1 systemd System and Service Manager
3 CHANGES WITH 239 in spe:
5 * NETWORK INTERFACE DEVICE NAMING CHANGES: elogind-udevd's "net_id"
6 builtin will name network interfaces differently than in previous
7 versions for virtual network interfaces created with SR-IOV and NPAR
8 and for devices where the PCI network controller device does not have
9 a slot number associated.
11 SR-IOV virtual devices are now named based on the name of the parent
12 interface, with a suffix of "v<N>", where <N> is the virtual device
13 number. Previously those virtual devices were named as if completely
16 The ninth and later NPAR virtual devices will be named following the
17 scheme used for the first eight NPAR partitions. Previously those
18 devices were not renamed and the kernel default (eth<n>) was used.
20 "net_id" will also generate names for PCI devices where the PCI
21 network controller device does not have an associated slot number
22 itself, but one of its parents does. Previously those devices were
23 not renamed and the kernel default (eth<n>) was used.
25 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
26 elogind.service. Since v235, IPAddressDeny=any has been set to
27 the unit. So, it is expected that the default behavior of
28 elogind is not changed. However, if distribution packagers or
29 administrators disabled or modified IPAddressDeny= setting by a
30 drop-in config file, then it may be necessary to update the file to
31 re-enable AF_INET and AF_INET6 to support network user name services,
34 * When the RestrictNamespaces= unit property is specified multiple
35 times, then the specified types are merged now. Previously, only the
36 last assignment was used. So, if distribution packagers or
37 administrators modified the setting by a drop-in config file, then it
38 may be necessary to update the file.
40 * When OnFailure= is used in combination with Restart= on a service
41 unit, then the specified units will no longer be triggered on
42 failures that result in restarting. Previously, the specified units
43 would be activated each time the unit failed, even when the unit was
44 going to be restarted automatically. This behaviour contradicted the
45 documentation. With this release the code is adjusted to match the
48 * elogind-tmpfiles will now print a notice whenever it encounters
49 tmpfiles.d/ lines referencing the /var/run/ directory. It will
50 recommend reworking them to use the /run/ directory instead (for
51 which /var/run/ is simply a symlinked compatibility alias). This way
52 elogind-tmpfiles can properly detect line conflicts and merge lines
53 referencing the same file by two paths, without having to access
56 * systemctl disable/unmask/preset/preset-all cannot be used with
57 --runtime. Previously this was allowed, but resulted in unintuitive
58 behaviour that wasn't useful. systemctl disable/unmask will now undo
59 both runtime and persistent enablement/masking, i.e. it will remove
60 any relevant symlinks both in /run and /etc.
62 * Note that all long-running system services shipped with elogind will
63 now default to a system call whitelist (rather than a blacklist, as
64 before). In particular, elogind-udevd will now enforce one too. For
65 most cases this should be safe, however downstream distributions
66 which disabled sandboxing of elogind-udevd (specifically the
67 MountFlags= setting), might want to disable this security feature
68 too, as the default whitelisting will prohibit all mount, swap,
69 reboot and clock changing operations from udev rules.
71 * sd-boot acquired new loader configuration settings to optionally turn
72 off Windows and MacOS boot partition discovery as well as
73 reboot-into-firmware menu items. It is also able to pick a better
74 screen resolution for HiDPI systems, and now provides loader
75 configuration settings to change the resolution explicitly.
77 * The elogind-resolve tool has been renamed to resolvectl (it also
78 * elogind-resolved now supports DNS-over-TLS. It's still
79 turned off by default, use DNSOverTLS=opportunistic to turn it on in
80 resolved.conf. We intend to make this the default as soon as couple
81 of additional techniques for optimizing the initial latency caused by
82 establishing a TLS/TCP connection are implemented.
84 * elogind-resolved.service and elogind-networkd.service now set
85 DynamicUser=yes. The users elogind-resolve and elogind-network are
86 not created by elogind-sysusers.
88 remains available under the old name, for compatibility), and its
89 interface is now verb-based, similar in style to the other <xyz>ctl
90 tools, such as systemctl or loginctl.
92 * The resolvectl/elogind-resolve tool also provides 'resolvconf'
93 compatibility. It may be symlinked under the 'resolvconf' name, in
94 which case it will take arguments and input compatible with the
95 Debian and FreeBSD resolvconf tool.
97 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
98 where the system initially suspends, and after a time-out resumes and
101 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
102 set the client will only send a DUID as client identifier.
104 * The nss-elogind glibc NSS module will now enumerate dynamic users and
105 groups in effect. Previously, it could resolve UIDs/GIDs to user
106 names/groups and vice versa, but did not support enumeration.
108 * journald's Compress= configuration setting now optionally accepts a
109 byte threshold value. All journal objects larger than this threshold
110 will be compressed, smaller ones will not. Previously this threshold
111 was not configurable and set to 512.
113 * A new system.conf setting NoNewPrivileges= is now available which may
114 be used to turn off acquisition of new privileges system-wide
115 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
116 for all its children). Note that turning this option on means setuid
117 binaries and file system capabilities lose their special powers.
118 While turning on this option is a big step towards a more secure
119 system, doing so is likely to break numerous pre-existing UNIX tools,
120 in particular su and sudo.
122 * A new service elogind-time-sync-wait.service has been added. If
123 enabled it will delay the time-sync.target unit at boot until time
124 synchronization has been received from the network. This
125 functionality is useful on systems lacking a local RTC or where it is
126 acceptable that the boot process shall be delayed by external network
129 * When hibernating, elogind will now inform the kernel of the image
130 write offset, on kernels new enough to support this. This means swap
131 files should work for hibernation now.
133 * When loading unit files, elogind will now look for drop-in unit files
134 extensions in additional places. Previously, for a unit file name
135 "foo-bar-baz.service" it would look for dropin files in
136 "foo-bar-baz.service.d/*.conf". Now, it will also look in
137 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
138 service name truncated after all inner dashes. This scheme allows
139 writing drop-ins easily that apply to a whole set of unit files at
140 once. It's particularly useful for mount and slice units (as their
141 naming is prefix based), but is also useful for service and other
142 units, for packages that install multiple unit files at once,
143 following a strict naming regime of beginning the unit file name with
144 the package's name. Two new specifiers are now supported in unit
145 files to match this: %j and %J are replaced by the part of the unit
146 name following the last dash.
148 * Unit files and other configuration files that support specifier
149 expansion now understand another three new specifiers: %T and %V will
150 resolve to /tmp and /var/tmp respectively, or whatever temporary
151 directory has been set for the calling user. %E will expand to either
152 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
154 * The ExecStart= lines of unit files are no longer required to
155 reference absolute paths. If non-absolute paths are specified the
156 specified binary name is searched within the service manager's
157 built-in $PATH, which may be queried with 'elogind-path
158 search-binaries-default'. It's generally recommended to continue to
159 use absolute paths for all binaries specified in unit files.
161 * Units gained a new load state "bad-setting", which is used when a
162 unit file was loaded, but contained fatal errors which prevent it
163 from being started (for example, a service unit has been defined
164 lacking both ExecStart= and ExecStop= lines).
166 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
167 support alternative debuggers, for example lldb. The old name
168 continues to be available however, for compatibility reasons. Use the
169 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
170 to pick an alternative debugger instead of the default gdb.
172 * systemctl and the other tools will now output escape sequences that
173 generate proper clickable hyperlinks in various terminal emulators
174 where useful (for example, in the "systemctl status" output you can
175 now click on the unit file name to quickly open it in the
176 editor/viewer of your choice). Note that not all terminal emulators
177 support this functionality yet, but many do. Unfortunately, the
178 "less" pager doesn't support this yet, hence this functionality is
179 currently automatically turned off when a pager is started (which
180 happens quite often due to auto-paging). We hope to remove this
181 limitation as soon as "less" learns these escape sequences. This new
182 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
183 environment variable. For details on these escape sequences see:
184 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
186 * networkd's .network files now support a new IPv6MTUBytes= option for
187 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
188 option in the [Route] section to configure the MTU to use for
189 specific routes. It also gained support for configuration of the DHCP
190 "UserClass" option through the new UserClass= setting. It gained
191 three new options in the new [CAN] section for configuring CAN
192 networks. The MULTICAST and ALLMULTI interface flags may now be
193 controlled explicitly with the new Multicast= and AllMulticast=
196 * networkd will now automatically make use of the kernel's route
197 expiration feature, if it is available.
199 * udevd's .link files now support setting the number of receive and
200 transmit channels, using the RxChannels=, TxChannels=,
201 OtherChannels=, CombinedChannels= settings.
203 * Support for UDPSegmentationOffload= has been removed, given its
204 limited support in hardware, and waning software support.
206 * networkd's .netdev files now support creating "netdevsim" interfaces.
208 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
209 to query the unit belonging to a specific kernel control group.
211 * elogind-analyze gained a new verb "cat-config", which may be used to
212 dump the contents of any configuration file, with all its matching
213 drop-in files added in, and honouring the usual search and masking
214 logic applied to elogind configuration files. For example use
215 "elogind-analyze cat-config elogind/system.conf" to get the complete
216 system configuration file of elogind how it would be loaded by PID 1
217 itself. Similar to this, various tools such as elogind-tmpfiles or
218 elogind-sysusers, gained a new option "--cat-config", which does the
219 corresponding operation for their own configuration settings. For
220 example, "elogind-tmpfiles --cat-config" will now output the full
221 list of tmpfiles.d/ lines in place.
223 * timedatectl gained two new verbs "timesync-status" (to show the
224 current NTP synchronization state of elogind-timesyncd) and
225 "show-timesync" (to show bus properties of elogind-timesyncd).
227 * elogind-timesyncd gained a bus interface on which it exposes details
230 * elogind-nspawn gained a new --rlimit= switch for setting initial
231 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
232 understood by elogind-timedated. It takes a colon-separated list of
233 unit names of NTP client services. The list is used by
234 "timedatectl set-ntp".
236 resource limits for the container payload. There's a new switch
237 --hostname= to explicitly override the container's hostname. A new
238 --no-new-privileges= switch may be used to control the
239 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
240 --oom-score-adjust= switch controls the OOM scoring adjustment value
241 for the payload. The new --cpu-affinity= switch controls the CPU
242 affinity of the container payload. The new --resolv-conf= switch
243 allows more detailed control of /etc/resolv.conf handling of the
244 container. Similarly, the new --timezone= switch allows more detailed
245 control of /etc/localtime handling of the container.
247 * elogind-detect-virt gained a new --list switch, which will print a
248 list of all currently known VM and container environments.
250 * Support for "Portable Services" has been added, see
251 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
252 experimental, but this is expected to change soon. Reflecting this
253 experimental state, the "portablectl" binary is not installed into
254 /usr/bin yet. The binary has to be called with the full path
255 /usr/lib/elogind/portablectl instead.
257 * journalctl's and systemctl's -o switch now knows a new log output
258 mode "with-unit". The output it generates is very similar to the
259 regular "short" mode, but displays the unit name instead of the
260 syslog tag for each log line. Also, the date is shown with timezone
261 information. This mode is probably more useful than the classic
262 "short" output mode for most purposes, except where pixel-perfect
263 compatibility with classic /var/log/messages formatting is required.
265 * A new --dump-bus-properties switch has been added to the elogind
266 binary, which may be used to dump all supported D-Bus properties.
267 (Options which are still supported, but are deprecated, are *not*
270 * sd-bus gained a set of new calls:
271 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
272 enable/disable the "floating" state of a bus slot object,
273 i.e. whether the slot object pins the bus it is allocated for into
274 memory or if the bus slot object gets disconnected when the bus goes
275 away. sd_bus_open_with_description(),
276 sd_bus_open_user_with_description(),
277 sd_bus_open_system_with_description() may be used to allocate bus
278 objects and set their description string already during allocation.
280 * sd-event gained support for watching inotify events from the event
281 loop, in an efficient way, sharing inotify handles between multiple
282 users. For this a new function sd_event_add_inotify() has been added.
284 * sd-event and sd-bus gained support for calling special user-supplied
285 destructor functions for userdata pointers associated with
286 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
287 functions sd_bus_slot_set_destroy_callback,
288 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
289 sd_bus_track_get_destroy_callback,
290 sd_event_source_set_destroy_callback,
291 sd_event_source_get_destroy_callback have been added.
293 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
295 * PID 1 will now automatically reschedule .timer units whenever the
296 local timezone changes. (They previously got rescheduled
297 automatically when the system clock changed.)
299 * New documentation has been added to document cgroups delegation,
300 portable services and the various code quality tools we have set up:
302 https://github.com/systemd/systemd/blob/master/doc/CGROUP_DELEGATION.md
303 https://github.com/systemd/systemd/blob/master/doc/PORTABLE_SERVICES.md
304 https://github.com/systemd/systemd/blob/master/doc/CODE_QUALITY.md
306 * The Boot Loader Specification has been added to the source tree.
308 https://github.com/systemd/systemd/blob/master/doc/BOOT_LOADER_SPECIFICATION.md
310 While moving it into our source tree we have updated it and further
311 changes are now accepted through the usual github PR workflow.
313 * pam_elogind will now look for PAM userdata fields elogind.memory_max,
314 elogind.tasks_max, elogind.cpu_weight, elogind.io_weight set by
315 earlier PAM modules. The data in these fields is used to initialize
316 the session scope's resource properties. Thus external PAM modules
317 may now configure per-session limits, for example sourced from
318 external user databases.
320 * socket units with Accept=yes will now maintain a "refused" counter in
321 addition to the existing "accepted" counter, counting connections
322 refused due to the enforced limits.
324 * The "elogind-path search-binaries-default" command may now be use to
325 query the default, built-in $PATH PID 1 will pass to the services it
328 * A new unit file setting PrivateMounts= has been added. It's a boolean
329 option. If enabled the unit's processes are invoked in their own file
330 system namespace. Note that this behaviour is also implied if any
331 other file system namespacing options (such as PrivateTmp=,
332 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
333 primarily useful for services that do not use any of the other file
334 system namespacing options. One such service is elogind-udevd.service
335 wher this is now used by default.
337 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
338 when the system is booted in UEFI "secure mode".
340 * A new unit "system-update-pre.target" is added, which defines an
341 optional synchronization point for offline system updates, as
342 implemented by the pre-existing "system-update.target" unit. It
343 allows ordering services before the service that executes the actual
344 update process in a generic way.
346 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
347 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
348 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
349 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
350 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
351 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
352 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
353 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
354 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
355 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
356 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
357 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
358 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
359 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
360 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
361 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
362 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
363 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
364 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
365 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
366 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
367 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
368 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
369 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
370 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
371 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
372 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
373 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
374 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
380 * The MemoryAccounting= unit property now defaults to on. After
381 discussions with the upstream control group maintainers we learnt
382 that the negative impact of cgroup memory accounting on current
383 kernels is finally relatively minimal, so that it should be safe to
384 enable this by default without affecting system performance. Besides
385 memory accounting only task accounting is turned on by default, all
386 other forms of resource accounting (CPU, IO, IP) remain off for now,
387 because it's not clear yet that their impact is small enough to move
388 from opt-in to opt-out. We recommend downstreams to leave memory
389 accounting on by default if kernel 4.14 or higher is primarily
390 used. On very resource constrained systems or when support for old
391 kernels is a necessity, -Dmemory-accounting-default=false can be used
392 to revert this change.
394 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
395 %udev_rules_update) and the journal catalog (%journal_catalog_update)
396 from the upgrade scriptlets of individual packages now do nothing.
397 Transfiletriggers have been added which will perform those updates
398 once at the end of the transaction.
400 Similar transfiletriggers have been added to execute any sysctl.d
401 and binfmt.d rules. Thus, it should be unnecessary to provide any
402 scriptlets to execute this configuration from package installation
405 * elogind-sysusers gained a mode where the configuration to execute is
406 specified on the command line, but this configuration is not executed
407 directly, but instead it is merged with the configuration on disk,
408 and the result is executed. This is useful for package installation
409 scripts which want to create the user before installing any files on
410 disk (in case some of those files are owned by that user), while
411 still allowing local admin overrides.
413 This functionality is exposed to rpm scriptlets through a new
414 %sysusers_create_package macro. Old %sysusers_create and
415 %sysusers_create_inline macros are deprecated.
417 A transfiletrigger for sysusers.d configuration is now installed,
418 which means that it should be unnecessary to call elogind-sysusers from
419 package installation scripts, unless the package installs any files
420 owned by those newly-created users, in which case
421 %sysusers_create_package should be used.
423 * Analogous change has been done for elogind-tmpfiles: it gained a mode
424 where the command-line configuration is merged with the configuration
425 on disk. This is exposed as the new %tmpfiles_create_package macro,
426 and %tmpfiles_create is deprecated. A transfiletrigger is installed
427 for tmpfiles.d, hence it should be unnecessary to call elogind-tmpfiles
428 from package installation scripts.
430 * sysusers.d configuration for a user may now also specify the group
431 number, in addition to the user number ("u username 123:456"), or
432 without the user number ("u username -:456").
434 * Configution items for elogind-sysusers can now be specified as
435 positional arguments when the new --inline switch is used.
437 * The login shell of users created through sysusers.d may now be
438 specified (previously, it was always /bin/sh for root and
439 /sbin/nologin for other users).
441 * elogind-analyze gained a new --global switch to look at global user
442 configuration. It also gained a unit-paths verb to list the unit load
443 paths that are compiled into elogind (which can be used with
444 --elogind, --user, or --global).
446 * udevadm trigger gained a new --settle/-w option to wait for any
447 triggered events to finish (but just those, and not any other events
448 which are triggered meanwhile).
450 * The action that elogind takes when the lid is closed and the
451 machine is connected to external power can now be configured using
452 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
453 was determined by HandleLidSwitch=, and, for backwards compatibility,
454 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
456 * journalctl will periodically call sd_journal_process() to make it
457 resilient against inotify queue overruns when journal files are
458 rotated very quickly.
460 * Two new functions in libelogind — sd_bus_get_n_queued_read and
461 sd_bus_get_n_queued_write — may be used to check the number of
462 pending bus messages.
464 * elogind gained a new
465 org.freedesktop.elogind1.Manager.AttachProcessesToUnit dbus call
466 which can be used to migrate foreign processes to scope and service
467 units. The primary user for this new API is elogind itself: the
468 elogind --user instance uses this call of the elogind --system
469 instance to migrate processes if it itself gets the request to
470 migrate processes and the kernel refuses this due to access
471 restrictions. Thanks to this "elogind-run --scope --user …" works
472 again in pure cgroups v2 environments when invoked from the user
475 * A new TemporaryFileSystem= setting can be used to mask out part of
476 the real file system tree with tmpfs mounts. This may be combined
477 with BindPaths= and BindReadOnlyPaths= to hide files or directories
478 not relevant to the unit, while still allowing some paths lower in
479 the tree to be accessed.
481 ProtectHome=tmpfs may now be used to hide user home and runtime
482 directories from units, in a way that is mostly equivalent to
483 "TemporaryFileSystem=/home /run/user /root".
485 * Non-service units are now started with KeyringMode=shared by default.
486 This means that mount and swapon and other mount tools have access
487 to keys in the main keyring.
489 * /sys/fs/bpf is now mounted automatically.
491 * QNX virtualization is now detected by elogind-detect-virt and may
492 be used in ConditionVirtualization=.
494 * IPAccounting= may now be enabled also for slice units.
496 * A new -Dsplit-bin= build configuration switch may be used to specify
497 whether bin and sbin directories are merged, or if they should be
498 included separately in $PATH and various listings of executable
499 directories. The build configuration scripts will try to autodetect
500 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
501 system, but distributions are encouraged to configure this
504 * A new -Dok-color= build configuration switch may be used to change
505 the colour of "OK" status messages.
507 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
508 PrivateNetwork=yes was buggy in previous versions of elogind. This
509 means that after the upgrade and daemon-reexec, any such units must
512 * INCOMPATIBILITY: as announced in the NEWS for 237, elogind-tmpfiles
513 will not exclude read-only files owned by root from cleanup.
515 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
516 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
517 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
518 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
519 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
520 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
521 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
522 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
523 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
524 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
525 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
526 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
527 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
528 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
529 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
530 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
536 * Some keyboards come with a zoom see-saw or rocker which until now got
537 mapped to the Linux "zoomin/out" keys in hwdb. However, these
538 keycodes are not recognized by any major desktop. They now produce
539 Up/Down key events so that they can be used for scrolling.
541 * INCOMPATIBILITY: elogind-tmpfiles' "f" lines changed behaviour
542 slightly: previously, if an argument was specified for lines of this
543 type (i.e. the right-most column was set) this string was appended to
544 existing files each time elogind-tmpfiles was run. This behaviour was
545 different from what the documentation said, and not particularly
546 useful, as repeated elogind-tmpfiles invocations would not be
547 idempotent and grow such files without bounds. With this release
548 behaviour has been altered slightly, to match what the documentation
549 says: lines of this type only have an effect if the indicated files
550 don't exist yet, and only then the argument string is written to the
553 * FUTURE INCOMPATIBILITY: In elogind v238 we intend to slightly change
554 elogind-tmpfiles behaviour: previously, read-only files owned by root
555 were always excluded from the file "aging" algorithm (i.e. the
556 automatic clean-up of directories like /tmp based on
557 atime/mtime/ctime). We intend to drop this restriction, and age files
558 by default even when owned by root and read-only. This behaviour was
559 inherited from older tools, but there have been requests to remove
560 it, and it's not obvious why this restriction was made in the first
561 place. Please speak up now, if you are aware of software that reqires
562 this behaviour, otherwise we'll remove the restriction in v238.
564 * A new environment variable $SYSTEMD_OFFLINE is now understood by
565 systemctl. It takes a boolean argument. If on, systemctl assumes it
566 operates on an "offline" OS tree, and will not attempt to talk to the
567 service manager. Previously, this mode was implicitly enabled if a
568 chroot() environment was detected, and this new environment variable
569 now provides explicit control.
571 * .path and .socket units may now be created transiently, too.
572 Previously only service, mount, automount and timer units were
573 supported as transient units. The elogind-run tool has been updated
574 to expose this new functionality, you may hence use it now to bind
575 arbitrary commands to path or socket activation on-the-fly from the
576 command line. Moreover, almost all properties are now exposed for the
577 unit types that already supported transient operation.
