiveOffload=, LargeReceiveOffload= options in the
1900 [Link] section of .link files.
1902 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
1903 Port VLAN ID can be configured for bridge devices using the new STP=,
1904 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
1905 section of .netdev files.
1907 * The route table to which routes received over DHCP or RA should be
1908 added can be configured with the new RouteTable= option in the [DHCP]
1909 and [IPv6AcceptRA] sections of .network files.
1911 * The Address Resolution Protocol can be disabled on links managed by
1912 systemd-networkd using the ARP=no setting in the [Link] section of
1915 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
1916 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
1917 encode information about the result and exit codes of the current
1918 service runtime cycle.
1920 * systemd-sysctl will now configure kernel parameters in the order
1921 they occur in the configuration files. This matches what sysctl
1922 has been traditionally doing.
1924 * kernel-install "plugins" that are executed to perform various
1925 tasks after a new kernel is added and before an old one is removed
1926 can now return a special value to terminate the procedure and
1927 prevent any later plugins from running.
1929 * Journald's SplitMode=login setting has been deprecated. It has been
1930 removed from documentation, and its use is discouraged. In a future
1931 release it will be completely removed, and made equivalent to current
1932 default of SplitMode=uid.
1934 * Storage=both option setting in /etc/systemd/coredump.conf has been
1935 removed. With fast LZ4 compression storing the core dump twice is not
1938 * The --share-system systemd-nspawn option has been replaced with an
1939 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
1940 this functionality is discouraged. In addition the variables
1941 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
1942 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
1943 individual namespaces.
1945 * "machinectl list" now shows the IP address of running containers in
1946 the output, as well as OS release information.
1948 * "loginctl list" now shows the TTY of each session in the output.
1950 * sd-bus gained new API calls sd_bus_track_set_recursive(),
1951 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
1952 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
1953 tracking objects in a "recursive" mode, where a single client can be
1954 counted multiple times, if it takes multiple references.
1956 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
1957 sd_bus_get_exit_on_disconnect(). They may be used to make a
1958 process using sd-bus automatically exit if the bus connection is
1961 * Bus clients of the service manager may now "pin" loaded units into
1962 memory, by taking an explicit reference on them. This is useful to
1963 ensure the client can retrieve runtime data about the service even
1964 after the service completed execution. Taking such a reference is
1965 available only for privileged clients and should be helpful to watch
1966 running services in a race-free manner, and in particular collect
1967 information about exit statuses and results.
1969 * The nss-resolve module has been changed to strictly return UNAVAIL
1970 when communication via D-Bus with resolved failed, and NOTFOUND when
1971 a lookup completed but was negative. This means it is now possible to
1972 neatly configure fallbacks using nsswitch.conf result checking
1973 expressions. Taking benefit of this, the new recommended
1974 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
1976 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
1978 * A new setting CtrlAltDelBurstAction= has been added to
1979 /etc/systemd/system.conf which may be used to configure the precise
1980 behaviour if the user on the console presses Ctrl-Alt-Del more often
1981 than 7 times in 2s. Previously this would unconditionally result in
1982 an expedited, immediate reboot. With this new setting the precise
1983 operation may be configured in more detail, and also turned off
1986 * In .netdev files two new settings RemoteChecksumTx= and
1987 RemoteChecksumRx= are now understood that permit configuring the
1988 remote checksumming logic for VXLAN networks.
1990 * The service manager learnt a new "invocation ID" concept for invoked
1991 services. Each runtime cycle of a service will get a new invocation
1992 ID (a 128bit random UUID) assigned that identifies the current
1993 run of the service uniquely and globally. A new invocation ID
1994 is generated each time a service starts up. The journal will store
1995 the invocation ID of a service along with any logged messages, thus
1996 making the invocation ID useful for matching the online runtime of a
1997 service with the offline log data it generated in a safe way without
1998 relying on synchronized timestamps. In many ways this new service
1999 invocation ID concept is similar to the kernel's boot ID concept that
2000 uniquely and globally identifies the runtime of each boot. The
2001 invocation ID of a service is passed to the service itself via an
2002 environment variable ($INVOCATION_ID). A new bus call
2003 GetUnitByInvocationID() has been added that is similar to GetUnit()
2004 but instead of retrieving the bus path for a unit by its name
2005 retrieves it by its invocation ID. The returned path is valid only as
2006 long as the passed invocation ID is current.
2008 * systemd-resolved gained a new "DNSStubListener" setting in
2009 resolved.conf. It either takes a boolean value or the special values
2010 "udp" and "tcp", and configures whether to enable the stub DNS
2011 listener on 127.0.0.53:53.
2013 * IP addresses configured via networkd may now carry additional
2014 configuration settings supported by the kernel. New options include:
2015 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
2016 PrefixRoute=, AutoJoin=.
2018 * The PAM configuration fragment file for "user@.service" shipped with
2019 systemd (i.e. the --user instance of systemd) has been stripped to
2020 the minimum necessary to make the system boot. Previously, it
2021 contained Fedora-specific stanzas that did not apply to other
2022 distributions. It is expected that downstream distributions add
2023 additional configuration lines, matching their needs to this file,
2024 using it only as rough template of what systemd itself needs. Note
2025 that this reduced fragment does not even include an invocation of
2026 pam_limits which most distributions probably want to add, even though
2027 systemd itself does not need it. (There's also the new build time
2028 option --with-pamconfdir=no to disable installation of the PAM
2031 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
2032 capability is now also dropped from its set (in addition to
2033 CAP_SYS_MKNOD as before).
2035 * In service unit files it is now possible to connect a specific named
2036 file descriptor with stdin/stdout/stdout of an executed service. The
2037 name may be specified in matching .socket units using the
2038 FileDescriptorName= setting.
2040 * A number of journal settings may now be configured on the kernel
2041 command line. Specifically, the following options are now understood:
2042 systemd.journald.max_level_console=,
2043 systemd.journald.max_level_store=,
2044 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
2045 systemd.journald.max_level_wall=.
2047 * "systemctl is-enabled --full" will now show by which symlinks a unit
2048 file is enabled in the unit dependency tree.
2050 * Support for VeraCrypt encrypted partitions has been added to the
2051 "cryptsetup" logic and /etc/crypttab.
2053 * systemd-detect-virt gained support for a new --private-users switch
2054 that checks whether the invoking processes are running inside a user
2055 namespace. Similar, a new special value "private-users" for the
2056 existing ConditionVirtualization= setting has been added, permitting
2057 skipping of specific units in user namespace environments.
2059 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
2060 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
2061 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
2062 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
2063 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
2064 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
2065 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
2066 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
2067 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
2068 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
2069 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
2070 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
2071 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
2072 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
2073 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
2074 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
2075 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
2076 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
2077 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
2078 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
2079 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
2080 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
2081 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
2082 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
2083 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
2084 Jędrzejewski-Szmek, Zeal Jagannatha
2086 — Santa Fe, 2016-11-03
2090 * In service units the various ExecXYZ= settings have been extended
2091 with an additional special character as first argument of the
2092 assigned value: if the character '+' is used the specified command
2093 line it will be run with full privileges, regardless of User=,
2094 Group=, CapabilityBoundingSet= and similar options. The effect is
2095 similar to the existing PermissionsStartOnly= option, but allows
2096 configuration of this concept for each executed command line
2099 * Services may now alter the service watchdog timeout at runtime by
2100 sending a WATCHDOG_USEC= message via sd_notify().
2102 * MemoryLimit= and related unit settings now optionally take percentage
2103 specifications. The percentage is taken relative to the amount of
2104 physical memory in the system (or in case of containers, the assigned
2105 amount of memory). This allows scaling service resources neatly with
2106 the amount of RAM available on the system. Similarly, systemd-logind's
2107 RuntimeDirectorySize= option now also optionally takes percentage
2110 * In similar fashion TasksMax= takes percentage values now, too. The
2111 value is taken relative to the configured maximum number of processes
2112 on the system. The per-service task maximum has been changed to 15%
2113 using this functionality. (Effectively this is an increase of 512 →
2114 4915 for service units, given the kernel's default pid_max setting.)
2116 * Calendar time specifications in .timer units now understand a ".."
2117 syntax for time ranges. Example: "4..7:10" may now be used for
2118 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
2121 * The InaccessableDirectories=, ReadOnlyDirectories= and
2122 ReadWriteDirectories= unit file settings have been renamed to
2123 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
2124 applied to all kinds of file nodes, and not just directories, with
2125 the exception of symlinks. Specifically these settings may now be
2126 used on block and character device nodes, UNIX sockets and FIFOS as
2127 well as regular files. The old names of these settings remain
2128 available for compatibility.
2130 * systemd will now log about all service processes it kills forcibly
2131 (using SIGKILL) because they remained after the clean shutdown phase
2132 of the service completed. This should help identifying services that
2133 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
2134 systemd-logind's configuration a similar log message is generated for
2135 processes killed at the end of each session due to this setting.
2137 * systemd will now set the $JOURNAL_STREAM environment variable for all
2138 services whose stdout/stderr are connected to the Journal (which
2139 effectively means by default: all services). The variable contains
2140 the device and inode number of the file descriptor used for
2141 stdout/stderr. This may be used by invoked programs to detect whether
2142 their stdout/stderr is connected to the Journal, in which case they
2143 can switch over to direct Journal communication, thus being able to
2144 pass extended, structured metadata along with their log messages. As
2145 one example, this is now used by glib's logging primitives.
2147 * When using systemd's default tmp.mount unit for /tmp, the mount point
2148 will now be established with the "nosuid" and "nodev" options. This
2149 avoids privilege escalation attacks that put traps and exploits into
2150 /tmp. However, this might cause problems if you e. g. put container
2151 images or overlays into /tmp; if you need this, override tmp.mount's
2152 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
2155 * systemd now supports the "memory" cgroup controller also on
2158 * The systemd-cgtop tool now optionally takes a control group path as
2159 command line argument. If specified, the control group list shown is
2160 limited to subgroups of that group.
2162 * The SystemCallFilter= unit file setting gained support for
2163 pre-defined, named system call filter sets. For example
2164 SystemCallFilter=@clock is now an effective way to make all clock
2165 changing-related system calls unavailable to a service. A number of
2166 similar pre-defined groups are defined. Writing system call filters
2167 for system services is simplified substantially with this new
2168 concept. Accordingly, all of systemd's own, long-running services now
2169 enable system call filtering based on this, by default.
2171 * A new service setting MemoryDenyWriteExecute= has been added, taking
2172 a boolean value. If turned on, a service may no longer create memory
2173 mappings that are writable and executable at the same time. This
2174 enhances security for services where this is enabled as it becomes
2175 harder to dynamically write and then execute memory in exploited
2176 service processes. This option has been enabled for all of systemd's
2177 own long-running services.
2179 * A new RestrictRealtime= service setting has been added, taking a
2180 boolean argument. If set the service's processes may no longer
2181 acquire realtime scheduling. This improves security as realtime
2182 scheduling may otherwise be used to easily freeze the system.
2184 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
2185 value. This may be used for requesting that the system manager inside
2186 of the container reports start-up completion to nspawn which then
2187 propagates this notification further to the service manager
2188 supervising nspawn itself. A related option NotifyReady= in .nspawn
2189 files has been added too. This functionality allows ordering of the
2190 start-up of multiple containers using the usual systemd ordering
2193 * machinectl gained a new command "stop" that is an alias for
2196 * systemd-resolved gained support for contacting DNS servers on
2197 link-local IPv6 addresses.
2199 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
2200 its caches. A method call for requesting the same operation has been
2201 added to the bus API too, and is made available via "systemd-resolve
2204 * systemd-resolve gained a new --status switch. If passed a brief
2205 summary of the used DNS configuration with per-interface information
2208 * resolved.conf gained a new Cache= boolean option, defaulting to
2209 on. If turned off local DNS caching is disabled. This comes with a
2210 performance penalty in particular when DNSSEC is enabled. Note that
2211 resolved disables its internal caching implicitly anyway, when the
2212 configured DNS server is on a host-local IP address such as ::1 or
2213 127.0.0.1, thus automatically avoiding double local caching.
2215 * systemd-resolved now listens on the local IP address 127.0.0.53:53
2216 for DNS requests. This improves compatibility with local programs
2217 that do not use the libc NSS or systemd-resolved's bus APIs for name
2218 resolution. This minimal DNS service is only available to local
2219 programs and does not implement the full DNS protocol, but enough to
2220 cover local DNS clients. A new, static resolv.conf file, listing just
2221 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
2222 now recommended to make /etc/resolv.conf a symlink to this file in
2223 order to route all DNS lookups to systemd-resolved, regardless if
2224 done via NSS, the bus API or raw DNS packets. Note that this local
2225 DNS service is not as fully featured as the libc NSS or
2226 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
2227 used to deliver link-local address information (as this implies
2228 sending a local interface index along), LLMNR/mDNS support via this
2229 interface is severely restricted. It is thus strongly recommended for
2230 all applications to use the libc NSS API or native systemd-resolved
2233 * systemd-networkd's bridge support learned a new setting
2234 VLANFiltering= for controlling VLAN filtering. Moreover a new section
2235 in .network files has been added for configuring VLAN bridging in
2236 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
2238 * systemd-networkd's IPv6 Router Advertisement code now makes use of
2239 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
2240 now be acquired without relying on DHCPv6. Two new options
2241 UseDomains= and UseDNS= have been added to configure this behaviour.
2243 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
2244 renamed IPv6AcceptRA=, without altering its behaviour. The old
2245 setting name remains available for compatibility reasons.
2247 * The systemd-networkd VTI/VTI6 tunneling support gained new options
2248 Key=, InputKey= and OutputKey=.
2250 * systemd-networkd gained support for VRF ("Virtual Routing Function")
2251 interface configuration.
2253 * "systemctl edit" may now be used to create new unit files by
2254 specifying the --force switch.
2256 * sd-event gained a new function sd_event_get_iteration() for
2257 requesting the current iteration counter of the event loop. It starts
2258 at zero and is increased by one with each event loop iteration.
2260 * A new rpm macro %systemd_ordering is provided by the macros.systemd
2261 file. It can be used in lieu of %systemd_requires in packages which
2262 don't use any systemd functionality and are intended to be installed
2263 in minimal containers without systemd present. This macro provides
2264 ordering dependencies to ensure that if the package is installed in
2265 the same rpm transaction as systemd, systemd will be installed before
2266 the scriptlets for the package are executed, allowing unit presets
2269 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
2270 been added to simplify packaging of generators.
2272 * The os-release file gained VERSION_CODENAME field for the
2273 distribution nickname (e.g. VERSION_CODENAME=woody).
2275 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
2276 can be set to disable parsing of metadata and the creation
2277 of persistent symlinks for that device.
2279 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
2280 to make them available to logged-in users has been reverted.
2282 * Much of the common code of the various systemd components is now
2283 built into an internal shared library libsystemd-shared-231.so
2284 (incorporating the systemd version number in the name, to be updated
2285 with future releases) that the components link to. This should
2286 decrease systemd footprint both in memory during runtime and on
2287 disk. Note that the shared library is not for public use, and is
2288 neither API nor ABI stable, but is likely to change with every new
2289 released update. Packagers need to make sure that binaries
2290 linking to libsystemd-shared.so are updated in step with the
2293 * Configuration for "mkosi" is now part of the systemd
2294 repository. mkosi is a tool to easily build legacy-free OS images,
2295 and is available on github: https://github.com/systemd/mkosi. If
2296 "mkosi" is invoked in the build tree a new raw OS image is generated
2297 incorporating the systemd sources currently being worked on and a
2298 clean, fresh distribution installation. The generated OS image may be
2299 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
2300 UEFI PC. This functionality is particularly useful to easily test
2301 local changes made to systemd in a pristine, defined environment. See
2302 doc/HACKING for details.
2304 * configure learned the --with-support-url= option to specify the
2305 distribution's bugtracker.
2307 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
2308 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
2309 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
2310 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
2311 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
2312 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
2313 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
2314 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
2315 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
2316 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
2317 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
2318 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
2319 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
2320 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
2321 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
2322 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
2323 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
2324 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
2325 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
2327 — Berlin, 2016-07-25
2331 * DNSSEC is now turned on by default in systemd-resolved (in
2332 "allow-downgrade" mode), but may be turned off during compile time by
2333 passing "--with-default-dnssec=no" to "configure" (and of course,
2334 during runtime with DNSSEC= in resolved.conf). We recommend
2335 downstreams to leave this on at least during development cycles and
2336 report any issues with the DNSSEC logic upstream. We are very
2337 interested in collecting feedback about the DNSSEC validator and its
2338 limitations in the wild. Note however, that DNSSEC support is
2339 probably nothing downstreams should turn on in stable distros just
2340 yet, as it might create incompatibilities with a few DNS servers and
2341 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
2342 automatically whenever we detect such incompatible setups, but there
2343 might be systems we do not cover yet. Hence: please help us testing
2344 the DNSSEC code, leave this on where you can, report back, but then
2345 again don't consider turning this on in your stable, LTS or
2346 production release just yet. (Note that you have to enable
2347 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
2348 and its DNSSEC mode for host name resolution from local
2351 * systemd-resolve conveniently resolves DANE records with the --tlsa
2352 option and OPENPGPKEY records with the --openpgp option. It also
2353 supports dumping raw DNS record data via the new --raw= switch.
2355 * systemd-logind will now by default terminate user processes that are
2356 part of the user session scope unit (session-XX.scope) when the user
2357 logs out. This behavior is controlled by the KillUserProcesses=
2358 setting in logind.conf, and the previous default of "no" is now
2359 changed to "yes". This means that user sessions will be properly
2360 cleaned up after, but additional steps are necessary to allow
2361 intentionally long-running processes to survive logout.
2363 While the user is logged in at least once, user@.service is running,
2364 and any service that should survive the end of any individual login
2365 session can be started at a user service or scope using systemd-run.
2366 systemd-run(1) man page has been extended with an example which shows
2367 how to run screen in a scope unit underneath user@.service. The same
2368 command works for tmux.
2370 After the user logs out of all sessions, user@.service will be
2371 terminated too, by default, unless the user has "lingering" enabled.
2372 To effectively allow users to run long-term tasks even if they are
2373 logged out, lingering must be enabled for them. See loginctl(1) for
2374 details. The default polkit policy was modified to allow users to
2375 set lingering for themselves without authentication.
2377 Previous defaults can be restored at compile time by the
2378 --without-kill-user-processes option to "configure".
2380 * systemd-logind gained new configuration settings SessionsMax= and
2381 InhibitorsMax=, both with a default of 8192. It will not register new
2382 user sessions or inhibitors above this limit.
2384 * systemd-logind will now reload configuration on SIGHUP.
2386 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
2387 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
2388 enable. Also, support for the "io" cgroup controller in the unified
2389 hierarchy has been added, so that the "memory", "pids" and "io" are
2390 now the controllers that are supported on the unified hierarchy.
2392 WARNING: it is not possible to use previous systemd versions with
2393 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
2394 is necessary to also update systemd in the initramfs if using the
2395 unified hierarchy. An updated SELinux policy is also required.
2397 * LLDP support has been extended, and both passive (receive-only) and
2398 active (sender) modes are supported. Passive mode ("routers-only") is
2399 enabled by default in systemd-networkd. Active LLDP mode is enabled
2400 by default for containers on the internal network. The "networkctl
2401 lldp" command may be used to list information gathered. "networkctl
2402 status" will also show basic LLDP information on connected peers now.
2404 * The IAID and DUID unique identifier sent in DHCP requests may now be
2405 configured for the system and each .network file managed by
2406 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
2408 * systemd-networkd gained support for configuring proxy ARP support for
2409 each interface, via the ProxyArp= setting in .network files. It also
2410 gained support for configuring the multicast querier feature of
2411 bridge devices, via the new MulticastQuerier= setting in .netdev
2412 files. Similarly, snooping on the IGMP traffic can be controlled
2413 via the new setting MulticastSnooping=.
2415 A new setting PreferredLifetime= has been added for addresses
2416 configured in .network file to configure the lifetime intended for an
2419 The systemd-networkd DHCP server gained the option EmitRouter=, which
2420 defaults to yes, to configure whether the DHCP Option 3 (Router)
2423 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
2424 systemd-socket-activate and installed into /usr/bin. It is now fully
2427 * systemd-journald now uses separate threads to flush changes to disk
2428 when closing journal files, thus reducing impact of slow disk I/O on
2429 logging performance.
2431 * The sd-journal API gained two new calls
2432 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
2433 can be used to open journal files using file descriptors instead of
2434 file or directory paths. sd_journal_open_container() has been
2435 deprecated, sd_journal_open_directory_fd() should be used instead
2436 with the flag SD_JOURNAL_OS_ROOT.
2438 * journalctl learned a new output mode "-o short-unix" that outputs log
2439 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
2440 UTC). It also gained support for a new --no-hostname setting to
2441 suppress the hostname column in the family of "short" output modes.
2443 * systemd-ask-password now optionally skips printing of the password to
2444 stdout with --no-output which can be useful in scripts.
2446 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
2447 (devices tagged with ID_MAKER_TOOL) are now tagged with
2448 "uaccess" and are available to logged in users.
2450 * The DeviceAllow= unit setting now supports specifiers (with "%").
2452 * "systemctl show" gained a new --value switch, which allows print a
2453 only the contents of a specific unit property, without also printing
2454 the property's name. Similar support was added to "show*" verbs
2455 of loginctl and machinectl that output "key=value" lists.
2457 * A new unit type "generated" was added for files dynamically generated
2458 by generator tools. Similarly, a new unit type "transient" is used
2459 for unit files created using the runtime API. "systemctl enable" will
2460 refuse to operate on such files.
2462 * A new command "systemctl revert" has been added that may be used to
2463 revert to the vendor version of a unit file, in case local changes
2464 have been made by adding drop-ins or overriding the unit file.
2466 * "machinectl clean" gained a new verb to automatically remove all or
2467 just hidden container images.
2469 * systemd-tmpfiles gained support for a new line type "e" for emptying
2470 directories, if they exist, without creating them if they don't.
2472 * systemd-nspawn gained support for automatically patching the UID/GIDs
2473 of the owners and the ACLs of all files and directories in a
2474 container tree to match the UID/GID user namespacing range selected
2475 for the container invocation. This mode is enabled via the new
2476 --private-users-chown switch. It also gained support for
2477 automatically choosing a free, previously unused UID/GID range when
2478 starting a container, via the new --private-users=pick setting (which
2479 implies --private-users-chown). Together, these options for the first
2480 time make user namespacing for nspawn containers fully automatic and
2481 thus deployable. The systemd-nspawn@.service template unit file has
2482 been changed to use this functionality by default.
2484 * systemd-nspawn gained a new --network-zone= switch, that allows
2485 creating ad-hoc virtual Ethernet links between multiple containers,
2486 that only exist as long as at least one container referencing them is
2487 running. This allows easy connecting of multiple containers with a
2488 common link that implements an Ethernet broadcast domain. Each of
2489 these network "zones" may be named relatively freely by the user, and
2490 may be referenced by any number of containers, but each container may
2491 only reference one of these "zones". On the lower level, this is
2492 implemented by an automatically managed bridge network interface for
2493 each zone, that is created when the first container referencing its
2494 zone is created and removed when the last one referencing its zone
2497 * The default start timeout may now be configured on the kernel command
2498 line via systemd.default_timeout_start_sec=. It was already
2499 configurable via the DefaultTimeoutStartSec= option in
2500 /etc/systemd/system.conf.
2502 * Socket units gained a new TriggerLimitIntervalSec= and
2503 TriggerLimitBurst= setting to configure a limit on the activation
2504 rate of the socket unit.
2506 * The LimitNICE= setting now optionally takes normal UNIX nice values
2507 in addition to the raw integer limit value. If the specified
2508 parameter is prefixed with "+" or "-" and is in the range -20..19 the
2509 value is understood as UNIX nice value. If not prefixed like this it
2510 is understood as raw RLIMIT_NICE limit.
2512 * Note that the effect of the PrivateDevices= unit file setting changed
2513 slightly with this release: the per-device /dev file system will be
2514 mounted read-only from this version on, and will have "noexec"
2515 set. This (minor) change of behavior might cause some (exceptional)
2516 legacy software to break, when PrivateDevices=yes is set for its
2517 service. Please leave PrivateDevices= off if you run into problems
2520 * systemd-bootchart has been split out to a separate repository:
2521 https://github.com/systemd/systemd-bootchart
2523 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
2524 merged into the kernel in its current form.
2526 * The compatibility libraries libsystemd-daemon.so,
2527 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
2528 which have been deprecated since systemd-209 have been removed along
2529 with the corresponding pkg-config files. All symbols provided by
2530 those libraries are provided by libsystemd.so.
2532 * The Capabilities= unit file setting has been removed (it is ignored
2533 for backwards compatibility). AmbientCapabilities= and
2534 CapabilityBoundingSet= should be used instead.
2536 * A new special target has been added, initrd-root-device.target,
2537 which creates a synchronization point for dependencies of the root
2538 device in early userspace. Initramfs builders must ensure that this
2539 target is now included in early userspace.
2541 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
2542 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
2543 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
2544 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
2545 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
2546 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
2547 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
2548 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
2549 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
2550 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
2551 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
2552 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
2553 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
2554 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
2555 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
2556 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
2557 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
2558 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
2559 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
2560 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
2561 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
2562 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
2563 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
2564 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
2565 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
2568 — Fairfax, 2016-05-21
2572 * The systemd-resolved DNS resolver service has gained a substantial
2573 set of new features, most prominently it may now act as a DNSSEC
2574 validating stub resolver. DNSSEC mode is currently turned off by
2575 default, but is expected to be turned on by default in one of the
2576 next releases. For now, we invite everybody to test the DNSSEC logic
2577 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
2578 service also gained a full set of D-Bus interfaces, including calls
2579 to configure DNS and DNSSEC settings per link (for use by external
2580 network management software). systemd-resolved and systemd-networkd
2581 now distinguish between "search" and "routing" domains. The former
2582 are used to qualify single-label names, the latter are used purely
2583 for routing lookups within certain domains to specific links.
2584 resolved now also synthesizes RRs for all entries from /etc/hosts.
2586 * The systemd-resolve tool (which is a client utility for
2587 systemd-resolved) has been improved considerably and is now fully
2588 supported and documented. Hence it has moved from /usr/lib/systemd to
2591 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
2594 * The coredump collection logic has been reworked: when a coredump is
2595 collected it is now written to disk, compressed and processed
2596 (including stacktrace extraction) from a new instantiated service
2597 systemd-coredump@.service, instead of directly from the
2598 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
2599 processing large coredumps can take up a substantial amount of
2600 resources and time, and this previously happened entirely outside of
2601 systemd's service supervision. With the new logic the core_pattern
2602 hook only does minimal metadata collection before passing off control
2603 to the new instantiated service, which is configured with a time
2604 limit, a nice level and other settings to minimize negative impact on
2605 the rest of the system. Also note that the new logic will honour the
2606 RLIMIT_CORE setting of the crashed process, which now allows users
2607 and processes to turn off coredumping for their processes by setting
2610 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
2611 and all forked processes by default. Previously, PID 1 would leave
2612 the setting at "0" for all processes, as set by the kernel. Note that
2613 the resource limit traditionally has no effect on the generated
2614 coredumps on the system if the /proc/sys/kernel/core_pattern hook
2615 logic is used. Since the limit is now honoured (see above) its
2616 default has been changed so that the coredumping logic is enabled by
2617 default for all processes, while allowing specific opt-out.
2619 * When the stacktrace is extracted from processes of system users, this
2620 is now done as "systemd-coredump" user, in order to sandbox this
2621 potentially security sensitive parsing operation. (Note that when
2622 processing coredumps of normal users this is done under the user ID
2623 of process that crashed, as before.) Packagers should take notice
2624 that it is now necessary to create the "systemd-coredump" system user
2625 and group at package installation time.
2627 * The systemd-activate socket activation testing tool gained support
2628 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
2629 and --seqpacket switches. It also has been extended to support both
2630 new-style and inetd-style file descriptor passing. Use the new
2631 --inetd switch to request inetd-style file descriptor passing.
2633 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
2634 variable, which takes a boolean value. If set to false, ANSI color
2635 output is disabled in the tools even when run on a terminal that
2638 * The VXLAN support in networkd now supports two new settings
2639 DestinationPort= and PortRange=.
2641 * A new systemd.machine_id= kernel command line switch has been added,
2642 that may be used to set the machine ID in /etc/machine-id if it is
2643 not initialized yet. This command line option has no effect if the
2644 file is already initialized.
2646 * systemd-nspawn gained a new --as-pid2 switch that invokes any
2647 specified command line as PID 2 rather than PID 1 in the
2648 container. In this mode PID 1 is a minimal stub init process that
2649 implements the special POSIX and Linux semantics of PID 1 regarding
2650 signal and child process management. Note that this stub init process
2651 is implemented in nspawn itself and requires no support from the
2652 container image. This new logic is useful to support running
2653 arbitrary commands in the container, as normal processes are
2654 generally not prepared to run as PID 1.
2656 * systemd-nspawn gained a new --chdir= switch for setting the current
2657 working directory for the process started in the container.
2659 * "journalctl /dev/sda" will now output all kernel log messages for
2660 specified device from the current boot, in addition to all devices
2661 that are parents of it. This should make log output about devices
2662 pretty useful, as long as kernel drivers attach enough metadata to
2663 the log messages. (The usual SATA drivers do.)
2665 * The sd-journal API gained two new calls
2666 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
2667 that report whether log data from /run or /var has been found.