579 * The elogind-mount command gained support for a new --owner= parameter
580 which takes a user name, which is then resolved and included in uid=
581 and gid= mount options string of the file system to mount.
583 * A new unit condition ConditionControlGroupController= has been added
584 that checks whether a specific cgroup controller is available.
586 * Unit files, udev's .link files, and elogind-networkd's .netdev and
587 .network files all gained support for a new condition
588 ConditionKernelVersion= for checking against specific kernel
591 * In elogind-networkd, the [IPVLAN] section in .netdev files gained
592 support for configuring device flags in the Flags= setting. In the
593 same files, the [Tunnel] section gained support for configuring
594 AllowLocalRemote=. The [Route] section in .network files gained
595 support for configuring InitialCongestionWindow=,
596 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
597 understands RapidCommit=.
599 * elogind-networkd's DHCPv6 support gained support for Prefix
602 * sd-bus gained support for a new "watch-bind" feature. When this
603 feature is enabled, an sd_bus connection may be set up to connect to
604 an AF_UNIX socket in the file system as soon as it is created. This
605 functionality is useful for writing early-boot services that
606 automatically connect to the system bus as soon as it is started,
607 without ugly time-based polling. elogind-networkd and
608 elogind-resolved have been updated to make use of this
609 functionality. busctl exposes this functionality in a new
610 --watch-bind= command line switch.
612 * sd-bus will now optionally synthesize a local "Connected" signal as
613 soon as a D-Bus connection is set up fully. This message mirrors the
614 already existing "Disconnected" signal which is synthesized when the
615 connection is terminated. This signal is generally useful but
616 particularly handy in combination with the "watch-bind" feature
617 described above. Synthesizing of this message has to be requested
618 explicitly through the new API call sd_bus_set_connected_signal(). In
619 addition a new call sd_bus_is_ready() has been added that checks
620 whether a connection is fully set up (i.e. between the "Connected" and
621 "Disconnected" signals).
623 * sd-bus gained two new calls sd_bus_request_name_async() and
624 sd_bus_release_name_async() for asynchronously registering bus
625 names. Similar, there is now sd_bus_add_match_async() for installing
626 a signal match asynchronously. All of elogind's own services have
627 been updated to make use of these calls. Doing these operations
628 asynchronously has two benefits: it reduces the risk of deadlocks in
629 case of cyclic dependencies between bus services, and it speeds up
630 service initialization since synchronization points for bus
631 round-trips are removed.
633 * sd-bus gained two new calls sd_bus_match_signal() and
634 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
635 and sd_bus_add_match_async() but instead of taking a D-Bus match
636 string take match fields as normal function parameters.
638 * sd-bus gained two new calls sd_bus_set_sender() and
639 sd_bus_message_set_sender() for setting the sender name of outgoing
640 messages (either for all outgoing messages or for just one specific
641 one). These calls are only useful in direct connections as on
642 brokered connections the broker fills in the sender anyway,
643 overwriting whatever the client filled in.
645 * sd-event gained a new pseudo-handle that may be specified on all API
646 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
647 used this refers to the default event loop object of the calling
648 thread. Note however that this does not implicitly allocate one —
649 which has to be done prior by using sd_event_default(). Similarly
650 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
651 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
652 to the default bus of the specified type of the calling thread. Here
653 too this does not implicitly allocate bus connection objects, this
654 has to be done prior with sd_bus_default() and friends.
656 * sd-event gained a new call pair
657 sd_event_source_{get|set}_io_fd_own(). This may be used to request
658 automatic closure of the file descriptor an IO event source watches
659 when the event source is destroyed.
661 * elogind-networkd gained support for natively configuring WireGuard
664 * In previous versions elogind synthesized user records both for the
665 "nobody" (UID 65534) and "root" (UID 0) users in nss-elogind and
666 internally. In order to simplify distribution-wide renames of the
667 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
668 new transitional flag file has been added: if
669 /etc/elogind/dont-synthesize-nobody exists synthesizing of the 65534
670 user and group record within the elogind codebase is disabled.
672 * elogind-notify gained a new --uid= option for selecting the source
673 user/UID to use for notification messages sent to the service
676 * journalctl gained a new --grep= option to list only entries in which
677 the message matches a certain pattern. By default matching is case
678 insensitive if the pattern is lowercase, and case sensitive
679 otherwise. Option --case-sensitive=yes|no can be used to override
680 this an specify case sensitivity or case insensitivity.
682 * There's now a "elogind-analyze service-watchdogs" command for printing
683 the current state of the service runtime watchdog, and optionally
684 enabling or disabling the per-service watchdogs system-wide if given a
685 boolean argument (i.e. the concept you configure in WatchdogSec=), for
686 debugging purposes. There's also a kernel command line option
687 elogind.service_watchdogs= for controlling the same.
689 * Two new "log-level" and "log-target" options for elogind-analyze were
690 added that merge the now deprecated get-log-level, set-log-level and
691 get-log-target, set-log-target pairs. The deprecated options are still
692 understood for backwards compatibility. The two new options print the
693 current value when no arguments are given, and set them when a
694 level/target is given as an argument.
696 * sysusers.d's "u" lines now optionally accept both a UID and a GID
697 specification, separated by a ":" character, in order to create users
698 where UID and GID do not match.
700 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
701 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
702 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
703 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
704 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
705 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
706 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
707 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
708 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
709 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
710 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
711 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
712 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
713 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
714 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
715 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
716 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
717 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
718 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
719 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
726 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
727 in v235 has been extended to also set the dummy.ko module option
728 numdummies=0, preventing the kernel from automatically creating
729 dummy0. All dummy interfaces must now be explicitly created.
731 * Unknown '%' specifiers in configuration files are now rejected. This
732 applies to units and tmpfiles.d configuration. Any percent characters
733 that are followed by a letter or digit that are not supposed to be
734 interpreted as the beginning of a specifier should be escaped by
735 doubling ("%%"). (So "size=5%" is still accepted, as well as
736 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
737 valid specifiers today.)
739 * systemd-resolved now maintains a new dynamic
740 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
741 recommended to make /etc/resolv.conf a symlink to it. This file
742 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
743 includes dynamically acquired search domains, achieving more correct
744 DNS resolution by software that bypasses local DNS APIs such as NSS.
746 * The "uaccess" udev tag has been dropped from /dev/kvm and
747 /dev/dri/renderD*. These devices now have the 0666 permissions by
748 default (but this may be changed at build-time). /dev/dri/renderD*
749 will now be owned by the "render" group along with /dev/kfd.
751 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
752 systemd-journal-gatewayd.service and
753 systemd-journal-upload.service. This means "nss-systemd" must be
754 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
755 services are resolved properly.
757 * In /etc/fstab two new mount options are now understood:
758 x-systemd.makefs and x-systemd.growfs. The former has the effect that
759 the configured file system is formatted before it is mounted, the
760 latter that the file system is resized to the full block device size
761 after it is mounted (i.e. if the file system is smaller than the
762 partition it resides on, it's grown). This is similar to the fsck
763 logic in /etc/fstab, and pulls in systemd-makefs@.service and
764 systemd-growfs@.service as necessary, similar to
765 systemd-fsck@.service. Resizing is currently only supported on ext4
768 * In systemd-networkd, the IPv6 RA logic now optionally may announce
769 DNS server and domain information.
771 * Support for the LUKS2 on-disk format for encrypted partitions has
772 been added. This requires libcryptsetup2 during compilation and
775 * The systemd --user instance will now signal "readiness" when its
776 basic.target unit has been reached, instead of when the run queue ran
777 empty for the first time.
779 * Tmpfiles.d with user configuration are now also supported.
780 systemd-tmpfiles gained a new --user switch, and snippets placed in
781 ~/.config/user-tmpfiles.d/ and corresponding directories will be
782 executed by systemd-tmpfiles --user running in the new
783 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
784 running in the user session.
786 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
787 %S resolves to the top-level state directory (/var/lib for the system
788 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
789 top-level cache directory (/var/cache for the system instance,
790 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
791 logs directory (/var/log for the system instance,
792 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
793 existing %t specifier, that resolves to the top-level runtime
794 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
797 * journalctl learnt a new parameter --output-fields= for limiting the
798 set of journal fields to output in verbose and JSON output modes.
800 * systemd-timesyncd's configuration file gained a new option
801 RootDistanceMaxSec= for setting the maximum root distance of servers
802 it'll use, as well as the new options PollIntervalMinSec= and
803 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
805 * bootctl gained a new command "list" for listing all available boot
806 menu items on systems that follow the boot loader specification.
808 * systemctl gained a new --dry-run switch that shows what would be done
809 instead of doing it, and is currently supported by the shutdown and
812 * ConditionSecurity= can now detect the TOMOYO security module.
814 * Unit file [Install] sections are now also respected in unit drop-in
815 files. This is intended to be used by drop-ins under /usr/lib/.
817 * systemd-firstboot may now also set the initial keyboard mapping.
819 * Udev "changed" events for devices which are exposed as systemd
820 .device units are now propagated to units specified in
821 ReloadPropagatedFrom= as reload requests.
823 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
824 unit template name (i.e. a name in the form of 'foobar@.service',
825 without the instance component between the '@' and - the '.'), then
826 the escaped sysfs path of the device is automatically used as the
829 * SystemCallFilter= in unit files has been extended so that an "errno"
830 can be specified individually for each system call. Example:
831 SystemCallFilter=~uname:EILSEQ.
833 * The cgroup delegation logic has been substantially updated. Delegate=
834 now optionally takes a list of controllers (instead of a boolean, as
835 before), which lists the controllers to delegate at least.
837 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
839 * A new LogLevelMax= setting configures the maximum log level any
840 process of the service may log at (i.e. anything with a lesser
841 priority than what is specified is automatically dropped). A new
842 LogExtraFields= setting allows configuration of additional journal
843 fields to attach to all log records generated by any of the unit's
846 * New StandardInputData= and StandardInputText= settings along with the
847 new option StandardInput=data may be used to configure textual or
848 binary data that shall be passed to the executed service process via
849 standard input, encoded in-line in the unit file.
851 * StandardInput=, StandardOutput= and StandardError= may now be used to
852 connect stdin/stdout/stderr of executed processes directly with a
853 file or AF_UNIX socket in the file system, using the new "file:" option.
855 * A new unit file option CollectMode= has been added, that allows
856 tweaking the garbage collection logic for units. It may be used to
857 tell systemd to garbage collect units that have failed automatically
858 (normally it only GCs units that exited successfully). systemd-run
859 and systemd-mount expose this new functionality with a new -G option.
861 * "machinectl bind" may now be used to bind mount non-directories
862 (i.e. regularfiles, devices, fifos, sockets).
864 * systemd-analyze gained a new verb "calendar" for validating and
865 testing calendar time specifications to use for OnCalendar= in timer
866 units. Besides validating the expression it will calculate the next
867 time the specified expression would elapse.
869 * In addition to the pre-existing FailureAction= unit file setting
870 there's now SuccessAction=, for configuring a shutdown action to
871 execute when a unit completes successfully. This is useful in
872 particular inside containers that shall terminate after some workload
873 has been completed. Also, both options are now supported for all unit
874 types, not just services.
876 * networkds's IP rule support gained two new options
877 IncomingInterface= and OutgoingInterface= for configuring the incoming
878 and outgoing interfaces of configured rules. systemd-networkd also
879 gained support for "vxcan" network devices.
881 * networkd gained a new setting RequiredForOnline=, taking a
882 boolean. If set, systemd-wait-online will take it into consideration
883 when determining that the system is up, otherwise it will ignore the
884 interface for this purpose.
886 * The sd_notify() protocol gained support for a new operation: with
887 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
888 store again, ahead of POLLHUP or POLLERR when they are removed
891 * A new document doc/UIDS-GIDS.md has been added to the source tree,
892 that documents the UID/GID range and assignment assumptions and
893 requirements of systemd.
895 * The watchdog device PID 1 will ping may now be configured through the
896 WatchdogDevice= configuration file setting, or by setting the
897 systemd.watchdog_service= kernel commandline option.
899 * systemd-resolved's gained support for registering DNS-SD services on
900 the local network using MulticastDNS. Services may either be
901 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
902 the same dir below /run, /usr/lib), or through its D-Bus API.
904 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
905 extend the effective start, runtime, and stop time. The service must
906 continue to send EXTEND_TIMEOUT_USEC within the period specified to
907 prevent the service manager from making the service as timedout.
909 * elogind-resolved's DNSSEC support gained support for RFC 8080
910 (Ed25519 keys and signatures).
912 * The elogind-resolve command line tool gained a new set of options
913 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
914 --set-nta= and --revert to configure per-interface DNS configuration
915 dynamically during runtime. It's useful for pushing DNS information
916 into elogind-resolved from DNS hook scripts that various interface
917 managing software supports (such as pppd).
919 * elogind-nspawn gained a new --network-namespace-path= command line
920 option, which may be used to make a container join an existing
921 network namespace, by specifying a path to a "netns" file.
923 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
924 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
925 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
926 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
927 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
928 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
929 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
930 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
931 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
932 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
933 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
934 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
935 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
936 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
937 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
938 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
939 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
940 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
941 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
942 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
943 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
944 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
945 Jędrzejewski-Szmek, Zeal Jagannatha
951 * INCOMPATIBILITY: systemd-logind.service and other long-running
952 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
953 communication with the outside. This generally improves security of
954 the system, and is in almost all cases a safe and good choice, as
955 these services do not and should not provide any network-facing
956 functionality. However, systemd-logind uses the glibc NSS API to
957 query the user database. This creates problems on systems where NSS
958 is set up to directly consult network services for user database
959 lookups. In particular, this creates incompatibilities with the
960 "nss-nis" module, which attempts to directly contact the NIS/YP
961 network servers it is configured for, and will now consistently
962 fail. In such cases, it is possible to turn off IP sandboxing for
963 systemd-logind.service (set IPAddressDeny= in its [Service] section
964 to the empty string, via a .d/ unit file drop-in). Downstream
965 distributions might want to update their nss-nis packaging to include
966 such a drop-in snippet, accordingly, to hide this incompatibility
967 from the user. Another option is to make use of glibc's nscd service
968 to proxy such network requests through a privilege-separated, minimal
969 local caching daemon, or to switch to more modern technologies such
970 sssd, whose NSS hook-ups generally do not involve direct network
971 access. In general, we think it's definitely time to question the
972 implementation choices of nss-nis, i.e. whether it's a good idea
973 today to embed a network-facing loadable module into all local
974 processes that need to query the user database, including the most
975 trivial and benign ones, such as "ls". For more details about
976 IPAddressDeny= see below.
978 * A new modprobe.d drop-in is now shipped by default that sets the
979 bonding module option max_bonds=0. This overrides the kernel default,
980 to avoid conflicts and ambiguity as to whether or not bond0 should be
981 managed by systemd-networkd or not. This resolves multiple issues
982 with bond0 properties not being applied, when bond0 is configured
983 with systemd-networkd. Distributors may choose to not package this,
984 however in that case users will be prevented from correctly managing
985 bond0 interface using systemd-networkd.
987 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
988 which print the logging level and target of the system manager. They
989 complement the existing "set-log-level" and "set-log-target" verbs
990 used to change those values.
992 * journald.conf gained a new boolean setting ReadKMsg= which defaults
993 to on. If turned off kernel log messages will not be read by
994 systemd-journald or included in the logs. It also gained a new
995 setting LineMax= for configuring the maximum line length in
996 STDOUT/STDERR log streams. The new default for this value is 48K, up
997 from the previous hardcoded 2048.
999 * A new unit setting RuntimeDirectoryPreserve= has been added, which
1000 allows more detailed control of what to do with a runtime directory
1001 configured with RuntimeDirectory= (i.e. a directory below /run or
1002 $XDG_RUNTIME_DIR) after a unit is stopped.
1004 * The RuntimeDirectory= setting for units gained support for creating
1005 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
1006 one top-level directory.
1008 * Units gained new options StateDirectory=, CacheDirectory=,
1009 LogsDirectory= and ConfigurationDirectory= which are closely related
1010 to RuntimeDirectory= but manage per-service directories below
1011 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
1012 possible to write unit files which when activated automatically gain
1013 properly owned service specific directories in these locations, thus
1014 making unit files self-contained and increasing compatibility with
1015 stateless systems and factory reset where /etc or /var are
1016 unpopulated at boot. Matching these new settings there's also
1017 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
1018 ConfigurationDirectoryMode= for configuring the access mode of these
1019 directories. These settings are particularly useful in combination
1020 with DynamicUser=yes as they provide secure, properly-owned,
1021 writable, and stateful locations for storage, excluded from the
1022 sandbox that such services live in otherwise.
1024 * Automake support has been removed from this release. systemd is now
1027 * systemd-journald will now aggressively cache client metadata during
1028 runtime, speeding up log write performance under pressure. This comes
1029 at a small price though: as much of the metadata is read
1030 asynchronously from /proc/ (and isn't implicitly attached to log
1031 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
1032 metadata stored alongside a log entry might be slightly
1033 out-of-date. Previously it could only be slightly newer than the log
1034 message. The time window is small however, and given that the kernel
1035 is unlikely to be improved anytime soon in this regard, this appears
1038 * nss-myhostname/systemd-resolved will now by default synthesize an
1039 A/AAAA resource record for the "_gateway" hostname, pointing to the
1040 current default IP gateway. Previously it did that for the "gateway"
1041 name, hampering adoption, as some distributions wanted to leave that
1042 host name open for local use. The old behaviour may still be
1043 requested at build time.
1045 * systemd-networkd's [Address] section in .network files gained a new
1046 Scope= setting for configuring the IP address scope. The [Network]
1047 section gained a new boolean setting ConfigureWithoutCarrier= that
1048 tells systemd-networkd to ignore link sensing when configuring the
1049 device. The [DHCP] section gained a new Anonymize= boolean option for
1050 turning on a number of options suggested in RFC 7844. A new
1051 [RoutingPolicyRule] section has been added for configuring the IP
1052 routing policy. The [Route] section has gained support for a new
1053 Type= setting which permits configuring
1054 blackhole/unreachable/prohibit routes.
1056 * The [VRF] section in .netdev files gained a new Table= setting for
1057 configuring the routing table to use. The [Tunnel] section gained a
1058 new Independent= boolean field for configuring tunnels independent of
1059 an underlying network interface. The [Bridge] section gained a new
1060 GroupForwardMask= option for configuration of propagation of link
1061 local frames between bridge ports.
1063 * The WakeOnLan= setting in .link files gained support for a number of
1064 new modes. A new TCP6SegmentationOffload= setting has been added for
1065 configuring TCP/IPv6 hardware segmentation offload.
1067 * The IPv6 RA sender implementation may now optionally send out RDNSS
1068 and RDNSSL records to supply DNS configuration to peers.
1070 * systemd-nspawn gained support for a new --system-call-filter= command
1071 line option for adding and removing entries in the default system
1072 call filter it applies. Moreover systemd-nspawn has been changed to
1073 implement a system call whitelist instead of a blacklist.
1075 * systemd-run gained support for a new --pipe command line option. If
1076 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
1077 are directly passed on to the activated transient service
1078 executable. This allows invoking arbitrary processes as systemd
1079 services (for example to take benefit of dependency management,
1080 accounting management, resource management or log management that is
1081 done automatically for services) — while still allowing them to be
1082 integrated in a classic UNIX shell pipeline.
1084 * When a service sends RELOAD=1 via sd_notify() and reload propagation
1085 using ReloadPropagationTo= is configured, a reload is now propagated
1086 to configured units. (Previously this was only done on explicitly
1087 requested reloads, using "systemctl reload" or an equivalent
1090 * For each service unit a restart counter is now kept: it is increased
1091 each time the service is restarted due to Restart=, and may be
1092 queried using "systemctl show -p NRestarts …".
1094 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
1095 @signal and @timer have been added, for usage with SystemCallFilter=
1096 in unit files and the new --system-call-filter= command line option
1097 of systemd-nspawn (see above).
1099 * ExecStart= lines in unit files gained two new modifiers: when a
1100 command line is prefixed with "!" the command will be executed as
1101 configured, except for the credentials applied by
1102 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
1103 "+", but does still apply namespacing options unlike "+". There's
1104 also "!!" now, which is mostly identical, but becomes a NOP on
1105 systems that support ambient capabilities. This is useful to write
1106 unit files that work with ambient capabilities where possible but
1107 automatically fall back to traditional privilege dropping mechanisms
1108 on systems where this is not supported.
1110 * ListenNetlink= settings in socket units now support RDMA netlink
1113 * A new unit file setting LockPersonality= has been added which permits
1114 locking down the chosen execution domain ("personality") of a service
1117 * A new special target "getty-pre.target" has been added, which is
1118 ordered before all text logins, and may be used to order services
1119 before textual logins acquire access to the console.
1121 * systemd will now attempt to load the virtio-rng.ko kernel module very
1122 early on if a VM environment supporting this is detected. This should
1123 improve entropy during early boot in virtualized environments.
1125 * A _netdev option is now supported in /etc/crypttab that operates in a
1126 similar way as the same option in /etc/fstab: it permits configuring
1127 encrypted devices that need to be ordered after the network is up.
1128 Following this logic, two new special targets
1129 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
1130 added that are to cryptsetup.target what remote-fs.target and
1131 remote-fs-pre.target are to local-fs.target.
1133 * Service units gained a new UnsetEnvironment= setting which permits
1134 unsetting specific environment variables for services that are
1135 normally passed to it (for example in order to mask out locale
1136 settings for specific services that can't deal with it).
1138 * Units acquired a new boolean option IPAccounting=. When turned on, IP
1139 traffic accounting (packet count as well as byte count) is done for
1140 the service, and shown as part of "systemctl status" or "systemd-run
1143 * Service units acquired two new options IPAddressAllow= and
1144 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
1145 for configuring a simple IP access control list for all sockets of
1146 the unit. These options are available also on .slice and .socket
1147 units, permitting flexible access list configuration for individual
1148 services as well as groups of services (as defined by a slice unit),
1149 including system-wide. Note that IP ACLs configured this way are
1150 enforced on every single IPv4 and IPv6 socket created by any process
1151 of the service unit, and apply to ingress as well as egress traffic.