2669 * journalctl gained a new switch "--fields" that prints all journal
2670 record field names currently in use in the journal. This is backed
2671 by two new sd-journal API calls sd_journal_enumerate_fields() and
2672 sd_journal_restart_fields().
2674 * Most configurable timeouts in systemd now expect an argument of
2675 "infinity" to turn them off, instead of "0" as before. The semantics
2676 from now on is that a timeout of "0" means "now", and "infinity"
2677 means "never". To maintain backwards compatibility, "0" continues to
2678 turn off previously existing timeout settings.
2680 * "systemctl reload-or-try-restart" has been renamed to "systemctl
2681 try-reload-or-restart" to clarify what it actually does: the "try"
2682 logic applies to both reloading and restarting, not just restarting.
2683 The old name continues to be accepted for compatibility.
2685 * On boot-up, when PID 1 detects that the system clock is behind the
2686 release date of the systemd version in use, the clock is now set
2687 to the latter. Previously, this was already done in timesyncd, in order
2688 to avoid running with clocks set to the various clock epochs such as
2689 1902, 1938 or 1970. With this change the logic is now done in PID 1
2690 in addition to timesyncd during early boot-up, so that it is enforced
2691 before the first process is spawned by systemd. Note that the logic
2692 in timesyncd remains, as it is more comprehensive and ensures
2693 clock monotonicity by maintaining a persistent timestamp file in
2694 /var. Since /var is generally not available in earliest boot or the
2695 initrd, this part of the logic remains in timesyncd, and is not done
2698 * Support for tweaking details in net_cls.class_id through the
2699 NetClass= configuration directive has been removed, as the kernel
2700 people have decided to deprecate that controller in cgroup v2.
2701 Userspace tools such as nftables are moving over to setting rules
2702 that are specific to the full cgroup path of a task, which obsoletes
2703 these controllers anyway. The NetClass= directive is kept around for
2704 legacy compatibility reasons. For a more in-depth description of the
2705 kernel change, please refer to the respective upstream commit:
2707 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
2709 * A new service setting RuntimeMaxSec= has been added that may be used
2710 to specify a maximum runtime for a service. If the timeout is hit, the
2711 service is terminated and put into a failure state.
2713 * A new service setting AmbientCapabilities= has been added. It allows
2714 configuration of additional Linux process capabilities that are
2715 passed to the activated processes. This is only available on very
2718 * The process resource limit settings in service units may now be used
2719 to configure hard and soft limits individually.
2721 * The various libsystemd APIs such as sd-bus or sd-event now publicly
2722 expose support for gcc's __attribute__((cleanup())) C extension.
2723 Specifically, for many object destructor functions alternative
2724 versions have been added that have names suffixed with "p" and take a
2725 pointer to a pointer to the object to destroy, instead of just a
2726 pointer to the object itself. This is useful because these destructor
2727 functions may be used directly as parameters to the cleanup
2728 construct. Internally, systemd has been a heavy user of this GCC
2729 extension for a long time, and with this change similar support is
2730 now available to consumers of the library outside of systemd. Note
2731 that by using this extension in your sources compatibility with old
2732 and strictly ANSI compatible C compilers is lost. However, all gcc or
2733 LLVM versions of recent years support this extension.
2735 * Timer units gained support for a new setting RandomizedDelaySec= that
2736 allows configuring some additional randomized delay to the configured
2737 time. This is useful to spread out timer events to avoid load peaks in
2738 clusters or larger setups.
2740 * Calendar time specifications now support sub-second accuracy.
2742 * Socket units now support listening on SCTP and UDP-lite protocol
2745 * The sd-event API now comes with a full set of man pages.
2747 * Older versions of systemd contained experimental support for
2748 compressing journal files and coredumps with the LZ4 compressor that
2749 was not compatible with the lz4 binary (due to API limitations of the
2750 lz4 library). This support has been removed; only support for files
2751 compatible with the lz4 binary remains. This LZ4 logic is now
2752 officially supported and no longer considered experimental.
2754 * The dkr image import logic has been removed again from importd. dkr's
2755 micro-services focus doesn't fit into the machine image focus of
2756 importd, and quickly got out of date with the upstream dkr API.
2758 * Creation of the /run/lock/lockdev/ directory was dropped from
2759 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
2760 been available for many years. If you still need this, you need to
2761 create your own tmpfiles.d config file with:
2763 d /run/lock/lockdev 0775 root lock -
2765 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
2766 and RebootArgument= have been moved from the [Service] section of
2767 unit files to [Unit], and they are now supported on all unit types,
2768 not just service units. Of course, elogind will continue to
2769 understand these settings also at the old location, in order to
2770 maintain compatibility.
2772 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
2773 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
2774 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
2775 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
2776 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
2777 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
2778 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
2779 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
2780 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
2781 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
2782 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
2783 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
2784 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
2785 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
2786 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
2787 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
2788 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
2789 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
2790 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2792 — Berlin, 2016-02-11
2796 * A number of properties previously only settable in unit
2797 files are now also available as properties to set when
2798 creating transient units programmatically via the bus, as it
2799 is exposed with systemd-run's --property=
2800 setting. Specifically, these are: SyslogIdentifier=,
2801 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
2802 EnvironmentFile=, ReadWriteDirectories=,
2803 ReadOnlyDirectories=, InaccessibleDirectories=,
2804 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
2806 * When creating transient services via the bus API it is now
2807 possible to pass in a set of file descriptors to use as
2808 STDIN/STDOUT/STDERR for the invoked process.
2810 * Slice units may now be created transiently via the bus APIs,
2811 similar to the way service and scope units may already be
2812 created transiently.
2814 * Wherever systemd expects a calendar timestamp specification
2815 (like in journalctl's --since= and --until= switches) UTC
2816 timestamps are now supported. Timestamps suffixed with "UTC"
2817 are now considered to be in Universal Time Coordinated
2818 instead of the local timezone. Also, timestamps may now
2819 optionally be specified with sub-second accuracy. Both of
2820 these additions also apply to recurring calendar event
2821 specification, such as OnCalendar= in timer units.
2823 * journalctl gained a new "--sync" switch that asks the
2824 journal daemon to write all so far unwritten log messages to
2825 disk and sync the files, before returning.
2827 * systemd-tmpfiles learned two new line types "q" and "Q" that
2828 operate like "v", but also set up a basic btrfs quota
2829 hierarchy when used on a btrfs file system with quota
2832 * tmpfiles' "v", "q" and "Q" will now create a plain directory
2833 instead of a subvolume (even on a btrfs file system) if the
2834 root directory is a plain directory, and not a
2835 subvolume. This should simplify things with certain chroot()
2836 environments which are not aware of the concept of btrfs
2839 * systemd-detect-virt gained a new --chroot switch to detect
2840 whether execution takes place in a chroot() environment.
2842 * CPUAffinity= now takes CPU index ranges in addition to
2845 * The various memory-related resource limit settings (such as
2846 LimitAS=) now understand the usual K, M, G, ... suffixes to
2847 the base of 1024 (IEC). Similar, the time-related resource
2848 limit settings understand the usual min, h, day, ...
2851 * There's a new system.conf setting DefaultTasksMax= to
2852 control the default TasksMax= setting for services and
2853 scopes running on the system. (TasksMax= is the primary
2854 setting that exposes the "pids" cgroup controller on systemd
2855 and was introduced in the previous systemd release.) The
2856 setting now defaults to 512, which means services that are
2857 not explicitly configured otherwise will only be able to
2858 create 512 processes or threads at maximum, from this
2859 version on. Note that this means that thread- or
2860 process-heavy services might need to be reconfigured to set
2861 TasksMax= to a higher value. It is sufficient to set
2862 TasksMax= in these specific unit files to a higher value, or
2863 even "infinity". Similar, there's now a logind.conf setting
2864 UserTasksMax= that defaults to 4096 and limits the total
2865 number of processes or tasks each user may own
2866 concurrently. nspawn containers also have the TasksMax=
2867 value set by default now, to 8192. Note that all of this
2868 only has an effect if the "pids" cgroup controller is
2869 enabled in the kernel. The general benefit of these changes
2870 should be a more robust and safer system, that provides a
2871 certain amount of per-service fork() bomb protection.
2873 * systemd-nspawn gained the new --network-veth-extra= switch
2874 to define additional and arbitrarily-named virtual Ethernet
2875 links between the host and the container.
2877 * A new service execution setting PassEnvironment= has been
2878 added that allows importing select environment variables
2879 from PID1's environment block into the environment block of
2882 * Timer units gained support for a new RemainAfterElapse=
2883 setting which takes a boolean argument. It defaults to on,
2884 exposing behaviour unchanged to previous releases. If set to
2885 off, timer units are unloaded after they elapsed if they
2886 cannot elapse again. This is particularly useful for
2887 transient timer units, which shall not stay around longer
2888 than until they first elapse.
2890 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
2891 default now (the kernel default is 16). This is beneficial
2892 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
2893 allows substantially larger numbers of queued
2894 datagrams. This should increase the capability of systemd to
2895 parallelize boot-up, as logging and sd_notify() are unlikely
2896 to stall execution anymore. If you need to change the value
2897 from the new defaults, use the usual sysctl.d/ snippets.
2899 * The compression framing format used by the journal or
2900 coredump processing has changed to be in line with what the
2901 official LZ4 tools generate. LZ4 compression support in
2902 systemd was considered unsupported previously, as the format
2903 was not compatible with the normal tools. With this release
2904 this has changed now, and it is hence safe for downstream
2905 distributions to turn it on. While not compressing as well
2906 as the XZ, LZ4 is substantially faster, which makes
2907 it a good default choice for the compression logic in the
2908 journal and in coredump handling.
2910 * Any reference to /etc/mtab has been dropped from
2911 systemd. The file has been obsolete since a while, but
2912 systemd refused to work on systems where it was incorrectly
2913 set up (it should be a symlink or non-existent). Please make
2914 sure to update to util-linux 2.27.1 or newer in conjunction
2915 with this systemd release, which also drops any reference to
2916 /etc/mtab. If you maintain a distribution make sure that no
2917 software you package still references it, as this is a
2918 likely source of bugs. There's also a glibc bug pending,
2919 asking for removal of any reference to this obsolete file:
2921 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
2923 Note that only util-linux versions built with
2924 --enable-libmount-force-mountinfo are supported.
2926 * Support for the ".snapshot" unit type has been removed. This
2927 feature turned out to be little useful and little used, and
2928 has now been removed from the core and from systemctl.
2930 * The dependency types RequiresOverridable= and
2931 RequisiteOverridable= have been removed from systemd. They
2932 have been used only very sparingly to our knowledge and
2933 other options that provide a similar effect (such as
2934 systemctl --mode=ignore-dependencies) are much more useful
2935 and commonly used. Moreover, they were only half-way
2936 implemented as the option to control behaviour regarding
2937 these dependencies was never added to systemctl. By removing
2938 these dependency types the execution engine becomes a bit
2939 simpler. Unit files that use these dependencies should be
2940 changed to use the non-Overridable dependency types
2941 instead. In fact, when parsing unit files with these
2942 options, that's what systemd will automatically convert them
2943 too, but it will also warn, asking users to fix the unit
2944 files accordingly. Removal of these dependency types should
2945 only affect a negligible number of unit files in the wild.
2947 * Behaviour of networkd's IPForward= option changed
2948 (again). It will no longer maintain a per-interface setting,
2949 but propagate one way from interfaces where this is enabled
2950 to the global kernel setting. The global setting will be
2951 enabled when requested by a network that is set up, but
2952 never be disabled again. This change was made to make sure
2953 IPv4 and IPv6 behaviour regarding packet forwarding is
2954 similar (as the Linux IPv6 stack does not support
2955 per-interface control of this setting) and to minimize
2958 * In unit files the behaviour of %u, %U, %h, %s has
2959 changed. These specifiers will now unconditionally resolve
2960 to the various user database fields of the user that the
2961 systemd instance is running as, instead of the user
2962 configured in the specific unit via User=. Note that this
2963 effectively doesn't change much, as resolving of these
2964 specifiers was already turned off in the --system instance
2965 of systemd, as we cannot do NSS lookups from PID 1. In the
2966 --user instance of systemd these specifiers where correctly
2967 resolved, but hardly made any sense, since the user instance
2968 lacks privileges to do user switches anyway, and User= is
2969 hence useless. Moreover, even in the --user instance of
2970 systemd behaviour was awkward as it would only take settings
2971 from User= assignment placed before the specifier into
2972 account. In order to unify and simplify the logic around
2973 this the specifiers will now always resolve to the
2974 credentials of the user invoking the manager (which in case
2975 of PID 1 is the root user).
2977 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
2978 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
2979 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
2980 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
2981 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
2982 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
2983 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
2984 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
2985 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
2986 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
2987 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
2988 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
2989 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
2990 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
2993 — Berlin, 2015-11-18
2997 * systemd now depends on util-linux v2.27. More specifically,
2998 the newly added mount monitor feature in libmount now
2999 replaces systemd's former own implementation.
3001 * libmount mandates /etc/mtab not to be regular file, and
3002 systemd now enforces this condition at early boot.
3003 /etc/mtab has been deprecated and warned about for a very
3004 long time, so systems running systemd should already have
3005 stopped having this file around as anything else than a
3006 symlink to /proc/self/mounts.
3008 * Support for the "pids" cgroup controller has been added. It
3009 allows accounting the number of tasks in a cgroup and
3010 enforcing limits on it. This adds two new setting
3011 TasksAccounting= and TasksMax= to each unit, as well as a
3012 global option DefaultTasksAccounting=.
3014 * Support for the "net_cls" cgroup controller has been added.
3015 It allows assigning a net class ID to each task in the
3016 cgroup, which can then be used in firewall rules and traffic
3017 shaping configurations. Note that the kernel netfilter net
3018 class code does not currently work reliably for ingress
3019 packets on unestablished sockets.
3021 This adds a new config directive called NetClass= to CGroup
3022 enabled units. Allowed values are positive numbers for fixed
3023 assignments and "auto" for picking a free value
3026 * 'systemctl is-system-running' now returns 'offline' if the
3027 system is not booted with systemd. This command can now be
3028 used as a substitute for 'systemd-notify --booted'.
3030 * Watchdog timeouts have been increased to 3 minutes for all
3031 in-tree service files. Apparently, disk IO issues are more
3032 frequent than we hoped, and user reported >1 minute waiting
3035 * 'machine-id-commit' functionality has been merged into
3036 'machine-id-setup --commit'. The separate binary has been
3039 * The WorkingDirectory= directive in unit files may now be set
3040 to the special value '~'. In this case, the working
3041 directory is set to the home directory of the user
3042 configured in User=.
3044 * "machinectl shell" will now open the shell in the home
3045 directory of the selected user by default.
3047 * The CrashChVT= configuration file setting is renamed to
3048 CrashChangeVT=, following our usual logic of not
3049 abbreviating unnecessarily. The old directive is still
3050 supported for compat reasons. Also, this directive now takes
3051 an integer value between 1 and 63, or a boolean value. The
3052 formerly supported '-1' value for disabling stays around for
3055 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
3056 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
3057 RootDirectory= properties can now be set for transient
3060 * The systemd-analyze tool gained a new "set-log-target" verb
3061 to change the logging target the system manager logs to
3062 dynamically during runtime. This is similar to how
3063 "systemd-analyze set-log-level" already changes the log
3066 * In nspawn /sys is now mounted as tmpfs, with only a selected
3067 set of subdirectories mounted in from the real sysfs. This
3068 enhances security slightly, and is useful for ensuring user
3069 namespaces work correctly.
3071 * Support for USB FunctionFS activation has been added. This
3072 allows implementation of USB gadget services that are
3073 activated as soon as they are requested, so that they don't
3074 have to run continuously, similar to classic socket
3077 * The "systemctl exit" command now optionally takes an
3078 additional parameter that sets the exit code to return from
3079 the systemd manager when exiting. This is only relevant when
3080 running the systemd user instance, or when running the
3081 system instance in a container.
3083 * sd-bus gained the new API calls sd_bus_path_encode_many()
3084 and sd_bus_path_decode_many() that allow easy encoding and
3085 decoding of multiple identifier strings inside a D-Bus
3086 object path. Another new call sd_bus_default_flush_close()
3087 has been added to flush and close per-thread default
3090 * systemd-cgtop gained support for a -M/--machine= switch to
3091 show the control groups within a certain container only.
3093 * "systemctl kill" gained support for an optional --fail
3094 switch. If specified the requested operation will fail of no
3095 processes have been killed, because the unit had no
3096 processes attached, or similar.
3098 * A new systemd.crash_reboot=1 kernel command line option has
3099 been added that triggers a reboot after crashing. This can
3100 also be set through CrashReboot= in systemd.conf.
3102 * The RuntimeDirectory= setting now understands unit
3103 specifiers like %i or %f.
3105 * A new (still internal) library API sd-ipv4acd has been added,
3106 that implements address conflict detection for IPv4. It's
3107 based on code from sd-ipv4ll, and will be useful for
3108 detecting DHCP address conflicts.
3110 * File descriptors passed during socket activation may now be
3111 named. A new API sd_listen_fds_with_names() is added to
3112 access the names. The default names may be overridden,
3113 either in the .socket file using the FileDescriptorName=
3114 parameter, or by passing FDNAME= when storing the file
3115 descriptors using sd_notify().
3117 * systemd-networkd gained support for:
3119 - Setting the IPv6 Router Advertisement settings via
3120 IPv6AcceptRouterAdvertisements= in .network files.
3122 - Configuring the HelloTimeSec=, MaxAgeSec= and
3123 ForwardDelaySec= bridge parameters in .netdev files.
3125 - Configuring PreferredSource= for static routes in
3128 * The "ask-password" framework used to query for LUKS harddisk
3129 passwords or SSL passwords during boot gained support for
3130 caching passwords in the kernel keyring, if it is
3131 available. This makes sure that the user only has to type in
3132 a passphrase once if there are multiple objects to unlock
3133 with the same one. Previously, such password caching was
3134 available only when Plymouth was used; this moves the
3135 caching logic into the systemd codebase itself. The
3136 "systemd-ask-password" utility gained a new --keyname=
3137 switch to control which kernel keyring key to use for
3138 caching a password in. This functionality is also useful for
3139 enabling display managers such as gdm to automatically
3140 unlock the user's GNOME keyring if its passphrase, the
3141 user's password and the harddisk password are the same, if
3142 gdm-autologin is used.
3144 * When downloading tar or raw images using "machinectl
3145 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
3146 file is now also downloaded, if it is available and stored
3147 next to the image file.
3149 * Units of type ".socket" gained a new boolean setting
3150 Writable= which is only useful in conjunction with
3151 ListenSpecial=. If true, enables opening the specified
3152 special file in O_RDWR mode rather than O_RDONLY mode.
3154 * systemd-rfkill has been reworked to become a singleton
3155 service that is activated through /dev/rfkill on each rfkill
3156 state change and saves the settings to disk. This way,
3157 systemd-rfkill is now compatible with devices that exist
3158 only intermittendly, and even restores state if the previous
3159 system shutdown was abrupt rather than clean.
3161 * The journal daemon gained support for vacuuming old journal
3162 files controlled by the number of files that shall remain,
3163 in addition to the already existing control by size and by
3164 date. This is useful as journal interleaving performance
3165 degrades with too many separate journal files, and allows
3166 putting an effective limit on them. The new setting defaults
3167 to 100, but this may be changed by setting SystemMaxFiles=
3168 and RuntimeMaxFiles= in journald.conf. Also, the
3169 "journalctl" tool gained the new --vacuum-files= switch to
3170 manually vacuum journal files to leave only the specified
3171 number of files in place.
3173 * udev will now create /dev/disk/by-path links for ATA devices
3174 on kernels where that is supported.
3176 * Galician, Serbian, Turkish and Korean translations were added.
3178 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
3179 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
3180 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
3181 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
3182 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
3183 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
3184 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
3185 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
3186 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
3187 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
3188 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
3189 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
3190 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
3191 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
3192 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
3193 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
3194 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
3195 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
3197 — Berlin, 2015-10-07
3201 * The DHCP implementation of systemd-networkd gained a set of
3204 - The DHCP server now supports emitting DNS and NTP
3205 information. It may be enabled and configured via
3206 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
3207 and NTP information is enabled, but no servers are
3208 configured, the corresponding uplink information (if there
3209 is any) is propagated.
3211 - Server and client now support transmission and reception
3212 of timezone information. It can be configured via the
3213 newly introduced network options UseTimezone=,
3214 EmitTimezone=, and Timezone=. Transmission of timezone
3215 information is enabled between host and containers by
3216 default now: the container will change its local timezone
3217 to what the host has set.
3219 - Lease timeouts can now be configured via
3220 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
3222 - The DHCP server improved on the stability of
3223 leases. Clients are more likely to get the same lease
3224 information back, even if the server loses state.
3226 - The DHCP server supports two new configuration options to
3227 control the lease address pool metrics, PoolOffset= and
3230 * The encapsulation limit of tunnels in systemd-networkd may
3231 now be configured via 'EncapsulationLimit='. It allows
3232 modifying the maximum additional levels of encapsulation
3233 that are permitted to be prepended to a packet.
3235 * systemd now supports the concept of user buses replacing
3236 session buses, if used with dbus-1.10 (and enabled via dbus
3237 --enable-user-session). It previously only supported this on
3238 kdbus-enabled systems, and this release expands this to
3239 'dbus-daemon' systems.
3241 * systemd-networkd now supports predictable interface names
3244 * systemd now optionally supports the new Linux kernel
3245 "unified" control group hierarchy. If enabled via the kernel
3246 command-line option 'systemd.unified_cgroup_hierarchy=1',
3247 systemd will try to mount the unified cgroup hierarchy
3248 directly on /sys/fs/cgroup. If not enabled, or not
3249 available, systemd will fall back to the legacy cgroup
3250 hierarchy setup, as before. Host system and containers can
3251 mix and match legacy and unified hierarchies as they
3252 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
3253 environment variable to individually select the hierarchy to
3254 use for executed containers. By default, nspawn will use the
3255 unified hierarchy for the containers if the host uses the
3256 unified hierarchy, and the legacy hierarchy otherwise.
3257 Please note that at this point the unified hierarchy is an
3258 experimental kernel feature and is likely to change in one
3259 of the next kernel releases. Therefore, it should not be
3260 enabled by default in downstream distributions yet. The
3261 minimum required kernel version for the unified hierarchy to
3262 work is 4.2. Note that when the unified hierarchy is used
3263 for the first time delegated access to controllers is
3264 safe. Because of this systemd-nspawn containers will get
3265 access to controllers now, as will systemd user
3266 sessions. This means containers and user sessions may now
3267 manage their own resources, partitioning up what the system
3270 * A new special scope unit "init.scope" has been introduced
3271 that encapsulates PID 1 of the system. It may be used to
3272 determine resource usage and enforce resource limits on PID
3273 1 itself. PID 1 hence moved out of the root of the control
3276 * The cgtop tool gained support for filtering out kernel
3277 threads when counting tasks in a control group. Also, the
3278 count of processes is now recursively summed up by
3279 default. Two options -k and --recursive= have been added to
3280 revert to old behaviour. The tool has also been updated to
3281 work correctly in containers now.
3283 * systemd-nspawn's --bind= and --bind-ro= options have been
3284 extended to allow creation of non-recursive bind mounts.
3286 * libsystemd gained two new calls sd_pid_get_cgroup() and
3287 sd_peer_get_cgroup() which return the control group path of
3288 a process or peer of a connected AF_UNIX socket. This
3289 function call is particularly useful when implementing
3290 delegated subtrees support in the control group hierarchy.
3292 * The "sd-event" event loop API of libsystemd now supports
3293 correct dequeuing of real-time signals, without losing
3296 * When systemd requests a PolicyKit decision when managing
3297 units it will now add additional fields to the request,
3298 including unit name and desired operation. This enables more
3299 powerful PolicyKit policies, that make decisions depending
3300 on these parameters.
3302 * nspawn learnt support for .nspawn settings files, that may
3303 accompany the image files or directories of containers, and
3304 may contain additional settings for the container. This is
3305 an alternative to configuring container parameters via the
3306 nspawn command line.
3308 Contributions from: Cristian Rodríguez, Daniel Mack, David
3309 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
3310 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
3311 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
3312 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
3313 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
3314 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
3315 Andersen, Tom Gundersen, Torstein Husebø
3317 — Berlin, 2015-09-08
3321 * machinectl gained a new verb 'shell' which opens a fresh
3322 shell on the target container or the host. It is similar to
3323 the existing 'login' command of machinectl, but spawns the
3324 shell directly without prompting for username or
3325 password. The pseudo machine '.host' now refers to the local
3326 host and is used by default. Hence, 'machinectl shell' can
3327 be used as replacement for 'su -' which spawns a session as
3328 a fresh systemd unit in a way that is fully isolated from
3329 the originating session.
3331 * systemd-networkd learned to cope with private-zone DHCP
3332 options and allows other programs to query the values.
3334 * SELinux access control when enabling/disabling units is no
3335 longer enforced with this release. The previous
3336 implementation was incorrect, and a new corrected
3337 implementation is not yet available. As unit file operations
3338 are still protected via PolicyKit and D-Bus policy this is
3339 not a security problem. Yet, distributions which care about
3340 optimal SELinux support should probably not stabilize on
3343 * sd-bus gained support for matches of type "arg0has=", that
3344 test for membership of strings in string arrays sent in bus
3347 * systemd-resolved now dumps the contents of its DNS and LLMNR
3348 caches to the logs on reception of the SIGUSR1 signal. This
3349 is useful to debug DNS behaviour.
3351 * The coredumpctl tool gained a new --directory= option to
3352 operate on journal files in a specific directory.
3354 * "systemctl reboot" and related commands gained a new
3355 "--message=" option which may be used to set a free-text
3356 wall message when shutting down or rebooting the
3357 system. This message is also logged, which is useful for
3358 figuring out the reason for a reboot or shutdown a
3361 * The "systemd-resolve-host" tool's -i switch now takes
3362 network interface numbers as alternative to interface names.
3364 * A new unit file setting for services has been introduced:
3365 UtmpMode= allows configuration of how precisely systemd
3366 handles utmp and wtmp entries for the service if this is
3367 enabled. This allows writing services that appear similar to
3368 user sessions in the output of the "w", "who", "last" and
3371 * systemd-resolved will now locally synthesize DNS resource
3372 records for the "localhost" and "gateway" domains as well as
3373 the local hostname. This should ensure that clients querying
3374 RRs via resolved will get similar results as those going via
3375 NSS, if nss-myhostname is enabled.
3377 Contributions from: Alastair Hughes, Alex Crawford, Daniel
3378 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
3379 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
3380 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
3381 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
3382 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
3383 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
3384 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
3385 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
3386 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
3387 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
3388 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3390 — Berlin, 2015-08-27
3394 * The systemd-efi-boot-generator functionality was merged into
3395 systemd-gpt-auto-generator.
3397 * systemd-networkd now supports Group Policy for vxlan
3398 devices. It can be enabled via the new boolean configuration
3399 option called 'GroupPolicyExtension='.
3401 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
3402 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
3403 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
3405 — Berlin, 2015-07-31
3409 * The python-systemd code has been removed from the systemd repository.
3410 A new repository has been created which accommodates the code from
3411 now on, and we kindly ask distributions to create a separate package
3412 for this: https://github.com/systemd/python-systemd
3414 * The systemd daemon will now reload its main configuration
3415 (/etc/systemd/system.conf) on daemon-reload.
3417 * sd-dhcp now exposes vendor specific extensions via
3418 sd_dhcp_lease_get_vendor_specific().
3420 * systemd-networkd gained a number of new configuration options.
3422 - A new boolean configuration option for TAP devices called
3423 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
3424 device, thus allowing to send and receive GSO packets.
3426 - A new tunnel configuration option called 'CopyDSCP='.
3427 If enabled, the DSCP field of ip6 tunnels is copied into the
3428 decapsulated packet.
3430 - A set of boolean bridge configuration options were added.
3431 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
3432 and 'UnicastFlood=' are now parsed by networkd and applied to the
3433 respective bridge link device via the respective IFLA_BRPORT_*
3436 - A new string configuration option to override the hostname sent
3437 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
3438 is true, networkd will use the configured hostname instead of the
3439 system hostname when sending DHCP requests.
3441 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
3442 networkd will configure the IPv6 flow-label of the tunnel device
3443 according to RFC2460.
3445 - The 'macvtap' virtual network devices are now supported, similar to
3446 the already supported 'macvlan' devices.
3448 * systemd-resolved now implements RFC5452 to improve resilience against
3449 cache poisoning. Additionally, source port randomization is enabled
3450 by default to further protect against DNS spoofing attacks.
3452 * nss-mymachines now supports translating UIDs and GIDs of running
3453 containers with user-namespaces enabled. If a container 'foo'
3454 translates a host uid 'UID' to the container uid 'TUID', then
3455 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
3456 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
3457 mapped as 'vg-foo-TGID'.
3459 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
3460 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
3461 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
3462 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
3463 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
3464 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
3465 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
3466 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
3467 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
3468 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
3470 — Berlin, 2015-07-29
3474 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
3475 There are no known issues with current sysfs, and udev does not need
3476 or should be used to work around such bugs.
3478 * udev does no longer enable USB HID power management. Several reports
3479 indicate, that some devices cannot handle that setting.