1153 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
1154 structured log message is generated each time the unit is stopped,
1155 containing information about the consumed resources of this
1158 * A new setting KeyringMode= has been added to unit files, which may be
1159 used to control how the kernel keyring is set up for executed
1162 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
1163 "systemctl kexec" and "systemctl exit" are now always asynchronous in
1164 behaviour (that is: these commands return immediately after the
1165 operation was enqueued instead of waiting for the operation to
1166 complete). Previously, "systemctl poweroff" and "systemctl reboot"
1167 were asynchronous on systems using systemd-logind (i.e. almost
1168 always, and like they were on sysvinit), and the other three commands
1169 were unconditionally synchronous. With this release this is cleaned
1170 up, and callers will see the same asynchronous behaviour on all
1171 systems for all five operations.
1173 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
1176 * .timer units now accept calendar specifications in other timezones
1177 than UTC or the local timezone.
1179 * The tmpfiles snippet var.conf has been changed to create
1180 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
1181 the "utmp" group already, and it appears to be generally understood
1182 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
1183 databases. Previously this was implemented correctly for all these
1184 databases excepts btmp, which has been opened up like this now
1185 too. Note that while the other databases are world-readable
1186 (i.e. 0644), btmp is not and remains more restrictive.
1188 * The systemd-resolve tool gained a new --reset-server-features
1189 switch. When invoked like this systemd-resolved will forget
1190 everything it learnt about the features supported by the configured
1191 upstream DNS servers, and restarts the feature probing logic on the
1192 next resolver look-up for them at the highest feature level
1195 * The status dump systemd-resolved sends to the logs upon receiving
1196 SIGUSR1 now also includes information about all DNS servers it is
1197 configured to use, and the features levels it probed for them.
1199 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
1200 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
1201 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
1202 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
1203 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
1204 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
1205 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
1206 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
1207 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
1208 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
1209 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
1210 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
1211 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
1212 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
1213 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
1214 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
1215 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
1216 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
1217 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
1218 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1220 — Berlin, 2017-10-06
1224 * Meson is now supported as build system in addition to Automake. It is
1225 our plan to remove Automake in one of our next releases, so that
1226 Meson becomes our exclusive build system. Hence, please start using
1227 the Meson build system in your downstream packaging. There's plenty
1228 of documentation around how to use Meson, the extremely brief
1231 ./autogen.sh && ./configure && make && sudo make install
1235 meson build && ninja -C build && sudo ninja -C build install
1237 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
1238 which permits configuring a timeout on the time a job is
1239 running. This is particularly useful for setting timeouts on jobs for
1242 * Unit files gained two new options ConditionUser= and ConditionGroup=
1243 for conditionalizing units based on the identity of the user/group
1244 running a systemd user instance.
1246 * systemd-networkd now understands a new FlowLabel= setting in the
1247 [VXLAN] section of .network files, as well as a Priority= in
1248 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
1249 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
1250 gained support for configuration of GENEVE links, and IPv6 address
1251 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
1253 * .link files now understand a new Port= setting.
1255 * systemd-networkd's DHCP support gained support for DHCP option 119
1256 (domain search list).
1258 * systemd-networkd gained support for serving IPv6 address ranges using
1259 the Router Advertisement protocol. The new .network configuration
1260 section [IPv6Prefix] may be used to configure the ranges to
1261 serve. This is implemented based on a new, minimal, native server
1262 implementation of RA.
1264 * journalctl's --output= switch gained support for a new parameter
1265 "short-iso-precise" for a mode where timestamps are shown as precise
1268 * systemd-udevd's "net_id" builtin may now generate stable network
1269 interface names from IBM PowerVM VIO devices as well as ACPI platform
1272 * MulticastDNS support in systemd-resolved may now be explicitly
1273 enabled/disabled using the new MulticastDNS= configuration file
1276 * systemd-resolved may now optionally use libidn2 instead of the libidn
1277 for processing internationalized domain names. Support for libidn2
1278 should be considered experimental and should not be enabled by
1281 * "machinectl pull-tar" and related call may now do verification of
1282 downloaded images using SUSE-style .sha256 checksum files in addition
1283 to the already existing support for validating using Ubuntu-style
1286 * sd-bus gained support for a new sd_bus_message_appendv() call which
1287 is va_list equivalent of sd_bus_message_append().
1289 * sd-boot gained support for validating images using SHIM/MOK.
1291 * The SMACK code learnt support for "onlycap".
1293 * systemd-mount --umount is now much smarter in figuring out how to
1294 properly unmount a device given its mount or device path.
1296 * The code to call libnss_dns as a fallback from libnss_resolve when
1297 the communication with systemd-resolved fails was removed. This
1298 fallback was redundant and interfered with the [!UNAVAIL=return]
1299 suffix. See nss-resolve(8) for the recommended configuration.
1301 * systemd-logind may now be restarted without losing state. It stores
1302 the file descriptors for devices it manages in the system manager
1303 using the FDSTORE= mechanism. Please note that further changes in
1304 other components may be required to make use of this (for example
1305 Xorg has code to listen for stops of systemd-logind and terminate
1306 itself when logind is stopped or restarted, in order to avoid using
1307 stale file descriptors for graphical devices, which is now
1308 counterproductive and must be reverted in order for restarts of
1309 systemd-logind to be safe. See
1310 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
1312 * All kernel install plugins are called with the environment variable
1313 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
1314 /etc/machine-id. If the file is missing or empty, the variable is
1315 empty and BOOT_DIR_ABS is the path of a temporary directory which is
1316 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
1317 is empty, all plugins should not put anything in BOOT_DIR_ABS.
1319 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
1320 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
1321 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
1322 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
1323 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
1324 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
1325 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
1326 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
1327 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
1328 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
1329 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
1330 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
1331 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
1332 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
1333 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
1334 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
1335 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
1336 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
1337 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
1338 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
1339 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
1340 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
1341 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
1342 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
1343 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
1344 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
1345 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
1346 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
1347 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
1350 — Berlin, 2017-07-12
1354 * The "hybrid" control group mode has been modified to improve
1355 compatibility with "legacy" cgroups-v1 setups. Specifically, the
1356 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
1357 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
1358 cgroups-v1 hierarchy), the only externally visible change being that
1359 the cgroups-v2 hierarchy is also mounted, to
1360 /sys/fs/cgroup/unified. This should provide a large degree of
1361 compatibility with "legacy" cgroups-v1, while taking benefit of the
1362 better management capabilities of cgroups-v2.
1364 * The default control group setup mode may be selected both a boot-time
1365 via a set of kernel command line parameters (specifically:
1366 systemd.unified_cgroup_hierarchy= and
1367 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
1368 default selected on the configure command line
1369 (--with-default-hierarchy=). The upstream default is "hybrid"
1370 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
1371 this will change in a future systemd version to be "unified" (pure
1372 cgroups-v2 mode). The third option for the compile time option is
1373 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
1374 distributions to default to "hybrid" mode for release distributions,
1375 starting with v233. We recommend "unified" for development
1376 distributions (specifically: distributions such as Fedora's rawhide)
1377 as that's where things are headed in the long run. Use "legacy" for
1378 greatest stability and compatibility only.
1380 * Note one current limitation of "unified" and "hybrid" control group
1381 setup modes: the kernel currently does not permit the systemd --user
1382 instance (i.e. unprivileged code) to migrate processes between two
1383 disconnected cgroup subtrees, even if both are managed and owned by
1384 the user. This effectively means "systemd-run --user --scope" doesn't
1385 work when invoked from outside of any "systemd --user" service or
1386 scope. Specifically, it is not supported from session scopes. We are
1387 working on fixing this in a future systemd version. (See #3388 for
1388 further details about this.)
1390 * DBus policy files are now installed into /usr rather than /etc. Make
1391 sure your system has dbus >= 1.9.18 running before upgrading to this
1392 version, or override the install path with --with-dbuspolicydir= .
1394 * All python scripts shipped with systemd (specifically: the various
1395 tests written in Python) now require Python 3.
1397 * systemd unit tests can now run standalone (without the source or
1398 build directories), and can be installed into /usr/lib/systemd/tests/
1399 with 'make install-tests'.
1401 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
1402 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
1405 * Support for the %c, %r, %R specifiers in unit files has been
1406 removed. Specifiers are not supposed to be dependent on configuration
1407 in the unit file itself (so that they resolve the same regardless
1408 where used in the unit files), but these specifiers were influenced
1409 by the Slice= option.
1411 * The shell invoked by debug-shell.service now defaults to /bin/sh in
1412 all cases. If distributions want to use a different shell for this
1413 purpose (for example Fedora's /sbin/sushell) they need to specify
1414 this explicitly at configure time using --with-debug-shell=.
1416 * The confirmation spawn prompt has been reworked to offer the
1419 (c)ontinue, proceed without asking anymore
1420 (D)ump, show the state of the unit
1421 (f)ail, don't execute the command and pretend it failed
1423 (i)nfo, show a short summary of the unit
1424 (j)obs, show jobs that are in progress
1425 (s)kip, don't execute the command and pretend it succeeded
1426 (y)es, execute the command
1428 The 'n' choice for the confirmation spawn prompt has been removed,
1429 because its meaning was confusing.
1431 The prompt may now also be redirected to an alternative console by
1432 specifying the console as parameter to systemd.confirm_spawn=.
1434 * Services of Type=notify require a READY=1 notification to be sent
1435 during startup. If no such message is sent, the service now fails,
1436 even if the main process exited with a successful exit code.
1438 * Services that fail to start up correctly now always have their
1439 ExecStopPost= commands executed. Previously, they'd enter "failed"
1440 state directly, without executing these commands.
1442 * The option MulticastDNS= of network configuration files has acquired
1443 an actual implementation. With MulticastDNS=yes a host can resolve
1444 names of remote hosts and reply to mDNS A and AAAA requests.
1446 * When units are about to be started an additional check is now done to
1447 ensure that all dependencies of type BindsTo= (when used in
1448 combination with After=) have been started.
1450 * systemd-analyze gained a new verb "syscall-filter" which shows which
1451 system call groups are defined for the SystemCallFilter= unit file
1452 setting, and which system calls they contain.
1454 * A new system call filter group "@filesystem" has been added,
1455 consisting of various file system related system calls. Group
1456 "@reboot" has been added, covering reboot, kexec and shutdown related
1457 calls. Finally, group "@swap" has been added covering swap
1458 configuration related calls.
1460 * A new unit file option RestrictNamespaces= has been added that may be
1461 used to restrict access to the various process namespace types the
1462 Linux kernel provides. Specifically, it may be used to take away the
1463 right for a service unit to create additional file system, network,
1464 user, and other namespaces. This sandboxing option is particularly
1465 relevant due to the high amount of recently discovered namespacing
1466 related vulnerabilities in the kernel.
1468 * systemd-udev's .link files gained support for a new AutoNegotiation=
1469 setting for configuring Ethernet auto-negotiation.
1471 * systemd-networkd's .network files gained support for a new
1472 ListenPort= setting in the [DHCP] section to explicitly configure the
1473 UDP client port the DHCP client shall listen on.
1475 * .network files gained a new Unmanaged= boolean setting for explicitly
1476 excluding one or more interfaces from management by systemd-networkd.
1478 * The systemd-networkd ProxyARP= option has been renamed to
1479 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
1480 renamed to ReduceARPProxy=. The old names continue to be available
1483 * systemd-networkd gained support for configuring IPv6 Proxy NDP
1484 addresses via the new IPv6ProxyNDPAddress= .network file setting.
1486 * systemd-networkd's bonding device support gained support for two new
1487 configuration options ActiveSlave= and PrimarySlave=.
1489 * The various options in the [Match] section of .network files gained
1490 support for negative matching.
1492 * New systemd-specific mount options are now understood in /etc/fstab:
1494 x-systemd.mount-timeout= may be used to configure the maximum
1495 permitted runtime of the mount command.
1497 x-systemd.device-bound may be set to bind a mount point to its
1498 backing device unit, in order to automatically remove a mount point
1499 if its backing device is unplugged. This option may also be
1500 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
1501 on the block device, which is now automatically set for all CDROM
1502 drives, so that mounted CDs are automatically unmounted when they are
1503 removed from the drive.
1505 x-systemd.after= and x-systemd.before= may be used to explicitly
1506 order a mount after or before another unit or mount point.
1508 * Enqueued start jobs for device units are now automatically garbage
1509 collected if there are no jobs waiting for them anymore.
1511 * systemctl list-jobs gained two new switches: with --after, for every
1512 queued job the jobs it's waiting for are shown; with --before the
1513 jobs which it's blocking are shown.
1515 * systemd-nspawn gained support for ephemeral boots from disk images
1516 (or in other words: --ephemeral and --image= may now be
1517 combined). Moreover, ephemeral boots are now supported for normal
1518 directories, even if the backing file system is not btrfs. Of course,
1519 if the file system does not support file system snapshots or
1520 reflinks, the initial copy operation will be relatively expensive, but
1521 this should still be suitable for many use cases.
1523 * Calendar time specifications in .timer units now support
1524 specifications relative to the end of a month by using "~" instead of
1525 "-" as separator between month and day. For example, "*-02~03" means
1526 "the third last day in February". In addition a new syntax for
1527 repeated events has been added using the "/" character. For example,
1528 "9..17/2:00" means "every two hours from 9am to 5pm".
1530 * systemd-socket-proxyd gained a new parameter --connections-max= for
1531 configuring the maximum number of concurrent connections.
1533 * sd-id128 gained a new API for generating unique IDs for the host in a
1534 way that does not leak the machine ID. Specifically,
1535 sd_id128_get_machine_app_specific() derives an ID based on the
1536 machine ID a in well-defined, non-reversible, stable way. This is
1537 useful whenever an identifier for the host is needed but where the
1538 identifier shall not be useful to identify the system beyond the
1539 scope of the application itself. (Internally this uses HMAC-SHA256 as
1540 keyed hash function using the machine ID as input.)
1542 * NotifyAccess= gained a new supported value "exec". When set
1543 notifications are accepted from all processes systemd itself invoked,
1544 including all control processes.
1546 * .nspawn files gained support for defining overlay mounts using the
1547 Overlay= and OverlayReadOnly= options. Previously this functionality
1548 was only available on the systemd-nspawn command line.
1550 * systemd-nspawn's --bind= and --overlay= options gained support for
1551 bind/overlay mounts whose source lies within the container tree by
1552 prefixing the source path with "+".
1554 * systemd-nspawn's --bind= and --overlay= options gained support for
1555 automatically allocating a temporary source directory in /var/tmp
1556 that is removed when the container dies. Specifically, if the source
1557 directory is specified as empty string this mechanism is selected. An
1558 example usage is --overlay=+/var::/var, which creates an overlay
1559 mount based on the original /var contained in the image, overlayed
1560 with a temporary directory in the host's /var/tmp. This way changes
1561 to /var are automatically flushed when the container shuts down.
1563 * systemd-nspawn --image= option does now permit raw file system block
1564 devices (in addition to images containing partition tables, as
1567 * The disk image dissection logic in systemd-nspawn gained support for
1568 automatically setting up LUKS encrypted as well as Verity protected
1569 partitions. When a container is booted from an encrypted image the
1570 passphrase is queried at start-up time. When a container with Verity
1571 data is started, the root hash is search in a ".roothash" file
1572 accompanying the disk image (alternatively, pass the root hash via
1573 the new --root-hash= command line option).
1575 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
1576 be used to dissect disk images the same way as systemd-nspawn does
1577 it, following the Bootable Partition Specification. It may even be
1578 used to mount disk images with complex partition setups (including
1579 LUKS and Verity partitions) to a local host directory, in order to
1580 inspect them. This tool is not considered public API (yet), and is
1581 thus not installed into /usr/bin. Please do not rely on its
1582 existence, since it might go away or be changed in later systemd
1585 * A new generator "systemd-verity-generator" has been added, similar in
1586 style to "systemd-cryptsetup-generator", permitting automatic setup of
1587 Verity root partitions when systemd boots up. In order to make use of
1588 this your partition setup should follow the Discoverable Partitions
1589 Specification, and the GPT partition ID of the root file system
1590 partition should be identical to the upper 128bit of the Verity root
1591 hash. The GPT partition ID of the Verity partition protecting it
1592 should be the lower 128bit of the Verity root hash. If the partition
1593 image follows this model it is sufficient to specify a single
1594 "roothash=" kernel command line argument to both configure which root
1595 image and verity partition to use as well as the root hash for
1596 it. Note that systemd-nspawn's Verity support follows the same
1597 semantics, meaning that disk images with proper Verity data in place
1598 may be booted in containers with systemd-nspawn as well as on
1599 physical systems via the verity generator. Also note that the "mkosi"
1600 tool available at https://github.com/systemd/mkosi has been updated
1601 to generate Verity protected disk images following this scheme. In
1602 fact, it has been updated to generate disk images that optionally
1603 implement a complete UEFI SecureBoot trust chain, involving a signed
1604 kernel and initrd image that incorporates such a root hash as well as
1605 a Verity-enabled root partition.
1607 * The hardware database (hwdb) udev supports has been updated to carry
1608 accelerometer quirks.
1610 * All system services are now run with a fresh kernel keyring set up
1611 for them. The invocation ID is stored by default in it, thus
1612 providing a safe, non-overridable way to determine the invocation
1615 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
1616 options for bind mounting arbitrary paths in a service-specific
1617 way. When these options are used, arbitrary host or service files and
1618 directories may be mounted to arbitrary locations in the service's
1621 * Documentation has been added that lists all of systemd's low-level
1622 environment variables:
1624 https://github.com/systemd/systemd/blob/master/doc/ENVIRONMENT.md
1626 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
1627 whether a specific socket file descriptor matches a specified socket
1630 * systemd-firstboot has been updated to check for the
1631 systemd.firstboot= kernel command line option. It accepts a boolean
1632 and when set to false the first boot questions are skipped.
1634 * systemd-fstab-generator has been updated to check for the
1635 systemd.volatile= kernel command line option, which either takes an
1636 optional boolean parameter or the special value "state". If used the
1637 system may be booted in a "volatile" boot mode. Specifically,
1638 "systemd.volatile" is used, the root directory will be mounted as
1639 tmpfs, and only /usr is mounted from the actual root file system. If
1640 "systemd.volatile=state" is used, the root directory will be mounted
1641 as usual, but /var is mounted as tmpfs. This concept provides similar
1642 functionality as systemd-nspawn's --volatile= option, but provides it
1643 on physical boots. Use this option for implementing stateless
1644 systems, or testing systems with all state and/or configuration reset
1645 to the defaults. (Note though that many distributions are not
1646 prepared to boot up without a populated /etc or /var, though.)
1648 * systemd-gpt-auto-generator gained support for LUKS encrypted root
1649 partitions. Previously it only supported LUKS encrypted partitions
1650 for all other uses, except for the root partition itself.
1652 * Socket units gained support for listening on AF_VSOCK sockets for
1653 communication in virtualized QEMU environments.
1655 * The "configure" script gained a new option --with-fallback-hostname=
1656 for specifying the fallback hostname to use if none is configured in
1657 /etc/hostname. For example, by specifying
1658 --with-fallback-hostname=fedora it is possible to default to a
1659 hostname of "fedora" on pristine installations.
1661 * systemd-cgls gained support for a new --unit= switch for listing only
1662 the control groups of a specific unit. Similar --user-unit= has been
1663 added for listing only the control groups of a specific user unit.
1665 * systemd-mount gained a new --umount switch for unmounting a mount or
1666 automount point (and all mount/automount points below it).
1668 * systemd will now refuse full configuration reloads (via systemctl
1669 daemon-reload and related calls) unless at least 16MiB of free space
1670 are available in /run. This is a safety precaution in order to ensure
1671 that generators can safely operate after the reload completed.
1673 * A new unit file option RootImage= has been added, which has a similar
1674 effect as RootDirectory= but mounts the service's root directory from
1675 a disk image instead of plain directory. This logic reuses the same
1676 image dissection and mount logic that systemd-nspawn already uses,
1677 and hence supports any disk images systemd-nspawn supports, including
1678 those following the Discoverable Partition Specification, as well as
1679 Verity enabled images. This option enables systemd to run system
1680 services directly off disk images acting as resource bundles,
1681 possibly even including full integrity data.
1683 * A new MountAPIVFS= unit file option has been added, taking a boolean
1684 argument. If enabled /proc, /sys and /dev (collectively called the
1685 "API VFS") will be mounted for the service. This is only relevant if
1686 RootDirectory= or RootImage= is used for the service, as these mounts
1687 are of course in place in the host mount namespace anyway.
1689 * systemd-nspawn gained support for a new --pivot-root= switch. If
1690 specified the root directory within the container image is pivoted to
1691 the specified mount point, while the original root disk is moved to a
1692 different place. This option enables booting of ostree images
1693 directly with systemd-nspawn.
1695 * The systemd build scripts will no longer complain if the NTP server
1696 addresses are not changed from the defaults. Google now supports
1697 these NTP servers officially. We still recommend downstreams to
1698 properly register an NTP pool with the NTP pool project though.
1700 * coredumpctl gained a new "--reverse" option for printing the list
1701 of coredumps in reverse order.
1703 * coredumpctl will now show additional information about truncated and
1704 inaccessible coredumps, as well as coredumps that are still being
1705 processed. It also gained a new --quiet switch for suppressing
1706 additional informational message in its output.
1708 * coredumpctl gained support for only showing coredumps newer and/or
1709 older than specific timestamps, using the new --since= and --until=
1710 options, reminiscent of journalctl's options by the same name.
1712 * The systemd-coredump logic has been improved so that it may be reused
1713 to collect backtraces in non-compiled languages, for example in
1714 scripting languages such as Python.
1716 * machinectl will now show the UID shift of local containers, if user
1717 namespacing is enabled for them.
1719 * systemd will now optionally run "environment generator" binaries at
1720 configuration load time. They may be used to add environment
1721 variables to the environment block passed to services invoked. One
1722 user environment generator is shipped by default that sets up
1723 environment variables based on files dropped into /etc/environment.d
1724 and ~/.config/environment.d/.
1726 * systemd-resolved now includes the new, recently published 2017 DNSSEC
1729 * hostnamed has been updated to report a new chassis type of
1730 "convertible" to cover "foldable" laptops that can both act as a
1731 tablet and as a laptop, such as various Lenovo Yoga devices.