3481 * The udev accelerometer helper was removed. The functionality
3482 is now fully included in iio-sensor-proxy. But this means,
3483 older iio-sensor-proxy versions will no longer provide
3484 accelerometer/orientation data with this systemd version.
3485 Please upgrade iio-sensor-proxy to version 1.0.
3487 * networkd gained a new configuration option IPv6PrivacyExtensions=
3488 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
3489 for Stateless Address") on selected networks.
3491 * For the sake of fewer build-time dependencies and less code in the
3492 main repository, the python bindings are about to be removed in the
3493 next release. A new repository has been created which accommodates
3494 the code from now on, and we kindly ask distributions to create a
3495 separate package for this. The removal will take place in v223.
3497 https://github.com/systemd/python-systemd
3499 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
3500 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
3501 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
3502 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
3503 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
3504 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
3505 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
3506 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
3507 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3508 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3510 — Berlin, 2015-07-07
3514 * The sd-bus.h and sd-event.h APIs have now been declared
3515 stable and have been added to the official interface of
3516 libsystemd.so. sd-bus implements an alternative D-Bus client
3517 library, that is relatively easy to use, very efficient and
3518 supports both classic D-Bus as well as kdbus as transport
3519 backend. sd-event is a generic event loop abstraction that
3520 is built around Linux epoll, but adds features such as event
3521 prioritization or efficient timer handling. Both APIs are good
3522 choices for C programs looking for a bus and/or event loop
3523 implementation that is minimal and does not have to be
3524 portable to other kernels.
3526 * kdbus support is no longer compile-time optional. It is now
3527 always built-in. However, it can still be disabled at
3528 runtime using the kdbus=0 kernel command line setting, and
3529 that setting may be changed to default to off, by specifying
3530 --disable-kdbus at build-time. Note though that the kernel
3531 command line setting has no effect if the kdbus.ko kernel
3532 module is not installed, in which case kdbus is (obviously)
3533 also disabled. We encourage all downstream distributions to
3534 begin testing kdbus by adding it to the kernel images in the
3535 development distributions, and leaving kdbus support in
3538 * The minimal required util-linux version has been bumped to
3541 * Support for chkconfig (--enable-chkconfig) was removed in
3542 favor of calling an abstraction tool
3543 /lib/systemd/systemd-sysv-install. This needs to be
3544 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
3545 in README for details.
3547 * If there's a systemd unit and a SysV init script for the
3548 same service name, and the user executes "systemctl enable"
3549 for it (or a related call), then this will now enable both
3550 (or execute the related operation on both), not just the
3553 * The libudev API documentation has been converted from gtkdoc
3556 * gudev has been removed from the systemd tree, it is now an
3559 * The systemd-cgtop tool learnt a new --raw switch to generate
3560 "raw" (machine parsable) output.
3562 * networkd's IPForwarding= .network file setting learnt the
3563 new setting "kernel", which ensures that networkd does not
3564 change the IP forwarding sysctl from the default kernel
3567 * The systemd-logind bus API now exposes a new boolean
3568 property "Docked" that reports whether logind considers the
3569 system "docked", i.e. connected to a docking station or not.
3571 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
3572 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
3573 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
3574 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
3575 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
3576 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
3577 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
3578 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
3579 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
3580 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
3581 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
3582 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
3583 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
3584 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3585 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
3586 Fink, Zbigniew Jędrzejewski-Szmek
3588 — Berlin, 2015-06-19
3592 * The gudev library has been extracted into a separate repository
3593 available at: https://git.gnome.org/browse/libgudev/
3594 It is now managed as part of the Gnome project. Distributions
3595 are recommended to pass --disable-gudev to systemd and use
3596 gudev from the Gnome project instead. gudev is still included
3597 in systemd, for now. It will be removed soon, though. Please
3598 also see the announcement-thread on systemd-devel:
3599 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
3601 * systemd now exposes a CPUUsageNSec= property for each
3602 service unit on the bus, that contains the overall consumed
3603 CPU time of a service (the sum of what each process of the
3604 service consumed). This value is only available if
3605 CPUAccounting= is turned on for a service, and is then shown
3606 in the "systemctl status" output.
3608 * Support for configuring alternative mappings of the old SysV
3609 runlevels to systemd targets has been removed. They are now
3610 hardcoded in a way that runlevels 2, 3, 4 all map to
3611 multi-user.target and 5 to graphical.target (which
3612 previously was already the default behaviour).
3614 * The auto-mounter logic gained support for mount point
3615 expiry, using a new TimeoutIdleSec= setting in .automount
3616 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
3618 * The EFI System Partition (ESP) as mounted to /boot by
3619 systemd-efi-boot-generator will now be unmounted
3620 automatically after 2 minutes of not being used. This should
3621 minimize the risk of ESP corruptions.
3623 * New /etc/fstab options x-systemd.requires= and
3624 x-systemd.requires-mounts-for= are now supported to express
3625 additional dependencies for mounts. This is useful for
3626 journalling file systems that support external journal
3627 devices or overlay file systems that require underlying file
3628 systems to be mounted.
3630 * systemd does not support direct live-upgrades (via systemctl
3631 daemon-reexec) from versions older than v44 anymore. As no
3632 distribution we are aware of shipped such old versions in a
3633 stable release this should not be problematic.
3635 * When systemd forks off a new per-connection service instance
3636 it will now set the $REMOTE_ADDR environment variable to the
3637 remote IP address, and $REMOTE_PORT environment variable to
3638 the remote IP port. This behaviour is similar to the
3639 corresponding environment variables defined by CGI.
3641 * systemd-networkd gained support for uplink failure
3642 detection. The BindCarrier= option allows binding interface
3643 configuration dynamically to the link sense of other
3644 interfaces. This is useful to achieve behaviour like in
3647 * systemd-networkd gained support for configuring the DHCP
3648 client identifier to use when requesting leases.
3650 * systemd-networkd now has a per-network UseNTP= option to
3651 configure whether NTP server information acquired via DHCP
3652 is passed on to services like systemd-timesyncd.
3654 * systemd-networkd gained support for vti6 tunnels.
3656 * Note that systemd-networkd manages the sysctl variable
3657 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
3658 it is configured for since v219. The variable controls IP
3659 forwarding, and is a per-interface alternative to the global
3660 /proc/sys/net/ipv[46]/ip_forward. This setting is
3661 configurable in the IPForward= option, which defaults to
3662 "no". This means if networkd is used for an interface it is
3663 no longer sufficient to set the global sysctl option to turn
3664 on IP forwarding! Instead, the .network file option
3665 IPForward= needs to be turned on! Note that the
3666 implementation of this behaviour was broken in v219 and has
3669 * Many bonding and vxlan options are now configurable in
3672 * systemd-nspawn gained a new --property= setting to set unit
3673 properties for the container scope. This is useful for
3674 setting resource parameters (e.g. "CPUShares=500") on
3675 containers started from the command line.
3677 * systemd-nspawn gained a new --private-users= switch to make
3678 use of user namespacing available on recent Linux kernels.
3680 * systemd-nspawn may now be called as part of a shell pipeline
3681 in which case the pipes used for stdin and stdout are passed
3682 directly to the process invoked in the container, without
3683 indirection via a pseudo tty.
3685 * systemd-nspawn gained a new switch to control the UNIX
3686 signal to use when killing the init process of the container
3689 * systemd-nspawn gained a new --overlay= switch for mounting
3690 overlay file systems into the container using the new kernel
3693 * When a container image is imported via systemd-importd and
3694 the host file system is not btrfs, a loopback block device
3695 file is created in /var/lib/machines.raw with a btrfs file
3696 system inside. It is then mounted to /var/lib/machines to
3697 enable btrfs features for container management. The loopback
3698 file and btrfs file system is grown as needed when container
3699 images are imported via systemd-importd.
3701 * systemd-machined/systemd-importd gained support for btrfs
3702 quota, to enforce container disk space limits on disk. This
3703 is exposed in "machinectl set-limit".
3705 * systemd-importd now can import containers from local .tar,
3706 .raw and .qcow2 images, and export them to .tar and .raw. It
3707 can also import dkr v2 images now from the network (on top
3710 * systemd-importd gained support for verifying downloaded
3711 images with gpg2 (previously only gpg1 was supported).
3713 * systemd-machined, systemd-logind, systemd: most bus calls
3714 are now accessible to unprivileged processes via
3715 PolicyKit. Also, systemd-logind will now allow users to kill
3716 their own sessions without further privileges or
3719 * systemd-shutdownd has been removed. This service was
3720 previously responsible for implementing scheduled shutdowns
3721 as exposed in /usr/bin/shutdown's time parameter. This
3722 functionality has now been moved into systemd-logind and is
3723 accessible via a bus interface.
3725 * "systemctl reboot" gained a new switch --firmware-setup that
3726 can be used to reboot into the EFI firmware setup, if that
3727 is available. systemd-logind now exposes an API on the bus
3728 to trigger such reboots, in case graphical desktop UIs want
3729 to cover this functionality.
3731 * "systemctl enable", "systemctl disable" and "systemctl mask"
3732 now support a new "--now" switch. If specified the units
3733 that are enabled will also be started, and the ones
3734 disabled/masked also stopped.
3736 * The Gummiboot EFI boot loader tool has been merged into
3737 systemd, and renamed to "systemd-boot". The bootctl tool has been
3738 updated to support systemd-boot.
3740 * An EFI kernel stub has been added that may be used to create
3741 kernel EFI binaries that contain not only the actual kernel,
3742 but also an initrd, boot splash, command line and OS release
3743 information. This combined binary can then be signed as a
3744 single image, so that the firmware can verify it all in one
3745 step. systemd-boot has special support for EFI binaries created
3746 like this and can extract OS release information from them
3747 and show them in the boot menu. This functionality is useful
3748 to implement cryptographically verified boot schemes.
3750 * Optional support has been added to systemd-fsck to pass
3751 fsck's progress report to an AF_UNIX socket in the file
3754 * udev will no longer create device symlinks for all block
3755 devices by default. A blacklist for excluding special block
3756 devices from this logic has been turned into a whitelist
3757 that requires picking block devices explicitly that require
3760 * A new (currently still internal) API sd-device.h has been
3761 added to libsystemd. This modernized API is supposed to
3762 replace libudev eventually. In fact, already much of libudev
3763 is now just a wrapper around sd-device.h.
3765 * A new hwdb database for storing metadata about pointing
3766 stick devices has been added.
3768 * systemd-tmpfiles gained support for setting file attributes
3769 similar to the "chattr" tool with new 'h' and 'H' lines.
3771 * systemd-journald will no longer unconditionally set the
3772 btrfs NOCOW flag on new journal files. This is instead done
3773 with tmpfiles snippet using the new 'h' line type. This
3774 allows easy disabling of this logic, by masking the
3775 journal-nocow.conf tmpfiles file.
3777 * systemd-journald will now translate audit message types to
3778 human readable identifiers when writing them to the
3779 journal. This should improve readability of audit messages.
3781 * The LUKS logic gained support for the offset= and skip=
3782 options in /etc/crypttab, as previously implemented by
3785 * /usr/lib/os-release gained a new optional field VARIANT= for
3786 distributions that support multiple variants (such as a
3787 desktop edition, a server edition, ...)
3789 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
3790 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
3791 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
3792 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
3793 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
3794 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
3795 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
3796 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
3797 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
3798 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
3799 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
3800 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
3801 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
3802 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
3803 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
3804 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
3805 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
3806 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
3807 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
3808 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
3809 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
3810 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
3811 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
3812 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
3813 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
3814 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
3815 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
3817 — Berlin, 2015-05-22
3821 * Introduce a new API "sd-hwdb.h" for querying the hardware
3822 metadata database. With this minimal interface one can query
3823 and enumerate the udev hwdb, decoupled from the old libudev
3824 library. libudev's interface for this is now only a wrapper
3825 around sd-hwdb. A new tool systemd-hwdb has been added to
3826 interface with and update the database.
3828 * When any of systemd's tools copies files (for example due to
3829 tmpfiles' C lines) a btrfs reflink will attempted first,
3830 before bytewise copying is done.
3832 * systemd-nspawn gained a new --ephemeral switch. When
3833 specified a btrfs snapshot is taken of the container's root
3834 directory, and immediately removed when the container
3835 terminates again. Thus, a container can be started whose
3836 changes never alter the container's root directory, and are
3837 lost on container termination. This switch can also be used
3838 for starting a container off the root file system of the
3839 host without affecting the host OS. This switch is only
3840 available on btrfs file systems.
3842 * systemd-nspawn gained a new --template= switch. It takes the
3843 path to a container tree to use as template for the tree
3844 specified via --directory=, should that directory be
3845 missing. This allows instantiating containers dynamically,
3846 on first run. This switch is only available on btrfs file
3849 * When a .mount unit refers to a mount point on which multiple
3850 mounts are stacked, and the .mount unit is stopped all of
3851 the stacked mount points will now be unmounted until no
3852 mount point remains.
3854 * systemd now has an explicit notion of supported and
3855 unsupported unit types. Jobs enqueued for unsupported unit
3856 types will now fail with an "unsupported" error code. More
3857 specifically .swap, .automount and .device units are not
3858 supported in containers, .busname units are not supported on
3859 non-kdbus systems. .swap and .automount are also not
3860 supported if their respective kernel compile time options
3863 * machinectl gained support for two new "copy-from" and
3864 "copy-to" commands for copying files from a running
3865 container to the host or vice versa.
3867 * machinectl gained support for a new "bind" command to bind
3868 mount host directories into local containers. This is
3869 currently only supported for nspawn containers.
3871 * networkd gained support for configuring bridge forwarding
3872 database entries (fdb) from .network files.
3874 * A new tiny daemon "systemd-importd" has been added that can
3875 download container images in tar, raw, qcow2 or dkr formats,
3876 and make them available locally in /var/lib/machines, so
3877 that they can run as nspawn containers. The daemon can GPG
3878 verify the downloads (not supported for dkr, since it has no
3879 provisions for verifying downloads). It will transparently
3880 decompress bz2, xz, gzip compressed downloads if necessary,
3881 and restore sparse files on disk. The daemon uses privilege
3882 separation to ensure the actual download logic runs with
3883 fewer privileges than the daemon itself. machinectl has
3884 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
3885 make the functionality of importd available to the
3886 user. With this in place the Fedora and Ubuntu "Cloud"
3887 images can be downloaded and booted as containers unmodified
3888 (the Fedora images lack the appropriate GPG signature files
3889 currently, so they cannot be verified, but this will change
3890 soon, hopefully). Note that downloading images is currently
3891 only fully supported on btrfs.
3893 * machinectl is now able to list container images found in
3894 /var/lib/machines, along with some metadata about sizes of
3895 disk and similar. If the directory is located on btrfs and
3896 quota is enabled, this includes quota display. A new command
3897 "image-status" has been added that shows additional
3898 information about images.
3900 * machinectl is now able to clone container images
3901 efficiently, if the underlying file system (btrfs) supports
3902 it, with the new "machinectl clone" command. It also
3903 gained commands for renaming and removing images, as well as
3904 marking them read-only or read-write (supported also on
3905 legacy file systems).
3907 * networkd gained support for collecting LLDP network
3908 announcements, from hardware that supports this. This is
3909 shown in networkctl output.
3911 * systemd-run gained support for a new -t (--pty) switch for
3912 invoking a binary on a pty whose input and output is
3913 connected to the invoking terminal. This allows executing
3914 processes as system services while interactively
3915 communicating with them via the terminal. Most interestingly
3916 this is supported across container boundaries. Invoking
3917 "systemd-run -t /bin/bash" is an alternative to running a
3918 full login session, the difference being that the former
3919 will not register a session, nor go through the PAM session
3922 * tmpfiles gained support for a new "v" line type for creating
3923 btrfs subvolumes. If the underlying file system is a legacy
3924 file system, this automatically degrades to creating a
3925 normal directory. Among others /var/lib/machines is now
3926 created like this at boot, should it be missing.
3928 * The directory /var/lib/containers/ has been deprecated and
3929 been replaced by /var/lib/machines. The term "machines" has
3930 been used in the systemd context as generic term for both
3931 VMs and containers, and hence appears more appropriate for
3932 this, as the directory can also contain raw images bootable
3935 * systemd-nspawn when invoked with -M but without --directory=
3936 or --image= is now capable of searching for the container
3937 root directory, subvolume or disk image automatically, in
3938 /var/lib/machines. systemd-nspawn@.service has been updated
3939 to make use of this, thus allowing it to be used for raw
3942 * A new machines.target unit has been introduced that is
3943 supposed to group all containers/VMs invoked as services on
3944 the system. systemd-nspawn@.service has been updated to
3945 integrate with that.
3947 * machinectl gained a new "start" command, for invoking a
3948 container as a service. "machinectl start foo" is mostly
3949 equivalent to "systemctl start systemd-nspawn@foo.service",
3950 but handles escaping in a nicer way.
3952 * systemd-nspawn will now mount most of the cgroupfs tree
3953 read-only into each container, with the exception of the
3954 container's own subtree in the name=systemd hierarchy.
3956 * journald now sets the special FS_NOCOW file flag for its
3957 journal files. This should improve performance on btrfs, by
3958 avoiding heavy fragmentation when journald's write-pattern
3959 is used on COW file systems. It degrades btrfs' data
3960 integrity guarantees for the files to the same levels as for
3961 ext3/ext4 however. This should be OK though as journald does
3962 its own data integrity checks and all its objects are
3963 checksummed on disk. Also, journald should handle btrfs disk
3964 full events a lot more gracefully now, by processing SIGBUS
3965 errors, and not relying on fallocate() anymore.
3967 * When journald detects that journal files it is writing to
3968 have been deleted it will immediately start new journal
3971 * systemd now provides a way to store file descriptors
3972 per-service in PID 1. This is useful for daemons to ensure
3973 that fds they require are not lost during a daemon
3974 restart. The fds are passed to the daemon on the next
3975 invocation in the same way socket activation fds are
3976 passed. This is now used by journald to ensure that the
3977 various sockets connected to all the system's stdout/stderr
3978 are not lost when journald is restarted. File descriptors
3979 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
3980 an extension to sd_notify(). Note that a limit is enforced
3981 on the number of fds a service can store in PID 1, and it
3982 defaults to 0, so that no fds may be stored, unless this is
3983 explicitly turned on.
3985 * The default TERM variable to use for units connected to a
3986 terminal, when no other value is explicitly is set is now
3987 vt220 rather than vt102. This should be fairly safe still,
3988 but allows PgUp/PgDn work.
3990 * The /etc/crypttab option header= as known from Debian is now
3993 * "loginctl user-status" and "loginctl session-status" will
3994 now show the last 10 lines of log messages of the
3995 user/session following the status output. Similar,
3996 "machinectl status" will show the last 10 log lines
3997 associated with a virtual machine or container
3998 service. (Note that this is usually not the log messages
3999 done in the VM/container itself, but simply what the
4000 container manager logs. For nspawn this includes all console
4003 * "loginctl session-status" without further argument will now
4004 show the status of the session of the caller. Similar,
4005 "lock-session", "unlock-session", "activate",
4006 "enable-linger", "disable-linger" may now be called without
4007 session/user parameter in which case they apply to the
4008 caller's session/user.
4010 * An X11 session scriptlet is now shipped that uploads
4011 $DISPLAY and $XAUTHORITY into the environment of the systemd
4012 --user daemon if a session begins. This should improve
4013 compatibility with X11 enabled applications run as systemd
4016 * Generators are now subject to masking via /etc and /run, the
4017 same way as unit files.
4019 * networkd .network files gained support for configuring
4020 per-link IPv4/IPv6 packet forwarding as well as IPv4
4021 masquerading. This is by default turned on for veth links to
4022 containers, as registered by systemd-nspawn. This means that
4023 nspawn containers run with --network-veth will now get
4024 automatic routed access to the host's networks without any
4025 further configuration or setup, as long as networkd runs on
4028 * systemd-nspawn gained the --port= (-p) switch to expose TCP
4029 or UDP posts of a container on the host. With this in place
4030 it is possible to run containers with private veth links
4031 (--network-veth), and have their functionality exposed on
4032 the host as if their services were running directly on the
4035 * systemd-nspawn's --network-veth switch now gained a short
4036 version "-n", since with the changes above it is now truly
4037 useful out-of-the-box. The systemd-nspawn@.service has been
4038 updated to make use of it too by default.
4040 * systemd-nspawn will now maintain a per-image R/W lock, to
4041 ensure that the same image is not started more than once
4042 writable. (It's OK to run an image multiple times
4043 simultaneously in read-only mode.)
4045 * systemd-nspawn's --image= option is now capable of
4046 dissecting and booting MBR and GPT disk images that contain
4047 only a single active Linux partition. Previously it
4048 supported only GPT disk images with proper GPT type
4049 IDs. This allows running cloud images from major
4050 distributions directly with systemd-nspawn, without
4053 * In addition to collecting mouse dpi data in the udev
4054 hardware database, there's now support for collecting angle
4055 information for mouse scroll wheels. The database is
4056 supposed to guarantee similar scrolling behavior on mice
4057 that it knows about. There's also support for collecting
4058 information about Touchpad types.
4060 * udev's input_id built-in will now also collect touch screen
4061 dimension data and attach it to probed devices.
4063 * /etc/os-release gained support for a Distribution Privacy
4066 * networkd gained support for creating "ipvlan", "gretap",
4067 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
4069 * systemd-tmpfiles gained support for "a" lines for setting
4072 * systemd-nspawn will now mount /tmp in the container to
4073 tmpfs, automatically.
4075 * systemd now exposes the memory.usage_in_bytes cgroup
4076 attribute and shows it for each service in the "systemctl
4077 status" output, if available.
4079 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
4080 immediate reboot is triggered. This useful if shutdown is
4081 hung and is unable to complete, to expedite the
4082 operation. Note that this kind of reboot will still unmount
4083 all file systems, and hence should not result in fsck being
4086 * A .device unit for an optical block device will now be
4087 considered active only when a medium is in the drive. Also,
4088 mount units are now bound to their backing devices thus
4089 triggering automatic unmounting when devices become
4090 unavailable. With this in place systemd will now
4091 automatically unmount left-over mounts when a CD-ROM is
4092 ejected or an USB stick is yanked from the system.
4094 * networkd-wait-online now has support for waiting for
4095 specific interfaces only (with globbing), and for giving up
4096 after a configurable timeout.
4098 * networkd now exits when idle. It will be automatically
4099 restarted as soon as interfaces show up, are removed or
4100 change state. networkd will stay around as long as there is
4101 at least one DHCP state machine or similar around, that keep
4104 * networkd may now configure IPv6 link-local addressing in
4105 addition to IPv4 link-local addressing.
4107 * The IPv6 "token" for use in SLAAC may now be configured for
4108 each .network interface in networkd.
4110 * Routes configured with networkd may now be assigned a scope
4113 * networkd's [Match] sections now support globbing and lists
4114 of multiple space-separated matches per item.
4116 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
4117 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
4118 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
4119 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
4120 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
4121 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
4122 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
4123 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
4124 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
4125 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
4126 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
4127 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
4128 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
4129 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
4130 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
4131 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
4132 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
4133 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
4134 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
4135 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
4136 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
4137 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
4138 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
4139 Hoffmann, Zbigniew Jędrzejewski-Szmek
4141 — Berlin, 2015-02-16
4145 * When querying unit file enablement status (for example via
4146 "systemctl is-enabled"), a new state "indirect" is now known
4147 which indicates that a unit might not be enabled itself, but
4148 another unit listed in its Also= setting might be.
4150 * Similar to the various existing ConditionXYZ= settings for
4151 units, there are now matching AssertXYZ= settings. While
4152 failing conditions cause a unit to be skipped, but its job
4153 to succeed, failing assertions declared like this will cause
4154 a unit start operation and its job to fail.
4156 * hostnamed now knows a new chassis type "embedded".
4158 * systemctl gained a new "edit" command. When used on a unit
4159 file, this allows extending unit files with .d/ drop-in
4160 configuration snippets or editing the full file (after
4161 copying it from /usr/lib to /etc). This will invoke the
4162 user's editor (as configured with $EDITOR), and reload the
4163 modified configuration after editing.
4165 * "systemctl status" now shows the suggested enablement state
4166 for a unit, as declared in the (usually vendor-supplied)
4167 system preset files.
4169 * nss-myhostname will now resolve the single-label host name
4170 "gateway" to the locally configured default IP routing
4171 gateways, ordered by their metrics. This assigns a stable
4172 name to the used gateways, regardless which ones are
4173 currently configured. Note that the name will only be
4174 resolved after all other name sources (if nss-myhostname is
4175 configured properly) and should hence not negatively impact
4176 systems that use the single-label host name "gateway" in
4179 * systemd-inhibit now allows filtering by mode when listing
4182 * Scope and service units gained a new "Delegate" boolean
4183 property, which, when set, allows processes running inside the
4184 unit to further partition resources. This is primarily
4185 useful for systemd user instances as well as container
4188 * journald will now pick up audit messages directly from
4189 the kernel, and log them like any other log message. The
4190 audit fields are split up and fully indexed. This means that
4191 journalctl in many ways is now a (nicer!) alternative to
4192 ausearch, the traditional audit client. Note that this
4193 implements only a minimal audit client. If you want the
4194 special audit modes like reboot-on-log-overflow, please use
4195 the traditional auditd instead, which can be used in
4196 parallel to journald.
4198 * The ConditionSecurity= unit file option now understands the
4199 special string "audit" to check whether auditing is
4202 * journalctl gained two new commands --vacuum-size= and
4203 --vacuum-time= to delete old journal files until the
4204 remaining ones take up no more than the specified size on disk,
4205 or are not older than the specified time.
4207 * A new, native PPPoE library has been added to sd-network,
4208 systemd's library of light-weight networking protocols. This
4209 library will be used in a future version of networkd to
4210 enable PPPoE communication without an external pppd daemon.
4212 * The busctl tool now understands a new "capture" verb that
4213 works similar to "monitor", but writes a packet capture
4214 trace to STDOUT that can be redirected to a file which is
4215 compatible with libcap's capture file format. This can then
4216 be loaded in Wireshark and similar tools to inspect bus
4219 * The busctl tool now understands a new "tree" verb that shows
4220 the object trees of a specific service on the bus, or of all
4223 * The busctl tool now understands a new "introspect" verb that
4224 shows all interfaces and members of objects on the bus,
4225 including their signature and values. This is particularly
4226 useful to get more information about bus objects shown by
4227 the new "busctl tree" command.
4229 * The busctl tool now understands new verbs "call",
4230 "set-property" and "get-property" for invoking bus method
4231 calls, setting and getting bus object properties in a
4234 * busctl gained a new --augment-creds= argument that controls
4235 whether the tool shall augment credential information it
4236 gets from the bus with data from /proc, in a possibly
4239 * nspawn's --link-journal= switch gained two new values
4240 "try-guest" and "try-host" that work like "guest" and
4241 "host", but do not fail if the host has no persistent
4242 journalling enabled. -j is now equivalent to
4243 --link-journal=try-guest.
4245 * macvlan network devices created by nspawn will now have
4246 stable MAC addresses.
4248 * A new SmackProcessLabel= unit setting has been added, which
4249 controls the SMACK security label processes forked off by
4250 the respective unit shall use.
4252 * If compiled with --enable-xkbcommon, systemd-localed will
4253 verify x11 keymap settings by compiling the given keymap. It
4254 will spew out warnings if the compilation fails. This
4255 requires libxkbcommon to be installed.
4257 * When a coredump is collected, a larger number of metadata
4258 fields is now collected and included in the journal records
4259 created for it. More specifically, control group membership,
4260 environment variables, memory maps, working directory,
4261 chroot directory, /proc/$PID/status, and a list of open file
4262 descriptors is now stored in the log entry.
4264 * The udev hwdb now contains DPI information for mice. For
4267 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
4269 * All systemd programs that read standalone configuration
4270 files in /etc now also support a corresponding series of
4271 .conf.d configuration directories in /etc/, /run/,
4272 /usr/local/lib/, /usr/lib/, and (if configured with
4273 --enable-split-usr) /lib/. In particular, the following
4274 configuration files now have corresponding configuration
4275 directories: system.conf user.conf, logind.conf,
4276 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
4277 resolved.conf, timesyncd.conf, journal-remote.conf, and
4278 journal-upload.conf. Note that distributions should use the
4279 configuration directories in /usr/lib/; the directories in
4280 /etc/ are reserved for the system administrator.
4282 * systemd-rfkill will no longer take the rfkill device name
4283 into account when storing rfkill state on disk, as the name
4284 might be dynamically assigned and not stable. Instead, the
4285 ID_PATH udev variable combined with the rfkill type (wlan,
4286 bluetooth, ...) is used.
4288 * A new service systemd-machine-id-commit.service has been
4289 added. When used on systems where /etc is read-only during
4290 boot, and /etc/machine-id is not initialized (but an empty
4291 file), this service will copy the temporary machine ID
4292 created as replacement into /etc after the system is fully
4293 booted up. This is useful for systems that are freshly
4294 installed with a non-initialized machine ID, but should get
4295 a fixed machine ID for subsequent boots.
4297 * networkd's .netdev files now provide a large set of
4298 configuration parameters for VXLAN devices. Similarly, the
4299 bridge port cost parameter is now configurable in .network
4300 files. There's also new support for configuring IP source
4301 routing. networkd .link files gained support for a new
4302 OriginalName= match that is useful to match against the
4303 original interface name the kernel assigned. .network files
4304 may include MTU= and MACAddress= fields for altering the MTU
4305 and MAC address while being connected to a specific network
4308 * The LUKS logic gained supported for configuring
4309 UUID-specific key files. There's also new support for naming
4310 LUKS device from the kernel command line, using the new
4311 luks.name= argument.