1733 Contributions from: Adrián López, Alexander Galanin, Alexander
1734 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
1735 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
1736 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
1737 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
1738 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
1739 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
1740 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
1741 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
1742 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
1743 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
1744 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
1745 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
1746 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
1747 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
1748 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
1749 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
1750 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
1751 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
1752 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
1753 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
1754 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
1755 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
1756 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
1757 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
1758 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
1759 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
1760 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
1763 — Berlin, 2017-03-01
1767 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
1768 RestrictAddressFamilies= enabled. These sandboxing options should
1769 generally be compatible with the various external udev call-out
1770 binaries we are aware of, however there may be exceptions, in
1771 particular when exotic languages for these call-outs are used. In
1772 this case, consider turning off these settings locally.
1774 * The new RemoveIPC= option can be used to remove IPC objects owned by
1775 the user or group of a service when that service exits.
1777 * The new ProtectKernelModules= option can be used to disable explicit
1778 load and unload operations of kernel modules by a service. In
1779 addition access to /usr/lib/modules is removed if this option is set.
1781 * ProtectSystem= option gained a new value "strict", which causes the
1782 whole file system tree with the exception of /dev, /proc, and /sys,
1783 to be remounted read-only for a service.
1785 * The new ProtectKernelTunables= option can be used to disable
1786 modification of configuration files in /sys and /proc by a service.
1787 Various directories and files are remounted read-only, so access is
1788 restricted even if the file permissions would allow it.
1790 * The new ProtectControlGroups= option can be used to disable write
1791 access by a service to /sys/fs/cgroup.
1793 * Various systemd services have been hardened with
1794 ProtectKernelTunables=yes, ProtectControlGroups=yes,
1795 RestrictAddressFamilies=.
1797 * Support for dynamically creating users for the lifetime of a service
1798 has been added. If DynamicUser=yes is specified, user and group IDs
1799 will be allocated from the range 61184..65519 for the lifetime of the
1800 service. They can be resolved using the new nss-systemd.so NSS
1801 module. The module must be enabled in /etc/nsswitch.conf. Services
1802 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
1803 any resources allocated by the service will be cleaned up when the
1804 service exits. They also have ProtectHome=read-only and
1805 ProtectSystem=strict enabled, so they are not able to make any
1806 permanent modifications to the system.
1808 * The nss-systemd module also always resolves root and nobody, making
1809 it possible to have no /etc/passwd or /etc/group files in minimal
1810 container or chroot environments.
1812 * Services may be started with their own user namespace using the new
1813 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
1814 under which the service is running are mapped. All other users are
1817 * Support for the cgroup namespace has been added to systemd-nspawn. If
1818 supported by kernel, the container system started by systemd-nspawn
1819 will have its own view of the cgroup hierarchy. This new behaviour
1820 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
1822 * The new MemorySwapMax= option can be used to limit the maximum swap
1823 usage under the unified cgroup hierarchy.
1825 * Support for the CPU controller in the unified cgroup hierarchy has
1826 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
1827 options. This controller requires out-of-tree patches for the kernel
1828 and the support is provisional.
1830 * Mount and automount units may now be created transiently
1831 (i.e. dynamically at runtime via the bus API, instead of requiring
1832 unit files in the file system).
1834 * systemd-mount is a new tool which may mount file systems – much like
1835 mount(8), optionally pulling in additional dependencies through
1836 transient .mount and .automount units. For example, this tool
1837 automatically runs fsck on a backing block device before mounting,
1838 and allows the automount logic to be used dynamically from the
1839 command line for establishing mount points. This tool is particularly
1840 useful when dealing with removable media, as it will ensure fsck is
1841 run – if necessary – before the first access and that the file system
1842 is quickly unmounted after each access by utilizing the automount
1843 logic. This maximizes the chance that the file system on the
1844 removable media stays in a clean state, and if it isn't in a clean
1845 state is fixed automatically.
1847 * LazyUnmount=yes option for mount units has been added to expose the
1848 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
1851 * /efi will be used as the mount point of the EFI boot partition, if
1852 the directory is present, and the mount point was not configured
1853 through other means (e.g. fstab). If /efi directory does not exist,
1854 /boot will be used as before. This makes it easier to automatically
1855 mount the EFI partition on systems where /boot is used for something
1858 * When operating on GPT disk images for containers, systemd-nspawn will
1859 now mount the ESP to /boot or /efi according to the same rules as PID
1860 1 running on a host. This allows tools like "bootctl" to operate
1861 correctly within such containers, in order to make container images
1862 bootable on physical systems.
1864 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
1866 * Two new user session targets have been added to support running
1867 graphical sessions under the systemd --user instance:
1868 graphical-session.target and graphical-session-pre.target. See
1869 systemd.special(7) for a description of how those targets should be
1872 * The vconsole initialization code has been significantly reworked to
1873 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
1874 support unicode keymaps. Font and keymap configuration will now be
1875 copied to all allocated virtual consoles.
1877 * FreeBSD's bhyve virtualization is now detected.
1879 * Information recorded in the journal for core dumps now includes the
1880 contents of /proc/mountinfo and the command line of the process at
1881 the top of the process hierarchy (which is usually the init process
1884 * systemd-journal-gatewayd learned the --directory= option to serve
1885 files from the specified location.
1887 * journalctl --root=… can be used to peruse the journal in the
1888 /var/log/ directories inside of a container tree. This is similar to
1889 the existing --machine= option, but does not require the container to
1892 * The hardware database has been extended to support
1893 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
1896 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
1897 specify the click rate for mice which include a horizontal wheel with
1898 a click rate that is different than the one for the vertical wheel.
1900 * systemd-run gained a new --wait option that makes service execution
1901 synchronous. (Specifically, the command will not return until the
1902 specified service binary exited.)
1904 * systemctl gained a new --wait option that causes the start command to
1905 wait until the units being started have terminated again.
1907 * A new journal output mode "short-full" has been added which displays
1908 timestamps with abbreviated English day names and adds a timezone
1909 suffix. Those timestamps include more information than the default
1910 "short" output mode, and can be passed directly to journalctl's
1911 --since= and --until= options.
1913 * /etc/resolv.conf will be bind-mounted into containers started by
1914 systemd-nspawn, if possible, so any changes to resolv.conf contents
1915 are automatically propagated to the container.
1917 * The number of instances for socket-activated services originating
1918 from a single IP address can be limited with
1919 MaxConnectionsPerSource=, extending the existing setting of
1922 * systemd-networkd gained support for vcan ("Virtual CAN") interface
1925 * .netdev and .network configuration can now be extended through
1928 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
1929 Segmentation Offload, Generic Receive Offload, Large Receive Offload
1930 can be enabled and disabled using the new UDPSegmentationOffload=,
1931 TCPSegmentationOffload=, GenericSegmentationOffload=,
1932 GenericReceiveOffload=, LargeReceiveOffload= options in the
1933 [Link] section of .link files.
1935 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
1936 Port VLAN ID can be configured for bridge devices using the new STP=,
1937 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
1938 section of .netdev files.
1940 * The route table to which routes received over DHCP or RA should be
1941 added can be configured with the new RouteTable= option in the [DHCP]
1942 and [IPv6AcceptRA] sections of .network files.
1944 * The Address Resolution Protocol can be disabled on links managed by
1945 systemd-networkd using the ARP=no setting in the [Link] section of
1948 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
1949 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
1950 encode information about the result and exit codes of the current
1951 service runtime cycle.
1953 * systemd-sysctl will now configure kernel parameters in the order
1954 they occur in the configuration files. This matches what sysctl
1955 has been traditionally doing.
1957 * kernel-install "plugins" that are executed to perform various
1958 tasks after a new kernel is added and before an old one is removed
1959 can now return a special value to terminate the procedure and
1960 prevent any later plugins from running.
1962 * Journald's SplitMode=login setting has been deprecated. It has been
1963 removed from documentation, and its use is discouraged. In a future
1964 release it will be completely removed, and made equivalent to current
1965 default of SplitMode=uid.
1967 * Storage=both option setting in /etc/systemd/coredump.conf has been
1968 removed. With fast LZ4 compression storing the core dump twice is not
1971 * The --share-system systemd-nspawn option has been replaced with an
1972 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
1973 this functionality is discouraged. In addition the variables
1974 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
1975 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
1976 individual namespaces.
1978 * "machinectl list" now shows the IP address of running containers in
1979 the output, as well as OS release information.
1981 * "loginctl list" now shows the TTY of each session in the output.
1983 * sd-bus gained new API calls sd_bus_track_set_recursive(),
1984 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
1985 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
1986 tracking objects in a "recursive" mode, where a single client can be
1987 counted multiple times, if it takes multiple references.
1989 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
1990 sd_bus_get_exit_on_disconnect(). They may be used to make a
1991 process using sd-bus automatically exit if the bus connection is
1994 * Bus clients of the service manager may now "pin" loaded units into
1995 memory, by taking an explicit reference on them. This is useful to
1996 ensure the client can retrieve runtime data about the service even
1997 after the service completed execution. Taking such a reference is
1998 available only for privileged clients and should be helpful to watch
1999 running services in a race-free manner, and in particular collect
2000 information about exit statuses and results.
2002 * The nss-resolve module has been changed to strictly return UNAVAIL
2003 when communication via D-Bus with resolved failed, and NOTFOUND when
2004 a lookup completed but was negative. This means it is now possible to
2005 neatly configure fallbacks using nsswitch.conf result checking
2006 expressions. Taking benefit of this, the new recommended
2007 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
2009 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
2011 * A new setting CtrlAltDelBurstAction= has been added to
2012 /etc/systemd/system.conf which may be used to configure the precise
2013 behaviour if the user on the console presses Ctrl-Alt-Del more often
2014 than 7 times in 2s. Previously this would unconditionally result in
2015 an expedited, immediate reboot. With this new setting the precise
2016 operation may be configured in more detail, and also turned off
2019 * In .netdev files two new settings RemoteChecksumTx= and
2020 RemoteChecksumRx= are now understood that permit configuring the
2021 remote checksumming logic for VXLAN networks.
2023 * The service manager learnt a new "invocation ID" concept for invoked
2024 services. Each runtime cycle of a service will get a new invocation
2025 ID (a 128bit random UUID) assigned that identifies the current
2026 run of the service uniquely and globally. A new invocation ID
2027 is generated each time a service starts up. The journal will store
2028 the invocation ID of a service along with any logged messages, thus
2029 making the invocation ID useful for matching the online runtime of a
2030 service with the offline log data it generated in a safe way without
2031 relying on synchronized timestamps. In many ways this new service
2032 invocation ID concept is similar to the kernel's boot ID concept that
2033 uniquely and globally identifies the runtime of each boot. The
2034 invocation ID of a service is passed to the service itself via an
2035 environment variable ($INVOCATION_ID). A new bus call
2036 GetUnitByInvocationID() has been added that is similar to GetUnit()
2037 but instead of retrieving the bus path for a unit by its name
2038 retrieves it by its invocation ID. The returned path is valid only as
2039 long as the passed invocation ID is current.
2041 * systemd-resolved gained a new "DNSStubListener" setting in
2042 resolved.conf. It either takes a boolean value or the special values
2043 "udp" and "tcp", and configures whether to enable the stub DNS
2044 listener on 127.0.0.53:53.
2046 * IP addresses configured via networkd may now carry additional
2047 configuration settings supported by the kernel. New options include:
2048 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
2049 PrefixRoute=, AutoJoin=.
2051 * The PAM configuration fragment file for "user@.service" shipped with
2052 systemd (i.e. the --user instance of systemd) has been stripped to
2053 the minimum necessary to make the system boot. Previously, it
2054 contained Fedora-specific stanzas that did not apply to other
2055 distributions. It is expected that downstream distributions add
2056 additional configuration lines, matching their needs to this file,
2057 using it only as rough template of what systemd itself needs. Note
2058 that this reduced fragment does not even include an invocation of
2059 pam_limits which most distributions probably want to add, even though
2060 systemd itself does not need it. (There's also the new build time
2061 option --with-pamconfdir=no to disable installation of the PAM
2064 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
2065 capability is now also dropped from its set (in addition to
2066 CAP_SYS_MKNOD as before).
2068 * In service unit files it is now possible to connect a specific named
2069 file descriptor with stdin/stdout/stdout of an executed service. The
2070 name may be specified in matching .socket units using the
2071 FileDescriptorName= setting.
2073 * A number of journal settings may now be configured on the kernel
2074 command line. Specifically, the following options are now understood:
2075 systemd.journald.max_level_console=,
2076 systemd.journald.max_level_store=,
2077 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
2078 systemd.journald.max_level_wall=.
2080 * "systemctl is-enabled --full" will now show by which symlinks a unit
2081 file is enabled in the unit dependency tree.
2083 * Support for VeraCrypt encrypted partitions has been added to the
2084 "cryptsetup" logic and /etc/crypttab.
2086 * systemd-detect-virt gained support for a new --private-users switch
2087 that checks whether the invoking processes are running inside a user
2088 namespace. Similar, a new special value "private-users" for the
2089 existing ConditionVirtualization= setting has been added, permitting
2090 skipping of specific units in user namespace environments.
2092 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
2093 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
2094 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
2095 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
2096 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
2097 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
2098 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
2099 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
2100 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
2101 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
2102 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
2103 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
2104 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
2105 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
2106 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
2107 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
2108 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
2109 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
2110 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
2111 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
2112 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
2113 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
2114 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
2115 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
2116 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
2117 Jędrzejewski-Szmek, Zeal Jagannatha
2119 — Santa Fe, 2016-11-03
2123 * In service units the various ExecXYZ= settings have been extended
2124 with an additional special character as first argument of the
2125 assigned value: if the character '+' is used the specified command
2126 line it will be run with full privileges, regardless of User=,
2127 Group=, CapabilityBoundingSet= and similar options. The effect is
2128 similar to the existing PermissionsStartOnly= option, but allows
2129 configuration of this concept for each executed command line
2132 * Services may now alter the service watchdog timeout at runtime by
2133 sending a WATCHDOG_USEC= message via sd_notify().
2135 * MemoryLimit= and related unit settings now optionally take percentage
2136 specifications. The percentage is taken relative to the amount of
2137 physical memory in the system (or in case of containers, the assigned
2138 amount of memory). This allows scaling service resources neatly with
2139 the amount of RAM available on the system. Similarly, systemd-logind's
2140 RuntimeDirectorySize= option now also optionally takes percentage
2143 * In similar fashion TasksMax= takes percentage values now, too. The
2144 value is taken relative to the configured maximum number of processes
2145 on the system. The per-service task maximum has been changed to 15%
2146 using this functionality. (Effectively this is an increase of 512 →
2147 4915 for service units, given the kernel's default pid_max setting.)
2149 * Calendar time specifications in .timer units now understand a ".."
2150 syntax for time ranges. Example: "4..7:10" may now be used for
2151 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
2154 * The InaccessableDirectories=, ReadOnlyDirectories= and
2155 ReadWriteDirectories= unit file settings have been renamed to
2156 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
2157 applied to all kinds of file nodes, and not just directories, with
2158 the exception of symlinks. Specifically these settings may now be
2159 used on block and character device nodes, UNIX sockets and FIFOS as
2160 well as regular files. The old names of these settings remain
2161 available for compatibility.
2163 * systemd will now log about all service processes it kills forcibly
2164 (using SIGKILL) because they remained after the clean shutdown phase
2165 of the service completed. This should help identifying services that
2166 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
2167 systemd-logind's configuration a similar log message is generated for
2168 processes killed at the end of each session due to this setting.
2170 * systemd will now set the $JOURNAL_STREAM environment variable for all
2171 services whose stdout/stderr are connected to the Journal (which
2172 effectively means by default: all services). The variable contains
2173 the device and inode number of the file descriptor used for
2174 stdout/stderr. This may be used by invoked programs to detect whether
2175 their stdout/stderr is connected to the Journal, in which case they
2176 can switch over to direct Journal communication, thus being able to
2177 pass extended, structured metadata along with their log messages. As
2178 one example, this is now used by glib's logging primitives.
2180 * When using systemd's default tmp.mount unit for /tmp, the mount point
2181 will now be established with the "nosuid" and "nodev" options. This
2182 avoids privilege escalation attacks that put traps and exploits into
2183 /tmp. However, this might cause problems if you e. g. put container
2184 images or overlays into /tmp; if you need this, override tmp.mount's
2185 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
2188 * systemd now supports the "memory" cgroup controller also on
2191 * The systemd-cgtop tool now optionally takes a control group path as
2192 command line argument. If specified, the control group list shown is
2193 limited to subgroups of that group.
2195 * The SystemCallFilter= unit file setting gained support for
2196 pre-defined, named system call filter sets. For example
2197 SystemCallFilter=@clock is now an effective way to make all clock
2198 changing-related system calls unavailable to a service. A number of
2199 similar pre-defined groups are defined. Writing system call filters
2200 for system services is simplified substantially with this new
2201 concept. Accordingly, all of systemd's own, long-running services now
2202 enable system call filtering based on this, by default.
2204 * A new service setting MemoryDenyWriteExecute= has been added, taking
2205 a boolean value. If turned on, a service may no longer create memory
2206 mappings that are writable and executable at the same time. This
2207 enhances security for services where this is enabled as it becomes
2208 harder to dynamically write and then execute memory in exploited
2209 service processes. This option has been enabled for all of systemd's
2210 own long-running services.
2212 * A new RestrictRealtime= service setting has been added, taking a
2213 boolean argument. If set the service's processes may no longer
2214 acquire realtime scheduling. This improves security as realtime
2215 scheduling may otherwise be used to easily freeze the system.
2217 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
2218 value. This may be used for requesting that the system manager inside
2219 of the container reports start-up completion to nspawn which then
2220 propagates this notification further to the service manager
2221 supervising nspawn itself. A related option NotifyReady= in .nspawn
2222 files has been added too. This functionality allows ordering of the
2223 start-up of multiple containers using the usual systemd ordering
2226 * machinectl gained a new command "stop" that is an alias for
2229 * systemd-resolved gained support for contacting DNS servers on
2230 link-local IPv6 addresses.
2232 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
2233 its caches. A method call for requesting the same operation has been
2234 added to the bus API too, and is made available via "systemd-resolve
2237 * systemd-resolve gained a new --status switch. If passed a brief
2238 summary of the used DNS configuration with per-interface information
2241 * resolved.conf gained a new Cache= boolean option, defaulting to
2242 on. If turned off local DNS caching is disabled. This comes with a
2243 performance penalty in particular when DNSSEC is enabled. Note that
2244 resolved disables its internal caching implicitly anyway, when the
2245 configured DNS server is on a host-local IP address such as ::1 or
2246 127.0.0.1, thus automatically avoiding double local caching.
2248 * systemd-resolved now listens on the local IP address 127.0.0.53:53
2249 for DNS requests. This improves compatibility with local programs
2250 that do not use the libc NSS or systemd-resolved's bus APIs for name
2251 resolution. This minimal DNS service is only available to local
2252 programs and does not implement the full DNS protocol, but enough to
2253 cover local DNS clients. A new, static resolv.conf file, listing just
2254 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
2255 now recommended to make /etc/resolv.conf a symlink to this file in
2256 order to route all DNS lookups to systemd-resolved, regardless if
2257 done via NSS, the bus API or raw DNS packets. Note that this local
2258 DNS service is not as fully featured as the libc NSS or
2259 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
2260 used to deliver link-local address information (as this implies
2261 sending a local interface index along), LLMNR/mDNS support via this
2262 interface is severely restricted. It is thus strongly recommended for
2263 all applications to use the libc NSS API or native systemd-resolved
2266 * systemd-networkd's bridge support learned a new setting
2267 VLANFiltering= for controlling VLAN filtering. Moreover a new section
2268 in .network files has been added for configuring VLAN bridging in
2269 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
2271 * systemd-networkd's IPv6 Router Advertisement code now makes use of
2272 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
2273 now be acquired without relying on DHCPv6. Two new options
2274 UseDomains= and UseDNS= have been added to configure this behaviour.
2276 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
2277 renamed IPv6AcceptRA=, without altering its behaviour. The old
2278 setting name remains available for compatibility reasons.
2280 * The systemd-networkd VTI/VTI6 tunneling support gained new options
2281 Key=, InputKey= and OutputKey=.
2283 * systemd-networkd gained support for VRF ("Virtual Routing Function")
2284 interface configuration.
2286 * "systemctl edit" may now be used to create new unit files by
2287 specifying the --force switch.
2289 * sd-event gained a new function sd_event_get_iteration() for
2290 requesting the current iteration counter of the event loop. It starts
2291 at zero and is increased by one with each event loop iteration.
2293 * A new rpm macro %systemd_ordering is provided by the macros.systemd
2294 file. It can be used in lieu of %systemd_requires in packages which
2295 don't use any systemd functionality and are intended to be installed
2296 in minimal containers without systemd present. This macro provides
2297 ordering dependencies to ensure that if the package is installed in
2298 the same rpm transaction as systemd, systemd will be installed before
2299 the scriptlets for the package are executed, allowing unit presets
2302 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
2303 been added to simplify packaging of generators.
2305 * The os-release file gained VERSION_CODENAME field for the
2306 distribution nickname (e.g. VERSION_CODENAME=woody).
2308 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
2309 can be set to disable parsing of metadata and the creation
2310 of persistent symlinks for that device.
2312 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
2313 to make them available to logged-in users has been reverted.
2315 * Much of the common code of the various systemd components is now
2316 built into an internal shared library libsystemd-shared-231.so
2317 (incorporating the systemd version number in the name, to be updated
2318 with future releases) that the components link to. This should
2319 decrease systemd footprint both in memory during runtime and on
2320 disk. Note that the shared library is not for public use, and is
2321 neither API nor ABI stable, but is likely to change with every new
2322 released update. Packagers need to make sure that binaries
2323 linking to libsystemd-shared.so are updated in step with the
2326 * Configuration for "mkosi" is now part of the systemd
2327 repository. mkosi is a tool to easily build legacy-free OS images,
2328 and is available on github: https://github.com/systemd/mkosi. If
2329 "mkosi" is invoked in the build tree a new raw OS image is generated
2330 incorporating the systemd sources currently being worked on and a
2331 clean, fresh distribution installation. The generated OS image may be
2332 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
2333 UEFI PC. This functionality is particularly useful to easily test
2334 local changes made to systemd in a pristine, defined environment. See
2335 doc/HACKING for details.
2337 * configure learned the --with-support-url= option to specify the
2338 distribution's bugtracker.
2340 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
2341 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
2342 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
2343 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
2344 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
2345 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
2346 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
2347 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
2348 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
2349 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
2350 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
2351 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
2352 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
2353 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
2354 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
2355 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
2356 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
2357 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
2358 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
2360 — Berlin, 2016-07-25
2364 * DNSSEC is now turned on by default in systemd-resolved (in
2365 "allow-downgrade" mode), but may be turned off during compile time by
2366 passing "--with-default-dnssec=no" to "configure" (and of course,
2367 during runtime with DNSSEC= in resolved.conf). We recommend
2368 downstreams to leave this on at least during development cycles and
2369 report any issues with the DNSSEC logic upstream. We are very
2370 interested in collecting feedback about the DNSSEC validator and its
2371 limitations in the wild. Note however, that DNSSEC support is
2372 probably nothing downstreams should turn on in stable distros just
2373 yet, as it might create incompatibilities with a few DNS servers and
2374 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
2375 automatically whenever we detect such incompatible setups, but there
2376 might be systems we do not cover yet. Hence: please help us testing
2377 the DNSSEC code, leave this on where you can, report back, but then
2378 again don't consider turning this on in your stable, LTS or
2379 production release just yet. (Note that you have to enable
2380 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
2381 and its DNSSEC mode for host name resolution from local
2384 * systemd-resolve conveniently resolves DANE records with the --tlsa
2385 option and OPENPGPKEY records with the --openpgp option. It also
2386 supports dumping raw DNS record data via the new --raw= switch.
2388 * systemd-logind will now by default terminate user processes that are
2389 part of the user session scope unit (session-XX.scope) when the user
2390 logs out. This behavior is controlled by the KillUserProcesses=
2391 setting in logind.conf, and the previous default of "no" is now
2392 changed to "yes". This means that user sessions will be properly
2393 cleaned up after, but additional steps are necessary to allow
2394 intentionally long-running processes to survive logout.
2396 While the user is logged in at least once, user@.service is running,
2397 and any service that should survive the end of any individual login
2398 session can be started at a user service or scope using systemd-run.
2399 systemd-run(1) man page has been extended with an example which shows
2400 how to run screen in a scope unit underneath user@.service. The same
2401 command works for tmux.
2403 After the user logs out of all sessions, user@.service will be
2404 terminated too, by default, unless the user has "lingering" enabled.
2405 To effectively allow users to run long-term tasks even if they are
2406 logged out, lingering must be enabled for them. See loginctl(1) for
2407 details. The default polkit policy was modified to allow users to
2408 set lingering for themselves without authentication.
2410 Previous defaults can be restored at compile time by the
2411 --without-kill-user-processes option to "configure".
2413 * systemd-logind gained new configuration settings SessionsMax= and
2414 InhibitorsMax=, both with a default of 8192. It will not register new
2415 user sessions or inhibitors above this limit.
2417 * systemd-logind will now reload configuration on SIGHUP.
2419 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
2420 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
2421 enable. Also, support for the "io" cgroup controller in the unified
2422 hierarchy has been added, so that the "memory", "pids" and "io" are
2423 now the controllers that are supported on the unified hierarchy.
2425 WARNING: it is not possible to use previous systemd versions with
2426 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
2427 is necessary to also update systemd in the initramfs if using the
2428 unified hierarchy. An updated SELinux policy is also required.
2430 * LLDP support has been extended, and both passive (receive-only) and
2431 active (sender) modes are supported. Passive mode ("routers-only") is
2432 enabled by default in systemd-networkd. Active LLDP mode is enabled
2433 by default for containers on the internal network. The "networkctl
2434 lldp" command may be used to list information gathered. "networkctl
2435 status" will also show basic LLDP information on connected peers now.
2437 * The IAID and DUID unique identifier sent in DHCP requests may now be
2438 configured for the system and each .network file managed by
2439 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
2441 * systemd-networkd gained support for configuring proxy ARP support for
2442 each interface, via the ProxyArp= setting in .network files. It also
2443 gained support for configuring the multicast querier feature of
2444 bridge devices, via the new MulticastQuerier= setting in .netdev
2445 files. Similarly, snooping on the IGMP traffic can be controlled
2446 via the new setting MulticastSnooping=.
2448 A new setting PreferredLifetime= has been added for addresses
2449 configured in .network file to configure the lifetime intended for an
2452 The systemd-networkd DHCP server gained the option EmitRouter=, which
2453 defaults to yes, to configure whether the DHCP Option 3 (Router)
2456 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
2457 systemd-socket-activate and installed into /usr/bin. It is now fully
2460 * systemd-journald now uses separate threads to flush changes to disk
2461 when closing journal files, thus reducing impact of slow disk I/O on
2462 logging performance.
2464 * The sd-journal API gained two new calls
2465 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
2466 can be used to open journal files using file descriptors instead of
2467 file or directory paths. sd_journal_open_container() has been
2468 deprecated, sd_journal_open_directory_fd() should be used instead
2469 with the flag SD_JOURNAL_OS_ROOT.
2471 * journalctl learned a new output mode "-o short-unix" that outputs log
2472 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
2473 UTC). It also gained support for a new --no-hostname setting to
2474 suppress the hostname column in the family of "short" output modes.
2476 * systemd-ask-password now optionally skips printing of the password to
2477 stdout with --no-output which can be useful in scripts.
2479 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
2480 (devices tagged with ID_MAKER_TOOL) are now tagged with
2481 "uaccess" and are available to logged in users.
2483 * The DeviceAllow= unit setting now supports specifiers (with "%").
2485 * "systemctl show" gained a new --value switch, which allows print a
2486 only the contents of a specific unit property, without also printing
2487 the property's name. Similar support was added to "show*" verbs
2488 of loginctl and machinectl that output "key=value" lists.
2490 * A new unit type "generated" was added for files dynamically generated
2491 by generator tools. Similarly, a new unit type "transient" is used
2492 for unit files created using the runtime API. "systemctl enable" will
2493 refuse to operate on such files.
2495 * A new command "systemctl revert" has been added that may be used to
2496 revert to the vendor version of a unit file, in case local changes
2497 have been made by adding drop-ins or overriding the unit file.
2499 * "machinectl clean" gained a new verb to automatically remove all or
2500 just hidden container images.
2502 * systemd-tmpfiles gained support for a new line type "e" for emptying
2503 directories, if they exist, without creating them if they don't.
2505 * systemd-nspawn gained support for automatically patching the UID/GIDs
2506 of the owners and the ACLs of all files and directories in a
2507 container tree to match the UID/GID user namespacing range selected
2508 for the container invocation. This mode is enabled via the new
2509 --private-users-chown switch. It also gained support for
2510 automatically choosing a free, previously unused UID/GID range when
2511 starting a container, via the new --private-users=pick setting (which
2512 implies --private-users-chown). Together, these options for the first
2513 time make user namespacing for nspawn containers fully automatic and
2514 thus deployable. The systemd-nspawn@.service template unit file has
2515 been changed to use this functionality by default.
2517 * systemd-nspawn gained a new --network-zone= switch, that allows
2518 creating ad-hoc virtual Ethernet links between multiple containers,
2519 that only exist as long as at least one container referencing them is
2520 running. This allows easy connecting of multiple containers with a
2521 common link that implements an Ethernet broadcast domain. Each of
2522 these network "zones" may be named relatively freely by the user, and
2523 may be referenced by any number of containers, but each container may
2524 only reference one of these "zones". On the lower level, this is
2525 implemented by an automatically managed bridge network interface for
2526 each zone, that is created when the first container referencing its
2527 zone is created and removed when the last one referencing its zone
2530 * The default start timeout may now be configured on the kernel command
2531 line via systemd.default_timeout_start_sec=. It was already
2532 configurable via the DefaultTimeoutStartSec= option in
2533 /etc/systemd/system.conf.
2535 * Socket units gained a new TriggerLimitIntervalSec= and
2536 TriggerLimitBurst= setting to configure a limit on the activation
2537 rate of the socket unit.
2539 * The LimitNICE= setting now optionally takes normal UNIX nice values
2540 in addition to the raw integer limit value. If the specified
2541 parameter is prefixed with "+" or "-" and is in the range -20..19 the
2542 value is understood as UNIX nice value. If not prefixed like this it
2543 is understood as raw RLIMIT_NICE limit.
2545 * Note that the effect of the PrivateDevices= unit file setting changed
2546 slightly with this release: the per-device /dev file system will be
2547 mounted read-only from this version on, and will have "noexec"
2548 set. This (minor) change of behavior might cause some (exceptional)
2549 legacy software to break, when PrivateDevices=yes is set for its
2550 service. Please leave PrivateDevices= off if you run into problems
2553 * systemd-bootchart has been split out to a separate repository:
2554 https://github.com/systemd/systemd-bootchart
2556 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
2557 merged into the kernel in its current form.
2559 * The compatibility libraries libsystemd-daemon.so,
2560 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
2561 which have been deprecated since systemd-209 have been removed along
2562 with the corresponding pkg-config files. All symbols provided by
2563 those libraries are provided by libsystemd.so.
2565 * The Capabilities= unit file setting has been removed (it is ignored
2566 for backwards compatibility). AmbientCapabilities= and
2567 CapabilityBoundingSet= should be used instead.
2569 * A new special target has been added, initrd-root-device.target,
2570 which creates a synchronization point for dependencies of the root
2571 device in early userspace. Initramfs builders must ensure that this
2572 target is now included in early userspace.
2574 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
2575 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
2576 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
2577 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
2578 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
2579 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
2580 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
2581 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
2582 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
2583 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
2584 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
2585 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
2586 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
2587 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
2588 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
2589 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
2590 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
2591 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
2592 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
2593 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
2594 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
2595 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
2596 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
2597 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
2598 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
2601 — Fairfax, 2016-05-21
2605 * The systemd-resolved DNS resolver service has gained a substantial
2606 set of new features, most prominently it may now act as a DNSSEC
2607 validating stub resolver. DNSSEC mode is currently turned off by
2608 default, but is expected to be turned on by default in one of the
2609 next releases. For now, we invite everybody to test the DNSSEC logic
2610 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
2611 service also gained a full set of D-Bus interfaces, including calls
2612 to configure DNS and DNSSEC settings per link (for use by external
2613 network management software). systemd-resolved and systemd-networkd
2614 now distinguish between "search" and "routing" domains. The former
2615 are used to qualify single-label names, the latter are used purely
2616 for routing lookups within certain domains to specific links.
2617 resolved now also synthesizes RRs for all entries from /etc/hosts.
2619 * The systemd-resolve tool (which is a client utility for
2620 systemd-resolved) has been improved considerably and is now fully
2621 supported and documented. Hence it has moved from /usr/lib/systemd to
2624 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
2627 * The coredump collection logic has been reworked: when a coredump is
2628 collected it is now written to disk, compressed and processed
2629 (including stacktrace extraction) from a new instantiated service
2630 systemd-coredump@.service, instead of directly from the
2631 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
2632 processing large coredumps can take up a substantial amount of
2633 resources and time, and this previously happened entirely outside of
2634 systemd's service supervision. With the new logic the core_pattern
2635 hook only does minimal metadata collection before passing off control
2636 to the new instantiated service, which is configured with a time
2637 limit, a nice level and other settings to minimize negative impact on
2638 the rest of the system. Also note that the new logic will honour the
2639 RLIMIT_CORE setting of the crashed process, which now allows users
2640 and processes to turn off coredumping for their processes by setting
2643 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
2644 and all forked processes by default. Previously, PID 1 would leave
2645 the setting at "0" for all processes, as set by the kernel. Note that
2646 the resource limit traditionally has no effect on the generated
2647 coredumps on the system if the /proc/sys/kernel/core_pattern hook
2648 logic is used. Since the limit is now honoured (see above) its
2649 default has been changed so that the coredumping logic is enabled by
2650 default for all processes, while allowing specific opt-out.
2652 * When the stacktrace is extracted from processes of system users, this
2653 is now done as "systemd-coredump" user, in order to sandbox this
2654 potentially security sensitive parsing operation. (Note that when
2655 processing coredumps of normal users this is done under the user ID
2656 of process that crashed, as before.) Packagers should take notice
2657 that it is now necessary to create the "systemd-coredump" system user
2658 and group at package installation time.
2660 * The systemd-activate socket activation testing tool gained support
2661 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
2662 and --seqpacket switches. It also has been extended to support both
2663 new-style and inetd-style file descriptor passing. Use the new
2664 --inetd switch to request inetd-style file descriptor passing.
2666 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
2667 variable, which takes a boolean value. If set to false, ANSI color
2668 output is disabled in the tools even when run on a terminal that
2671 * The VXLAN support in networkd now supports two new settings
2672 DestinationPort= and PortRange=.
2674 * A new systemd.machine_id= kernel command line switch has been added,
2675 that may be used to set the machine ID in /etc/machine-id if it is
2676 not initialized yet. This command line option has no effect if the
2677 file is already initialized.
2679 * systemd-nspawn gained a new --as-pid2 switch that invokes any
2680 specified command line as PID 2 rather than PID 1 in the
2681 container. In this mode PID 1 is a minimal stub init process that
2682 implements the special POSIX and Linux semantics of PID 1 regarding
2683 signal and child process management. Note that this stub init process
2684 is implemented in nspawn itself and requires no support from the
2685 container image. This new logic is useful to support running
2686 arbitrary commands in the container, as normal processes are
2687 generally not prepared to run as PID 1.
2689 * systemd-nspawn gained a new --chdir= switch for setting the current
2690 working directory for the process started in the container.
2692 * "journalctl /dev/sda" will now output all kernel log messages for
2693 specified device from the current boot, in addition to all devices
2694 that are parents of it. This should make log output about devices
2695 pretty useful, as long as kernel drivers attach enough metadata to
2696 the log messages. (The usual SATA drivers do.)
2698 * The sd-journal API gained two new calls
2699 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
2700 that report whether log data from /run or /var has been found.
2702 * journalctl gained a new switch "--fields" that prints all journal
2703 record field names currently in use in the journal. This is backed
2704 by two new sd-journal API calls sd_journal_enumerate_fields() and
2705 sd_journal_restart_fields().
2707 * Most configurable timeouts in systemd now expect an argument of
2708 "infinity" to turn them off, instead of "0" as before. The semantics
2709 from now on is that a timeout of "0" means "now", and "infinity"
2710 means "never". To maintain backwards compatibility, "0" continues to
2711 turn off previously existing timeout settings.
2713 * "systemctl reload-or-try-restart" has been renamed to "systemctl
2714 try-reload-or-restart" to clarify what it actually does: the "try"
2715 logic applies to both reloading and restarting, not just restarting.
2716 The old name continues to be accepted for compatibility.
2718 * On boot-up, when PID 1 detects that the system clock is behind the
2719 release date of the systemd version in use, the clock is now set
2720 to the latter. Previously, this was already done in timesyncd, in order
2721 to avoid running with clocks set to the various clock epochs such as
2722 1902, 1938 or 1970. With this change the logic is now done in PID 1
2723 in addition to timesyncd during early boot-up, so that it is enforced
2724 before the first process is spawned by systemd. Note that the logic
2725 in timesyncd remains, as it is more comprehensive and ensures
2726 clock monotonicity by maintaining a persistent timestamp file in
2727 /var. Since /var is generally not available in earliest boot or the
2728 initrd, this part of the logic remains in timesyncd, and is not done
2731 * Support for tweaking details in net_cls.class_id through the
2732 NetClass= configuration directive has been removed, as the kernel
2733 people have decided to deprecate that controller in cgroup v2.
2734 Userspace tools such as nftables are moving over to setting rules
2735 that are specific to the full cgroup path of a task, which obsoletes
2736 these controllers anyway. The NetClass= directive is kept around for
2737 legacy compatibility reasons. For a more in-depth description of the
2738 kernel change, please refer to the respective upstream commit:
2740 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
2742 * A new service setting RuntimeMaxSec= has been added that may be used
2743 to specify a maximum runtime for a service. If the timeout is hit, the
2744 service is terminated and put into a failure state.
2746 * A new service setting AmbientCapabilities= has been added. It allows
2747 configuration of additional Linux process capabilities that are
2748 passed to the activated processes. This is only available on very
2751 * The process resource limit settings in service units may now be used
2752 to configure hard and soft limits individually.
2754 * The various libsystemd APIs such as sd-bus or sd-event now publicly
2755 expose support for gcc's __attribute__((cleanup())) C extension.
2756 Specifically, for many object destructor functions alternative
2757 versions have been added that have names suffixed with "p" and take a
2758 pointer to a pointer to the object to destroy, instead of just a
2759 pointer to the object itself. This is useful because these destructor
2760 functions may be used directly as parameters to the cleanup
2761 construct. Internally, systemd has been a heavy user of this GCC
2762 extension for a long time, and with this change similar support is
2763 now available to consumers of the library outside of systemd. Note
2764 that by using this extension in your sources compatibility with old
2765 and strictly ANSI compatible C compilers is lost. However, all gcc or
2766 LLVM versions of recent years support this extension.
2768 * Timer units gained support for a new setting RandomizedDelaySec= that
2769 allows configuring some additional randomized delay to the configured
2770 time. This is useful to spread out timer events to avoid load peaks in
2771 clusters or larger setups.
2773 * Calendar time specifications now support sub-second accuracy.
2775 * Socket units now support listening on SCTP and UDP-lite protocol
2778 * The sd-event API now comes with a full set of man pages.
2780 * Older versions of systemd contained experimental support for
2781 compressing journal files and coredumps with the LZ4 compressor that
2782 was not compatible with the lz4 binary (due to API limitations of the
2783 lz4 library). This support has been removed; only support for files
2784 compatible with the lz4 binary remains. This LZ4 logic is now
2785 officially supported and no longer considered experimental.
2787 * The dkr image import logic has been removed again from importd. dkr's
2788 micro-services focus doesn't fit into the machine image focus of
2789 importd, and quickly got out of date with the upstream dkr API.
2791 * Creation of the /run/lock/lockdev/ directory was dropped from
2792 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
2793 been available for many years. If you still need this, you need to
2794 create your own tmpfiles.d config file with:
2796 d /run/lock/lockdev 0775 root lock -
2798 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
2799 and RebootArgument= have been moved from the [Service] section of
2800 unit files to [Unit], and they are now supported on all unit types,
2801 not just service units. Of course, elogind will continue to
2802 understand these settings also at the old location, in order to
2803 maintain compatibility.
2805 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
2806 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
2807 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
2808 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
2809 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
2810 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
2811 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
2812 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
2813 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
2814 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
2815 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
2816 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
2817 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
2818 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
2819 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
2820 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
2821 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
2822 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
2823 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2825 — Berlin, 2016-02-11
2829 * A number of properties previously only settable in unit
2830 files are now also available as properties to set when
2831 creating transient units programmatically via the bus, as it
2832 is exposed with systemd-run's --property=
2833 setting. Specifically, these are: SyslogIdentifier=,
2834 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
2835 EnvironmentFile=, ReadWriteDirectories=,
2836 ReadOnlyDirectories=, InaccessibleDirectories=,
2837 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
2839 * When creating transient services via the bus API it is now
2840 possible to pass in a set of file descriptors to use as
2841 STDIN/STDOUT/STDERR for the invoked process.
2843 * Slice units may now be created transiently via the bus APIs,
2844 similar to the way service and scope units may already be
2845 created transiently.
2847 * Wherever systemd expects a calendar timestamp specification
2848 (like in journalctl's --since= and --until= switches) UTC
2849 timestamps are now supported. Timestamps suffixed with "UTC"
2850 are now considered to be in Universal Time Coordinated
2851 instead of the local timezone. Also, timestamps may now
2852 optionally be specified with sub-second accuracy. Both of
2853 these additions also apply to recurring calendar event
2854 specification, such as OnCalendar= in timer units.
2856 * journalctl gained a new "--sync" switch that asks the
2857 journal daemon to write all so far unwritten log messages to
2858 disk and sync the files, before returning.
2860 * systemd-tmpfiles learned two new line types "q" and "Q" that
2861 operate like "v", but also set up a basic btrfs quota
2862 hierarchy when used on a btrfs file system with quota
2865 * tmpfiles' "v", "q" and "Q" will now create a plain directory
2866 instead of a subvolume (even on a btrfs file system) if the
2867 root directory is a plain directory, and not a
2868 subvolume. This should simplify things with certain chroot()
2869 environments which are not aware of the concept of btrfs
2872 * systemd-detect-virt gained a new --chroot switch to detect
2873 whether execution takes place in a chroot() environment.
2875 * CPUAffinity= now takes CPU index ranges in addition to
2878 * The various memory-related resource limit settings (such as
2879 LimitAS=) now understand the usual K, M, G, ... suffixes to
2880 the base of 1024 (IEC). Similar, the time-related resource
2881 limit settings understand the usual min, h, day, ...
2884 * There's a new system.conf setting DefaultTasksMax= to
2885 control the default TasksMax= setting for services and
2886 scopes running on the system. (TasksMax= is the primary
2887 setting that exposes the "pids" cgroup controller on systemd
2888 and was introduced in the previous systemd release.) The
2889 setting now defaults to 512, which means services that are
2890 not explicitly configured otherwise will only be able to
2891 create 512 processes or threads at maximum, from this
2892 version on. Note that this means that thread- or
2893 process-heavy services might need to be reconfigured to set
2894 TasksMax= to a higher value. It is sufficient to set
2895 TasksMax= in these specific unit files to a higher value, or
2896 even "infinity". Similar, there's now a logind.conf setting
2897 UserTasksMax= that defaults to 4096 and limits the total
2898 number of processes or tasks each user may own
2899 concurrently. nspawn containers also have the TasksMax=
2900 value set by default now, to 8192. Note that all of this
2901 only has an effect if the "pids" cgroup controller is
2902 enabled in the kernel. The general benefit of these changes
2903 should be a more robust and safer system, that provides a
2904 certain amount of per-service fork() bomb protection.
2906 * systemd-nspawn gained the new --network-veth-extra= switch
2907 to define additional and arbitrarily-named virtual Ethernet
2908 links between the host and the container.
2910 * A new service execution setting PassEnvironment= has been
2911 added that allows importing select environment variables
2912 from PID1's environment block into the environment block of
2915 * Timer units gained support for a new RemainAfterElapse=
2916 setting which takes a boolean argument. It defaults to on,
2917 exposing behaviour unchanged to previous releases. If set to
2918 off, timer units are unloaded after they elapsed if they
2919 cannot elapse again. This is particularly useful for
2920 transient timer units, which shall not stay around longer
2921 than until they first elapse.