4313 * Timer units may now be transiently created via the bus API
4314 (this was previously already available for scope and service
4315 units). In addition it is now possible to create multiple
4316 transient units at the same time with a single bus call. The
4317 "systemd-run" tool has been updated to make use of this for
4318 running commands on a specified time, in at(1)-style.
4320 * tmpfiles gained support for "t" lines, for assigning
4321 extended attributes to files. Among other uses this may be
4322 used to assign SMACK labels to files.
4324 Contributions from: Alin Rauta, Alison Chaiken, Andrej
4325 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
4326 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
4327 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
4328 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
4329 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
4330 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
4331 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
4332 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
4333 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
4334 Michael Biebl, Michael Chapman, Michael Marineau, Michal
4335 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
4336 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
4337 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
4338 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
4339 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4340 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
4341 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
4343 — Berlin, 2014-12-10
4347 * journalctl gained the new options -t/--identifier= to match
4348 on the syslog identifier (aka "tag"), as well as --utc to
4349 show log timestamps in the UTC timezone. journalctl now also
4350 accepts -n/--lines=all to disable line capping in a pager.
4352 * journalctl gained a new switch, --flush, that synchronously
4353 flushes logs from /run/log/journal to /var/log/journal if
4354 persistent storage is enabled. systemd-journal-flush.service
4355 now waits until the operation is complete.
4357 * Services can notify the manager before they start a reload
4358 (by sending RELOADING=1) or shutdown (by sending
4359 STOPPING=1). This allows the manager to track and show the
4360 internal state of daemons and closes a race condition when
4361 the process is still running but has closed its D-Bus
4364 * Services with Type=oneshot do not have to have any ExecStart
4367 * User units are now loaded also from
4368 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
4369 /run/systemd/user directory that was already previously
4370 supported, but is under the control of the user.
4372 * Job timeouts (i.e. time-outs on the time a job that is
4373 queued stays in the run queue) can now optionally result in
4374 immediate reboot or power-off actions (JobTimeoutAction= and
4375 JobTimeoutRebootArgument=). This is useful on ".target"
4376 units, to limit the maximum time a target remains
4377 undispatched in the run queue, and to trigger an emergency
4378 operation in such a case. This is now used by default to
4379 turn off the system if boot-up (as defined by everything in
4380 basic.target) hangs and does not complete for at least
4381 15min. Also, if power-off or reboot hang for at least 30min
4382 an immediate power-off/reboot operation is triggered. This
4383 functionality is particularly useful to increase reliability
4384 on embedded devices, but also on laptops which might
4385 accidentally get powered on when carried in a backpack and
4386 whose boot stays stuck in a hard disk encryption passphrase
4389 * systemd-logind can be configured to also handle lid switch
4390 events even when the machine is docked or multiple displays
4391 are attached (HandleLidSwitchDocked= option).
4393 * A helper binary and a service have been added which can be
4394 used to resume from hibernation in the initramfs. A
4395 generator will parse the resume= option on the kernel
4396 command line to trigger resume.
4398 * A user console daemon systemd-consoled has been
4399 added. Currently, it is a preview, and will so far open a
4400 single terminal on each session of the user marked as
4401 Desktop=systemd-console.
4403 * Route metrics can be specified for DHCP routes added by
4406 * The SELinux context of socket-activated services can be set
4407 from the information provided by the networking stack
4408 (SELinuxContextFromNet= option).
4410 * Userspace firmware loading support has been removed and
4411 the minimum supported kernel version is thus bumped to 3.7.
4413 * Timeout for udev workers has been increased from 1 to 3
4414 minutes, but a warning will be printed after 1 minute to
4415 help diagnose kernel modules that take a long time to load.
4417 * Udev rules can now remove tags on devices with TAG-="foobar".
4419 * systemd's readahead implementation has been removed. In many
4420 circumstances it didn't give expected benefits even for
4421 rotational disk drives and was becoming less relevant in the
4422 age of SSDs. As none of the developers has been using
4423 rotating media anymore, and nobody stepped up to actively
4424 maintain this component of systemd it has now been removed.
4426 * Swap units can use Options= to specify discard options.
4427 Discard options specified for swaps in /etc/fstab are now
4430 * Docker containers are now detected as a separate type of
4433 * The Password Agent protocol gained support for queries where
4434 the user input is shown, useful e.g. for user names.
4435 systemd-ask-password gained a new --echo option to turn that
4438 * The default sysctl.d/ snippets will now set:
4440 net.core.default_qdisc = fq_codel
4442 This selects Fair Queuing Controlled Delay as the default
4443 queuing discipline for network interfaces. fq_codel helps
4444 fight the network bufferbloat problem. It is believed to be
4445 a good default with no tuning required for most workloads.
4446 Downstream distributions may override this choice. On 10Gbit
4447 servers that do not do forwarding, "fq" may perform better.
4448 Systems without a good clocksource should use "pfifo_fast".
4450 * If kdbus is enabled during build a new option BusPolicy= is
4451 available for service units, that allows locking all service
4452 processes into a stricter bus policy, in order to limit
4453 access to various bus services, or even hide most of them
4454 from the service's view entirely.
4456 * networkctl will now show the .network and .link file
4457 networkd has applied to a specific interface.
4459 * sd-login gained a new API call sd_session_get_desktop() to
4460 query which desktop environment has been selected for a
4463 * UNIX utmp support is now compile-time optional to support
4464 legacy-free systems.
4466 * systemctl gained two new commands "add-wants" and
4467 "add-requires" for pulling in units from specific targets
4470 * If the word "rescue" is specified on the kernel command line
4471 the system will now boot into rescue mode (aka
4472 rescue.target), which was previously available only by
4473 specifying "1" or "systemd.unit=rescue.target" on the kernel
4474 command line. This new kernel command line option nicely
4475 mirrors the already existing "emergency" kernel command line
4478 * New kernel command line options mount.usr=, mount.usrflags=,
4479 mount.usrfstype= have been added that match root=, rootflags=,
4480 rootfstype= but allow mounting a specific file system to
4483 * The $NOTIFY_SOCKET is now also passed to control processes of
4484 services, not only the main process.
4486 * This version reenables support for fsck's -l switch. This
4487 means at least version v2.25 of util-linux is required for
4488 operation, otherwise dead-locks on device nodes may
4489 occur. Again: you need to update util-linux to at least
4490 v2.25 when updating systemd to v217.
4492 * The "multi-seat-x" tool has been removed from systemd, as
4493 its functionality has been integrated into X servers 1.16,
4494 and the tool is hence redundant. It is recommended to update
4495 display managers invoking this tool to simply invoke X
4496 directly from now on, again.
4498 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
4499 message flag has been added for all of systemd's PolicyKit
4500 authenticated method calls has been added. In particular
4501 this now allows optional interactive authorization via
4502 PolicyKit for many of PID1's privileged operations such as
4503 unit file enabling and disabling.
4505 * "udevadm hwdb --update" learnt a new switch "--usr" for
4506 placing the rebuilt hardware database in /usr instead of
4507 /etc. When used only hardware database entries stored in
4508 /usr will be used, and any user database entries in /etc are
4509 ignored. This functionality is useful for vendors to ship a
4510 pre-built database on systems where local configuration is
4511 unnecessary or unlikely.
4513 * Calendar time specifications in .timer units now also
4514 understand the strings "semi-annually", "quarterly" and
4515 "minutely" as shortcuts (in addition to the preexisting
4516 "anually", "hourly", ...).
4518 * systemd-tmpfiles will now correctly create files in /dev
4519 at boot which are marked for creation only at boot. It is
4520 recommended to always create static device nodes with 'c!'
4521 and 'b!', so that they are created only at boot and not
4522 overwritten at runtime.
4524 * When the watchdog logic is used for a service (WatchdogSec=)
4525 and the watchdog timeout is hit the service will now be
4526 terminated with SIGABRT (instead of just SIGTERM), in order
4527 to make sure a proper coredump and backtrace is
4528 generated. This ensures that hanging services will result in
4529 similar coredump/backtrace behaviour as services that hit a
4532 Contributions from: Andreas Henriksson, Andrei Borzenkov,
4533 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
4534 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
4535 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
4536 Herrmann, David Sommerseth, David Strauss, Emil Renner
4537 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
4538 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
4539 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
4540 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
4541 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
4542 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
4543 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
4544 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
4545 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
4546 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
4547 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
4548 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
4549 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
4550 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
4551 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
4552 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
4555 — Berlin, 2014-10-28
4559 * timedated no longer reads NTP implementation unit names from
4560 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
4561 implementations should add a
4563 Conflicts=systemd-timesyncd.service
4565 to their unit files to take over and replace systemd's NTP
4566 default functionality.
4568 * systemd-sysusers gained a new line type "r" for configuring
4569 which UID/GID ranges to allocate system users/groups
4570 from. Lines of type "u" may now add an additional column
4571 that specifies the home directory for the system user to be
4572 created. Also, systemd-sysusers may now optionally read user
4573 information from STDIN instead of a file. This is useful for
4574 invoking it from RPM preinst scriptlets that need to create
4575 users before the first RPM file is installed since these
4576 files might need to be owned by them. A new
4577 %sysusers_create_inline RPM macro has been introduced to do
4578 just that. systemd-sysusers now updates the shadow files as
4579 well as the user/group databases, which should enhance
4580 compatibility with certain tools like grpck.
4582 * A number of bus APIs of PID 1 now optionally consult
4583 PolicyKit to permit access for otherwise unprivileged
4584 clients under certain conditions. Note that this currently
4585 doesn't support interactive authentication yet, but this is
4586 expected to be added eventually, too.
4588 * /etc/machine-info now has new fields for configuring the
4589 deployment environment of the machine, as well as the
4590 location of the machine. hostnamectl has been updated with
4591 new command to update these fields.
4593 * systemd-timesyncd has been updated to automatically acquire
4594 NTP server information from systemd-networkd, which might
4595 have been discovered via DHCP.
4597 * systemd-resolved now includes a caching DNS stub resolver
4598 and a complete LLMNR name resolution implementation. A new
4599 NSS module "nss-resolve" has been added which can be used
4600 instead of glibc's own "nss-dns" to resolve hostnames via
4601 systemd-resolved. Hostnames, addresses and arbitrary RRs may
4602 be resolved via systemd-resolved D-Bus APIs. In contrast to
4603 the glibc internal resolver systemd-resolved is aware of
4604 multi-homed system, and keeps DNS server and caches separate
4605 and per-interface. Queries are sent simultaneously on all
4606 interfaces that have DNS servers configured, in order to
4607 properly handle VPNs and local LANs which might resolve
4608 separate sets of domain names. systemd-resolved may acquire
4609 DNS server information from systemd-networkd automatically,
4610 which in turn might have discovered them via DHCP. A tool
4611 "systemd-resolve-host" has been added that may be used to
4612 query the DNS logic in resolved. systemd-resolved implements
4613 IDNA and automatically uses IDNA or UTF-8 encoding depending
4614 on whether classic DNS or LLMNR is used as transport. In the
4615 next releases we intend to add a DNSSEC and mDNS/DNS-SD
4616 implementation to systemd-resolved.
4618 * A new NSS module nss-mymachines has been added, that
4619 automatically resolves the names of all local registered
4620 containers to their respective IP addresses.
4622 * A new client tool "networkctl" for systemd-networkd has been
4623 added. It currently is entirely passive and will query
4624 networking configuration from udev, rtnetlink and networkd,
4625 and present it to the user in a very friendly
4626 way. Eventually, we hope to extend it to become a full
4627 control utility for networkd.
4629 * .socket units gained a new DeferAcceptSec= setting that
4630 controls the kernels' TCP_DEFER_ACCEPT sockopt for
4631 TCP. Similarly, support for controlling TCP keep-alive
4632 settings has been added (KeepAliveTimeSec=,
4633 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
4634 turning off Nagle's algorithm on TCP has been added
4637 * logind learned a new session type "web", for use in projects
4638 like Cockpit which register web clients as PAM sessions.
4640 * timer units with at least one OnCalendar= setting will now
4641 be started only after time-sync.target has been
4642 reached. This way they will not elapse before the system
4643 clock has been corrected by a local NTP client or
4644 similar. This is particular useful on RTC-less embedded
4645 machines, that come up with an invalid system clock.
4647 * systemd-nspawn's --network-veth= switch should now result in
4648 stable MAC addresses for both the outer and the inner side
4651 * systemd-nspawn gained a new --volatile= switch for running
4652 container instances with /etc or /var unpopulated.
4654 * The kdbus client code has been updated to use the new Linux
4655 3.17 memfd subsystem instead of the old kdbus-specific one.
4657 * systemd-networkd's DHCP client and server now support
4658 FORCERENEW. There are also new configuration options to
4659 configure the vendor client identifier and broadcast mode
4662 * systemd will no longer inform the kernel about the current
4663 timezone, as this is necessarily incorrect and racy as the
4664 kernel has no understanding of DST and similar
4665 concepts. This hence means FAT timestamps will be always
4666 considered UTC, similar to what Android is already
4667 doing. Also, when the RTC is configured to the local time
4668 (rather than UTC) systemd will never synchronize back to it,
4669 as this might confuse Windows at a later boot.
4671 * systemd-analyze gained a new command "verify" for offline
4672 validation of unit files.
4674 * systemd-networkd gained support for a couple of additional
4675 settings for bonding networking setups. Also, the metric for
4676 statically configured routes may now be configured. For
4677 network interfaces where this is appropriate the peer IP
4678 address may now be configured.
4680 * systemd-networkd's DHCP client will no longer request
4681 broadcasting by default, as this tripped up some networks.
4682 For hardware where broadcast is required the feature should
4683 be switched back on using RequestBroadcast=yes.
4685 * systemd-networkd will now set up IPv4LL addresses (when
4686 enabled) even if DHCP is configured successfully.
4688 * udev will now default to respect network device names given
4689 by the kernel when the kernel indicates that these are
4690 predictable. This behavior can be tweaked by changing
4691 NamePolicy= in the relevant .link file.
4693 * A new library systemd-terminal has been added that
4694 implements full TTY stream parsing and rendering. This
4695 library is supposed to be used later on for implementing a
4696 full userspace VT subsystem, replacing the current kernel
4699 * A new tool systemd-journal-upload has been added to push
4700 journal data to a remote system running
4701 systemd-journal-remote.
4703 * journald will no longer forward all local data to another
4704 running syslog daemon. This change has been made because
4705 rsyslog (which appears to be the most commonly used syslog
4706 implementation these days) no longer makes use of this, and
4707 instead pulls the data out of the journal on its own. Since
4708 forwarding the messages to a non-existent syslog server is
4709 more expensive than we assumed we have now turned this
4710 off. If you run a syslog server that is not a recent rsyslog
4711 version, you have to turn this option on again
4712 (ForwardToSyslog= in journald.conf).
4714 * journald now optionally supports the LZ4 compressor for
4715 larger journal fields. This compressor should perform much
4716 better than XZ which was the previous default.
4718 * machinectl now shows the IP addresses of local containers,
4719 if it knows them, plus the interface name of the container.
4721 * A new tool "systemd-escape" has been added that makes it
4722 easy to escape strings to build unit names and similar.
4724 * sd_notify() messages may now include a new ERRNO= field
4725 which is parsed and collected by systemd and shown among the
4726 "systemctl status" output for a service.
4728 * A new component "systemd-firstboot" has been added that
4729 queries the most basic systemd information (timezone,
4730 hostname, root password) interactively on first
4731 boot. Alternatively it may also be used to provision these
4732 things offline on OS images installed into directories.
4734 * The default sysctl.d/ snippets will now set
4736 net.ipv4.conf.default.promote_secondaries=1
4738 This has the benefit of no flushing secondary IP addresses
4739 when primary addresses are removed.
4741 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
4742 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
4743 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
4744 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
4745 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
4746 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
4747 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
4748 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
4749 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
4750 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
4751 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
4752 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
4753 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
4754 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
4755 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
4757 — Berlin, 2014-08-19
4761 * A new tool systemd-sysusers has been added. This tool
4762 creates system users and groups in /etc/passwd and
4763 /etc/group, based on static declarative system user/group
4764 definitions in /usr/lib/sysusers.d/. This is useful to
4765 enable factory resets and volatile systems that boot up with
4766 an empty /etc directory, and thus need system users and
4767 groups created during early boot. systemd now also ships
4768 with two default sysusers.d/ files for the most basic
4769 users and groups systemd and the core operating system
4772 * A new tmpfiles snippet has been added that rebuilds the
4773 essential files in /etc on boot, should they be missing.
4775 * A directive for ensuring automatic clean-up of
4776 /var/cache/man/ has been removed from the default
4777 configuration. This line should now be shipped by the man
4778 implementation. The necessary change has been made to the
4779 man-db implementation. Note that you need to update your man
4780 implementation to one that ships this line, otherwise no
4781 automatic clean-up of /var/cache/man will take place.
4783 * A new condition ConditionNeedsUpdate= has been added that
4784 may conditionalize services to only run when /etc or /var
4785 are "older" than the vendor operating system resources in
4786 /usr. This is useful for reconstructing or updating /etc
4787 after an offline update of /usr or a factory reset, on the
4788 next reboot. Services that want to run once after such an
4789 update or reset should use this condition and order
4790 themselves before the new systemd-update-done.service, which
4791 will mark the two directories as fully updated. A number of
4792 service files have been added making use of this, to rebuild
4793 the udev hardware database, the journald message catalog and
4794 dynamic loader cache (ldconfig). The systemd-sysusers tool
4795 described above also makes use of this now. With this in
4796 place it is now possible to start up a minimal operating
4797 system with /etc empty cleanly. For more information on the
4798 concepts involved see this recent blog story:
4800 http://0pointer.de/blog/projects/stateless.html
4802 * A new system group "input" has been introduced, and all
4803 input device nodes get this group assigned. This is useful
4804 for system-level software to get access to input devices. It
4805 complements what is already done for "audio" and "video".
4807 * systemd-networkd learnt minimal DHCPv4 server support in
4808 addition to the existing DHCPv4 client support. It also
4809 learnt DHCPv6 client and IPv6 Router Solicitation client
4810 support. The DHCPv4 client gained support for static routes
4811 passed in from the server. Note that the [DHCPv4] section
4812 known in older systemd-networkd versions has been renamed to
4813 [DHCP] and is now also used by the DHCPv6 client. Existing
4814 .network files using settings of this section should be
4815 updated, though compatibility is maintained. Optionally, the
4816 client hostname may now be sent to the DHCP server.
4818 * networkd gained support for vxlan virtual networks as well
4819 as tun/tap and dummy devices.
4821 * networkd gained support for automatic allocation of address
4822 ranges for interfaces from a system-wide pool of
4823 addresses. This is useful for dynamically managing a large
4824 number of interfaces with a single network configuration
4825 file. In particular this is useful to easily assign
4826 appropriate IP addresses to the veth links of a large number
4827 of nspawn instances.
4829 * RPM macros for processing sysusers, sysctl and binfmt
4830 drop-in snippets at package installation time have been
4833 * The /etc/os-release file should now be placed in
4834 /usr/lib/os-release. The old location is automatically
4835 created as symlink. /usr/lib is the more appropriate
4836 location of this file, since it shall actually describe the
4837 vendor operating system shipped in /usr, and not the
4838 configuration stored in /etc.
4840 * .mount units gained a new boolean SloppyOptions= setting
4841 that maps to mount(8)'s -s option which enables permissive
4842 parsing of unknown mount options.
4844 * tmpfiles learnt a new "L+" directive which creates a symlink
4845 but (unlike "L") deletes a pre-existing file first, should
4846 it already exist and not already be the correct
4847 symlink. Similarly, "b+", "c+" and "p+" directives have been
4848 added as well, which create block and character devices, as
4849 well as fifos in the filesystem, possibly removing any
4850 pre-existing files of different types.
4852 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
4853 'argument' field (which so far specified the source to
4854 symlink/copy the files from) is now optional. If omitted the
4855 same file os copied from /usr/share/factory/ suffixed by the
4856 full destination path. This is useful for populating /etc
4857 with essential files, by copying them from vendor defaults
4858 shipped in /usr/share/factory/etc.
4860 * A new command "systemctl preset-all" has been added that
4861 applies the service preset settings to all installed unit
4862 files. A new switch --preset-mode= has been added that
4863 controls whether only enable or only disable operations
4866 * A new command "systemctl is-system-running" has been added
4867 that allows checking the overall state of the system, for
4868 example whether it is fully up and running.
4870 * When the system boots up with an empty /etc, the equivalent
4871 to "systemctl preset-all" is executed during early boot, to
4872 make sure all default services are enabled after a factory
4875 * systemd now contains a minimal preset file that enables the
4876 most basic services systemd ships by default.
4878 * Unit files' [Install] section gained a new DefaultInstance=
4879 field for defining the default instance to create if a
4880 template unit is enabled with no instance specified.
4882 * A new passive target cryptsetup-pre.target has been added
4883 that may be used by services that need to make they run and
4884 finish before the first LUKS cryptographic device is set up.
4886 * The /dev/loop-control and /dev/btrfs-control device nodes
4887 are now owned by the "disk" group by default, opening up
4888 access to this group.
4890 * systemd-coredump will now automatically generate a
4891 stack trace of all core dumps taking place on the system,
4892 based on elfutils' libdw library. This stack trace is logged
4895 * systemd-coredump may now optionally store coredumps directly
4896 on disk (in /var/lib/systemd/coredump, possibly compressed),
4897 instead of storing them unconditionally in the journal. This
4898 mode is the new default. A new configuration file
4899 /etc/systemd/coredump.conf has been added to configure this
4900 and other parameters of systemd-coredump.
4902 * coredumpctl gained a new "info" verb to show details about a
4903 specific coredump. A new switch "-1" has also been added
4904 that makes sure to only show information about the most
4905 recent entry instead of all entries. Also, as the tool is
4906 generally useful now the "systemd-" prefix of the binary
4907 name has been removed. Distributions that want to maintain
4908 compatibility with the old name should add a symlink from
4909 the old name to the new name.
4911 * journald's SplitMode= now defaults to "uid". This makes sure
4912 that unprivileged users can access their own coredumps with
4913 coredumpctl without restrictions.
4915 * New kernel command line options "systemd.wants=" (for
4916 pulling an additional unit during boot), "systemd.mask="
4917 (for masking a specific unit for the boot), and
4918 "systemd.debug-shell" (for enabling the debug shell on tty9)
4919 have been added. This is implemented in the new generator
4920 "systemd-debug-generator".
4922 * systemd-nspawn will now by default filter a couple of
4923 syscalls for containers, among them those required for
4924 kernel module loading, direct x86 IO port access, swap
4925 management, and kexec. Most importantly though
4926 open_by_handle_at() is now prohibited for containers,
4927 closing a hole similar to a recently discussed vulnerability
4928 in docker regarding access to files on file hierarchies the
4929 container should normally not have access to. Note that, for
4930 nspawn, we generally make no security claims anyway (and
4931 this is explicitly documented in the man page), so this is
4932 just a fix for one of the most obvious problems.
4934 * A new man page file-hierarchy(7) has been added that
4935 contains a minimized, modernized version of the file system
4936 layout systemd expects, similar in style to the FHS
4937 specification or hier(5). A new tool systemd-path(1) has
4938 been added to query many of these paths for the local
4941 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
4942 longer done. Since the directory now has a per-user size
4943 limit, and is cleaned on logout this appears unnecessary,
4944 in particular since this now brings the lifecycle of this
4945 directory closer in line with how IPC objects are handled.
4947 * systemd.pc now exports a number of additional directories,
4948 including $libdir (which is useful to identify the library
4949 path for the primary architecture of the system), and a
4950 couple of drop-in directories.
4952 * udev's predictable network interface names now use the dev_port
4953 sysfs attribute, introduced in linux 3.15 instead of dev_id to
4954 distinguish between ports of the same PCI function. dev_id should
4955 only be used for ports using the same HW address, hence the need
4958 * machined has been updated to export the OS version of a
4959 container (read from /etc/os-release and
4960 /usr/lib/os-release) on the bus. This is now shown in
4961 "machinectl status" for a machine.
4963 * A new service setting RestartForceExitStatus= has been
4964 added. If configured to a set of exit signals or process
4965 return values, the service will be restarted when the main
4966 daemon process exits with any of them, regardless of the
4969 * systemctl's -H switch for connecting to remote systemd
4970 machines has been extended so that it may be used to
4971 directly connect to a specific container on the
4972 host. "systemctl -H root@foobar:waldi" will now connect as
4973 user "root" to host "foobar", and then proceed directly to
4974 the container named "waldi". Note that currently you have to
4975 authenticate as user "root" for this to work, as entering
4976 containers is a privileged operation.
4978 Contributions from: Andreas Henriksson, Benjamin Steinwender,
4979 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
4980 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
4981 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
4982 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
4983 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
4984 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
4985 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
4986 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
4987 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
4988 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
4989 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
4991 — Berlin, 2014-07-03
4995 * As an experimental feature, udev now tries to lock the
4996 disk device node (flock(LOCK_SH|LOCK_NB)) while it
4997 executes events for the disk or any of its partitions.
4998 Applications like partitioning programs can lock the
4999 disk device node (flock(LOCK_EX)) and claim temporary
5000 device ownership that way; udev will entirely skip all event
5001 handling for this disk and its partitions. If the disk
5002 was opened for writing, the close will trigger a partition
5003 table rescan in udev's "watch" facility, and if needed
5004 synthesize "change" events for the disk and all its partitions.
5005 This is now unconditionally enabled, and if it turns out to
5006 cause major problems, we might turn it on only for specific
5007 devices, or might need to disable it entirely. Device Mapper
5008 devices are excluded from this logic.
5010 * We temporarily dropped the "-l" switch for fsck invocations,
5011 since they collide with the flock() logic above. util-linux
5012 upstream has been changed already to avoid this conflict,
5013 and we will readd "-l" as soon as util-linux with this
5014 change has been released.
5016 * The dependency on libattr has been removed. Since a long
5017 time, the extended attribute calls have moved to glibc, and
5018 libattr is thus unnecessary.
5020 * Virtualization detection works without privileges now. This
5021 means the systemd-detect-virt binary no longer requires
5022 CAP_SYS_PTRACE file capabilities, and our daemons can run
5023 with fewer privileges.
5025 * systemd-networkd now runs under its own "systemd-network"
5026 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
5027 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
5028 loses the ability to write to files owned by root this way.
5030 * Similarly, systemd-resolved now runs under its own
5031 "systemd-resolve" user with no capabilities remaining.
5033 * Similarly, systemd-bus-proxyd now runs under its own
5034 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
5036 * systemd-networkd gained support for setting up "veth"
5037 virtual Ethernet devices for container connectivity, as well
5038 as GRE and VTI tunnels.
5040 * systemd-networkd will no longer automatically attempt to
5041 manually load kernel modules necessary for certain tunnel
5042 transports. Instead, it is assumed the kernel loads them
5043 automatically when required. This only works correctly on
5044 very new kernels. On older kernels, please consider adding
5045 the kernel modules to /etc/modules-load.d/ as a work-around.
5047 * The resolv.conf file systemd-resolved generates has been
5048 moved to /run/systemd/resolve/. If you have a symlink from
5049 /etc/resolv.conf, it might be necessary to correct it.
5051 * Two new service settings, ProtectHome= and ProtectSystem=,
5052 have been added. When enabled, they will make the user data
5053 (such as /home) inaccessible or read-only and the system
5054 (such as /usr) read-only, for specific services. This allows
5055 very light-weight per-service sandboxing to avoid
5056 modifications of user data or system files from
5057 services. These two new switches have been enabled for all
5058 of systemd's long-running services, where appropriate.
5060 * Socket units gained new SocketUser= and SocketGroup=
5061 settings to set the owner user and group of AF_UNIX sockets
5062 and FIFOs in the file system.
5064 * Socket units gained a new RemoveOnStop= setting. If enabled,
5065 all FIFOS and sockets in the file system will be removed
5066 when the specific socket unit is stopped.
5068 * Socket units gained a new Symlinks= setting. It takes a list
5069 of symlinks to create to file system sockets or FIFOs
5070 created by the specific Unix sockets. This is useful to
5071 manage symlinks to socket nodes with the same life-cycle as
5074 * The /dev/log socket and /dev/initctl FIFO have been moved to
5075 /run, and have been replaced by symlinks. This allows
5076 connecting to these facilities even if PrivateDevices=yes is
5077 used for a service (which makes /dev/log itself unavailable,
5078 but /run is left). This also has the benefit of ensuring
5079 that /dev only contains device nodes, directories and
5080 symlinks, and nothing else.
5082 * sd-daemon gained two new calls sd_pid_notify() and
5083 sd_pid_notifyf(). They are similar to sd_notify() and
5084 sd_notifyf(), but allow overriding of the source PID of
5085 notification messages if permissions permit this. This is
5086 useful to send notify messages on behalf of a different
5087 process (for example, the parent process). The
5088 systemd-notify tool has been updated to make use of this
5089 when sending messages (so that notification messages now
5090 originate from the shell script invoking systemd-notify and
5091 not the systemd-notify process itself. This should minimize
5092 a race where systemd fails to associate notification
5093 messages to services when the originating process already
5096 * A new "on-abnormal" setting for Restart= has been added. If
5097 set, it will result in automatic restarts on all "abnormal"
5098 reasons for a process to exit, which includes unclean
5099 signals, core dumps, timeouts and watchdog timeouts, but
5100 does not include clean and unclean exit codes or clean
5101 signals. Restart=on-abnormal is an alternative for
5102 Restart=on-failure for services that shall be able to
5103 terminate and avoid restarts on certain errors, by
5104 indicating so with an unclean exit code. Restart=on-failure
5105 or Restart=on-abnormal is now the recommended setting for
5106 all long-running services.