2923 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
2924 default now (the kernel default is 16). This is beneficial
2925 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
2926 allows substantially larger numbers of queued
2927 datagrams. This should increase the capability of systemd to
2928 parallelize boot-up, as logging and sd_notify() are unlikely
2929 to stall execution anymore. If you need to change the value
2930 from the new defaults, use the usual sysctl.d/ snippets.
2932 * The compression framing format used by the journal or
2933 coredump processing has changed to be in line with what the
2934 official LZ4 tools generate. LZ4 compression support in
2935 systemd was considered unsupported previously, as the format
2936 was not compatible with the normal tools. With this release
2937 this has changed now, and it is hence safe for downstream
2938 distributions to turn it on. While not compressing as well
2939 as the XZ, LZ4 is substantially faster, which makes
2940 it a good default choice for the compression logic in the
2941 journal and in coredump handling.
2943 * Any reference to /etc/mtab has been dropped from
2944 systemd. The file has been obsolete since a while, but
2945 systemd refused to work on systems where it was incorrectly
2946 set up (it should be a symlink or non-existent). Please make
2947 sure to update to util-linux 2.27.1 or newer in conjunction
2948 with this systemd release, which also drops any reference to
2949 /etc/mtab. If you maintain a distribution make sure that no
2950 software you package still references it, as this is a
2951 likely source of bugs. There's also a glibc bug pending,
2952 asking for removal of any reference to this obsolete file:
2954 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
2956 Note that only util-linux versions built with
2957 --enable-libmount-force-mountinfo are supported.
2959 * Support for the ".snapshot" unit type has been removed. This
2960 feature turned out to be little useful and little used, and
2961 has now been removed from the core and from systemctl.
2963 * The dependency types RequiresOverridable= and
2964 RequisiteOverridable= have been removed from systemd. They
2965 have been used only very sparingly to our knowledge and
2966 other options that provide a similar effect (such as
2967 systemctl --mode=ignore-dependencies) are much more useful
2968 and commonly used. Moreover, they were only half-way
2969 implemented as the option to control behaviour regarding
2970 these dependencies was never added to systemctl. By removing
2971 these dependency types the execution engine becomes a bit
2972 simpler. Unit files that use these dependencies should be
2973 changed to use the non-Overridable dependency types
2974 instead. In fact, when parsing unit files with these
2975 options, that's what systemd will automatically convert them
2976 too, but it will also warn, asking users to fix the unit
2977 files accordingly. Removal of these dependency types should
2978 only affect a negligible number of unit files in the wild.
2980 * Behaviour of networkd's IPForward= option changed
2981 (again). It will no longer maintain a per-interface setting,
2982 but propagate one way from interfaces where this is enabled
2983 to the global kernel setting. The global setting will be
2984 enabled when requested by a network that is set up, but
2985 never be disabled again. This change was made to make sure
2986 IPv4 and IPv6 behaviour regarding packet forwarding is
2987 similar (as the Linux IPv6 stack does not support
2988 per-interface control of this setting) and to minimize
2991 * In unit files the behaviour of %u, %U, %h, %s has
2992 changed. These specifiers will now unconditionally resolve
2993 to the various user database fields of the user that the
2994 systemd instance is running as, instead of the user
2995 configured in the specific unit via User=. Note that this
2996 effectively doesn't change much, as resolving of these
2997 specifiers was already turned off in the --system instance
2998 of systemd, as we cannot do NSS lookups from PID 1. In the
2999 --user instance of systemd these specifiers where correctly
3000 resolved, but hardly made any sense, since the user instance
3001 lacks privileges to do user switches anyway, and User= is
3002 hence useless. Moreover, even in the --user instance of
3003 systemd behaviour was awkward as it would only take settings
3004 from User= assignment placed before the specifier into
3005 account. In order to unify and simplify the logic around
3006 this the specifiers will now always resolve to the
3007 credentials of the user invoking the manager (which in case
3008 of PID 1 is the root user).
3010 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
3011 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
3012 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
3013 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
3014 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
3015 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
3016 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
3017 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
3018 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
3019 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
3020 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
3021 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
3022 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
3023 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
3026 — Berlin, 2015-11-18
3030 * systemd now depends on util-linux v2.27. More specifically,
3031 the newly added mount monitor feature in libmount now
3032 replaces systemd's former own implementation.
3034 * libmount mandates /etc/mtab not to be regular file, and
3035 systemd now enforces this condition at early boot.
3036 /etc/mtab has been deprecated and warned about for a very
3037 long time, so systems running systemd should already have
3038 stopped having this file around as anything else than a
3039 symlink to /proc/self/mounts.
3041 * Support for the "pids" cgroup controller has been added. It
3042 allows accounting the number of tasks in a cgroup and
3043 enforcing limits on it. This adds two new setting
3044 TasksAccounting= and TasksMax= to each unit, as well as a
3045 global option DefaultTasksAccounting=.
3047 * Support for the "net_cls" cgroup controller has been added.
3048 It allows assigning a net class ID to each task in the
3049 cgroup, which can then be used in firewall rules and traffic
3050 shaping configurations. Note that the kernel netfilter net
3051 class code does not currently work reliably for ingress
3052 packets on unestablished sockets.
3054 This adds a new config directive called NetClass= to CGroup
3055 enabled units. Allowed values are positive numbers for fixed
3056 assignments and "auto" for picking a free value
3059 * 'systemctl is-system-running' now returns 'offline' if the
3060 system is not booted with systemd. This command can now be
3061 used as a substitute for 'systemd-notify --booted'.
3063 * Watchdog timeouts have been increased to 3 minutes for all
3064 in-tree service files. Apparently, disk IO issues are more
3065 frequent than we hoped, and user reported >1 minute waiting
3068 * 'machine-id-commit' functionality has been merged into
3069 'machine-id-setup --commit'. The separate binary has been
3072 * The WorkingDirectory= directive in unit files may now be set
3073 to the special value '~'. In this case, the working
3074 directory is set to the home directory of the user
3075 configured in User=.
3077 * "machinectl shell" will now open the shell in the home
3078 directory of the selected user by default.
3080 * The CrashChVT= configuration file setting is renamed to
3081 CrashChangeVT=, following our usual logic of not
3082 abbreviating unnecessarily. The old directive is still
3083 supported for compat reasons. Also, this directive now takes
3084 an integer value between 1 and 63, or a boolean value. The
3085 formerly supported '-1' value for disabling stays around for
3088 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
3089 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
3090 RootDirectory= properties can now be set for transient
3093 * The systemd-analyze tool gained a new "set-log-target" verb
3094 to change the logging target the system manager logs to
3095 dynamically during runtime. This is similar to how
3096 "systemd-analyze set-log-level" already changes the log
3099 * In nspawn /sys is now mounted as tmpfs, with only a selected
3100 set of subdirectories mounted in from the real sysfs. This
3101 enhances security slightly, and is useful for ensuring user
3102 namespaces work correctly.
3104 * Support for USB FunctionFS activation has been added. This
3105 allows implementation of USB gadget services that are
3106 activated as soon as they are requested, so that they don't
3107 have to run continuously, similar to classic socket
3110 * The "systemctl exit" command now optionally takes an
3111 additional parameter that sets the exit code to return from
3112 the systemd manager when exiting. This is only relevant when
3113 running the systemd user instance, or when running the
3114 system instance in a container.
3116 * sd-bus gained the new API calls sd_bus_path_encode_many()
3117 and sd_bus_path_decode_many() that allow easy encoding and
3118 decoding of multiple identifier strings inside a D-Bus
3119 object path. Another new call sd_bus_default_flush_close()
3120 has been added to flush and close per-thread default
3123 * systemd-cgtop gained support for a -M/--machine= switch to
3124 show the control groups within a certain container only.
3126 * "systemctl kill" gained support for an optional --fail
3127 switch. If specified the requested operation will fail of no
3128 processes have been killed, because the unit had no
3129 processes attached, or similar.
3131 * A new systemd.crash_reboot=1 kernel command line option has
3132 been added that triggers a reboot after crashing. This can
3133 also be set through CrashReboot= in systemd.conf.
3135 * The RuntimeDirectory= setting now understands unit
3136 specifiers like %i or %f.
3138 * A new (still internal) library API sd-ipv4acd has been added,
3139 that implements address conflict detection for IPv4. It's
3140 based on code from sd-ipv4ll, and will be useful for
3141 detecting DHCP address conflicts.
3143 * File descriptors passed during socket activation may now be
3144 named. A new API sd_listen_fds_with_names() is added to
3145 access the names. The default names may be overridden,
3146 either in the .socket file using the FileDescriptorName=
3147 parameter, or by passing FDNAME= when storing the file
3148 descriptors using sd_notify().
3150 * systemd-networkd gained support for:
3152 - Setting the IPv6 Router Advertisement settings via
3153 IPv6AcceptRouterAdvertisements= in .network files.
3155 - Configuring the HelloTimeSec=, MaxAgeSec= and
3156 ForwardDelaySec= bridge parameters in .netdev files.
3158 - Configuring PreferredSource= for static routes in
3161 * The "ask-password" framework used to query for LUKS harddisk
3162 passwords or SSL passwords during boot gained support for
3163 caching passwords in the kernel keyring, if it is
3164 available. This makes sure that the user only has to type in
3165 a passphrase once if there are multiple objects to unlock
3166 with the same one. Previously, such password caching was
3167 available only when Plymouth was used; this moves the
3168 caching logic into the systemd codebase itself. The
3169 "systemd-ask-password" utility gained a new --keyname=
3170 switch to control which kernel keyring key to use for
3171 caching a password in. This functionality is also useful for
3172 enabling display managers such as gdm to automatically
3173 unlock the user's GNOME keyring if its passphrase, the
3174 user's password and the harddisk password are the same, if
3175 gdm-autologin is used.
3177 * When downloading tar or raw images using "machinectl
3178 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
3179 file is now also downloaded, if it is available and stored
3180 next to the image file.
3182 * Units of type ".socket" gained a new boolean setting
3183 Writable= which is only useful in conjunction with
3184 ListenSpecial=. If true, enables opening the specified
3185 special file in O_RDWR mode rather than O_RDONLY mode.
3187 * systemd-rfkill has been reworked to become a singleton
3188 service that is activated through /dev/rfkill on each rfkill
3189 state change and saves the settings to disk. This way,
3190 systemd-rfkill is now compatible with devices that exist
3191 only intermittendly, and even restores state if the previous
3192 system shutdown was abrupt rather than clean.
3194 * The journal daemon gained support for vacuuming old journal
3195 files controlled by the number of files that shall remain,
3196 in addition to the already existing control by size and by
3197 date. This is useful as journal interleaving performance
3198 degrades with too many separate journal files, and allows
3199 putting an effective limit on them. The new setting defaults
3200 to 100, but this may be changed by setting SystemMaxFiles=
3201 and RuntimeMaxFiles= in journald.conf. Also, the
3202 "journalctl" tool gained the new --vacuum-files= switch to
3203 manually vacuum journal files to leave only the specified
3204 number of files in place.
3206 * udev will now create /dev/disk/by-path links for ATA devices
3207 on kernels where that is supported.
3209 * Galician, Serbian, Turkish and Korean translations were added.
3211 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
3212 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
3213 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
3214 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
3215 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
3216 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
3217 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
3218 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
3219 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
3220 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
3221 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
3222 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
3223 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
3224 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
3225 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
3226 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
3227 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
3228 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
3230 — Berlin, 2015-10-07
3234 * The DHCP implementation of systemd-networkd gained a set of
3237 - The DHCP server now supports emitting DNS and NTP
3238 information. It may be enabled and configured via
3239 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
3240 and NTP information is enabled, but no servers are
3241 configured, the corresponding uplink information (if there
3242 is any) is propagated.
3244 - Server and client now support transmission and reception
3245 of timezone information. It can be configured via the
3246 newly introduced network options UseTimezone=,
3247 EmitTimezone=, and Timezone=. Transmission of timezone
3248 information is enabled between host and containers by
3249 default now: the container will change its local timezone
3250 to what the host has set.
3252 - Lease timeouts can now be configured via
3253 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
3255 - The DHCP server improved on the stability of
3256 leases. Clients are more likely to get the same lease
3257 information back, even if the server loses state.
3259 - The DHCP server supports two new configuration options to
3260 control the lease address pool metrics, PoolOffset= and
3263 * The encapsulation limit of tunnels in systemd-networkd may
3264 now be configured via 'EncapsulationLimit='. It allows
3265 modifying the maximum additional levels of encapsulation
3266 that are permitted to be prepended to a packet.
3268 * systemd now supports the concept of user buses replacing
3269 session buses, if used with dbus-1.10 (and enabled via dbus
3270 --enable-user-session). It previously only supported this on
3271 kdbus-enabled systems, and this release expands this to
3272 'dbus-daemon' systems.
3274 * systemd-networkd now supports predictable interface names
3277 * systemd now optionally supports the new Linux kernel
3278 "unified" control group hierarchy. If enabled via the kernel
3279 command-line option 'systemd.unified_cgroup_hierarchy=1',
3280 systemd will try to mount the unified cgroup hierarchy
3281 directly on /sys/fs/cgroup. If not enabled, or not
3282 available, systemd will fall back to the legacy cgroup
3283 hierarchy setup, as before. Host system and containers can
3284 mix and match legacy and unified hierarchies as they
3285 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
3286 environment variable to individually select the hierarchy to
3287 use for executed containers. By default, nspawn will use the
3288 unified hierarchy for the containers if the host uses the
3289 unified hierarchy, and the legacy hierarchy otherwise.
3290 Please note that at this point the unified hierarchy is an
3291 experimental kernel feature and is likely to change in one
3292 of the next kernel releases. Therefore, it should not be
3293 enabled by default in downstream distributions yet. The
3294 minimum required kernel version for the unified hierarchy to
3295 work is 4.2. Note that when the unified hierarchy is used
3296 for the first time delegated access to controllers is
3297 safe. Because of this systemd-nspawn containers will get
3298 access to controllers now, as will systemd user
3299 sessions. This means containers and user sessions may now
3300 manage their own resources, partitioning up what the system
3303 * A new special scope unit "init.scope" has been introduced
3304 that encapsulates PID 1 of the system. It may be used to
3305 determine resource usage and enforce resource limits on PID
3306 1 itself. PID 1 hence moved out of the root of the control
3309 * The cgtop tool gained support for filtering out kernel
3310 threads when counting tasks in a control group. Also, the
3311 count of processes is now recursively summed up by
3312 default. Two options -k and --recursive= have been added to
3313 revert to old behaviour. The tool has also been updated to
3314 work correctly in containers now.
3316 * systemd-nspawn's --bind= and --bind-ro= options have been
3317 extended to allow creation of non-recursive bind mounts.
3319 * libsystemd gained two new calls sd_pid_get_cgroup() and
3320 sd_peer_get_cgroup() which return the control group path of
3321 a process or peer of a connected AF_UNIX socket. This
3322 function call is particularly useful when implementing
3323 delegated subtrees support in the control group hierarchy.
3325 * The "sd-event" event loop API of libsystemd now supports
3326 correct dequeuing of real-time signals, without losing
3329 * When systemd requests a PolicyKit decision when managing
3330 units it will now add additional fields to the request,
3331 including unit name and desired operation. This enables more
3332 powerful PolicyKit policies, that make decisions depending
3333 on these parameters.
3335 * nspawn learnt support for .nspawn settings files, that may
3336 accompany the image files or directories of containers, and
3337 may contain additional settings for the container. This is
3338 an alternative to configuring container parameters via the
3339 nspawn command line.
3341 Contributions from: Cristian Rodríguez, Daniel Mack, David
3342 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
3343 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
3344 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
3345 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
3346 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
3347 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
3348 Andersen, Tom Gundersen, Torstein Husebø
3350 — Berlin, 2015-09-08
3354 * machinectl gained a new verb 'shell' which opens a fresh
3355 shell on the target container or the host. It is similar to
3356 the existing 'login' command of machinectl, but spawns the
3357 shell directly without prompting for username or
3358 password. The pseudo machine '.host' now refers to the local
3359 host and is used by default. Hence, 'machinectl shell' can
3360 be used as replacement for 'su -' which spawns a session as
3361 a fresh systemd unit in a way that is fully isolated from
3362 the originating session.
3364 * systemd-networkd learned to cope with private-zone DHCP
3365 options and allows other programs to query the values.
3367 * SELinux access control when enabling/disabling units is no
3368 longer enforced with this release. The previous
3369 implementation was incorrect, and a new corrected
3370 implementation is not yet available. As unit file operations
3371 are still protected via PolicyKit and D-Bus policy this is
3372 not a security problem. Yet, distributions which care about
3373 optimal SELinux support should probably not stabilize on
3376 * sd-bus gained support for matches of type "arg0has=", that
3377 test for membership of strings in string arrays sent in bus
3380 * systemd-resolved now dumps the contents of its DNS and LLMNR
3381 caches to the logs on reception of the SIGUSR1 signal. This
3382 is useful to debug DNS behaviour.
3384 * The coredumpctl tool gained a new --directory= option to
3385 operate on journal files in a specific directory.
3387 * "systemctl reboot" and related commands gained a new
3388 "--message=" option which may be used to set a free-text
3389 wall message when shutting down or rebooting the
3390 system. This message is also logged, which is useful for
3391 figuring out the reason for a reboot or shutdown a
3394 * The "systemd-resolve-host" tool's -i switch now takes
3395 network interface numbers as alternative to interface names.
3397 * A new unit file setting for services has been introduced:
3398 UtmpMode= allows configuration of how precisely systemd
3399 handles utmp and wtmp entries for the service if this is
3400 enabled. This allows writing services that appear similar to
3401 user sessions in the output of the "w", "who", "last" and
3404 * systemd-resolved will now locally synthesize DNS resource
3405 records for the "localhost" and "gateway" domains as well as
3406 the local hostname. This should ensure that clients querying
3407 RRs via resolved will get similar results as those going via
3408 NSS, if nss-myhostname is enabled.
3410 Contributions from: Alastair Hughes, Alex Crawford, Daniel
3411 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
3412 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
3413 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
3414 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
3415 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
3416 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
3417 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
3418 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
3419 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
3420 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
3421 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3423 — Berlin, 2015-08-27
3427 * The systemd-efi-boot-generator functionality was merged into
3428 systemd-gpt-auto-generator.
3430 * systemd-networkd now supports Group Policy for vxlan
3431 devices. It can be enabled via the new boolean configuration
3432 option called 'GroupPolicyExtension='.
3434 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
3435 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
3436 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
3438 — Berlin, 2015-07-31
3442 * The python-systemd code has been removed from the systemd repository.
3443 A new repository has been created which accommodates the code from
3444 now on, and we kindly ask distributions to create a separate package
3445 for this: https://github.com/systemd/python-systemd
3447 * The systemd daemon will now reload its main configuration
3448 (/etc/systemd/system.conf) on daemon-reload.
3450 * sd-dhcp now exposes vendor specific extensions via
3451 sd_dhcp_lease_get_vendor_specific().
3453 * systemd-networkd gained a number of new configuration options.
3455 - A new boolean configuration option for TAP devices called
3456 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
3457 device, thus allowing to send and receive GSO packets.
3459 - A new tunnel configuration option called 'CopyDSCP='.
3460 If enabled, the DSCP field of ip6 tunnels is copied into the
3461 decapsulated packet.
3463 - A set of boolean bridge configuration options were added.
3464 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
3465 and 'UnicastFlood=' are now parsed by networkd and applied to the
3466 respective bridge link device via the respective IFLA_BRPORT_*
3469 - A new string configuration option to override the hostname sent
3470 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
3471 is true, networkd will use the configured hostname instead of the
3472 system hostname when sending DHCP requests.
3474 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
3475 networkd will configure the IPv6 flow-label of the tunnel device
3476 according to RFC2460.
3478 - The 'macvtap' virtual network devices are now supported, similar to
3479 the already supported 'macvlan' devices.
3481 * systemd-resolved now implements RFC5452 to improve resilience against
3482 cache poisoning. Additionally, source port randomization is enabled
3483 by default to further protect against DNS spoofing attacks.
3485 * nss-mymachines now supports translating UIDs and GIDs of running
3486 containers with user-namespaces enabled. If a container 'foo'
3487 translates a host uid 'UID' to the container uid 'TUID', then
3488 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
3489 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
3490 mapped as 'vg-foo-TGID'.
3492 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
3493 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
3494 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
3495 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
3496 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
3497 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
3498 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
3499 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
3500 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
3501 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
3503 — Berlin, 2015-07-29
3507 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
3508 There are no known issues with current sysfs, and udev does not need
3509 or should be used to work around such bugs.
3511 * udev does no longer enable USB HID power management. Several reports
3512 indicate, that some devices cannot handle that setting.
3514 * The udev accelerometer helper was removed. The functionality
3515 is now fully included in iio-sensor-proxy. But this means,
3516 older iio-sensor-proxy versions will no longer provide
3517 accelerometer/orientation data with this systemd version.
3518 Please upgrade iio-sensor-proxy to version 1.0.
3520 * networkd gained a new configuration option IPv6PrivacyExtensions=
3521 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
3522 for Stateless Address") on selected networks.
3524 * For the sake of fewer build-time dependencies and less code in the
3525 main repository, the python bindings are about to be removed in the
3526 next release. A new repository has been created which accommodates
3527 the code from now on, and we kindly ask distributions to create a
3528 separate package for this. The removal will take place in v223.
3530 https://github.com/systemd/python-systemd
3532 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
3533 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
3534 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
3535 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
3536 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
3537 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
3538 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
3539 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
3540 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3541 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3543 — Berlin, 2015-07-07
3547 * The sd-bus.h and sd-event.h APIs have now been declared
3548 stable and have been added to the official interface of
3549 libsystemd.so. sd-bus implements an alternative D-Bus client
3550 library, that is relatively easy to use, very efficient and
3551 supports both classic D-Bus as well as kdbus as transport
3552 backend. sd-event is a generic event loop abstraction that
3553 is built around Linux epoll, but adds features such as event
3554 prioritization or efficient timer handling. Both APIs are good
3555 choices for C programs looking for a bus and/or event loop
3556 implementation that is minimal and does not have to be
3557 portable to other kernels.
3559 * kdbus support is no longer compile-time optional. It is now
3560 always built-in. However, it can still be disabled at
3561 runtime using the kdbus=0 kernel command line setting, and
3562 that setting may be changed to default to off, by specifying
3563 --disable-kdbus at build-time. Note though that the kernel
3564 command line setting has no effect if the kdbus.ko kernel
3565 module is not installed, in which case kdbus is (obviously)
3566 also disabled. We encourage all downstream distributions to
3567 begin testing kdbus by adding it to the kernel images in the
3568 development distributions, and leaving kdbus support in
3571 * The minimal required util-linux version has been bumped to
3574 * Support for chkconfig (--enable-chkconfig) was removed in
3575 favor of calling an abstraction tool
3576 /lib/systemd/systemd-sysv-install. This needs to be
3577 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
3578 in README for details.
3580 * If there's a systemd unit and a SysV init script for the
3581 same service name, and the user executes "systemctl enable"
3582 for it (or a related call), then this will now enable both
3583 (or execute the related operation on both), not just the
3586 * The libudev API documentation has been converted from gtkdoc
3589 * gudev has been removed from the systemd tree, it is now an
3592 * The systemd-cgtop tool learnt a new --raw switch to generate
3593 "raw" (machine parsable) output.
3595 * networkd's IPForwarding= .network file setting learnt the
3596 new setting "kernel", which ensures that networkd does not
3597 change the IP forwarding sysctl from the default kernel
3600 * The systemd-logind bus API now exposes a new boolean
3601 property "Docked" that reports whether logind considers the
3602 system "docked", i.e. connected to a docking station or not.
3604 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
3605 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
3606 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
3607 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
3608 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
3609 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
3610 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
3611 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
3612 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
3613 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
3614 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
3615 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
3616 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
3617 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3618 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
3619 Fink, Zbigniew Jędrzejewski-Szmek
3621 — Berlin, 2015-06-19
3625 * The gudev library has been extracted into a separate repository
3626 available at: https://git.gnome.org/browse/libgudev/
3627 It is now managed as part of the Gnome project. Distributions
3628 are recommended to pass --disable-gudev to systemd and use
3629 gudev from the Gnome project instead. gudev is still included
3630 in systemd, for now. It will be removed soon, though. Please
3631 also see the announcement-thread on systemd-devel:
3632 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
3634 * systemd now exposes a CPUUsageNSec= property for each
3635 service unit on the bus, that contains the overall consumed
3636 CPU time of a service (the sum of what each process of the
3637 service consumed). This value is only available if
3638 CPUAccounting= is turned on for a service, and is then shown
3639 in the "systemctl status" output.
3641 * Support for configuring alternative mappings of the old SysV
3642 runlevels to systemd targets has been removed. They are now
3643 hardcoded in a way that runlevels 2, 3, 4 all map to
3644 multi-user.target and 5 to graphical.target (which
3645 previously was already the default behaviour).
3647 * The auto-mounter logic gained support for mount point
3648 expiry, using a new TimeoutIdleSec= setting in .automount
3649 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
3651 * The EFI System Partition (ESP) as mounted to /boot by
3652 systemd-efi-boot-generator will now be unmounted
3653 automatically after 2 minutes of not being used. This should
3654 minimize the risk of ESP corruptions.
3656 * New /etc/fstab options x-systemd.requires= and
3657 x-systemd.requires-mounts-for= are now supported to express
3658 additional dependencies for mounts. This is useful for
3659 journalling file systems that support external journal
3660 devices or overlay file systems that require underlying file
3661 systems to be mounted.
3663 * systemd does not support direct live-upgrades (via systemctl
3664 daemon-reexec) from versions older than v44 anymore. As no
3665 distribution we are aware of shipped such old versions in a
3666 stable release this should not be problematic.
3668 * When systemd forks off a new per-connection service instance
3669 it will now set the $REMOTE_ADDR environment variable to the
3670 remote IP address, and $REMOTE_PORT environment variable to
3671 the remote IP port. This behaviour is similar to the
3672 corresponding environment variables defined by CGI.
3674 * systemd-networkd gained support for uplink failure
3675 detection. The BindCarrier= option allows binding interface
3676 configuration dynamically to the link sense of other
3677 interfaces. This is useful to achieve behaviour like in
3680 * systemd-networkd gained support for configuring the DHCP
3681 client identifier to use when requesting leases.
3683 * systemd-networkd now has a per-network UseNTP= option to
3684 configure whether NTP server information acquired via DHCP
3685 is passed on to services like systemd-timesyncd.
3687 * systemd-networkd gained support for vti6 tunnels.
3689 * Note that systemd-networkd manages the sysctl variable
3690 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
3691 it is configured for since v219. The variable controls IP
3692 forwarding, and is a per-interface alternative to the global
3693 /proc/sys/net/ipv[46]/ip_forward. This setting is
3694 configurable in the IPForward= option, which defaults to
3695 "no". This means if networkd is used for an interface it is
3696 no longer sufficient to set the global sysctl option to turn
3697 on IP forwarding! Instead, the .network file option
3698 IPForward= needs to be turned on! Note that the
3699 implementation of this behaviour was broken in v219 and has
3702 * Many bonding and vxlan options are now configurable in
3705 * systemd-nspawn gained a new --property= setting to set unit
3706 properties for the container scope. This is useful for
3707 setting resource parameters (e.g. "CPUShares=500") on
3708 containers started from the command line.
3710 * systemd-nspawn gained a new --private-users= switch to make
3711 use of user namespacing available on recent Linux kernels.
3713 * systemd-nspawn may now be called as part of a shell pipeline
3714 in which case the pipes used for stdin and stdout are passed
3715 directly to the process invoked in the container, without
3716 indirection via a pseudo tty.
3718 * systemd-nspawn gained a new switch to control the UNIX
3719 signal to use when killing the init process of the container
3722 * systemd-nspawn gained a new --overlay= switch for mounting
3723 overlay file systems into the container using the new kernel
3726 * When a container image is imported via systemd-importd and
3727 the host file system is not btrfs, a loopback block device
3728 file is created in /var/lib/machines.raw with a btrfs file
3729 system inside. It is then mounted to /var/lib/machines to
3730 enable btrfs features for container management. The loopback
3731 file and btrfs file system is grown as needed when container
3732 images are imported via systemd-importd.
3734 * systemd-machined/systemd-importd gained support for btrfs
3735 quota, to enforce container disk space limits on disk. This
3736 is exposed in "machinectl set-limit".
3738 * systemd-importd now can import containers from local .tar,
3739 .raw and .qcow2 images, and export them to .tar and .raw. It
3740 can also import dkr v2 images now from the network (on top
3743 * systemd-importd gained support for verifying downloaded
3744 images with gpg2 (previously only gpg1 was supported).
3746 * systemd-machined, systemd-logind, systemd: most bus calls
3747 are now accessible to unprivileged processes via
3748 PolicyKit. Also, systemd-logind will now allow users to kill
3749 their own sessions without further privileges or
3752 * systemd-shutdownd has been removed. This service was
3753 previously responsible for implementing scheduled shutdowns
3754 as exposed in /usr/bin/shutdown's time parameter. This
3755 functionality has now been moved into systemd-logind and is
3756 accessible via a bus interface.
3758 * "systemctl reboot" gained a new switch --firmware-setup that
3759 can be used to reboot into the EFI firmware setup, if that
3760 is available. systemd-logind now exposes an API on the bus
3761 to trigger such reboots, in case graphical desktop UIs want
3762 to cover this functionality.
3764 * "systemctl enable", "systemctl disable" and "systemctl mask"
3765 now support a new "--now" switch. If specified the units
3766 that are enabled will also be started, and the ones
3767 disabled/masked also stopped.
3769 * The Gummiboot EFI boot loader tool has been merged into
3770 systemd, and renamed to "systemd-boot". The bootctl tool has been
3771 updated to support systemd-boot.
3773 * An EFI kernel stub has been added that may be used to create
3774 kernel EFI binaries that contain not only the actual kernel,
3775 but also an initrd, boot splash, command line and OS release
3776 information. This combined binary can then be signed as a
3777 single image, so that the firmware can verify it all in one
3778 step. systemd-boot has special support for EFI binaries created
3779 like this and can extract OS release information from them
3780 and show them in the boot menu. This functionality is useful
3781 to implement cryptographically verified boot schemes.
3783 * Optional support has been added to systemd-fsck to pass
3784 fsck's progress report to an AF_UNIX socket in the file
3787 * udev will no longer create device symlinks for all block
3788 devices by default. A blacklist for excluding special block
3789 devices from this logic has been turned into a whitelist
3790 that requires picking block devices explicitly that require
3793 * A new (currently still internal) API sd-device.h has been
3794 added to libsystemd. This modernized API is supposed to
3795 replace libudev eventually. In fact, already much of libudev
3796 is now just a wrapper around sd-device.h.
3798 * A new hwdb database for storing metadata about pointing
3799 stick devices has been added.
3801 * systemd-tmpfiles gained support for setting file attributes
3802 similar to the "chattr" tool with new 'h' and 'H' lines.
3804 * systemd-journald will no longer unconditionally set the
3805 btrfs NOCOW flag on new journal files. This is instead done
3806 with tmpfiles snippet using the new 'h' line type. This
3807 allows easy disabling of this logic, by masking the
3808 journal-nocow.conf tmpfiles file.
3810 * systemd-journald will now translate audit message types to
3811 human readable identifiers when writing them to the
3812 journal. This should improve readability of audit messages.
3814 * The LUKS logic gained support for the offset= and skip=
3815 options in /etc/crypttab, as previously implemented by
3818 * /usr/lib/os-release gained a new optional field VARIANT= for
3819 distributions that support multiple variants (such as a
3820 desktop edition, a server edition, ...)
3822 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
3823 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
3824 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
3825 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
3826 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
3827 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
3828 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
3829 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
3830 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
3831 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
3832 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
3833 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
3834 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
3835 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
3836 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
3837 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
3838 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
3839 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
3840 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
3841 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
3842 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
3843 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
3844 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
3845 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
3846 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
3847 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
3848 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
3850 — Berlin, 2015-05-22
3854 * Introduce a new API "sd-hwdb.h" for querying the hardware
3855 metadata database. With this minimal interface one can query
3856 and enumerate the udev hwdb, decoupled from the old libudev
3857 library. libudev's interface for this is now only a wrapper
3858 around sd-hwdb. A new tool systemd-hwdb has been added to
3859 interface with and update the database.
3861 * When any of systemd's tools copies files (for example due to
3862 tmpfiles' C lines) a btrfs reflink will attempted first,
3863 before bytewise copying is done.
3865 * systemd-nspawn gained a new --ephemeral switch. When
3866 specified a btrfs snapshot is taken of the container's root
3867 directory, and immediately removed when the container
3868 terminates again. Thus, a container can be started whose
3869 changes never alter the container's root directory, and are
3870 lost on container termination. This switch can also be used
3871 for starting a container off the root file system of the
3872 host without affecting the host OS. This switch is only
3873 available on btrfs file systems.
3875 * systemd-nspawn gained a new --template= switch. It takes the
3876 path to a container tree to use as template for the tree
3877 specified via --directory=, should that directory be
3878 missing. This allows instantiating containers dynamically,
3879 on first run. This switch is only available on btrfs file
3882 * When a .mount unit refers to a mount point on which multiple
3883 mounts are stacked, and the .mount unit is stopped all of
3884 the stacked mount points will now be unmounted until no
3885 mount point remains.
3887 * systemd now has an explicit notion of supported and
3888 unsupported unit types. Jobs enqueued for unsupported unit
3889 types will now fail with an "unsupported" error code. More
3890 specifically .swap, .automount and .device units are not
3891 supported in containers, .busname units are not supported on
3892 non-kdbus systems. .swap and .automount are also not
3893 supported if their respective kernel compile time options
3896 * machinectl gained support for two new "copy-from" and
3897 "copy-to" commands for copying files from a running
3898 container to the host or vice versa.
3900 * machinectl gained support for a new "bind" command to bind
3901 mount host directories into local containers. This is
3902 currently only supported for nspawn containers.
3904 * networkd gained support for configuring bridge forwarding
3905 database entries (fdb) from .network files.
3907 * A new tiny daemon "systemd-importd" has been added that can
3908 download container images in tar, raw, qcow2 or dkr formats,
3909 and make them available locally in /var/lib/machines, so
3910 that they can run as nspawn containers. The daemon can GPG
3911 verify the downloads (not supported for dkr, since it has no
3912 provisions for verifying downloads). It will transparently
3913 decompress bz2, xz, gzip compressed downloads if necessary,
3914 and restore sparse files on disk. The daemon uses privilege
3915 separation to ensure the actual download logic runs with
3916 fewer privileges than the daemon itself. machinectl has
3917 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
3918 make the functionality of importd available to the
3919 user. With this in place the Fedora and Ubuntu "Cloud"
3920 images can be downloaded and booted as containers unmodified
3921 (the Fedora images lack the appropriate GPG signature files
3922 currently, so they cannot be verified, but this will change
3923 soon, hopefully). Note that downloading images is currently
3924 only fully supported on btrfs.
3926 * machinectl is now able to list container images found in
3927 /var/lib/machines, along with some metadata about sizes of
3928 disk and similar. If the directory is located on btrfs and
3929 quota is enabled, this includes quota display. A new command
3930 "image-status" has been added that shows additional
3931 information about images.
3933 * machinectl is now able to clone container images
3934 efficiently, if the underlying file system (btrfs) supports
3935 it, with the new "machinectl clone" command. It also
3936 gained commands for renaming and removing images, as well as
3937 marking them read-only or read-write (supported also on
3938 legacy file systems).
3940 * networkd gained support for collecting LLDP network
3941 announcements, from hardware that supports this. This is
3942 shown in networkctl output.
3944 * systemd-run gained support for a new -t (--pty) switch for
3945 invoking a binary on a pty whose input and output is
3946 connected to the invoking terminal. This allows executing
3947 processes as system services while interactively
3948 communicating with them via the terminal. Most interestingly
3949 this is supported across container boundaries. Invoking
3950 "systemd-run -t /bin/bash" is an alternative to running a
3951 full login session, the difference being that the former
3952 will not register a session, nor go through the PAM session
3955 * tmpfiles gained support for a new "v" line type for creating
3956 btrfs subvolumes. If the underlying file system is a legacy
3957 file system, this automatically degrades to creating a
3958 normal directory. Among others /var/lib/machines is now
3959 created like this at boot, should it be missing.
3961 * The directory /var/lib/containers/ has been deprecated and
3962 been replaced by /var/lib/machines. The term "machines" has
3963 been used in the systemd context as generic term for both
3964 VMs and containers, and hence appears more appropriate for
3965 this, as the directory can also contain raw images bootable
3968 * systemd-nspawn when invoked with -M but without --directory=
3969 or --image= is now capable of searching for the container
3970 root directory, subvolume or disk image automatically, in
3971 /var/lib/machines. systemd-nspawn@.service has been updated
3972 to make use of this, thus allowing it to be used for raw
3975 * A new machines.target unit has been introduced that is
3976 supposed to group all containers/VMs invoked as services on
3977 the system. systemd-nspawn@.service has been updated to
3978 integrate with that.
3980 * machinectl gained a new "start" command, for invoking a
3981 container as a service. "machinectl start foo" is mostly
3982 equivalent to "systemctl start systemd-nspawn@foo.service",
3983 but handles escaping in a nicer way.
3985 * systemd-nspawn will now mount most of the cgroupfs tree
3986 read-only into each container, with the exception of the
3987 container's own subtree in the name=systemd hierarchy.
3989 * journald now sets the special FS_NOCOW file flag for its
3990 journal files. This should improve performance on btrfs, by
3991 avoiding heavy fragmentation when journald's write-pattern
3992 is used on COW file systems. It degrades btrfs' data
3993 integrity guarantees for the files to the same levels as for
3994 ext3/ext4 however. This should be OK though as journald does
3995 its own data integrity checks and all its objects are
3996 checksummed on disk. Also, journald should handle btrfs disk
3997 full events a lot more gracefully now, by processing SIGBUS
3998 errors, and not relying on fallocate() anymore.
4000 * When journald detects that journal files it is writing to
4001 have been deleted it will immediately start new journal
4004 * systemd now provides a way to store file descriptors
4005 per-service in PID 1. This is useful for daemons to ensure
4006 that fds they require are not lost during a daemon
4007 restart. The fds are passed to the daemon on the next
4008 invocation in the same way socket activation fds are
4009 passed. This is now used by journald to ensure that the
4010 various sockets connected to all the system's stdout/stderr
4011 are not lost when journald is restarted. File descriptors
4012 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
4013 an extension to sd_notify(). Note that a limit is enforced
4014 on the number of fds a service can store in PID 1, and it
4015 defaults to 0, so that no fds may be stored, unless this is
4016 explicitly turned on.
4018 * The default TERM variable to use for units connected to a
4019 terminal, when no other value is explicitly is set is now
4020 vt220 rather than vt102. This should be fairly safe still,
4021 but allows PgUp/PgDn work.
4023 * The /etc/crypttab option header= as known from Debian is now
4026 * "loginctl user-status" and "loginctl session-status" will
4027 now show the last 10 lines of log messages of the
4028 user/session following the status output. Similar,
4029 "machinectl status" will show the last 10 log lines
4030 associated with a virtual machine or container
4031 service. (Note that this is usually not the log messages
4032 done in the VM/container itself, but simply what the
4033 container manager logs. For nspawn this includes all console
4036 * "loginctl session-status" without further argument will now
4037 show the status of the session of the caller. Similar,
4038 "lock-session", "unlock-session", "activate",
4039 "enable-linger", "disable-linger" may now be called without
4040 session/user parameter in which case they apply to the
4041 caller's session/user.
4043 * An X11 session scriptlet is now shipped that uploads
4044 $DISPLAY and $XAUTHORITY into the environment of the systemd
4045 --user daemon if a session begins. This should improve
4046 compatibility with X11 enabled applications run as systemd
4049 * Generators are now subject to masking via /etc and /run, the
4050 same way as unit files.
4052 * networkd .network files gained support for configuring
4053 per-link IPv4/IPv6 packet forwarding as well as IPv4
4054 masquerading. This is by default turned on for veth links to
4055 containers, as registered by systemd-nspawn. This means that
4056 nspawn containers run with --network-veth will now get
4057 automatic routed access to the host's networks without any
4058 further configuration or setup, as long as networkd runs on
4061 * systemd-nspawn gained the --port= (-p) switch to expose TCP
4062 or UDP posts of a container on the host. With this in place
4063 it is possible to run containers with private veth links
4064 (--network-veth), and have their functionality exposed on
4065 the host as if their services were running directly on the
4068 * systemd-nspawn's --network-veth switch now gained a short
4069 version "-n", since with the changes above it is now truly
4070 useful out-of-the-box. The systemd-nspawn@.service has been
4071 updated to make use of it too by default.
4073 * systemd-nspawn will now maintain a per-image R/W lock, to
4074 ensure that the same image is not started more than once
4075 writable. (It's OK to run an image multiple times
4076 simultaneously in read-only mode.)
4078 * systemd-nspawn's --image= option is now capable of
4079 dissecting and booting MBR and GPT disk images that contain
4080 only a single active Linux partition. Previously it
4081 supported only GPT disk images with proper GPT type
4082 IDs. This allows running cloud images from major
4083 distributions directly with systemd-nspawn, without
4086 * In addition to collecting mouse dpi data in the udev
4087 hardware database, there's now support for collecting angle
4088 information for mouse scroll wheels. The database is
4089 supposed to guarantee similar scrolling behavior on mice
4090 that it knows about. There's also support for collecting
4091 information about Touchpad types.
4093 * udev's input_id built-in will now also collect touch screen
4094 dimension data and attach it to probed devices.
4096 * /etc/os-release gained support for a Distribution Privacy
4099 * networkd gained support for creating "ipvlan", "gretap",
4100 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
4102 * systemd-tmpfiles gained support for "a" lines for setting
4105 * systemd-nspawn will now mount /tmp in the container to
4106 tmpfs, automatically.
4108 * systemd now exposes the memory.usage_in_bytes cgroup
4109 attribute and shows it for each service in the "systemctl
4110 status" output, if available.
4112 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
4113 immediate reboot is triggered. This useful if shutdown is
4114 hung and is unable to complete, to expedite the
4115 operation. Note that this kind of reboot will still unmount
4116 all file systems, and hence should not result in fsck being
4119 * A .device unit for an optical block device will now be
4120 considered active only when a medium is in the drive. Also,
4121 mount units are now bound to their backing devices thus
4122 triggering automatic unmounting when devices become
4123 unavailable. With this in place systemd will now
4124 automatically unmount left-over mounts when a CD-ROM is
4125 ejected or an USB stick is yanked from the system.
4127 * networkd-wait-online now has support for waiting for
4128 specific interfaces only (with globbing), and for giving up
4129 after a configurable timeout.
4131 * networkd now exits when idle. It will be automatically
4132 restarted as soon as interfaces show up, are removed or
4133 change state. networkd will stay around as long as there is
4134 at least one DHCP state machine or similar around, that keep
4137 * networkd may now configure IPv6 link-local addressing in
4138 addition to IPv4 link-local addressing.
4140 * The IPv6 "token" for use in SLAAC may now be configured for
4141 each .network interface in networkd.
4143 * Routes configured with networkd may now be assigned a scope
4146 * networkd's [Match] sections now support globbing and lists
4147 of multiple space-separated matches per item.
4149 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
4150 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
4151 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
4152 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
4153 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
4154 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
4155 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
4156 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
4157 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
4158 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
4159 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
4160 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
4161 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
4162 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
4163 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
4164 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
4165 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
4166 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
4167 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
4168 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
4169 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
4170 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
4171 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
4172 Hoffmann, Zbigniew Jędrzejewski-Szmek
4174 — Berlin, 2015-02-16
4178 * When querying unit file enablement status (for example via
4179 "systemctl is-enabled"), a new state "indirect" is now known
4180 which indicates that a unit might not be enabled itself, but
4181 another unit listed in its Also= setting might be.
4183 * Similar to the various existing ConditionXYZ= settings for
4184 units, there are now matching AssertXYZ= settings. While
4185 failing conditions cause a unit to be skipped, but its job
4186 to succeed, failing assertions declared like this will cause
4187 a unit start operation and its job to fail.