5108 * If the InaccessibleDirectories= service setting points to a
5109 mount point (or if there are any submounts contained within
5110 it), it is now attempted to completely unmount it, to make
5111 the file systems truly unavailable for the respective
5114 * The ReadOnlyDirectories= service setting and
5115 systemd-nspawn's --read-only parameter are now recursively
5116 applied to all submounts, too.
5118 * Mount units may now be created transiently via the bus APIs.
5120 * The support for SysV and LSB init scripts has been removed
5121 from the systemd daemon itself. Instead, it is now
5122 implemented as a generator that creates native systemd units
5123 from these scripts when needed. This enables us to remove a
5124 substantial amount of legacy code from PID 1, following the
5125 fact that many distributions only ship a very small number
5126 of LSB/SysV init scripts nowadays.
5128 * Privileged Xen (dom0) domains are not considered
5129 virtualization anymore by the virtualization detection
5130 logic. After all, they generally have unrestricted access to
5131 the hardware and usually are used to manage the unprivileged
5134 * systemd-tmpfiles gained a new "C" line type, for copying
5135 files or entire directories.
5137 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
5138 lines. So far, they have been non-globbing versions of the
5139 latter, and have thus been redundant. In future, it is
5140 recommended to only use "z". "m" has hence been removed
5141 from the documentation, even though it stays supported.
5143 * A tmpfiles snippet to recreate the most basic structure in
5144 /var has been added. This is enough to create the /var/run →
5145 /run symlink and create a couple of structural
5146 directories. This allows systems to boot up with an empty or
5147 volatile /var. Of course, while with this change, the core OS
5148 now is capable with dealing with a volatile /var, not all
5149 user services are ready for it. However, we hope that sooner
5150 or later, many service daemons will be changed upstream so
5151 that they are able to automatically create their necessary
5152 directories in /var at boot, should they be missing. This is
5153 the first step to allow state-less systems that only require
5154 the vendor image for /usr to boot.
5156 * systemd-nspawn has gained a new --tmpfs= switch to mount an
5157 empty tmpfs instance to a specific directory. This is
5158 particularly useful for making use of the automatic
5159 reconstruction of /var (see above), by passing --tmpfs=/var.
5161 * Access modes specified in tmpfiles snippets may now be
5162 prefixed with "~", which indicates that they shall be masked
5163 by whether the existing file or directory is currently
5164 writable, readable or executable at all. Also, if specified,
5165 the sgid/suid/sticky bits will be masked for all
5168 * A new passive target unit "network-pre.target" has been
5169 added which is useful for services that shall run before any
5170 network is configured, for example firewall scripts.
5172 * The "floppy" group that previously owned the /dev/fd*
5173 devices is no longer used. The "disk" group is now used
5174 instead. Distributions should probably deprecate usage of
5177 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
5178 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
5179 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
5180 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
5181 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
5182 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
5185 — Berlin, 2014-06-11
5189 * A new "systemd-timesyncd" daemon has been added for
5190 synchronizing the system clock across the network. It
5191 implements an SNTP client. In contrast to NTP
5192 implementations such as chrony or the NTP reference server,
5193 this only implements a client side, and does not bother with
5194 the full NTP complexity, focusing only on querying time from
5195 one remote server and synchronizing the local clock to
5196 it. Unless you intend to serve NTP to networked clients or
5197 want to connect to local hardware clocks, this simple NTP
5198 client should be more than appropriate for most
5199 installations. The daemon runs with minimal privileges, and
5200 has been hooked up with networkd to only operate when
5201 network connectivity is available. The daemon saves the
5202 current clock to disk every time a new NTP sync has been
5203 acquired, and uses this to possibly correct the system clock
5204 early at bootup, in order to accommodate for systems that
5205 lack an RTC such as the Raspberry Pi and embedded devices,
5206 and to make sure that time monotonically progresses on these
5207 systems, even if it is not always correct. To make use of
5208 this daemon, a new system user and group "systemd-timesync"
5209 needs to be created on installation of systemd.
5211 * The queue "seqnum" interface of libudev has been disabled, as
5212 it was generally incompatible with device namespacing as
5213 sequence numbers of devices go "missing" if the devices are
5214 part of a different namespace.
5216 * "systemctl list-timers" and "systemctl list-sockets" gained
5217 a --recursive switch for showing units of these types also
5218 for all local containers, similar in style to the already
5219 supported --recursive switch for "systemctl list-units".
5221 * A new RebootArgument= setting has been added for service
5222 units, which may be used to specify a kernel reboot argument
5223 to use when triggering reboots with StartLimitAction=.
5225 * A new FailureAction= setting has been added for service
5226 units which may be used to specify an operation to trigger
5227 when a service fails. This works similarly to
5228 StartLimitAction=, but unlike it, controls what is done
5229 immediately rather than only after several attempts to
5230 restart the service in question.
5232 * hostnamed got updated to also expose the kernel name,
5233 release, and version on the bus. This is useful for
5234 executing commands like hostnamectl with the -H switch.
5235 systemd-analyze makes use of this to properly display
5236 details when running non-locally.
5238 * The bootchart tool can now show cgroup information in the
5239 graphs it generates.
5241 * The CFS CPU quota cgroup attribute is now exposed for
5242 services. The new CPUQuota= switch has been added for this
5243 which takes a percentage value. Setting this will have the
5244 result that a service may never get more CPU time than the
5245 specified percentage, even if the machine is otherwise idle.
5247 * systemd-networkd learned IPIP and SIT tunnel support.
5249 * LSB init scripts exposing a dependency on $network will now
5250 get a dependency on network-online.target rather than simply
5251 network.target. This should bring LSB handling closer to
5252 what it was on SysV systems.
5254 * A new fsck.repair= kernel option has been added to control
5255 how fsck shall deal with unclean file systems at boot.
5257 * The (.ini) configuration file parser will now silently
5258 ignore sections whose name begins with "X-". This may be
5259 used to maintain application-specific extension sections in unit
5262 * machined gained a new API to query the IP addresses of
5263 registered containers. "machinectl status" has been updated
5264 to show these addresses in its output.
5266 * A new call sd_uid_get_display() has been added to the
5267 sd-login APIs for querying the "primary" session of a
5268 user. The "primary" session of the user is elected from the
5269 user's sessions and generally a graphical session is
5270 preferred over a text one.
5272 * A minimal systemd-resolved daemon has been added. It
5273 currently simply acts as a companion to systemd-networkd and
5274 manages resolv.conf based on per-interface DNS
5275 configuration, possibly supplied via DHCP. In the long run
5276 we hope to extend this into a local DNSSEC enabled DNS and
5279 * The systemd-networkd-wait-online tool is now enabled by
5280 default. It will delay network-online.target until a network
5281 connection has been configured. The tool primarily integrates
5282 with networkd, but will also make a best effort to make sense
5283 of network configuration performed in some other way.
5285 * Two new service options StartupCPUShares= and
5286 StartupBlockIOWeight= have been added that work similarly to
5287 CPUShares= and BlockIOWeight= however only apply during
5288 system startup. This is useful to prioritize certain services
5289 differently during bootup than during normal runtime.
5291 * hostnamed has been changed to prefer the statically
5292 configured hostname in /etc/hostname (unless set to
5293 'localhost' or empty) over any dynamic one supplied by
5294 dhcp. With this change, the rules for picking the hostname
5295 match more closely the rules of other configuration settings
5296 where the local administrator's configuration in /etc always
5297 overrides any other settings.
5299 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
5300 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
5301 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
5302 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
5303 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
5304 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
5305 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
5306 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
5307 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
5308 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
5309 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
5310 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
5311 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
5312 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
5313 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
5314 Lindskog, WaLyong Cho, Will Woods, Zbigniew
5317 — Beijing, 2014-05-28
5321 * When restoring the screen brightness at boot, stay away from
5322 the darkest setting or from the lowest 5% of the available
5323 range, depending on which is the larger value of both. This
5324 should effectively protect the user from rebooting into a
5325 black screen, should the brightness have been set to minimum
5328 * sd-login gained a new sd_machine_get_class() call to
5329 determine the class ("vm" or "container") of a machine
5330 registered with machined.
5332 * sd-login gained new calls
5333 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
5334 to query the identity of the peer of a local AF_UNIX
5335 connection. They operate similarly to their sd_pid_get_xyz()
5338 * PID 1 will now maintain a system-wide system state engine
5339 with the states "starting", "running", "degraded",
5340 "maintenance", "stopping". These states are bound to system
5341 startup, normal runtime, runtime with at least one failed
5342 service, rescue/emergency mode and system shutdown. This
5343 state is shown in the "systemctl status" output when no unit
5344 name is passed. It is useful to determine system state, in
5345 particularly when doing so for many systems or containers at
5348 * A new command "list-machines" has been added to "systemctl"
5349 that lists all local OS containers and shows their system
5350 state (see above), if systemd runs inside of them.
5352 * systemctl gained a new "-r" switch to recursively enumerate
5353 units on all local containers, when used with the
5354 "list-unit" command (which is the default one that is
5355 executed when no parameters are specified).
5357 * The GPT automatic partition discovery logic will now honour
5358 two GPT partition flags: one may be set on a partition to
5359 cause it to be mounted read-only, and the other may be set
5360 on a partition to ignore it during automatic discovery.
5362 * Two new GPT type UUIDs have been added for automatic root
5363 partition discovery, for 32-bit and 64-bit ARM. This is not
5364 particularly useful for discovering the root directory on
5365 these architectures during bare-metal boots (since UEFI is
5366 not common there), but still very useful to allow booting of
5367 ARM disk images in nspawn with the -i option.
5369 * MAC addresses of interfaces created with nspawn's
5370 --network-interface= switch will now be generated from the
5371 machine name, and thus be stable between multiple invocations
5374 * logind will now automatically remove all IPC objects owned
5375 by a user if she or he fully logs out. This makes sure that
5376 users who are logged out cannot continue to consume IPC
5377 resources. This covers SysV memory, semaphores and message
5378 queues as well as POSIX shared memory and message
5379 queues. Traditionally, SysV and POSIX IPC had no life-cycle
5380 limits. With this functionality, that is corrected. This may
5381 be turned off by using the RemoveIPC= switch of logind.conf.
5383 * The systemd-machine-id-setup and tmpfiles tools gained a
5384 --root= switch to operate on a specific root directory,
5387 * journald can now forward logged messages to the TTYs of all
5388 logged in users ("wall"). This is the default for all
5389 emergency messages now.
5391 * A new tool systemd-journal-remote has been added to stream
5392 journal log messages across the network.
5394 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
5395 controller trees are mounted into it. Note that the
5396 directories mounted beneath it are not read-only. This is a
5397 security measure and is particularly useful because glibc
5398 actually includes a search logic to pick any tmpfs it can
5399 find to implement shm_open() if /dev/shm is not available
5400 (which it might very well be in namespaced setups).
5402 * machinectl gained a new "poweroff" command to cleanly power
5403 down a local OS container.
5405 * The PrivateDevices= unit file setting will now also drop the
5406 CAP_MKNOD capability from the capability bound set, and
5407 imply DevicePolicy=closed.
5409 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
5410 comprehensively on all long-running systemd services where
5411 this is appropriate.
5413 * systemd-udevd will now run in a disassociated mount
5414 namespace. To mount directories from udev rules, make sure to
5415 pull in mount units via SYSTEMD_WANTS properties.
5417 * The kdbus support gained support for uploading policy into
5418 the kernel. sd-bus gained support for creating "monitoring"
5419 connections that can eavesdrop into all bus communication
5420 for debugging purposes.
5422 * Timestamps may now be specified in seconds since the UNIX
5423 epoch Jan 1st, 1970 by specifying "@" followed by the value
5426 * Native tcpwrap support in systemd has been removed. tcpwrap
5427 is old code, not really maintained anymore and has serious
5428 shortcomings, and better options such as firewalls
5429 exist. For setups that require tcpwrap usage, please
5430 consider invoking your socket-activated service via tcpd,
5431 like on traditional inetd.
5433 * A new system.conf configuration option
5434 DefaultTimerAccuracySec= has been added that controls the
5435 default AccuracySec= setting of .timer units.
5437 * Timer units gained a new WakeSystem= switch. If enabled,
5438 timers configured this way will cause the system to resume
5439 from system suspend (if the system supports that, which most
5442 * Timer units gained a new Persistent= switch. If enabled,
5443 timers configured this way will save to disk when they have
5444 been last triggered. This information is then used on next
5445 reboot to possible execute overdue timer events, that
5446 could not take place because the system was powered off.
5447 This enables simple anacron-like behaviour for timer units.
5449 * systemctl's "list-timers" will now also list the time a
5450 timer unit was last triggered in addition to the next time
5451 it will be triggered.
5453 * systemd-networkd will now assign predictable IPv4LL
5454 addresses to its local interfaces.
5456 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
5457 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
5458 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
5459 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
5460 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
5461 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
5462 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
5463 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
5466 — Berlin, 2014-03-25
5470 * A new unit file setting RestrictAddressFamilies= has been
5471 added to restrict which socket address families unit
5472 processes gain access to. This takes address family names
5473 like "AF_INET" or "AF_UNIX", and is useful to minimize the
5474 attack surface of services via exotic protocol stacks. This
5475 is built on seccomp system call filters.
5477 * Two new unit file settings RuntimeDirectory= and
5478 RuntimeDirectoryMode= have been added that may be used to
5479 manage a per-daemon runtime directories below /run. This is
5480 an alternative for setting up directory permissions with
5481 tmpfiles snippets, and has the advantage that the runtime
5482 directory's lifetime is bound to the daemon runtime and that
5483 the daemon starts up with an empty directory each time. This
5484 is particularly useful when writing services that drop
5485 privileges using the User= or Group= setting.
5487 * The DeviceAllow= unit setting now supports globbing for
5488 matching against device group names.
5490 * The systemd configuration file system.conf gained new
5491 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
5492 DefaultMemoryAccounting= to globally turn on/off accounting
5493 for specific resources (cgroups) for all units. These
5494 settings may still be overridden individually in each unit
5497 * systemd-gpt-auto-generator is now able to discover /srv and
5498 root partitions in addition to /home and swap partitions. It
5499 also supports LUKS-encrypted partitions now. With this in
5500 place, automatic discovery of partitions to mount following
5501 the Discoverable Partitions Specification
5502 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
5503 is now a lot more complete. This allows booting without
5504 /etc/fstab and without root= on the kernel command line on
5505 systems prepared appropriately.
5507 * systemd-nspawn gained a new --image= switch which allows
5508 booting up disk images and Linux installations on any block
5509 device that follow the Discoverable Partitions Specification
5510 (see above). This means that installations made with
5511 appropriately updated installers may now be started and
5512 deployed using container managers, completely
5513 unmodified. (We hope that libvirt-lxc will add support for
5514 this feature soon, too.)
5516 * systemd-nspawn gained a new --network-macvlan= setting to
5517 set up a private macvlan interface for the
5518 container. Similarly, systemd-networkd gained a new
5519 Kind=macvlan setting in .netdev files.
5521 * systemd-networkd now supports configuring local addresses
5524 * A new tool systemd-network-wait-online has been added to
5525 synchronously wait for network connectivity using
5528 * The sd-bus.h bus API gained a new sd_bus_track object for
5529 tracking the life-cycle of bus peers. Note that sd-bus.h is
5530 still not a public API though (unless you specify
5531 --enable-kdbus on the configure command line, which however
5532 voids your warranty and you get no API stability guarantee).
5534 * The $XDG_RUNTIME_DIR runtime directories for each user are
5535 now individual tmpfs instances, which has the benefit of
5536 introducing separate pools for each user, with individual
5537 size limits, and thus making sure that unprivileged clients
5538 can no longer negatively impact the system or other users by
5539 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
5540 RuntimeDirectorySize= has been introduced that allows
5541 controlling the default size limit for all users. It
5542 defaults to 10% of the available physical memory. This is no
5543 replacement for quotas on tmpfs though (which the kernel
5544 still does not support), as /dev/shm and /tmp are still
5545 shared resources used by both the system and unprivileged
5548 * logind will now automatically turn off automatic suspending
5549 on laptop lid close when more than one display is
5550 connected. This was previously expected to be implemented
5551 individually in desktop environments (such as GNOME),
5552 however has been added to logind now, in order to fix a
5553 boot-time race where a desktop environment might not have
5554 been started yet and thus not been able to take an inhibitor
5555 lock at the time where logind already suspends the system
5556 due to a closed lid.
5558 * logind will now wait at least 30s after each system
5559 suspend/resume cycle, and 3min after system boot before
5560 suspending the system due to a closed laptop lid. This
5561 should give USB docking stations and similar enough time to
5562 be probed and configured after system resume and boot in
5563 order to then act as suspend blocker.
5565 * systemd-run gained a new --property= setting which allows
5566 initialization of resource control properties (and others)
5567 for the created scope or service unit. Example: "systemd-run
5568 --property=BlockIOWeight=10 updatedb" may be used to run
5569 updatedb at a low block IO scheduling weight.
5571 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
5572 now also work in --scope mode.
5574 * When systemd is compiled with kdbus support, basic support
5575 for enforced policies is now in place. (Note that enabling
5576 kdbus still voids your warranty and no API compatibility
5579 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
5580 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5581 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
5582 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
5583 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
5584 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
5585 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
5586 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
5587 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
5588 Zbigniew Jędrzejewski-Szmek
5590 — Berlin, 2014-03-12
5594 * systemd will now relabel /dev after loading the SMACK policy
5595 according to SMACK rules.
5597 * A new unit file option AppArmorProfile= has been added to
5598 set the AppArmor profile for the processes of a unit.
5600 * A new condition check ConditionArchitecture= has been added
5601 to conditionalize units based on the system architecture, as
5602 reported by uname()'s "machine" field.
5604 * systemd-networkd now supports matching on the system
5605 virtualization, architecture, kernel command line, host name
5608 * logind is now a lot more aggressive when suspending the
5609 machine due to a closed laptop lid. Instead of acting only
5610 on the lid close action, it will continuously watch the lid
5611 status and act on it. This is useful for laptops where the
5612 power button is on the outside of the chassis so that it can
5613 be reached without opening the lid (such as the Lenovo
5614 Yoga). On those machines, logind will now immediately
5615 re-suspend the machine if the power button has been
5616 accidentally pressed while the laptop was suspended and in a
5617 backpack or similar.
5619 * logind will now watch SW_DOCK switches and inhibit reaction
5620 to the lid switch if it is pressed. This means that logind
5621 will not suspend the machine anymore if the lid is closed
5622 and the system is docked, if the laptop supports SW_DOCK
5623 notifications via the input layer. Note that ACPI docking
5624 stations do not generate this currently. Also note that this
5625 logic is usually not fully sufficient and Desktop
5626 Environments should take a lid switch inhibitor lock when an
5627 external display is connected, as systemd will not watch
5630 * nspawn will now make use of the devices cgroup controller by
5631 default, and only permit creation of and access to the usual
5632 API device nodes like /dev/null or /dev/random, as well as
5633 access to (but not creation of) the pty devices.
5635 * We will now ship a default .network file for
5636 systemd-networkd that automatically configures DHCP for
5637 network interfaces created by nspawn's --network-veth or
5638 --network-bridge= switches.
5640 * systemd will now understand the usual M, K, G, T suffixes
5641 according to SI conventions (i.e. to the base 1000) when
5642 referring to throughput and hardware metrics. It will stay
5643 with IEC conventions (i.e. to the base 1024) for software
5644 metrics, according to what is customary according to
5645 Wikipedia. We explicitly document which base applies for
5646 each configuration option.
5648 * The DeviceAllow= setting in unit files now supports a syntax
5649 to whitelist an entire group of devices node majors at once,
5650 based on the /proc/devices listing. For example, with the
5651 string "char-pts", it is now possible to whitelist all
5652 current and future pseudo-TTYs at once.
5654 * sd-event learned a new "post" event source. Event sources of
5655 this type are triggered by the dispatching of any event
5656 source of a type that is not "post". This is useful for
5657 implementing clean-up and check event sources that are
5658 triggered by other work being done in the program.
5660 * systemd-networkd is no longer statically enabled, but uses
5661 the usual [Install] sections so that it can be
5662 enabled/disabled using systemctl. It still is enabled by
5665 * When creating a veth interface pair with systemd-nspawn, the
5666 host side will now be prefixed with "vb-" if
5667 --network-bridge= is used, and with "ve-" if --network-veth
5668 is used. This way, it is easy to distinguish these cases on
5669 the host, for example to apply different configuration to
5670 them with systemd-networkd.
5672 * The compatibility libraries for libsystemd-journal.so,
5673 libsystem-id128.so, libsystemd-login.so and
5674 libsystemd-daemon.so do not make use of IFUNC
5675 anymore. Instead, we now build libsystemd.so multiple times
5676 under these alternative names. This means that the footprint
5677 is drastically increased, but given that these are
5678 transitional compatibility libraries, this should not matter
5679 much. This change has been made necessary to support the ARM
5680 platform for these compatibility libraries, as the ARM
5681 toolchain is not really at the same level as the toolchain
5682 for other architectures like x86 and does not support
5683 IFUNC. Please make sure to use --enable-compat-libs only
5684 during a transitional period!
5686 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
5687 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5688 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
5689 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
5690 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
5691 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
5692 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
5693 Zbigniew Jędrzejewski-Szmek
5695 — Berlin, 2014-02-24
5699 * A new component "systemd-networkd" has been added that can
5700 be used to configure local network interfaces statically or
5701 via DHCP. It is capable of bringing up bridges, VLANs, and
5702 bonding. Currently, no hook-ups for interactive network
5703 configuration are provided. Use this for your initrd,
5704 container, embedded, or server setup if you need a simple,
5705 yet powerful, network configuration solution. This
5706 configuration subsystem is quite nifty, as it allows wildcard
5707 hotplug matching in interfaces. For example, with a single
5708 configuration snippet, you can configure that all Ethernet
5709 interfaces showing up are automatically added to a bridge,
5710 or similar. It supports link-sensing and more.
5712 * A new tool "systemd-socket-proxyd" has been added which can
5713 act as a bidirectional proxy for TCP sockets. This is
5714 useful for adding socket activation support to services that
5715 do not actually support socket activation, including virtual
5716 machines and the like.
5718 * Add a new tool to save/restore rfkill state on
5721 * Save/restore state of keyboard backlights in addition to
5722 display backlights on shutdown/boot.
5724 * udev learned a new SECLABEL{} construct to label device
5725 nodes with a specific security label when they appear. For
5726 now, only SECLABEL{selinux} is supported, but the syntax is
5727 prepared for additional security frameworks.
5729 * udev gained a new scheme to configure link-level attributes
5730 from files in /etc/systemd/network/*.link. These files can
5731 match against MAC address, device path, driver name and type,
5732 and will apply attributes like the naming policy, link speed,
5733 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
5734 address assignment policy (randomized, ...).
5736 * The configuration of network interface naming rules for
5737 "permanent interface names" has changed: a new NamePolicy=
5738 setting in the [Link] section of .link files determines the
5739 priority of possible naming schemes (onboard, slot, MAC,
5740 path). The default value of this setting is determined by
5741 /usr/lib/net/links/99-default.link. Old
5742 80-net-name-slot.rules udev configuration file has been
5743 removed, so local configuration overriding this file should
5744 be adapted to override 99-default.link instead.
5746 * When the User= switch is used in a unit file, also
5747 initialize $SHELL= based on the user database entry.
5749 * systemd no longer depends on libdbus. All communication is
5750 now done with sd-bus, systemd's low-level bus library
5753 * kdbus support has been added to PID 1 itself. When kdbus is
5754 enabled, this causes PID 1 to set up the system bus and
5755 enable support for a new ".busname" unit type that
5756 encapsulates bus name activation on kdbus. It works a little
5757 bit like ".socket" units, except for bus names. A new
5758 generator has been added that converts classic dbus1 service
5759 activation files automatically into native systemd .busname
5762 * sd-bus: add a light-weight vtable implementation that allows
5763 defining objects on the bus with a simple static const
5764 vtable array of its methods, signals and properties.
5766 * systemd will not generate or install static dbus
5767 introspection data anymore to /usr/share/dbus-1/interfaces,
5768 as the precise format of these files is unclear, and
5769 nothing makes use of it.
5771 * A proxy daemon is now provided to proxy clients connecting
5772 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
5773 compatibility with classic D-Bus.
5775 * A bus driver implementation has been added that supports the
5776 classic D-Bus bus driver calls on kdbus, also for
5777 compatibility purposes.
5779 * A new API "sd-event.h" has been added that implements a
5780 minimal event loop API built around epoll. It provides a
5781 couple of features that direct epoll usage is lacking:
5782 prioritization of events, scales to large numbers of timer
5783 events, per-event timer slack (accuracy), system-wide
5784 coalescing of timer events, exit handlers, watchdog
5785 supervision support using systemd's sd_notify() API, child
5788 * A new API "sd-rntl.h" has been added that provides an API
5789 around the route netlink interface of the kernel, similar in
5790 style to "sd-bus.h".
5792 * A new API "sd-dhcp-client.h" has been added that provides a
5793 small DHCPv4 client-side implementation. This is used by
5796 * There is a new kernel command line option
5797 "systemd.restore_state=0|1". When set to "0", none of the
5798 systemd tools will restore saved runtime state to hardware
5799 devices. More specifically, the rfkill and backlight states
5802 * The FsckPassNo= compatibility option in mount/service units
5803 has been removed. The fstab generator will now add the
5804 necessary dependencies automatically, and does not require
5805 PID1's support for that anymore.
5807 * journalctl gained a new switch, --list-boots, that lists
5808 recent boots with their times and boot IDs.
5810 * The various tools like systemctl, loginctl, timedatectl,
5811 busctl, systemd-run, ... have gained a new switch "-M" to
5812 connect to a specific, local OS container (as direct
5813 connection, without requiring SSH). This works on any
5814 container that is registered with machined, such as those
5815 created by libvirt-lxc or nspawn.
5817 * systemd-run and systemd-analyze also gained support for "-H"
5818 to connect to remote hosts via SSH. This is particularly
5819 useful for systemd-run because it enables queuing of jobs
5820 onto remote systems.
5822 * machinectl gained a new command "login" to open a getty
5823 login in any local container. This works with any container
5824 that is registered with machined (such as those created by
5825 libvirt-lxc or nspawn), and which runs systemd inside.
5827 * machinectl gained a new "reboot" command that may be used to
5828 trigger a reboot on a specific container that is registered
5829 with machined. This works on any container that runs an init
5830 system of some kind.
5832 * systemctl gained a new "list-timers" command to print a nice
5833 listing of installed timer units with the times they elapse
5836 * Alternative reboot() parameters may now be specified on the
5837 "systemctl reboot" command line and are passed to the
5838 reboot() system call.
5840 * systemctl gained a new --job-mode= switch to configure the
5841 mode to queue a job with. This is a more generic version of
5842 --fail, --irreversible, and --ignore-dependencies, which are
5843 still available but not advertised anymore.
5845 * /etc/systemd/system.conf gained new settings to configure
5846 various default timeouts of units, as well as the default
5847 start limit interval and burst. These may still be overridden
5850 * PID1 will now export on the bus profile data of the security
5851 policy upload process (such as the SELinux policy upload to
5854 * journald: when forwarding logs to the console, include
5855 timestamps (following the setting in
5856 /sys/module/printk/parameters/time).
5858 * OnCalendar= in timer units now understands the special
5859 strings "yearly" and "annually". (Both are equivalent)
5861 * The accuracy of timer units is now configurable with the new
5862 AccuracySec= setting. It defaults to 1min.
5864 * A new dependency type JoinsNamespaceOf= has been added that
5865 allows running two services within the same /tmp and network
5866 namespace, if PrivateNetwork= or PrivateTmp= are used.
5868 * A new command "cat" has been added to systemctl. It outputs
5869 the original unit file of a unit, and concatenates the
5870 contents of additional "drop-in" unit file snippets, so that
5871 the full configuration is shown.
5873 * systemctl now supports globbing on the various "list-xyz"
5874 commands, like "list-units" or "list-sockets", as well as on
5875 those commands which take multiple unit names.
5877 * journalctl's --unit= switch gained support for globbing.
5879 * All systemd daemons now make use of the watchdog logic so
5880 that systemd automatically notices when they hang.
5882 * If the $container_ttys environment variable is set,
5883 getty-generator will automatically spawn a getty for each
5884 listed tty. This is useful for container managers to request
5885 login gettys to be spawned on as many ttys as needed.
5887 * %h, %s, %U specifier support is not available anymore when
5888 used in unit files for PID 1. This is because NSS calls are
5889 not safe from PID 1. They stay available for --user
5890 instances of systemd, and as special case for the root user.
5892 * loginctl gained a new "--no-legend" switch to turn off output
5895 * The "sd-login.h" API gained three new calls:
5896 sd_session_is_remote(), sd_session_get_remote_user(),
5897 sd_session_get_remote_host() to query information about
5900 * The udev hardware database now also carries vendor/product
5901 information of SDIO devices.
5903 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
5904 determine whether watchdog notifications are requested by
5907 * Socket-activated per-connection services now include a
5908 short description of the connection parameters in the
5911 * tmpfiles gained a new "--boot" option. When this is not used,
5912 only lines where the command character is not suffixed with
5913 "!" are executed. When this option is specified, those
5914 options are executed too. This partitions tmpfiles
5915 directives into those that can be safely executed at any
5916 time, and those which should be run only at boot (for
5917 example, a line that creates /run/nologin).
5919 * A new API "sd-resolve.h" has been added which provides a simple
5920 asynchronous wrapper around glibc NSS host name resolution
5921 calls, such as getaddrinfo(). In contrast to glibc's
5922 getaddrinfo_a(), it does not use signals. In contrast to most
5923 other asynchronous name resolution libraries, this one does
5924 not reimplement DNS, but reuses NSS, so that alternate
5925 host name resolution systems continue to work, such as mDNS,
5926 LDAP, etc. This API is based on libasyncns, but it has been
5927 cleaned up for inclusion in systemd.
5929 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
5930 "sd-daemon.h" are no longer found in individual libraries
5931 libsystemd-journal.so, libsystemd-login.so,
5932 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
5933 merged them into a single library, libsystemd.so, which
5934 provides all symbols. The reason for this is cyclic
5935 dependencies, as these libraries tend to use each other's
5936 symbols. So far, we have managed to workaround that by linking
5937 a copy of a good part of our code into each of these
5938 libraries again and again, which, however, makes certain
5939 things hard to do, like sharing static variables. Also, it
5940 substantially increases footprint. With this change, there
5941 is only one library for the basic APIs systemd
5942 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
5943 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
5944 library as well, however are subject to the --enable-kdbus
5945 switch (see below). Note that "sd-dhcp-client.h" is not part
5946 of this library (this is because it only consumes, never
5947 provides, services of/to other APIs). To make the transition
5948 easy from the separate libraries to the unified one, we
5949 provide the --enable-compat-libs compile-time switch which
5950 will generate stub libraries that are compatible with the
5951 old ones but redirect all calls to the new one.
5953 * All of the kdbus logic and the new APIs "sd-bus.h",
5954 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
5955 and "sd-utf8.h" are compile-time optional via the
5956 "--enable-kdbus" switch, and they are not compiled in by
5957 default. To make use of kdbus, you have to explicitly enable
5958 the switch. Note however, that neither the kernel nor the
5959 userspace API for all of this is considered stable yet. We
5960 want to maintain the freedom to still change the APIs for
5961 now. By specifying this build-time switch, you acknowledge
5962 that you are aware of the instability of the current
5965 * Also, note that while kdbus is pretty much complete,
5966 it lacks one thing: proper policy support. This means you
5967 can build a fully working system with all features; however,
5968 it will be highly insecure. Policy support will be added in
5969 one of the next releases, at the same time that we will
5970 declare the APIs stable.
5972 * When the kernel command line argument "kdbus" is specified,
5973 systemd will automatically load the kdbus.ko kernel module. At
5974 this stage of development, it is only useful for testing kdbus
5975 and should not be used in production. Note: if "--enable-kdbus"
5976 is specified, and the kdbus.ko kernel module is available, and
5977 "kdbus" is added to the kernel command line, the entire system
5978 runs with kdbus instead of dbus-daemon, with the above mentioned
5979 problem of missing the system policy enforcement. Also a future
5980 version of kdbus.ko or a newer systemd will not be compatible with
5981 each other, and will unlikely be able to boot the machine if only
5982 one of them is updated.
5984 * systemctl gained a new "import-environment" command which
5985 uploads the caller's environment (or parts thereof) into the
5986 service manager so that it is inherited by services started
5987 by the manager. This is useful to upload variables like
5988 $DISPLAY into the user service manager.
5990 * A new PrivateDevices= switch has been added to service units
5991 which allows running a service with a namespaced /dev
5992 directory that does not contain any device nodes for
5993 physical devices. More specifically, it only includes devices
5994 such as /dev/null, /dev/urandom, and /dev/zero which are API
5997 * logind has been extended to support behaviour like VT
5998 switching on seats that do not support a VT. This makes
5999 multi-session available on seats that are not the first seat
6000 (seat0), and on systems where kernel support for VTs has
6001 been disabled at compile-time.
6003 * If a process holds a delay lock for system sleep or shutdown
6004 and fails to release it in time, we will now log its
6005 identity. This makes it easier to identify processes that
6006 cause slow suspends or power-offs.
6008 * When parsing /etc/crypttab, support for a new key-slot=
6009 option as supported by Debian is added. It allows indicating
6010 which LUKS slot to use on disk, speeding up key loading.
6012 * The sd_journal_sendv() API call has been checked and
6013 officially declared to be async-signal-safe so that it may
6014 be invoked from signal handlers for logging purposes.
6016 * Boot-time status output is now enabled automatically after a
6017 short timeout if boot does not progress, in order to give
6018 the user an indication what she or he is waiting for.
6020 * The boot-time output has been improved to show how much time
6021 remains until jobs expire.
6023 * The KillMode= switch in service units gained a new possible
6024 value "mixed". If set, and the unit is shut down, then the
6025 initial SIGTERM signal is sent only to the main daemon
6026 process, while the following SIGKILL signal is sent to
6027 all remaining processes of the service.
6029 * When a scope unit is registered, a new property "Controller"
6030 may be set. If set to a valid bus name, systemd will send a
6031 RequestStop() signal to this name when it would like to shut
6032 down the scope. This may be used to hook manager logic into
6033 the shutdown logic of scope units. Also, scope units may now
6034 be put in a special "abandoned" state, in which case the
6035 manager process which created them takes no further
6036 responsibilities for it.
6038 * When reading unit files, systemd will now verify
6039 the access mode of these files, and warn about certain
6040 suspicious combinations. This has been added to make it
6041 easier to track down packaging bugs where unit files are
6042 marked executable or world-writable.
6044 * systemd-nspawn gained a new "--setenv=" switch to set
6045 container-wide environment variables. The similar option in
6046 systemd-activate was renamed from "--environment=" to
6047 "--setenv=" for consistency.
6049 * systemd-nspawn has been updated to create a new kdbus domain
6050 for each container that is invoked, thus allowing each
6051 container to have its own set of system and user buses,
6052 independent of the host.
6054 * systemd-nspawn gained a new --drop-capability= switch to run
6055 the container with less capabilities than the default. Both
6056 --drop-capability= and --capability= now take the special
6057 string "all" for dropping or keeping all capabilities.
6059 * systemd-nspawn gained new switches for executing containers
6060 with specific SELinux labels set.
6062 * systemd-nspawn gained a new --quiet switch to not generate
6063 any additional output but the container's own console
6066 * systemd-nspawn gained a new --share-system switch to run a
6067 container without PID namespacing enabled.
6069 * systemd-nspawn gained a new --register= switch to control
6070 whether the container is registered with systemd-machined or
6071 not. This is useful for containers that do not run full
6072 OS images, but only specific apps.
6074 * systemd-nspawn gained a new --keep-unit which may be used
6075 when invoked as the only program from a service unit, and
6076 results in registration of the unit service itself in
6077 systemd-machined, instead of a newly opened scope unit.
6079 * systemd-nspawn gained a new --network-interface= switch for
6080 moving arbitrary interfaces to the container. The new
6081 --network-veth switch creates a virtual Ethernet connection
6082 between host and container. The new --network-bridge=
6083 switch then allows assigning the host side of this virtual
6084 Ethernet connection to a bridge device.
6086 * systemd-nspawn gained a new --personality= switch for
6087 setting the kernel personality for the container. This is
6088 useful when running a 32-bit container on a 64-bit host. A
6089 similar option Personality= is now also available for service
6092 * logind will now also track a "Desktop" identifier for each
6093 session which encodes the desktop environment of it. This is
6094 useful for desktop environments that want to identify
6095 multiple running sessions of itself easily.
6097 * A new SELinuxContext= setting for service units has been
6098 added that allows setting a specific SELinux execution
6099 context for a service.
6101 * Most systemd client tools will now honour $SYSTEMD_LESS for
6102 settings of the "less" pager. By default, these tools will
6103 override $LESS to allow certain operations to work, such as
6104 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
6105 influence this logic.
6107 * systemd's "seccomp" hook-up has been changed to make use of
6108 the libseccomp library instead of using its own
6109 implementation. This has benefits for portability among
6112 * For usage together with SystemCallFilter=, a new
6113 SystemCallErrorNumber= setting has been introduced that
6114 allows configuration of a system error number to be returned
6115 on filtered system calls, instead of immediately killing the
6116 process. Also, SystemCallArchitectures= has been added to
6117 limit access to system calls of a particular architecture
6118 (in order to turn off support for unused secondary
6119 architectures). There is also a global
6120 SystemCallArchitectures= setting in system.conf now to turn
6121 off support for non-native system calls system-wide.
6123 * systemd requires a kernel with a working name_to_handle_at(),
6124 please see the kernel config requirements in the README file.
6126 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
6127 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
6128 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
6129 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
6130 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
6131 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
6132 Elia Pinto, Florian Weimer, George McCollister, Goffredo
6133 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
6134 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
6135 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
6136 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
6137 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
6138 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
6139 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
6140 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
6141 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
6142 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
6143 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
6144 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
6145 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
6146 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
6147 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
6148 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
6149 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
6151 — Berlin, 2014-02-20
6155 * logind has gained support for facilitating privileged input
6156 and drm device access for unprivileged clients. This work is
6157 useful to allow Wayland display servers (and similar
6158 programs, such as kmscon) to run under the user's ID and
6159 access input and drm devices which are normally
6160 protected. When this is used (and the kernel is new enough)
6161 logind will "mute" IO on the file descriptors passed to
6162 Wayland as long as it is in the background and "unmute" it
6163 if it returns into the foreground. This allows secure
6164 session switching without allowing background sessions to
6165 eavesdrop on input and display data. This also introduces
6166 session switching support if VT support is turned off in the
6167 kernel, and on seats that are not seat0.
6169 * A new kernel command line option luks.options= is understood
6170 now which allows specifying LUKS options for usage for LUKS
6171 encrypted partitions specified with luks.uuid=.
6173 * tmpfiles.d(5) snippets may now use specifier expansion in
6174 path names. More specifically %m, %b, %H, %v, are now
6175 replaced by the local machine id, boot id, hostname, and
6176 kernel version number.
6178 * A new tmpfiles.d(5) command "m" has been introduced which
6179 may be used to change the owner/group/access mode of a file
6180 or directory if it exists, but do nothing if it does not.
6182 * This release removes high-level support for the
6183 MemorySoftLimit= cgroup setting. The underlying kernel
6184 cgroup attribute memory.soft_limit= is currently badly
6185 designed and likely to be removed from the kernel API in its
6186 current form, hence we should not expose it for now.
6188 * The memory.use_hierarchy cgroup attribute is now enabled for
6189 all cgroups systemd creates in the memory cgroup
6190 hierarchy. This option is likely to be come the built-in
6191 default in the kernel anyway, and the non-hierarchical mode
6192 never made much sense in the intrinsically hierarchical
6195 * A new field _SYSTEMD_SLICE= is logged along with all journal
6196 messages containing the slice a message was generated
6197 from. This is useful to allow easy per-customer filtering of
6198 logs among other things.
6200 * systemd-journald will no longer adjust the group of journal
6201 files it creates to the "systemd-journal" group. Instead we
6202 rely on the journal directory to be owned by the
6203 "systemd-journal" group, and its setgid bit set, so that the
6204 kernel file system layer will automatically enforce that
6205 journal files inherit this group assignment. The reason for
6206 this change is that we cannot allow NSS look-ups from
6207 journald which would be necessary to resolve
6208 "systemd-journal" to a numeric GID, because this might
6209 create deadlocks if NSS involves synchronous queries to
6210 other daemons (such as nscd, or sssd) which in turn are
6211 logging clients of journald and might block on it, which
6212 would then dead lock. A tmpfiles.d(5) snippet included in
6213 systemd will make sure the setgid bit and group are
6214 properly set on the journal directory if it exists on every
6215 boot. However, we recommend adjusting it manually after
6216 upgrades too (or from RPM scriptlets), so that the change is
6217 not delayed until next reboot.
6219 * Backlight and random seed files in /var/lib/ have moved into
6220 the /var/lib/systemd/ directory, in order to centralize all
6221 systemd generated files in one directory.
6223 * Boot time performance measurements (as displayed by
6224 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
6225 performance information if that's available to determine how
6226 much time BIOS and boot loader initialization required. With
6227 a sufficiently new BIOS you hence no longer need to boot
6228 with Gummiboot to get access to such information.
6230 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
6231 Cristian Rodríguez, Dave Reisner, David Herrmann, David
6232 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
6233 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
6234 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
6235 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
6236 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6238 — Berlin, 2013-10-02
6242 * The Restart= option for services now understands a new
6243 on-watchdog setting, which will restart the service
6244 automatically if the service stops sending out watchdog keep
6245 alive messages (as configured with WatchdogSec=).
6247 * The getty generator (which is responsible for bringing up a
6248 getty on configured serial consoles) will no longer only
6249 start a getty on the primary kernel console but on all
6250 others, too. This makes the order in which console= is
6251 specified on the kernel command line less important.
6253 * libsystemd-logind gained a new sd_session_get_vt() call to
6254 retrieve the VT number of a session.
6256 * If the option "tries=0" is set for an entry of /etc/crypttab
6257 its passphrase is queried indefinitely instead of any
6258 maximum number of tries.
6260 * If a service with a configure PID file terminates its PID
6261 file will now be removed automatically if it still exists
6262 afterwards. This should put an end to stale PID files.
6264 * systemd-run will now also take relative binary path names
6265 for execution and no longer insists on absolute paths.
6267 * InaccessibleDirectories= and ReadOnlyDirectories= now take
6268 paths that are optionally prefixed with "-" to indicate that
6269 it should not be considered a failure if they do not exist.
6271 * journalctl -o (and similar commands) now understands a new
6272 output mode "short-precise", it is similar to "short" but
6273 shows timestamps with usec accuracy.
6275 * The option "discard" (as known from Debian) is now
6276 synonymous to "allow-discards" in /etc/crypttab. In fact,
6277 "discard" is preferred now (since it is easier to remember
6280 * Some licensing clean-ups were made, so that more code is now
6281 LGPL-2.1 licensed than before.
6283 * A minimal tool to save/restore the display backlight
6284 brightness across reboots has been added. It will store the
6285 backlight setting as late as possible at shutdown, and
6286 restore it as early as possible during reboot.
6288 * A logic to automatically discover and enable home and swap
6289 partitions on GPT disks has been added. With this in place
6290 /etc/fstab becomes optional for many setups as systemd can
6291 discover certain partitions located on the root disk
6292 automatically. Home partitions are recognized under their
6293 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
6294 partitions are recognized under their GPT type ID
6295 0657fd6da4ab43c484e50933c84b4f4f.
6297 * systemd will no longer pass any environment from the kernel
6298 or initrd to system services. If you want to set an
6299 environment for all services, do so via the kernel command
6300 line systemd.setenv= assignment.
6302 * The systemd-sysctl tool no longer natively reads the file
6303 /etc/sysctl.conf. If desired, the file should be symlinked
6304 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
6305 legacy support by a symlink rather than built-in code, it
6306 also makes the otherwise hidden order of application of the
6307 different files visible. (Note that this partly reverts to a
6308 pre-198 application order of sysctl knobs!)
6310 * The "systemctl set-log-level" and "systemctl dump" commands
6311 have been moved to systemd-analyze.
6313 * systemd-run learned the new --remain-after-exit switch,
6314 which causes the scope unit not to be cleaned up
6315 automatically after the process terminated.
6317 * tmpfiles learned a new --exclude-prefix= switch to exclude
6318 certain paths from operation.
6320 * journald will now automatically flush all messages to disk
6321 as soon as a message at the log level CRIT, ALERT or EMERG
6324 Contributions from: Andrew Cook, Brandon Philips, Christian
6325 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
6326 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
6327 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
6328 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
6329 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
6330 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
6331 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
6332 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
6333 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
6334 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
6335 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
6336 William Giokas, Zbigniew Jędrzejewski-Szmek
6338 — Berlin, 2013-09-13
6342 * The documentation has been updated to cover the various new
6343 concepts introduced with 205.
6345 * Unit files now understand the new %v specifier which
6346 resolves to the kernel version string as returned by "uname
6349 * systemctl now supports filtering the unit list output by
6350 load state, active state and sub state, using the new
6353 * "systemctl status" will now show the results of the
6354 condition checks (like ConditionPathExists= and similar) of
6355 the last start attempts of the unit. They are also logged to
6358 * "journalctl -b" may now be used to look for boot output of a
6359 specific boot. Try "journalctl -b -1" for the previous boot,
6360 but the syntax is substantially more powerful.
6362 * "journalctl --show-cursor" has been added which prints the
6363 cursor string the last shown log line. This may then be used
6364 with the new "journalctl --after-cursor=" switch to continue
6365 browsing logs from that point on.
6367 * "journalctl --force" may now be used to force regeneration
6370 * Creation of "dead" device nodes has been moved from udev
6371 into kmod and tmpfiles. Previously, udev would read the kmod
6372 databases to pre-generate dead device nodes based on meta
6373 information contained in kernel modules, so that these would
6374 be auto-loaded on access rather then at boot. As this
6375 does not really have much to do with the exposing actual
6376 kernel devices to userspace this has always been slightly
6377 alien in the udev codebase. Following the new scheme kmod
6378 will now generate a runtime snippet for tmpfiles from the
6379 module meta information and it now is tmpfiles' job to the
6380 create the nodes. This also allows overriding access and
6381 other parameters for the nodes using the usual tmpfiles
6382 facilities. As side effect this allows us to remove the
6383 CAP_SYS_MKNOD capability bit from udevd entirely.
6385 * logind's device ACLs may now be applied to these "dead"
6386 devices nodes too, thus finally allowing managed access to
6387 devices such as /dev/snd/sequencer without loading the
6388 backing module right-away.
6390 * A new RPM macro has been added that may be used to apply
6391 tmpfiles configuration during package installation.
6393 * systemd-detect-virt and ConditionVirtualization= now can
6394 detect User-Mode-Linux machines (UML).
6396 * journald will now implicitly log the effective capabilities
6397 set of processes in the message metadata.
6399 * systemd-cryptsetup has gained support for TrueCrypt volumes.
6401 * The initrd interface has been simplified (more specifically,
6402 support for passing performance data via environment
6403 variables and fsck results via files in /run has been
6404 removed). These features were non-essential, and are
6405 nowadays available in a much nicer way by having systemd in
6406 the initrd serialize its state and have the hosts systemd
6407 deserialize it again.
6409 * The udev "keymap" data files and tools to apply keyboard
6410 specific mappings of scan to key codes, and force-release
6411 scan code lists have been entirely replaced by a udev
6412 "keyboard" builtin and a hwdb data file.
6414 * systemd will now honour the kernel's "quiet" command line
6415 argument also during late shutdown, resulting in a
6416 completely silent shutdown when used.
6418 * There's now an option to control the SO_REUSEPORT socket
6419 option in .socket units.
6421 * Instance units will now automatically get a per-template
6422 subslice of system.slice unless something else is explicitly
6423 configured. For example, instances of sshd@.service will now
6424 implicitly be placed in system-sshd.slice rather than
6425 system.slice as before.
6427 * Test coverage support may now be enabled at build time.
6429 Contributions from: Dave Reisner, Frederic Crozat, Harald
6430 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
6431 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
6432 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
6433 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
6434 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
6435 Giokas, Zbigniew Jędrzejewski-Szmek
6437 — Berlin, 2013-07-23
6441 * Two new unit types have been introduced:
6443 Scope units are very similar to service units, however, are
6444 created out of pre-existing processes — instead of PID 1
6445 forking off the processes. By using scope units it is
6446 possible for system services and applications to group their
6447 own child processes (worker processes) in a powerful way
6448 which then maybe used to organize them, or kill them
6449 together, or apply resource limits on them.
6451 Slice units may be used to partition system resources in an
6452 hierarchical fashion and then assign other units to them. By
6453 default there are now three slices: system.slice (for all
6454 system services), user.slice (for all user sessions),
6455 machine.slice (for VMs and containers).
6457 Slices and scopes have been introduced primarily in
6458 context of the work to move cgroup handling to a
6459 single-writer scheme, where only PID 1
6460 creates/removes/manages cgroups.
6462 * There's a new concept of "transient" units. In contrast to
6463 normal units these units are created via an API at runtime,
6464 not from configuration from disk. More specifically this
6465 means it is now possible to run arbitrary programs as
6466 independent services, with all execution parameters passed
6467 in via bus APIs rather than read from disk. Transient units
6468 make systemd substantially more dynamic then it ever was,
6469 and useful as a general batch manager.
6471 * logind has been updated to make use of scope and slice units
6472 for managing user sessions. As a user logs in he will get
6473 his own private slice unit, to which all sessions are added
6474 as scope units. We also added support for automatically
6475 adding an instance of user@.service for the user into the
6476 slice. Effectively logind will no longer create cgroup
6477 hierarchies on its own now, it will defer entirely to PID 1
6478 for this by means of scope, service and slice units. Since
6479 user sessions this way become entities managed by PID 1
6480 the output of "systemctl" is now a lot more comprehensive.
6482 * A new mini-daemon "systemd-machined" has been added which
6483 may be used by virtualization managers to register local
6484 VMs/containers. nspawn has been updated accordingly, and
6485 libvirt will be updated shortly. machined will collect a bit
6486 of meta information about the VMs/containers, and assign
6487 them their own scope unit (see above). The collected
6488 meta-data is then made available via the "machinectl" tool,
6489 and exposed in "ps" and similar tools. machined/machinectl
6490 is compile-time optional.
6492 * As discussed earlier, the low-level cgroup configuration
6493 options ControlGroup=, ControlGroupModify=,
6494 ControlGroupPersistent=, ControlGroupAttribute= have been
6495 removed. Please use high-level attribute settings instead as
6496 well as slice units.
6498 * A new bus call SetUnitProperties() has been added to alter
6499 various runtime parameters of a unit. This is primarily
6500 useful to alter cgroup parameters dynamically in a nice way,
6501 but will be extended later on to make more properties
6502 modifiable at runtime. systemctl gained a new set-properties
6503 command that wraps this call.
6505 * A new tool "systemd-run" has been added which can be used to
6506 run arbitrary command lines as transient services or scopes,
6507 while configuring a number of settings via the command
6508 line. This tool is currently very basic, however already
6509 very useful. We plan to extend this tool to even allow
6510 queuing of execution jobs with time triggers from the
6511 command line, similar in fashion to "at".
6513 * nspawn will now inform the user explicitly that kernels with
6514 audit enabled break containers, and suggest the user to turn
6517 * Support for detecting the IMA and AppArmor security
6518 frameworks with ConditionSecurity= has been added.
6520 * journalctl gained a new "-k" switch for showing only kernel
6521 messages, mimicking dmesg output; in addition to "--user"
6522 and "--system" switches for showing only user's own logs
6525 * systemd-delta can now show information about drop-in
6526 snippets extending unit files.
6528 * libsystemd-bus has been substantially updated but is still
6529 not available as public API.
6531 * systemd will now look for the "debug" argument on the kernel
6532 command line and enable debug logging, similar to what
6533 "systemd.log_level=debug" already did before.
6535 * "systemctl set-default", "systemctl get-default" has been
6536 added to configure the default.target symlink, which
6537 controls what to boot into by default.
6539 * "systemctl set-log-level" has been added as a convenient
6540 way to raise and lower systemd logging threshold.
6542 * "systemd-analyze plot" will now show the time the various
6543 generators needed for execution, as well as information
6544 about the unit file loading.
6546 * libsystemd-journal gained a new sd_journal_open_files() call
6547 for opening specific journal files. journactl also gained a
6548 new switch to expose this new functionality. Previously we
6549 only supported opening all files from a directory, or all
6550 files from the system, as opening individual files only is
6551 racy due to journal file rotation.
6553 * systemd gained the new DefaultEnvironment= setting in
6554 /etc/systemd/system.conf to set environment variables for
6557 * If a privileged process logs a journal message with the
6558 OBJECT_PID= field set, then journald will automatically
6559 augment this with additional OBJECT_UID=, OBJECT_GID=,
6560 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
6561 system services want to log events about specific client
6562 processes. journactl/systemctl has been updated to make use
6563 of this information if all log messages regarding a specific
6566 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
6567 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
6568 Reisner, David Coppa, David King, David Strauss, Eelco
6569 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
6570 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
6571 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
6572 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
6573 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
6574 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
6575 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
6576 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
6577 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
6578 Łukasz Stelmach, 장동준
6582 * The Python bindings gained some minimal support for the APIs
6583 exposed by libsystemd-logind.
6585 * ConditionSecurity= gained support for detecting SMACK. Since
6586 this condition already supports SELinux and AppArmor we only
6587 miss IMA for this. Patches welcome!
6589 Contributions from: Karol Lewandowski, Lennart Poettering,
6590 Zbigniew Jędrzejewski-Szmek
6594 * systemd-nspawn will now create /etc/resolv.conf if
6595 necessary, before bind-mounting the host's file onto it.
6597 * systemd-nspawn will now store meta information about a
6598 container on the container's cgroup as extended attribute
6599 fields, including the root directory.
6601 * The cgroup hierarchy has been reworked in many ways. All
6602 objects any of the components systemd creates in the cgroup
6603 tree are now suffixed. More specifically, user sessions are
6604 now placed in cgroups suffixed with ".session", users in
6605 cgroups suffixed with ".user", and nspawn containers in
6606 cgroups suffixed with ".nspawn". Furthermore, all cgroup
6607 names are now escaped in a simple scheme to avoid collision
6608 of userspace object names with kernel filenames. This work
6609 is preparation for making these objects relocatable in the
6610 cgroup tree, in order to allow easy resource partitioning of
6611 these objects without causing naming conflicts.
6613 * systemctl list-dependencies gained the new switches
6614 --plain, --reverse, --after and --before.
6616 * systemd-inhibit now shows the process name of processes that
6617 have taken an inhibitor lock.
6619 * nss-myhostname will now also resolve "localhost"
6620 implicitly. This makes /etc/hosts an optional file and
6621 nicely handles that on IPv6 ::1 maps to both "localhost" and
6624 * libsystemd-logind.so gained a new call
6625 sd_get_machine_names() to enumerate running containers and
6626 VMs (currently only supported by very new libvirt and
6627 nspawn). sd_login_monitor can now be used to watch
6628 VMs/containers coming and going.
6630 * .include is not allowed recursively anymore, and only in
6631 unit files. Usually it is better to use drop-in snippets in
6632 .d/*.conf anyway, as introduced with systemd 198.
6634 * systemd-analyze gained a new "critical-chain" command that
6635 determines the slowest chain of units run during system
6636 boot-up. It is very useful for tracking down where
6637 optimizing boot time is the most beneficial.
6639 * systemd will no longer allow manipulating service paths in
6640 the name=systemd:/system cgroup tree using ControlGroup= in
6641 units. (But is still fine with it in all other dirs.)
6643 * There's a new systemd-nspawn@.service service file that may
6644 be used to easily run nspawn containers as system
6645 services. With the container's root directory in
6646 /var/lib/container/foobar it is now sufficient to run
6647 "systemctl start systemd-nspawn@foobar.service" to boot it.
6649 * systemd-cgls gained a new parameter "--machine" to list only
6650 the processes within a certain container.
6652 * ConditionSecurity= now can check for "apparmor". We still
6653 are lacking checks for SMACK and IMA for this condition
6654 check though. Patches welcome!
6656 * A new configuration file /etc/systemd/sleep.conf has been
6657 added that may be used to configure which kernel operation
6658 systemd is supposed to execute when "suspend", "hibernate"
6659 or "hybrid-sleep" is requested. This makes the new kernel
6660 "freeze" state accessible to the user.
6662 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
6663 the passed argument if applicable.
6665 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
6666 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
6667 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
6668 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
6669 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
6670 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
6671 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
6676 * The output of 'systemctl list-jobs' got some polishing. The
6677 '--type=' argument may now be passed more than once. A new
6678 command 'systemctl list-sockets' has been added which shows
6679 a list of kernel sockets systemd is listening on with the
6680 socket units they belong to, plus the units these socket
6683 * The experimental libsystemd-bus library got substantial
6684 updates to work in conjunction with the (also experimental)
6685 kdbus kernel project. It works well enough to exchange
6686 messages with some sophistication. Note that kdbus is not
6687 ready yet, and the library is mostly an elaborate test case
6688 for now, and not installable.
6690 * systemd gained a new unit 'systemd-static-nodes.service'
6691 that generates static device nodes earlier during boot, and
6692 can run in conjunction with udev.
6694 * libsystemd-login gained a new call sd_pid_get_user_unit()
6695 to retrieve the user systemd unit a process is running
6696 in. This is useful for systems where systemd is used as
6699 * systemd-nspawn now places all containers in the new /machine
6700 top-level cgroup directory in the name=systemd
6701 hierarchy. libvirt will soon do the same, so that we get a
6702 uniform separation of /system, /user and /machine for system
6703 services, user processes and containers/virtual
6704 machines. This new cgroup hierarchy is also useful to stick
6705 stable names to specific container instances, which can be
6706 recognized later this way (this name may be controlled
6707 via systemd-nspawn's new -M switch). libsystemd-login also
6708 gained a new call sd_pid_get_machine_name() to retrieve the
6709 name of the container/VM a specific process belongs to.
6711 * bootchart can now store its data in the journal.
6713 * libsystemd-journal gained a new call
6714 sd_journal_add_conjunction() for AND expressions to the
6715 matching logic. This can be used to express more complex
6716 logical expressions.
6718 * journactl can now take multiple --unit= and --user-unit=
6721 * The cryptsetup logic now understands the "luks.key=" kernel
6722 command line switch for specifying a file to read the
6723 decryption key from. Also, if a configured key file is not
6724 found the tool will now automatically fall back to prompting
6727 * Python systemd.journal module was updated to wrap recently
6728 added functions from libsystemd-journal. The interface was
6729 changed to bring the low level interface in s.j._Reader
6730 closer to the C API, and the high level interface in
6731 s.j.Reader was updated to wrap and convert all data about
6734 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
6735 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
6736 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
6737 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
6738 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
6739 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6743 * journalctl --update-catalog now understands a new --root=
6744 option to operate on catalogs found in a different root
6747 * During shutdown after systemd has terminated all running
6748 services a final killing loop kills all remaining left-over
6749 processes. We will now print the name of these processes
6750 when we send SIGKILL to them, since this usually indicates a
6753 * If /etc/crypttab refers to password files stored on
6754 configured mount points automatic dependencies will now be
6755 generated to ensure the specific mount is established first
6756 before the key file is attempted to be read.
6758 * 'systemctl status' will now show information about the
6759 network sockets a socket unit is listening on.
6761 * 'systemctl status' will also shown information about any
6762 drop-in configuration file for units. (Drop-In configuration
6763 files in this context are files such as
6764 /etc/systemd/systemd/foobar.service.d/*.conf)
6766 * systemd-cgtop now optionally shows summed up CPU times of
6767 cgroups. Press '%' while running cgtop to switch between
6768 percentage and absolute mode. This is useful to determine
6769 which cgroups use up the most CPU time over the entire
6770 runtime of the system. systemd-cgtop has also been updated
6771 to be 'pipeable' for processing with further shell tools.
6773 * 'hostnamectl set-hostname' will now allow setting of FQDN
6776 * The formatting and parsing of time span values has been
6777 changed. The parser now understands fractional expressions
6778 such as "5.5h". The formatter will now output fractional
6779 expressions for all time spans under 1min, i.e. "5.123456s"
6780 rather than "5s 123ms 456us". For time spans under 1s
6781 millisecond values are shown, for those under 1ms
6782 microsecond values are shown. This should greatly improve
6783 all time-related output of systemd.
6785 * libsystemd-login and libsystemd-journal gained new
6786 functions for querying the poll() events mask and poll()
6787 timeout value for integration into arbitrary event
6790 * localectl gained the ability to list available X11 keymaps
6791 (models, layouts, variants, options).
6793 * 'systemd-analyze dot' gained the ability to filter for
6794 specific units via shell-style globs, to create smaller,
6795 more useful graphs. I.e. it is now possible to create simple
6796 graphs of all the dependencies between only target units, or
6797 of all units that Avahi has dependencies with.
6799 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
6800 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
6801 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
6802 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
6803 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
6804 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
6805 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
6809 * The boot-time readahead implementation for rotating media
6810 will now read the read-ahead data in multiple passes which
6811 consist of all read requests made in equidistant time
6812 intervals. This means instead of strictly reading read-ahead
6813 data in its physical order on disk we now try to find a
6814 middle ground between physical and access time order.
6816 * /etc/os-release files gained a new BUILD_ID= field for usage
6817 on operating systems that provide continuous builds of OS
6820 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
6821 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
6822 William Douglas, Zbigniew Jędrzejewski-Szmek
6826 * systemd-python gained an API exposing libsystemd-daemon.
6828 * The SMACK setup logic gained support for uploading CIPSO
6831 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
6832 ReadOnlyDirectories= and InaccessibleDirectories= has
6833 changed. The private /tmp and /var/tmp directories are now
6834 shared by all processes of a service (which means
6835 ExecStartPre= may now leave data in /tmp that ExecStart= of
6836 the same service can still access). When a service is
6837 stopped its temporary directories are immediately deleted
6838 (normal clean-up with tmpfiles is still done in addition to
6841 * By default, systemd will now set a couple of sysctl
6842 variables in the kernel: the safe sysrq options are turned
6843 on, IP route verification is turned on, and source routing
6844 disabled. The recently added hardlink and softlink
6845 protection of the kernel is turned on. These settings should
6846 be reasonably safe, and good defaults for all new systems.
6848 * The predictable network naming logic may now be turned off
6849 with a new kernel command line switch: net.ifnames=0.
6851 * A new libsystemd-bus module has been added that implements a
6852 pretty complete D-Bus client library. For details see:
6854 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
6856 * journald will now explicitly flush the journal files to disk
6857 at the latest 5min after each write. The file will then also
6858 be marked offline until the next write. This should increase
6859 reliability in case of a crash. The synchronization delay
6860 can be configured via SyncIntervalSec= in journald.conf.
6862 * There's a new remote-fs-setup.target unit that can be used
6863 to pull in specific services when at least one remote file
6864 system is to be mounted.
6866 * There are new targets timers.target and paths.target as
6867 canonical targets to pull user timer and path units in
6868 from. This complements sockets.target with a similar
6869 purpose for socket units.
6871 * libudev gained a new call udev_device_set_attribute_value()
6872 to set sysfs attributes of a device.
6874 * The udev daemon now sets the default number of worker
6875 processes executed in parallel based on the number of available
6876 CPUs instead of the amount of available RAM. This is supposed
6877 to provide a more reliable default and limit a too aggressive
6878 parallelism for setups with 1000s of devices connected.
6880 Contributions from: Auke Kok, Colin Walters, Cristian
6881 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
6882 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
6883 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
6884 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
6885 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
6886 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
6887 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
6888 Zbigniew Jędrzejewski-Szmek
6892 * Configuration of unit files may now be extended via drop-in
6893 files without having to edit/override the unit files
6894 themselves. More specifically, if the administrator wants to
6895 change one value for a service file foobar.service he can
6896 now do so by dropping in a configuration snippet into
6897 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
6898 will load all these snippets and apply them on top of the
6899 main unit configuration file, possibly extending or
6900 overriding its settings. Using these drop-in snippets is
6901 generally nicer than the two earlier options for changing
6902 unit files locally: copying the files from
6903 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
6904 them there; or creating a new file in /etc/systemd/system/
6905 that incorporates the original one via ".include". Drop-in
6906 snippets into these .d/ directories can be placed in any
6907 directory systemd looks for units in, and the usual
6908 overriding semantics between /usr/lib, /etc and /run apply
6911 * Most unit file settings which take lists of items can now be
6912 reset by assigning the empty string to them. For example,
6913 normally, settings such as Environment=FOO=BAR append a new
6914 environment variable assignment to the environment block,
6915 each time they are used. By assigning Environment= the empty
6916 string the environment block can be reset to empty. This is
6917 particularly useful with the .d/*.conf drop-in snippets
6918 mentioned above, since this adds the ability to reset list
6919 settings from vendor unit files via these drop-ins.
6921 * systemctl gained a new "list-dependencies" command for
6922 listing the dependencies of a unit recursively.
6924 * Inhibitors are now honored and listed by "systemctl
6925 suspend", "systemctl poweroff" (and similar) too, not only
6926 GNOME. These commands will also list active sessions by
6929 * Resource limits (as exposed by the various control group
6930 controllers) can now be controlled dynamically at runtime
6931 for all units. More specifically, you can now use a command
6932 like "systemctl set-cgroup-attr foobar.service cpu.shares
6933 2000" to alter the CPU shares a specific service gets. These
6934 settings are stored persistently on disk, and thus allow the
6935 administrator to easily adjust the resource usage of
6936 services with a few simple commands. This dynamic resource
6937 management logic is also available to other programs via the
6938 bus. Almost any kernel cgroup attribute and controller is
6941 * systemd-vconsole-setup will now copy all font settings to
6942 all allocated VTs, where it previously applied them only to
6945 * libsystemd-login gained the new sd_session_get_tty() API
6948 * This release drops support for a few legacy or
6949 distribution-specific LSB facility names when parsing init
6950 scripts: $x-display-manager, $mail-transfer-agent,
6951 $mail-transport-agent, $mail-transfer-agent, $smtp,
6952 $null. Also, the mail-transfer-agent.target unit backing
6953 this has been removed. Distributions which want to retain
6954 compatibility with this should carry the burden for
6955 supporting this themselves and patch support for these back
6956 in, if they really need to. Also, the facilities $syslog and
6957 $local_fs are now ignored, since systemd does not support
6958 early-boot LSB init scripts anymore, and these facilities
6959 are implied anyway for normal services. syslog.target has
6962 * There are new bus calls on PID1's Manager object for
6963 cancelling jobs, and removing snapshot units. Previously,
6964 both calls were only available on the Job and Snapshot
6967 * systemd-journal-gatewayd gained SSL support.
6969 * The various "environment" files, such as /etc/locale.conf
6970 now support continuation lines with a backslash ("\") as
6971 last character in the line, similarly in style (but different)
6972 to how this is supported in shells.
6974 * For normal user processes the _SYSTEMD_USER_UNIT= field is
6975 now implicitly appended to every log entry logged. systemctl
6976 has been updated to filter by this field when operating on a
6977 user systemd instance.
6979 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
6980 CAP_AUDIT_CONTROL capabilities to the capabilities set for
6981 the container. This makes it easier to boot unmodified
6982 Fedora systems in a container, which however still requires
6983 audit=0 to be passed on the kernel command line. Auditing in
6984 kernel and userspace is unfortunately still too broken in
6985 context of containers, hence we recommend compiling it out
6986 of the kernel or using audit=0. Hopefully this will be fixed
6987 one day for good in the kernel.
6989 * nspawn gained the new --bind= and --bind-ro= parameters to
6990 bind mount specific directories from the host into the
6993 * nspawn will now mount its own devpts file system instance
6994 into the container, in order not to leak pty devices from
6995 the host into the container.
6997 * systemd will now read the firmware boot time performance
6998 information from the EFI variables, if the used boot loader
6999 supports this, and takes it into account for boot performance
7000 analysis via "systemd-analyze". This is currently supported
7001 only in conjunction with Gummiboot, but could be supported
7002 by other boot loaders too. For details see:
7004 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
7006 * A new generator has been added that automatically mounts the
7007 EFI System Partition (ESP) to /boot, if that directory
7008 exists, is empty, and no other file system has been
7009 configured to be mounted there.
7011 * logind will now send out PrepareForSleep(false) out
7012 unconditionally, after coming back from suspend. This may be
7013 used by applications as asynchronous notification for
7014 system resume events.
7016 * "systemctl unlock-sessions" has been added, that allows
7017 unlocking the screens of all user sessions at once, similar
7018 to how "systemctl lock-sessions" already locked all users
7019 sessions. This is backed by a new D-Bus call UnlockSessions().
7021 * "loginctl seat-status" will now show the master device of a
7022 seat. (i.e. the device of a seat that needs to be around for
7023 the seat to be considered available, usually the graphics
7026 * tmpfiles gained a new "X" line type, that allows
7027 configuration of files and directories (with wildcards) that
7028 shall be excluded from automatic cleanup ("aging").
7030 * udev default rules set the device node permissions now only
7031 at "add" events, and do not change them any longer with a
7032 later "change" event.
7034 * The log messages for lid events and power/sleep keypresses
7035 now carry a message ID.
7037 * We now have a substantially larger unit test suite, but this
7038 continues to be work in progress.
7040 * udevadm hwdb gained a new --root= parameter to change the
7041 root directory to operate relative to.
7043 * logind will now issue a background sync() request to the kernel
7044 early at shutdown, so that dirty buffers are flushed to disk early
7045 instead of at the last moment, in order to optimize shutdown
7048 * A new bootctl tool has been added that is an interface for
7049 certain boot loader operations. This is currently a preview
7050 and is likely to be extended into a small mechanism daemon
7051 like timedated, localed, hostnamed, and can be used by
7052 graphical UIs to enumerate available boot options, and
7053 request boot into firmware operations.
7055 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
7056 the rest of the package. It also has been updated to work
7057 correctly in initrds.
7059 * Policykit previously has been runtime optional, and is now
7060 also compile time optional via a configure switch.
7062 * systemd-analyze has been reimplemented in C. Also "systemctl
7063 dot" has moved into systemd-analyze.
7065 * "systemctl status" with no further parameters will now print
7066 the status of all active or failed units.
7068 * Operations such as "systemctl start" can now be executed
7069 with a new mode "--irreversible" which may be used to queue
7070 operations that cannot accidentally be reversed by a later
7071 job queuing. This is by default used to make shutdown
7072 requests more robust.
7074 * The Python API of systemd now gained a new module for
7075 reading journal files.
7077 * A new tool kernel-install has been added that can install
7078 kernel images according to the Boot Loader Specification:
7080 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
7082 * Boot time console output has been improved to provide
7083 animated boot time output for hanging jobs.
7085 * A new tool systemd-activate has been added which can be used
7086 to test socket activation with, directly from the command
7087 line. This should make it much easier to test and debug
7088 socket activation in daemons.
7090 * journalctl gained a new "--reverse" (or -r) option to show
7091 journal output in reverse order (i.e. newest line first).
7093 * journalctl gained a new "--pager-end" (or -e) option to jump
7094 to immediately jump to the end of the journal in the
7095 pager. This is only supported in conjunction with "less".
7097 * journalctl gained a new "--user-unit=" option, that works
7098 similarly to "--unit=" but filters for user units rather than
7101 * A number of unit files to ease adoption of systemd in
7102 initrds has been added. This moves some minimal logic from
7103 the various initrd implementations into systemd proper.
7105 * The journal files are now owned by a new group
7106 "systemd-journal", which exists specifically to allow access
7107 to the journal, and nothing else. Previously, we used the
7108 "adm" group for that, which however possibly covers more
7109 than just journal/log file access. This new group is now
7110 already used by systemd-journal-gatewayd to ensure this
7111 daemon gets access to the journal files and as little else
7112 as possible. Note that "make install" will also set FS ACLs
7113 up for /var/log/journal to give "adm" and "wheel" read
7114 access to it, in addition to "systemd-journal" which owns
7115 the journal files. We recommend that packaging scripts also
7116 add read access to "adm" + "wheel" to /var/log/journal, and
7117 all existing/future journal files. To normal users and
7118 administrators little changes, however packagers need to
7119 ensure to create the "systemd-journal" system group at
7120 package installation time.
7122 * The systemd-journal-gatewayd now runs as unprivileged user
7123 systemd-journal-gateway:systemd-journal-gateway. Packaging
7124 scripts need to create these system user/group at
7127 * timedated now exposes a new boolean property CanNTP that
7128 indicates whether a local NTP service is available or not.
7130 * systemd-detect-virt will now also detect xen PVs
7132 * The pstore file system is now mounted by default, if it is
7135 * In addition to the SELinux and IMA policies we will now also
7136 load SMACK policies at early boot.
7138 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
7139 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
7140 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
7141 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
7142 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
7143 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
7144 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
7145 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
7146 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
7147 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
7148 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
7149 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
7150 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
7151 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
7155 * Timer units now support calendar time events in addition to
7156 monotonic time events. That means you can now trigger a unit
7157 based on a calendar time specification such as "Thu,Fri
7158 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
7159 or fifth day of any month of the year 2013, given that it is
7160 a thursday or friday. This brings timer event support
7161 considerably closer to cron's capabilities. For details on
7162 the supported calendar time specification language see
7165 * udev now supports a number of different naming policies for
7166 network interfaces for predictable names, and a combination
7167 of these policies is now the default. Please see this wiki
7168 document for details:
7170 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
7172 * Auke Kok's bootchart implementation has been added to the
7173 systemd tree. It is an optional component that can graph the
7174 boot in quite some detail. It is one of the best bootchart
7175 implementations around and minimal in its code and
7178 * nss-myhostname has been integrated into the systemd source
7179 tree. nss-myhostname guarantees that the local hostname
7180 always stays resolvable via NSS. It has been a weak
7181 requirement of systemd-hostnamed since a long time, and
7182 since its code is actually trivial we decided to just
7183 include it in systemd's source tree. It can be turned off
7184 with a configure switch.
7186 * The read-ahead logic is now capable of properly detecting
7187 whether a btrfs file system is on SSD or rotating media, in
7188 order to optimize the read-ahead scheme. Previously, it was
7189 only capable of detecting this on traditional file systems
7192 * In udev, additional device properties are now read from the
7193 IAB in addition to the OUI database. Also, Bluetooth company
7194 identities are attached to the devices as well.
7196 * In service files %U may be used as specifier that is
7197 replaced by the configured user name of the service.
7199 * nspawn may now be invoked without a controlling TTY. This
7200 makes it suitable for invocation as its own service. This
7201 may be used to set up a simple containerized server system
7202 using only core OS tools.
7204 * systemd and nspawn can now accept socket file descriptors
7205 when they are started for socket activation. This enables
7206 implementation of socket activated nspawn
7207 containers. i.e. think about autospawning an entire OS image
7208 when the first SSH or HTTP connection is received. We expect
7209 that similar functionality will also be added to libvirt-lxc
7212 * journalctl will now suppress ANSI color codes when
7213 presenting log data.
7215 * systemctl will no longer show control group information for
7216 a unit if the control group is empty anyway.
7218 * logind can now automatically suspend/hibernate/shutdown the
7221 * /etc/machine-info and hostnamed now also expose the chassis
7222 type of the system. This can be used to determine whether
7223 the local system is a laptop, desktop, handset or
7224 tablet. This information may either be configured by the
7225 user/vendor or is automatically determined from ACPI and DMI
7226 information if possible.
7228 * A number of PolicyKit actions are now bound together with
7229 "imply" rules. This should simplify creating UIs because
7230 many actions will now authenticate similar ones as well.
7232 * Unit files learnt a new condition ConditionACPower= which
7233 may be used to conditionalize a unit depending on whether an
7234 AC power source is connected or not, of whether the system
7235 is running on battery power.
7237 * systemctl gained a new "is-failed" verb that may be used in
7238 shell scripts and suchlike to check whether a specific unit
7239 is in the "failed" state.
7241 * The EnvironmentFile= setting in unit files now supports file
7242 globbing, and can hence be used to easily read a number of
7243 environment files at once.
7245 * systemd will no longer detect and recognize specific
7246 distributions. All distribution-specific #ifdeffery has been
7247 removed, systemd is now fully generic and
7248 distribution-agnostic. Effectively, not too much is lost as
7249 a lot of the code is still accessible via explicit configure
7250 switches. However, support for some distribution specific
7251 legacy configuration file formats has been dropped. We
7252 recommend distributions to simply adopt the configuration
7253 files everybody else uses now and convert the old
7254 configuration from packaging scripts. Most distributions
7255 already did that. If that's not possible or desirable,
7256 distributions are welcome to forward port the specific
7257 pieces of code locally from the git history.
7259 * When logging a message about a unit systemd will now always
7260 log the unit name in the message meta data.
7262 * localectl will now also discover system locale data that is
7263 not stored in locale archives, but directly unpacked.
7265 * logind will no longer unconditionally use framebuffer
7266 devices as seat masters, i.e. as devices that are required
7267 to be existing before a seat is considered preset. Instead,
7268 it will now look for all devices that are tagged as
7269 "seat-master" in udev. By default, framebuffer devices will
7270 be marked as such, but depending on local systems, other
7271 devices might be marked as well. This may be used to
7272 integrate graphics cards using closed source drivers (such
7273 as NVidia ones) more nicely into logind. Note however, that
7274 we recommend using the open source NVidia drivers instead,
7275 and no udev rules for the closed-source drivers will be
7276 shipped from us upstream.
7278 Contributions from: Adam Williamson, Alessandro Crismani, Auke
7279 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
7280 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
7281 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
7282 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
7283 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
7284 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
7285 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
7286 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
7287 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
7288 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
7289 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
7294 * udev gained support for loading additional device properties
7295 from an indexed database that is keyed by vendor/product IDs
7296 and similar device identifiers. For the beginning this
7297 "hwdb" is populated with data from the well-known PCI and
7298 USB database, but also includes PNP, ACPI and OID data. In
7299 the longer run this indexed database shall grow into
7300 becoming the one central database for non-essential
7301 userspace device metadata. Previously, data from the PCI/USB
7302 database was only attached to select devices, since the
7303 lookup was a relatively expensive operation due to O(n) time
7304 complexity (with n being the number of entries in the
7305 database). Since this is now O(1), we decided to add in this
7306 data for all devices where this is available, by
7307 default. Note that the indexed database needs to be rebuilt
7308 when new data files are installed. To achieve this you need
7309 to update your packaging scripts to invoke "udevadm hwdb
7310 --update" after installation of hwdb data files. For
7311 RPM-based distributions we introduced the new
7312 %udev_hwdb_update macro for this purpose.
7314 * The Journal gained support for the "Message Catalog", an
7315 indexed database to link up additional information with
7316 journal entries. For further details please check:
7318 https://www.freedesktop.org/wiki/Software/systemd/catalog
7320 The indexed message catalog database also needs to be
7321 rebuilt after installation of message catalog files. Use
7322 "journalctl --update-catalog" for this. For RPM-based
7323 distributions we introduced the %journal_catalog_update
7324 macro for this purpose.
7326 * The Python Journal bindings gained support for the standard
7327 Python logging framework.
7329 * The Journal API gained new functions for checking whether
7330 the underlying file system of a journal file is capable of
7331 properly reporting file change notifications, or whether
7332 applications that want to reflect journal changes "live"
7333 need to recheck journal files continuously in appropriate
7336 * It is now possible to set the "age" field for tmpfiles
7337 entries to 0, indicating that files matching this entry
7338 shall always be removed when the directories are cleaned up.
7340 * coredumpctl gained a new "gdb" verb which invokes gdb
7341 right-away on the selected coredump.
7343 * There's now support for "hybrid sleep" on kernels that
7344 support this, in addition to "suspend" and "hibernate". Use
7345 "systemctl hybrid-sleep" to make use of this.
7347 * logind's HandleSuspendKey= setting (and related settings)
7348 now gained support for a new "lock" setting to simply
7349 request the screen lock on all local sessions, instead of
7350 actually executing a suspend or hibernation.
7352 * systemd will now mount the EFI variables file system by
7355 * Socket units now gained support for configuration of the
7356 SMACK security label.
7358 * timedatectl will now output the time of the last and next
7359 daylight saving change.
7361 * We dropped support for various legacy and distro-specific
7362 concepts, such as insserv, early-boot SysV services
7363 (i.e. those for non-standard runlevels such as 'b' or 'S')
7364 or ArchLinux /etc/rc.conf support. We recommend the
7365 distributions who still need support this to either continue
7366 to maintain the necessary patches downstream, or find a
7367 different solution. (Talk to us if you have questions!)
7369 * Various systemd components will now bypass PolicyKit checks
7370 for root and otherwise handle properly if PolicyKit is not
7371 found to be around. This should fix most issues for
7372 PolicyKit-less systems. Quite frankly this should have been
7373 this way since day one. It is absolutely our intention to
7374 make systemd work fine on PolicyKit-less systems, and we
7375 consider it a bug if something does not work as it should if
7376 PolicyKit is not around.
7378 * For embedded systems it is now possible to build udev and
7379 systemd without blkid and/or kmod support.
7381 * "systemctl switch-root" is now capable of switching root
7382 more than once. I.e. in addition to transitions from the
7383 initrd to the host OS it is now possible to transition to
7384 further OS images from the host. This is useful to implement
7385 offline updating tools.
7387 * Various other additions have been made to the RPM macros
7388 shipped with systemd. Use %udev_rules_update() after
7389 installing new udev rules files. %_udevhwdbdir,
7390 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
7391 %_sysctldir are now available which resolve to the right
7392 directories for packages to place various data files in.
7394 * journalctl gained the new --full switch (in addition to
7395 --all, to disable ellipsation for long messages.
7397 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
7398 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
7399 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
7400 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
7401 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
7402 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
7403 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
7404 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
7405 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
7409 * journalctl gained new --since= and --until= switches to
7410 filter by time. It also now supports nice filtering for
7411 units via --unit=/-u.
7413 * Type=oneshot services may use ExecReload= and do the
7416 * The journal daemon now supports time-based rotation and
7417 vacuuming, in addition to the usual disk-space based
7420 * The journal will now index the available field values for
7421 each field name. This enables clients to show pretty drop
7422 downs of available match values when filtering. The bash
7423 completion of journalctl has been updated
7424 accordingly. journalctl gained a new switch -F to list all
7425 values a certain field takes in the journal database.
7427 * More service events are now written as structured messages
7428 to the journal, and made recognizable via message IDs.
7430 * The timedated, localed and hostnamed mini-services which
7431 previously only provided support for changing time, locale
7432 and hostname settings from graphical DEs such as GNOME now
7433 also have a minimal (but very useful) text-based client
7434 utility each. This is probably the nicest way to changing
7435 these settings from the command line now, especially since
7436 it lists available options and is fully integrated with bash
7439 * There's now a new tool "systemd-coredumpctl" to list and
7440 extract coredumps from the journal.
7442 * We now install a README each in /var/log/ and
7443 /etc/rc.d/init.d explaining where the system logs and init
7444 scripts went. This hopefully should help folks who go to
7445 that dirs and look into the otherwise now empty void and
7446 scratch their heads.
7448 * When user-services are invoked (by systemd --user) the
7449 $MANAGERPID env var is set to the PID of systemd.
7451 * SIGRTMIN+24 when sent to a --user instance will now result
7452 in immediate termination of systemd.
7454 * gatewayd received numerous feature additions such as a
7455 "follow" mode, for live syncing and filtering.
7457 * browse.html now allows filtering and showing detailed
7458 information on specific entries. Keyboard navigation and
7459 mouse screen support has been added.
7461 * gatewayd/journalctl now supports HTML5/JSON
7462 Server-Sent-Events as output.
7464 * The SysV init script compatibility logic will now
7465 heuristically determine whether a script supports the
7466 "reload" verb, and only then make this available as
7469 * "systemctl status --follow" has been removed, use "journalctl
7472 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
7473 have been removed since they are hardly useful to be
7476 * And I'd like to take the opportunity to specifically mention
7477 Zbigniew for his great contributions. Zbigniew, you rock!
7479 Contributions from: Andrew Eikum, Christian Hesse, Colin
7480 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
7481 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
7482 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
7483 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
7484 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
7485 Jędrzejewski-Szmek, Сковорода Никита Андреевич
7489 * If /etc/vconsole.conf is non-existent or empty we will no
7490 longer load any console font or key map at boot by
7491 default. Instead the kernel defaults will be left
7492 intact. This is definitely the right thing to do, as no
7493 configuration should mean no configuration, and hard-coding
7494 font names that are different on all archs is probably a bad
7495 idea. Also, the kernel default key map and font should be
7496 good enough for most cases anyway, and mostly identical to
7497 the userspace fonts/key maps we previously overloaded them
7498 with. If distributions want to continue to default to a
7499 non-kernel font or key map they should ship a default
7500 /etc/vconsole.conf with the appropriate contents.
7502 Contributions from: Colin Walters, Daniel J Walsh, Dave
7503 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
7504 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
7508 * journalctl gained a new --cursor= switch to show entries
7509 starting from the specified location in the journal.
7511 * We now enforce a size limit on journal entry fields exported
7512 with "-o json" in journalctl. Fields larger than 4K will be
7513 assigned null. This can be turned off with --all.
7515 * An (optional) journal gateway daemon is now available as
7516 "systemd-journal-gatewayd.service". This service provides
7517 access to the journal via HTTP and JSON. This functionality
7518 will be used to implement live log synchronization in both
7519 pull and push modes, but has various other users too, such
7520 as easy log access for debugging of embedded devices. Right
7521 now it is already useful to retrieve the journal via HTTP:
7523 # systemctl start systemd-journal-gatewayd.service
7524 # wget http://localhost:19531/entries
7526 This will download the journal contents in a
7527 /var/log/messages compatible format. The same as JSON:
7529 # curl -H"Accept: application/json" http://localhost:19531/entries
7531 This service is also accessible via a web browser where a
7532 single static HTML5 app is served that uses the JSON logic
7533 to enable the user to do some basic browsing of the
7534 journal. This will be extended later on. Here's an example
7535 screenshot of this app in its current state:
7537 http://0pointer.de/public/journal-gatewayd
7539 Contributions from: Kay Sievers, Lennart Poettering, Robert
7540 Milasan, Tom Gundersen
7544 * The bash completion logic is now available for journalctl
7547 * We do not mount the "cpuset" controller anymore together with
7548 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
7549 started if no parameters are assigned to it. "cpuset" hence
7550 broke code that assumed it could create "cpu" groups and
7553 * journalctl -f will now subscribe to terminal size changes,
7554 and line break accordingly.
7556 Contributions from: Dave Reisner, Kay Sievers, Lennart
7557 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
7561 * nspawn will now create a symlink /etc/localtime in the
7562 container environment, copying the host's timezone
7563 setting. Previously this has been done via a bind mount, but
7564 since symlinks cannot be bind mounted this has now been
7565 changed to create/update the appropriate symlink.
7567 * journalctl -n's line number argument is now optional, and
7568 will default to 10 if omitted.
7570 * journald will now log the maximum size the journal files may
7571 take up on disk. This is particularly useful if the default
7572 built-in logic of determining this parameter from the file
7573 system size is used. Use "systemctl status
7574 systemd-journald.service" to see this information.
7576 * The multi-seat X wrapper tool has been stripped down. As X
7577 is now capable of enumerating graphics devices via udev in a
7578 seat-aware way the wrapper is not strictly necessary
7579 anymore. A stripped down temporary stop-gap is still shipped
7580 until the upstream display managers have been updated to
7581 fully support the new X logic. Expect this wrapper to be
7582 removed entirely in one of the next releases.
7584 * HandleSleepKey= in logind.conf has been split up into
7585 HandleSuspendKey= and HandleHibernateKey=. The old setting
7586 is not available anymore. X11 and the kernel are
7587 distinguishing between these keys and we should too. This
7588 also means the inhibition lock for these keys has been split
7591 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
7592 Poettering, Lukas Nykryn, Václav Pavlín
7596 * Whenever a unit changes state we will now log this to the
7597 journal and show along the unit's own log output in
7600 * ConditionPathIsMountPoint= can now properly detect bind
7601 mount points too. (Previously, a bind mount of one file
7602 system to another place in the same file system could not be
7603 detected as mount, since they shared struct stat's st_dev
7606 * We will now mount the cgroup controllers cpu, cpuacct,
7607 cpuset and the controllers net_cls, net_prio together by
7610 * nspawn containers will now have a virtualized boot
7611 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
7612 over with a randomized ID at container initialization). This
7613 has the effect of making "journalctl -b" do the right thing
7616 * The JSON output journal serialization has been updated not
7617 to generate "endless" list objects anymore, but rather one
7618 JSON object per line. This is more in line how most JSON
7619 parsers expect JSON objects. The new output mode
7620 "json-pretty" has been added to provide similar output, but
7621 neatly aligned for readability by humans.
7623 * We dropped all explicit sync() invocations in the shutdown
7624 code. The kernel does this implicitly anyway in the kernel
7625 reboot() syscall. halt(8)'s -n option is now a compatibility
7628 * We now support virtualized reboot() in containers, as
7629 supported by newer kernels. We will fall back to exit() if
7630 CAP_SYS_REBOOT is not available to the container. Also,
7631 nspawn makes use of this now and will actually reboot the
7632 container if the containerized OS asks for that.
7634 * journalctl will only show local log output by default
7635 now. Use --merge (-m) to show remote log output, too.
7637 * libsystemd-journal gained the new sd_journal_get_usage()
7638 call to determine the current disk usage of all journal
7639 files. This is exposed in the new "journalctl --disk-usage"
7642 * journald gained a new configuration setting SplitMode= in
7643 journald.conf which may be used to control how user journals
7644 are split off. See journald.conf(5) for details.
7646 * A new condition type ConditionFileNotEmpty= has been added.
7648 * tmpfiles' "w" lines now support file globbing, to write
7649 multiple files at once.
7651 * We added Python bindings for the journal submission
7652 APIs. More Python APIs for a number of selected APIs will
7653 likely follow. Note that we intend to add native bindings
7654 only for the Python language, as we consider it common
7655 enough to deserve bindings shipped within systemd. There are
7656 various projects outside of systemd that provide bindings
7657 for languages such as PHP or Lua.
7659 * Many conditions will now resolve specifiers such as %i. In
7660 addition, PathChanged= and related directives of .path units
7661 now support specifiers as well.
7663 * There's now a new RPM macro definition for the system preset
7666 * journald will now warn if it ca not forward a message to the
7667 syslog daemon because its socket is full.
7669 * timedated will no longer write or process /etc/timezone,
7670 except on Debian. As we do not support late mounted /usr
7671 anymore /etc/localtime always being a symlink is now safe,
7672 and hence the information in /etc/timezone is not necessary
7675 * logind will now always reserve one VT for a text getty (VT6
7676 by default). Previously if more than 6 X sessions where
7677 started they took up all the VTs with auto-spawned gettys,
7678 so that no text gettys were available anymore.
7680 * udev will now automatically inform the btrfs kernel logic
7681 about btrfs RAID components showing up. This should make
7682 simple hotplug based btrfs RAID assembly work.
7684 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
7685 (but not for its children which will stay at the kernel
7686 default). This should allow setups with a lot more listening
7689 * systemd will now always pass the configured timezone to the
7690 kernel at boot. timedated will do the same when the timezone
7693 * logind's inhibition logic has been updated. By default,
7694 logind will now handle the lid switch, the power and sleep
7695 keys all the time, even in graphical sessions. If DEs want
7696 to handle these events on their own they should take the new
7697 handle-power-key, handle-sleep-key and handle-lid-switch
7698 inhibitors during their runtime. A simple way to achieve
7699 that is to invoke the DE wrapped in an invocation of:
7701 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
7703 * Access to unit operations is now checked via SELinux taking
7704 the unit file label and client process label into account.
7706 * systemd will now notify the administrator in the journal
7707 when he over-mounts a non-empty directory.
7709 * There are new specifiers that are resolved in unit files,
7710 for the host name (%H), the machine ID (%m) and the boot ID
7713 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
7714 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
7715 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
7716 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
7717 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
7718 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
7719 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
7723 * Support for reading structured kernel messages from
7724 /dev/kmsg has now been added and is enabled by default.
7726 * Support for reading kernel messages from /proc/kmsg has now
7727 been removed. If you want kernel messages in the journal
7728 make sure to run a recent kernel (>= 3.5) that supports
7729 reading structured messages from /dev/kmsg (see
7730 above). /proc/kmsg is now exclusive property of classic
7731 syslog daemons again.
7733 * The libudev API gained the new
7734 udev_device_new_from_device_id() call.
7736 * The logic for file system namespace (ReadOnlyDirectory=,
7737 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
7738 require pivot_root() anymore. This means fewer temporary
7739 directories are created below /tmp for this feature.
7741 * nspawn containers will now see and receive all submounts
7742 made on the host OS below the root file system of the
7745 * Forward Secure Sealing is now supported for Journal files,
7746 which provide cryptographical sealing of journal files so
7747 that attackers cannot alter log history anymore without this
7748 being detectable. Lennart will soon post a blog story about
7749 this explaining it in more detail.
7751 * There are two new service settings RestartPreventExitStatus=
7752 and SuccessExitStatus= which allow configuration of exit
7753 status (exit code or signal) which will be excepted from the
7754 restart logic, resp. consider successful.
7756 * journalctl gained the new --verify switch that can be used
7757 to check the integrity of the structure of journal files and
7758 (if Forward Secure Sealing is enabled) the contents of
7761 * nspawn containers will now be run with /dev/stdin, /dev/fd/
7762 and similar symlinks pre-created. This makes running shells
7763 as container init process a lot more fun.
7765 * The fstab support can now handle PARTUUID= and PARTLABEL=
7768 * A new ConditionHost= condition has been added to match
7769 against the hostname (with globs) and machine ID. This is
7770 useful for clusters where a single OS image is used to
7771 provision a large number of hosts which shall run slightly
7772 different sets of services.
7774 * Services which hit the restart limit will now be placed in a
7777 Contributions from: Bertram Poettering, Dave Reisner, Huang
7778 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
7779 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
7783 * When running in --user mode systemd will now become a
7784 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
7785 tree a lot more organized.
7787 * A new PartOf= unit dependency type has been introduced that
7788 may be used to group services in a natural way.
7790 * "systemctl enable" may now be used to enable instances of
7793 * journalctl now prints error log levels in red, and
7794 warning/notice log levels in bright white. It also supports
7795 filtering by log level now.
7797 * cgtop gained a new -n switch (similar to top), to configure
7798 the maximum number of iterations to run for. It also gained
7799 -b, to run in batch mode (accepting no input).
7801 * The suffix ".service" may now be omitted on most systemctl
7802 command lines involving service unit names.
7804 * There's a new bus call in logind to lock all sessions, as
7805 well as a loginctl verb for it "lock-sessions".
7807 * libsystemd-logind.so gained a new call sd_journal_perror()
7808 that works similar to libc perror() but logs to the journal
7809 and encodes structured information about the error number.
7811 * /etc/crypttab entries now understand the new keyfile-size=
7814 * shutdown(8) now can send a (configurable) wall message when
7815 a shutdown is cancelled.
7817 * The mount propagation mode for the root file system will now
7818 default to "shared", which is useful to make containers work
7819 nicely out-of-the-box so that they receive new mounts from
7820 the host. This can be undone locally by running "mount
7821 --make-rprivate /" if needed.
7823 * The prefdm.service file has been removed. Distributions
7824 should maintain this unit downstream if they intend to keep
7825 it around. However, we recommend writing normal unit files
7826 for display managers instead.
7828 * Since systemd is a crucial part of the OS we will now
7829 default to a number of compiler switches that improve
7830 security (hardening) such as read-only relocations, stack
7831 protection, and suchlike.
7833 * The TimeoutSec= setting for services is now split into
7834 TimeoutStartSec= and TimeoutStopSec= to allow configuration
7835 of individual time outs for the start and the stop phase of
7838 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
7839 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
7840 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
7841 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
7842 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
7843 Gundersen, Zbigniew Jędrzejewski-Szmek
7847 * The journal and id128 C APIs are now fully documented as man
7850 * Extra safety checks have been added when transitioning from
7851 the initial RAM disk to the main system to avoid accidental
7854 * /etc/crypttab entries now understand the new keyfile-offset=
7857 * systemctl -t can now be used to filter by unit load state.
7859 * The journal C API gained the new sd_journal_wait() call to
7860 make writing synchronous journal clients easier.
7862 * journalctl gained the new -D switch to show journals from a
7865 * journalctl now displays a special marker between log
7866 messages of two different boots.
7868 * The journal is now explicitly flushed to /var via a service
7869 systemd-journal-flush.service, rather than implicitly simply
7870 by seeing /var/log/journal to be writable.
7872 * journalctl (and the journal C APIs) can now match for much
7873 more complex expressions, with alternatives and
7876 * When transitioning from the initial RAM disk to the main
7877 system we will now kill all processes in a killing spree to
7878 ensure no processes stay around by accident.
7880 * Three new specifiers may be used in unit files: %u, %h, %s
7881 resolve to the user name, user home directory resp. user
7882 shell. This is useful for running systemd user instances.
7884 * We now automatically rotate journal files if their data
7885 object hash table gets a fill level > 75%. We also size the
7886 hash table based on the configured maximum file size. This
7887 together should lower hash collisions drastically and thus
7888 speed things up a bit.
7890 * journalctl gained the new "--header" switch to introspect
7891 header data of journal files.
7893 * A new setting SystemCallFilters= has been added to services
7894 which may be used to apply blacklists or whitelists to
7895 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
7897 * nspawn gained a new --link-journal= switch (and quicker: -j)
7898 to link the container journal with the host. This makes it
7899 very easy to centralize log viewing on the host for all
7900 guests while still keeping the journal files separated.
7902 * Many bugfixes and optimizations
7904 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
7905 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
7906 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
7911 * Several tools now understand kernel command line arguments,
7912 which are only read when run in an initial RAM disk. They
7913 usually follow closely their normal counterparts, but are
7916 * There's a new tool to analyze the readahead files that are
7917 automatically generated at boot. Use:
7919 /usr/lib/systemd/systemd-readahead analyze /.readahead
7921 * We now provide an early debug shell on tty9 if this enabled. Use:
7923 systemctl enable debug-shell.service
7925 * All plymouth related units have been moved into the Plymouth
7926 package. Please make sure to upgrade your Plymouth version
7929 * systemd-tmpfiles now supports getting passed the basename of
7930 a configuration file only, in which case it will look for it
7931 in all appropriate directories automatically.
7933 * udevadm info now takes a /dev or /sys path as argument, and
7934 does the right thing. Example:
7936 udevadm info /dev/sda
7937 udevadm info /sys/class/block/sda
7939 * systemctl now prints a warning if a unit is stopped but a
7940 unit that might trigger it continues to run. Example: a
7941 service is stopped but the socket that activates it is left
7944 * "systemctl status" will now mention if the log output was
7945 shortened due to rotation since a service has been started.
7947 * The journal API now exposes functions to determine the
7948 "cutoff" times due to rotation.
7950 * journald now understands SIGUSR1 and SIGUSR2 for triggering
7951 immediately flushing of runtime logs to /var if possible,
7952 resp. for triggering immediate rotation of the journal
7955 * It is now considered an error if a service is attempted to
7956 be stopped that is not loaded.
7958 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
7960 * systemd-analyze now supports Python 3
7962 * tmpfiles now supports cleaning up directories via aging
7963 where the first level dirs are always kept around but
7964 directories beneath it automatically aged. This is enabled
7965 by prefixing the age field with '~'.
7967 * Seat objects now expose CanGraphical, CanTTY properties
7968 which is required to deal with very fast bootups where the
7969 display manager might be running before the graphics drivers
7970 completed initialization.
7972 * Seat objects now expose a State property.
7974 * We now include RPM macros for service enabling/disabling
7975 based on the preset logic. We recommend RPM based
7976 distributions to make use of these macros if possible. This
7977 makes it simpler to reuse RPM spec files across
7980 * We now make sure that the collected systemd unit name is
7981 always valid when services log to the journal via
7984 * There's a new man page kernel-command-line(7) detailing all
7985 command line options we understand.
7987 * The fstab generator may now be disabled at boot by passing
7988 fstab=0 on the kernel command line.
7990 * A new kernel command line option modules-load= is now understood
7991 to load a specific kernel module statically, early at boot.
7993 * Unit names specified on the systemctl command line are now
7994 automatically escaped as needed. Also, if file system or
7995 device paths are specified they are automatically turned
7996 into the appropriate mount or device unit names. Example:
7998 systemctl status /home
7999 systemctl status /dev/sda
8001 * The SysVConsole= configuration option has been removed from
8002 system.conf parsing.
8004 * The SysV search path is no longer exported on the D-Bus
8007 * The Names= option has been removed from unit file parsing.
8009 * There's a new man page bootup(7) detailing the boot process.
8011 * Every unit and every generator we ship with systemd now
8012 comes with full documentation. The self-explanatory boot is
8015 * A couple of services gained "systemd-" prefixes in their
8016 name if they wrap systemd code, rather than only external
8017 code. Among them fsck@.service which is now
8018 systemd-fsck@.service.
8020 * The HaveWatchdog property has been removed from the D-Bus
8023 * systemd.confirm_spawn= on the kernel command line should now
8026 * There's a new man page crypttab(5) which details all options
8027 we actually understand.
8029 * systemd-nspawn gained a new --capability= switch to pass
8030 additional capabilities to the container.
8032 * timedated will now read known NTP implementation unit names
8033 from /usr/lib/systemd/ntp-units.d/*.list,
8034 systemd-timedated-ntp.target has been removed.
8036 * journalctl gained a new switch "-b" that lists log data of
8037 the current boot only.
8039 * The notify socket is in the abstract namespace again, in
8040 order to support daemons which chroot() at start-up.
8042 * There is a new Storage= configuration option for journald
8043 which allows configuration of where log data should go. This
8044 also provides a way to disable journal logging entirely, so
8045 that data collected is only forwarded to the console, the
8046 kernel log buffer or another syslog implementation.
8048 * Many bugfixes and optimizations
8050 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
8051 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
8052 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
8053 Shawn Landden, Tom Gundersen
8057 * "systemctl help <unit>" now shows the man page if one is
8060 * Several new man pages have been added.
8062 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
8063 MaxLevelConsole= can now be specified in
8064 journald.conf. These options allow reducing the amount of
8065 data stored on disk or forwarded by the log level.
8067 * TimerSlackNSec= can now be specified in system.conf for
8068 PID1. This allows system-wide power savings.
8070 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
8071 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
8076 * logind is now capable of (optionally) handling power and
8077 sleep keys as well as the lid switch.
8079 * journalctl now understands the syntax "journalctl
8080 /usr/bin/avahi-daemon" to get all log output of a specific
8083 * CapabilityBoundingSet= in system.conf now also influences
8084 the capability bound set of usermode helpers of the kernel.
8086 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
8087 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
8088 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
8089 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
8093 * Note that we skipped 139 releases here in order to set the
8094 new version to something that is greater than both udev's
8095 and systemd's most recent version number.
8097 * udev: all udev sources are merged into the systemd source tree now.
8098 All future udev development will happen in the systemd tree. It
8099 is still fully supported to use the udev daemon and tools without
8100 systemd running, like in initramfs or other init systems. Building
8101 udev though, will require the *build* of the systemd tree, but
8102 udev can be properly *run* without systemd.
8104 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
8105 should be used to create dead device nodes as workarounds for broken
8108 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
8109 no longer supported. udev_monitor_new_from_netlink() needs to be
8110 used to subscribe to events.
8112 * udev: when udevd is started by systemd, processes which are left
8113 behind by forking them off of udev rules, are unconditionally cleaned
8114 up and killed now after the event handling has finished. Services or
8115 daemons must be started as systemd services. Services can be
8116 pulled-in by udev to get started, but they can no longer be directly
8117 forked by udev rules.
8119 * udev: the daemon binary is called systemd-udevd now and installed
8120 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
8121 to adapt to that, create symlink, or rename the binary after building
8124 * libudev no longer provides these symbols:
8125 udev_monitor_from_socket()
8126 udev_queue_get_failed_list_entry()
8127 udev_get_{dev,sys,run}_path()
8128 The versions number was bumped and symbol versioning introduced.
8130 * systemd-loginctl and systemd-journalctl have been renamed
8131 to loginctl and journalctl to match systemctl.
8133 * The config files: /etc/systemd/systemd-logind.conf and
8134 /etc/systemd/systemd-journald.conf have been renamed to
8135 logind.conf and journald.conf. Package updates should rename
8136 the files to the new names on upgrade.
8138 * For almost all files the license is now LGPL2.1+, changed
8139 from the previous GPL2.0+. Exceptions are some minor stuff
8140 of udev (which will be changed to LGPL2.1 eventually, too),
8141 and the MIT licensed sd-daemon.[ch] library that is suitable
8142 to be used as drop-in files.
8144 * systemd and logind now handle system sleep states, in
8145 particular suspending and hibernating.
8147 * logind now implements a sleep/shutdown/idle inhibiting logic
8148 suitable for a variety of uses. Soonishly Lennart will blog
8149 about this in more detail.
8151 * var-run.mount and var-lock.mount are no longer provided
8152 (which previously bind mounted these directories to their new
8153 places). Distributions which have not converted these
8154 directories to symlinks should consider stealing these files
8155 from git history and add them downstream.
8157 * We introduced the Documentation= field for units and added
8158 this to all our shipped units. This is useful to make it
8159 easier to explore the boot and the purpose of the various
8162 * All smaller setup units (such as
8163 systemd-vconsole-setup.service) now detect properly if they
8164 are run in a container and are skipped when
8165 appropriate. This guarantees an entirely noise-free boot in
8166 Linux container environments such as systemd-nspawn.
8168 * A framework for implementing offline system updates is now
8169 integrated, for details see:
8170 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
8172 * A new service type Type=idle is available now which helps us
8173 avoiding ugly interleaving of getty output and boot status
8176 * There's now a system-wide CapabilityBoundingSet= option to
8177 globally reduce the set of capabilities for the
8178 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
8179 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
8180 even CAP_NET_ADMIN system-wide for secure systems.
8182 * There are now system-wide DefaultLimitXXX= options to
8183 globally change the defaults of the various resource limits
8184 for all units started by PID 1.
8186 * Harald Hoyer's systemd test suite has been integrated into
8187 systemd which allows easy testing of systemd builds in qemu
8188 and nspawn. (This is really awesome! Ask us for details!)
8190 * The fstab parser is now implemented as generator, not inside
8193 * systemctl will now warn you if .mount units generated from
8194 /etc/fstab are out of date due to changes in fstab that
8195 have not been read by systemd yet.
8197 * systemd is now suitable for usage in initrds. Dracut has
8198 already been updated to make use of this. With this in place
8199 initrds get a slight bit faster but primarily are much
8200 easier to introspect and debug since "systemctl status" in
8201 the host system can be used to introspect initrd services,
8202 and the journal from the initrd is kept around too.
8204 * systemd-delta has been added, a tool to explore differences
8205 between user/admin configuration and vendor defaults.
8207 * PrivateTmp= now affects both /tmp and /var/tmp.
8209 * Boot time status messages are now much prettier and feature
8210 proper english language. Booting up systemd has never been
8213 * Read-ahead pack files now include the inode number of all
8214 files to pre-cache. When the inode changes the pre-caching
8215 is not attempted. This should be nicer to deal with updated
8216 packages which might result in changes of read-ahead
8219 * We now temporaritly lower the kernel's read_ahead_kb variable
8220 when collecting read-ahead data to ensure the kernel's
8221 built-in read-ahead does not add noise to our measurements
8222 of necessary blocks to pre-cache.
8224 * There's now RequiresMountsFor= to add automatic dependencies
8225 for all mounts necessary for a specific file system path.
8227 * MountAuto= and SwapAuto= have been removed from
8228 system.conf. Mounting file systems at boot has to take place
8231 * nspawn now learned a new switch --uuid= to set the machine
8232 ID on the command line.
8234 * nspawn now learned the -b switch to automatically search
8237 * vt102 is now the default TERM for serial TTYs, upgraded from
8240 * systemd-logind now works on VT-less systems.
8242 * The build tree has been reorganized. The individual
8243 components now have directories of their own.
8245 * A new condition type ConditionPathIsReadWrite= is now available.
8247 * nspawn learned the new -C switch to create cgroups for the
8248 container in other hierarchies.
8250 * We now have support for hardware watchdogs, configurable in
8253 * The scheduled shutdown logic now has a public API.
8255 * We now mount /tmp as tmpfs by default, but this can be
8256 masked and /etc/fstab can override it.
8258 * Since udisks does not make use of /media anymore we are not
8259 mounting a tmpfs on it anymore.
8261 * journalctl gained a new --local switch to only interleave
8262 locally generated journal files.
8264 * We can now load the IMA policy at boot automatically.
8266 * The GTK tools have been split off into a systemd-ui.
8268 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
8269 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
8270 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
8271 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
8272 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
8273 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
8274 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
8275 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
8276 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
8281 * This is mostly a bugfix release
8283 * Support optional initialization of the machine ID from the
8284 KVM or container configured UUID.
8286 * Support immediate reboots with "systemctl reboot -ff"
8288 * Show /etc/os-release data in systemd-analyze output
8290 * Many bugfixes for the journal, including endianness fixes and
8291 ensuring that disk space enforcement works
8293 * sd-login.h is C++ compatible again
8295 * Extend the /etc/os-release format on request of the Debian
8298 * We now refuse non-UTF8 strings used in various configuration
8299 and unit files. This is done to ensure we do not pass invalid
8300 data over D-Bus or expose it elsewhere.
8302 * Register Mimo USB Screens as suitable for automatic seat
8305 * Read SELinux client context from journal clients in a race
8308 * Reorder configuration file lookup order. /etc now always
8309 overrides /run in order to allow the administrator to always
8310 and unconditionally override vendor-supplied or
8311 automatically generated data.
8313 * The various user visible bits of the journal now have man
8314 pages. We still lack man pages for the journal API calls
8317 * We now ship all man pages in HTML format again in the
8320 Contributions from: Dave Reisner, Dirk Eibach, Frederic
8321 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
8322 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
8327 * This is mostly a bugfix release
8329 * systems lacking /etc/os-release are no longer supported.
8331 * Various functionality updates to libsystemd-login.so
8333 * Track class of PAM logins to distinguish greeters from
8336 Contributions from: Kay Sievers, Lennart Poettering, Michael
8341 * This is an important bugfix release for v41.
8343 * Building man pages is now optional which should be useful
8344 for those building systemd from git but unwilling to install
8347 * Watchdog support for supervising services is now usable. In
8348 a future release support for hardware watchdogs
8349 (i.e. /dev/watchdog) will be added building on this.
8351 * Service start rate limiting is now configurable and can be
8352 turned off per service. When a start rate limit is hit a
8353 reboot can automatically be triggered.
8355 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
8357 Contributions from: Benjamin Franzke, Bill Nottingham,
8358 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
8359 Schmidt, Michał Górny, Piotr Drąg
8363 * The systemd binary is installed /usr/lib/systemd/systemd now;
8364 An existing /sbin/init symlink needs to be adapted with the
8367 * The code that loads kernel modules has been ported to invoke
8368 libkmod directly, instead of modprobe. This means we do not
8369 support systems with module-init-tools anymore.
8371 * Watchdog support is now already useful, but still not
8374 * A new kernel command line option systemd.setenv= is
8375 understood to set system wide environment variables
8376 dynamically at boot.
8378 * We now limit the set of capabilities of systemd-journald.
8380 * We now set SIGPIPE to ignore by default, since it only is
8381 useful in shell pipelines, and has little use in general
8382 code. This can be disabled with IgnoreSIPIPE=no in unit
8385 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
8386 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
8391 * This is mostly a bugfix release
8393 * We now expose the reason why a service failed in the
8394 "Result" D-Bus property.
8396 * Rudimentary service watchdog support (will be completed over
8397 the next few releases.)
8399 * When systemd forks off in order execute some service we will
8400 now immediately changes its argv[0] to reflect which process
8401 it will execute. This is useful to minimize the time window
8402 with a generic argv[0], which makes bootcharts more useful
8404 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
8405 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
8406 Mike Kazantsev, Ray Strode
8410 * This is mostly a test release, but incorporates many
8413 * New systemd-cgtop tool to show control groups by their
8416 * Linking against libacl for ACLs is optional again. If
8417 disabled, support tracking device access for active logins
8418 goes becomes unavailable, and so does access to the user
8419 journals by the respective users.
8421 * If a group "adm" exists, journal files are automatically
8422 owned by them, thus allow members of this group full access
8423 to the system journal as well as all user journals.
8425 * The journal now stores the SELinux context of the logging
8426 client for all entries.
8428 * Add C++ inclusion guards to all public headers
8430 * New output mode "cat" in the journal to print only text
8431 messages, without any meta data like date or time.
8433 * Include tiny X server wrapper as a temporary stop-gap to
8434 teach XOrg udev display enumeration. This is used by display
8435 managers such as gdm, and will go away as soon as XOrg
8436 learned native udev hotplugging for display devices.
8438 * Add new systemd-cat tool for executing arbitrary programs
8439 with STDERR/STDOUT connected to the journal. Can also act as
8440 BSD logger replacement, and does so by default.
8442 * Optionally store all locally generated coredumps in the
8443 journal along with meta data.
8445 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
8446 writing short strings to files (for usage for /sys), and for
8447 creating symlinks, character and block device nodes.
8449 * New unit file option ControlGroupPersistent= to make cgroups
8450 persistent, following the mechanisms outlined in
8451 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
8453 * Support multiple local RTCs in a sane way
8455 * No longer monopolize IO when replaying readahead data on
8456 rotating disks, since we might starve non-file-system IO to
8457 death, since fanotify() will not see accesses done by blkid,
8460 * Do not show kernel threads in systemd-cgls anymore, unless
8461 requested with new -k switch.
8463 Contributions from: Dan Horák, Kay Sievers, Lennart
8464 Poettering, Michal Schmidt
8468 * This is mostly a test release, but incorporates many
8471 * The git repository moved to:
8472 git://anongit.freedesktop.org/systemd/systemd
8473 ssh://git.freedesktop.org/git/systemd/systemd
8475 * First release with the journal
8476 http://0pointer.de/blog/projects/the-journal.html
8478 * The journal replaces both systemd-kmsg-syslogd and
8479 systemd-stdout-bridge.
8481 * New sd_pid_get_unit() API call in libsystemd-logind
8483 * Many systemadm clean-ups
8485 * Introduce remote-fs-pre.target which is ordered before all
8486 remote mounts and may be used to start services before all
8489 * Added Mageia support
8491 * Add bash completion for systemd-loginctl
8493 * Actively monitor PID file creation for daemons which exit in
8494 the parent process before having finished writing the PID
8495 file in the daemon process. Daemons which do this need to be
8496 fixed (i.e. PID file creation must have finished before the
8497 parent exits), but we now react a bit more gracefully to them.
8499 * Add colourful boot output, mimicking the well-known output
8500 of existing distributions.
8502 * New option PassCredentials= for socket units, for
8503 compatibility with a recent kernel ABI breakage.
8505 * /etc/rc.local is now hooked in via a generator binary, and
8506 thus will no longer act as synchronization point during
8509 * systemctl list-unit-files now supports --root=.
8511 * systemd-tmpfiles now understands two new commands: z, Z for
8512 relabelling files according to the SELinux database. This is
8513 useful to apply SELinux labels to specific files in /sys,
8516 * Output of SysV services is now forwarded to both the console
8517 and the journal by default, not only just the console.
8519 * New man pages for all APIs from libsystemd-login.
8521 * The build tree got reorganized and the build system is a
8522 lot more modular allowing embedded setups to specifically
8523 select the components of systemd they are interested in.
8525 * Support for Linux systems lacking the kernel VT subsystem is
8528 * configure's --with-rootdir= got renamed to
8529 --with-rootprefix= to follow the naming used by udev and
8532 * Unless specified otherwise we will now install to /usr instead
8533 of /usr/local by default.
8535 * Processes with '@' in argv[0][0] are now excluded from the
8536 final shut-down killing spree, following the logic explained
8538 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
8540 * All processes remaining in a service cgroup when we enter
8541 the START or START_PRE states are now killed with
8542 SIGKILL. That means it is no longer possible to spawn
8543 background processes from ExecStart= lines (which was never
8544 supported anyway, and bad style).
8546 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
8547 reloading of units together.
8549 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
8550 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
8551 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
8552 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
8553 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
~ianmdlvl / elogind.git / blob