4189 * hostnamed now knows a new chassis type "embedded".
4191 * systemctl gained a new "edit" command. When used on a unit
4192 file, this allows extending unit files with .d/ drop-in
4193 configuration snippets or editing the full file (after
4194 copying it from /usr/lib to /etc). This will invoke the
4195 user's editor (as configured with $EDITOR), and reload the
4196 modified configuration after editing.
4198 * "systemctl status" now shows the suggested enablement state
4199 for a unit, as declared in the (usually vendor-supplied)
4200 system preset files.
4202 * nss-myhostname will now resolve the single-label host name
4203 "gateway" to the locally configured default IP routing
4204 gateways, ordered by their metrics. This assigns a stable
4205 name to the used gateways, regardless which ones are
4206 currently configured. Note that the name will only be
4207 resolved after all other name sources (if nss-myhostname is
4208 configured properly) and should hence not negatively impact
4209 systems that use the single-label host name "gateway" in
4212 * systemd-inhibit now allows filtering by mode when listing
4215 * Scope and service units gained a new "Delegate" boolean
4216 property, which, when set, allows processes running inside the
4217 unit to further partition resources. This is primarily
4218 useful for systemd user instances as well as container
4221 * journald will now pick up audit messages directly from
4222 the kernel, and log them like any other log message. The
4223 audit fields are split up and fully indexed. This means that
4224 journalctl in many ways is now a (nicer!) alternative to
4225 ausearch, the traditional audit client. Note that this
4226 implements only a minimal audit client. If you want the
4227 special audit modes like reboot-on-log-overflow, please use
4228 the traditional auditd instead, which can be used in
4229 parallel to journald.
4231 * The ConditionSecurity= unit file option now understands the
4232 special string "audit" to check whether auditing is
4235 * journalctl gained two new commands --vacuum-size= and
4236 --vacuum-time= to delete old journal files until the
4237 remaining ones take up no more than the specified size on disk,
4238 or are not older than the specified time.
4240 * A new, native PPPoE library has been added to sd-network,
4241 systemd's library of light-weight networking protocols. This
4242 library will be used in a future version of networkd to
4243 enable PPPoE communication without an external pppd daemon.
4245 * The busctl tool now understands a new "capture" verb that
4246 works similar to "monitor", but writes a packet capture
4247 trace to STDOUT that can be redirected to a file which is
4248 compatible with libcap's capture file format. This can then
4249 be loaded in Wireshark and similar tools to inspect bus
4252 * The busctl tool now understands a new "tree" verb that shows
4253 the object trees of a specific service on the bus, or of all
4256 * The busctl tool now understands a new "introspect" verb that
4257 shows all interfaces and members of objects on the bus,
4258 including their signature and values. This is particularly
4259 useful to get more information about bus objects shown by
4260 the new "busctl tree" command.
4262 * The busctl tool now understands new verbs "call",
4263 "set-property" and "get-property" for invoking bus method
4264 calls, setting and getting bus object properties in a
4267 * busctl gained a new --augment-creds= argument that controls
4268 whether the tool shall augment credential information it
4269 gets from the bus with data from /proc, in a possibly
4272 * nspawn's --link-journal= switch gained two new values
4273 "try-guest" and "try-host" that work like "guest" and
4274 "host", but do not fail if the host has no persistent
4275 journalling enabled. -j is now equivalent to
4276 --link-journal=try-guest.
4278 * macvlan network devices created by nspawn will now have
4279 stable MAC addresses.
4281 * A new SmackProcessLabel= unit setting has been added, which
4282 controls the SMACK security label processes forked off by
4283 the respective unit shall use.
4285 * If compiled with --enable-xkbcommon, systemd-localed will
4286 verify x11 keymap settings by compiling the given keymap. It
4287 will spew out warnings if the compilation fails. This
4288 requires libxkbcommon to be installed.
4290 * When a coredump is collected, a larger number of metadata
4291 fields is now collected and included in the journal records
4292 created for it. More specifically, control group membership,
4293 environment variables, memory maps, working directory,
4294 chroot directory, /proc/$PID/status, and a list of open file
4295 descriptors is now stored in the log entry.
4297 * The udev hwdb now contains DPI information for mice. For
4300 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
4302 * All systemd programs that read standalone configuration
4303 files in /etc now also support a corresponding series of
4304 .conf.d configuration directories in /etc/, /run/,
4305 /usr/local/lib/, /usr/lib/, and (if configured with
4306 --enable-split-usr) /lib/. In particular, the following
4307 configuration files now have corresponding configuration
4308 directories: system.conf user.conf, logind.conf,
4309 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
4310 resolved.conf, timesyncd.conf, journal-remote.conf, and
4311 journal-upload.conf. Note that distributions should use the
4312 configuration directories in /usr/lib/; the directories in
4313 /etc/ are reserved for the system administrator.
4315 * systemd-rfkill will no longer take the rfkill device name
4316 into account when storing rfkill state on disk, as the name
4317 might be dynamically assigned and not stable. Instead, the
4318 ID_PATH udev variable combined with the rfkill type (wlan,
4319 bluetooth, ...) is used.
4321 * A new service systemd-machine-id-commit.service has been
4322 added. When used on systems where /etc is read-only during
4323 boot, and /etc/machine-id is not initialized (but an empty
4324 file), this service will copy the temporary machine ID
4325 created as replacement into /etc after the system is fully
4326 booted up. This is useful for systems that are freshly
4327 installed with a non-initialized machine ID, but should get
4328 a fixed machine ID for subsequent boots.
4330 * networkd's .netdev files now provide a large set of
4331 configuration parameters for VXLAN devices. Similarly, the
4332 bridge port cost parameter is now configurable in .network
4333 files. There's also new support for configuring IP source
4334 routing. networkd .link files gained support for a new
4335 OriginalName= match that is useful to match against the
4336 original interface name the kernel assigned. .network files
4337 may include MTU= and MACAddress= fields for altering the MTU
4338 and MAC address while being connected to a specific network
4341 * The LUKS logic gained supported for configuring
4342 UUID-specific key files. There's also new support for naming
4343 LUKS device from the kernel command line, using the new
4344 luks.name= argument.
4346 * Timer units may now be transiently created via the bus API
4347 (this was previously already available for scope and service
4348 units). In addition it is now possible to create multiple
4349 transient units at the same time with a single bus call. The
4350 "systemd-run" tool has been updated to make use of this for
4351 running commands on a specified time, in at(1)-style.
4353 * tmpfiles gained support for "t" lines, for assigning
4354 extended attributes to files. Among other uses this may be
4355 used to assign SMACK labels to files.
4357 Contributions from: Alin Rauta, Alison Chaiken, Andrej
4358 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
4359 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
4360 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
4361 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
4362 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
4363 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
4364 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
4365 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
4366 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
4367 Michael Biebl, Michael Chapman, Michael Marineau, Michal
4368 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
4369 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
4370 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
4371 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
4372 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4373 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
4374 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
4376 — Berlin, 2014-12-10
4380 * journalctl gained the new options -t/--identifier= to match
4381 on the syslog identifier (aka "tag"), as well as --utc to
4382 show log timestamps in the UTC timezone. journalctl now also
4383 accepts -n/--lines=all to disable line capping in a pager.
4385 * journalctl gained a new switch, --flush, that synchronously
4386 flushes logs from /run/log/journal to /var/log/journal if
4387 persistent storage is enabled. systemd-journal-flush.service
4388 now waits until the operation is complete.
4390 * Services can notify the manager before they start a reload
4391 (by sending RELOADING=1) or shutdown (by sending
4392 STOPPING=1). This allows the manager to track and show the
4393 internal state of daemons and closes a race condition when
4394 the process is still running but has closed its D-Bus
4397 * Services with Type=oneshot do not have to have any ExecStart
4400 * User units are now loaded also from
4401 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
4402 /run/systemd/user directory that was already previously
4403 supported, but is under the control of the user.
4405 * Job timeouts (i.e. time-outs on the time a job that is
4406 queued stays in the run queue) can now optionally result in
4407 immediate reboot or power-off actions (JobTimeoutAction= and
4408 JobTimeoutRebootArgument=). This is useful on ".target"
4409 units, to limit the maximum time a target remains
4410 undispatched in the run queue, and to trigger an emergency
4411 operation in such a case. This is now used by default to
4412 turn off the system if boot-up (as defined by everything in
4413 basic.target) hangs and does not complete for at least
4414 15min. Also, if power-off or reboot hang for at least 30min
4415 an immediate power-off/reboot operation is triggered. This
4416 functionality is particularly useful to increase reliability
4417 on embedded devices, but also on laptops which might
4418 accidentally get powered on when carried in a backpack and
4419 whose boot stays stuck in a hard disk encryption passphrase
4422 * systemd-logind can be configured to also handle lid switch
4423 events even when the machine is docked or multiple displays
4424 are attached (HandleLidSwitchDocked= option).
4426 * A helper binary and a service have been added which can be
4427 used to resume from hibernation in the initramfs. A
4428 generator will parse the resume= option on the kernel
4429 command line to trigger resume.
4431 * A user console daemon systemd-consoled has been
4432 added. Currently, it is a preview, and will so far open a
4433 single terminal on each session of the user marked as
4434 Desktop=systemd-console.
4436 * Route metrics can be specified for DHCP routes added by
4439 * The SELinux context of socket-activated services can be set
4440 from the information provided by the networking stack
4441 (SELinuxContextFromNet= option).
4443 * Userspace firmware loading support has been removed and
4444 the minimum supported kernel version is thus bumped to 3.7.
4446 * Timeout for udev workers has been increased from 1 to 3
4447 minutes, but a warning will be printed after 1 minute to
4448 help diagnose kernel modules that take a long time to load.
4450 * Udev rules can now remove tags on devices with TAG-="foobar".
4452 * systemd's readahead implementation has been removed. In many
4453 circumstances it didn't give expected benefits even for
4454 rotational disk drives and was becoming less relevant in the
4455 age of SSDs. As none of the developers has been using
4456 rotating media anymore, and nobody stepped up to actively
4457 maintain this component of systemd it has now been removed.
4459 * Swap units can use Options= to specify discard options.
4460 Discard options specified for swaps in /etc/fstab are now
4463 * Docker containers are now detected as a separate type of
4466 * The Password Agent protocol gained support for queries where
4467 the user input is shown, useful e.g. for user names.
4468 systemd-ask-password gained a new --echo option to turn that
4471 * The default sysctl.d/ snippets will now set:
4473 net.core.default_qdisc = fq_codel
4475 This selects Fair Queuing Controlled Delay as the default
4476 queuing discipline for network interfaces. fq_codel helps
4477 fight the network bufferbloat problem. It is believed to be
4478 a good default with no tuning required for most workloads.
4479 Downstream distributions may override this choice. On 10Gbit
4480 servers that do not do forwarding, "fq" may perform better.
4481 Systems without a good clocksource should use "pfifo_fast".
4483 * If kdbus is enabled during build a new option BusPolicy= is
4484 available for service units, that allows locking all service
4485 processes into a stricter bus policy, in order to limit
4486 access to various bus services, or even hide most of them
4487 from the service's view entirely.
4489 * networkctl will now show the .network and .link file
4490 networkd has applied to a specific interface.
4492 * sd-login gained a new API call sd_session_get_desktop() to
4493 query which desktop environment has been selected for a
4496 * UNIX utmp support is now compile-time optional to support
4497 legacy-free systems.
4499 * systemctl gained two new commands "add-wants" and
4500 "add-requires" for pulling in units from specific targets
4503 * If the word "rescue" is specified on the kernel command line
4504 the system will now boot into rescue mode (aka
4505 rescue.target), which was previously available only by
4506 specifying "1" or "systemd.unit=rescue.target" on the kernel
4507 command line. This new kernel command line option nicely
4508 mirrors the already existing "emergency" kernel command line
4511 * New kernel command line options mount.usr=, mount.usrflags=,
4512 mount.usrfstype= have been added that match root=, rootflags=,
4513 rootfstype= but allow mounting a specific file system to
4516 * The $NOTIFY_SOCKET is now also passed to control processes of
4517 services, not only the main process.
4519 * This version reenables support for fsck's -l switch. This
4520 means at least version v2.25 of util-linux is required for
4521 operation, otherwise dead-locks on device nodes may
4522 occur. Again: you need to update util-linux to at least
4523 v2.25 when updating systemd to v217.
4525 * The "multi-seat-x" tool has been removed from systemd, as
4526 its functionality has been integrated into X servers 1.16,
4527 and the tool is hence redundant. It is recommended to update
4528 display managers invoking this tool to simply invoke X
4529 directly from now on, again.
4531 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
4532 message flag has been added for all of systemd's PolicyKit
4533 authenticated method calls has been added. In particular
4534 this now allows optional interactive authorization via
4535 PolicyKit for many of PID1's privileged operations such as
4536 unit file enabling and disabling.
4538 * "udevadm hwdb --update" learnt a new switch "--usr" for
4539 placing the rebuilt hardware database in /usr instead of
4540 /etc. When used only hardware database entries stored in
4541 /usr will be used, and any user database entries in /etc are
4542 ignored. This functionality is useful for vendors to ship a
4543 pre-built database on systems where local configuration is
4544 unnecessary or unlikely.
4546 * Calendar time specifications in .timer units now also
4547 understand the strings "semi-annually", "quarterly" and
4548 "minutely" as shortcuts (in addition to the preexisting
4549 "anually", "hourly", ...).
4551 * systemd-tmpfiles will now correctly create files in /dev
4552 at boot which are marked for creation only at boot. It is
4553 recommended to always create static device nodes with 'c!'
4554 and 'b!', so that they are created only at boot and not
4555 overwritten at runtime.
4557 * When the watchdog logic is used for a service (WatchdogSec=)
4558 and the watchdog timeout is hit the service will now be
4559 terminated with SIGABRT (instead of just SIGTERM), in order
4560 to make sure a proper coredump and backtrace is
4561 generated. This ensures that hanging services will result in
4562 similar coredump/backtrace behaviour as services that hit a
4565 Contributions from: Andreas Henriksson, Andrei Borzenkov,
4566 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
4567 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
4568 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
4569 Herrmann, David Sommerseth, David Strauss, Emil Renner
4570 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
4571 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
4572 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
4573 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
4574 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
4575 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
4576 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
4577 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
4578 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
4579 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
4580 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
4581 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
4582 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
4583 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
4584 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
4585 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
4588 — Berlin, 2014-10-28
4592 * timedated no longer reads NTP implementation unit names from
4593 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
4594 implementations should add a
4596 Conflicts=systemd-timesyncd.service
4598 to their unit files to take over and replace systemd's NTP
4599 default functionality.
4601 * systemd-sysusers gained a new line type "r" for configuring
4602 which UID/GID ranges to allocate system users/groups
4603 from. Lines of type "u" may now add an additional column
4604 that specifies the home directory for the system user to be
4605 created. Also, systemd-sysusers may now optionally read user
4606 information from STDIN instead of a file. This is useful for
4607 invoking it from RPM preinst scriptlets that need to create
4608 users before the first RPM file is installed since these
4609 files might need to be owned by them. A new
4610 %sysusers_create_inline RPM macro has been introduced to do
4611 just that. systemd-sysusers now updates the shadow files as
4612 well as the user/group databases, which should enhance
4613 compatibility with certain tools like grpck.
4615 * A number of bus APIs of PID 1 now optionally consult
4616 PolicyKit to permit access for otherwise unprivileged
4617 clients under certain conditions. Note that this currently
4618 doesn't support interactive authentication yet, but this is
4619 expected to be added eventually, too.
4621 * /etc/machine-info now has new fields for configuring the
4622 deployment environment of the machine, as well as the
4623 location of the machine. hostnamectl has been updated with
4624 new command to update these fields.
4626 * systemd-timesyncd has been updated to automatically acquire
4627 NTP server information from systemd-networkd, which might
4628 have been discovered via DHCP.
4630 * systemd-resolved now includes a caching DNS stub resolver
4631 and a complete LLMNR name resolution implementation. A new
4632 NSS module "nss-resolve" has been added which can be used
4633 instead of glibc's own "nss-dns" to resolve hostnames via
4634 systemd-resolved. Hostnames, addresses and arbitrary RRs may
4635 be resolved via systemd-resolved D-Bus APIs. In contrast to
4636 the glibc internal resolver systemd-resolved is aware of
4637 multi-homed system, and keeps DNS server and caches separate
4638 and per-interface. Queries are sent simultaneously on all
4639 interfaces that have DNS servers configured, in order to
4640 properly handle VPNs and local LANs which might resolve
4641 separate sets of domain names. systemd-resolved may acquire
4642 DNS server information from systemd-networkd automatically,
4643 which in turn might have discovered them via DHCP. A tool
4644 "systemd-resolve-host" has been added that may be used to
4645 query the DNS logic in resolved. systemd-resolved implements
4646 IDNA and automatically uses IDNA or UTF-8 encoding depending
4647 on whether classic DNS or LLMNR is used as transport. In the
4648 next releases we intend to add a DNSSEC and mDNS/DNS-SD
4649 implementation to systemd-resolved.
4651 * A new NSS module nss-mymachines has been added, that
4652 automatically resolves the names of all local registered
4653 containers to their respective IP addresses.
4655 * A new client tool "networkctl" for systemd-networkd has been
4656 added. It currently is entirely passive and will query
4657 networking configuration from udev, rtnetlink and networkd,
4658 and present it to the user in a very friendly
4659 way. Eventually, we hope to extend it to become a full
4660 control utility for networkd.
4662 * .socket units gained a new DeferAcceptSec= setting that
4663 controls the kernels' TCP_DEFER_ACCEPT sockopt for
4664 TCP. Similarly, support for controlling TCP keep-alive
4665 settings has been added (KeepAliveTimeSec=,
4666 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
4667 turning off Nagle's algorithm on TCP has been added
4670 * logind learned a new session type "web", for use in projects
4671 like Cockpit which register web clients as PAM sessions.
4673 * timer units with at least one OnCalendar= setting will now
4674 be started only after time-sync.target has been
4675 reached. This way they will not elapse before the system
4676 clock has been corrected by a local NTP client or
4677 similar. This is particular useful on RTC-less embedded
4678 machines, that come up with an invalid system clock.
4680 * systemd-nspawn's --network-veth= switch should now result in
4681 stable MAC addresses for both the outer and the inner side
4684 * systemd-nspawn gained a new --volatile= switch for running
4685 container instances with /etc or /var unpopulated.
4687 * The kdbus client code has been updated to use the new Linux
4688 3.17 memfd subsystem instead of the old kdbus-specific one.
4690 * systemd-networkd's DHCP client and server now support
4691 FORCERENEW. There are also new configuration options to
4692 configure the vendor client identifier and broadcast mode
4695 * systemd will no longer inform the kernel about the current
4696 timezone, as this is necessarily incorrect and racy as the
4697 kernel has no understanding of DST and similar
4698 concepts. This hence means FAT timestamps will be always
4699 considered UTC, similar to what Android is already
4700 doing. Also, when the RTC is configured to the local time
4701 (rather than UTC) systemd will never synchronize back to it,
4702 as this might confuse Windows at a later boot.
4704 * systemd-analyze gained a new command "verify" for offline
4705 validation of unit files.
4707 * systemd-networkd gained support for a couple of additional
4708 settings for bonding networking setups. Also, the metric for
4709 statically configured routes may now be configured. For
4710 network interfaces where this is appropriate the peer IP
4711 address may now be configured.
4713 * systemd-networkd's DHCP client will no longer request
4714 broadcasting by default, as this tripped up some networks.
4715 For hardware where broadcast is required the feature should
4716 be switched back on using RequestBroadcast=yes.
4718 * systemd-networkd will now set up IPv4LL addresses (when
4719 enabled) even if DHCP is configured successfully.
4721 * udev will now default to respect network device names given
4722 by the kernel when the kernel indicates that these are
4723 predictable. This behavior can be tweaked by changing
4724 NamePolicy= in the relevant .link file.
4726 * A new library systemd-terminal has been added that
4727 implements full TTY stream parsing and rendering. This
4728 library is supposed to be used later on for implementing a
4729 full userspace VT subsystem, replacing the current kernel
4732 * A new tool systemd-journal-upload has been added to push
4733 journal data to a remote system running
4734 systemd-journal-remote.
4736 * journald will no longer forward all local data to another
4737 running syslog daemon. This change has been made because
4738 rsyslog (which appears to be the most commonly used syslog
4739 implementation these days) no longer makes use of this, and
4740 instead pulls the data out of the journal on its own. Since
4741 forwarding the messages to a non-existent syslog server is
4742 more expensive than we assumed we have now turned this
4743 off. If you run a syslog server that is not a recent rsyslog
4744 version, you have to turn this option on again
4745 (ForwardToSyslog= in journald.conf).
4747 * journald now optionally supports the LZ4 compressor for
4748 larger journal fields. This compressor should perform much
4749 better than XZ which was the previous default.
4751 * machinectl now shows the IP addresses of local containers,
4752 if it knows them, plus the interface name of the container.
4754 * A new tool "systemd-escape" has been added that makes it
4755 easy to escape strings to build unit names and similar.
4757 * sd_notify() messages may now include a new ERRNO= field
4758 which is parsed and collected by systemd and shown among the
4759 "systemctl status" output for a service.
4761 * A new component "systemd-firstboot" has been added that
4762 queries the most basic systemd information (timezone,
4763 hostname, root password) interactively on first
4764 boot. Alternatively it may also be used to provision these
4765 things offline on OS images installed into directories.
4767 * The default sysctl.d/ snippets will now set
4769 net.ipv4.conf.default.promote_secondaries=1
4771 This has the benefit of no flushing secondary IP addresses
4772 when primary addresses are removed.
4774 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
4775 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
4776 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
4777 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
4778 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
4779 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
4780 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
4781 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
4782 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
4783 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
4784 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
4785 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
4786 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
4787 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
4788 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
4790 — Berlin, 2014-08-19
4794 * A new tool systemd-sysusers has been added. This tool
4795 creates system users and groups in /etc/passwd and
4796 /etc/group, based on static declarative system user/group
4797 definitions in /usr/lib/sysusers.d/. This is useful to
4798 enable factory resets and volatile systems that boot up with
4799 an empty /etc directory, and thus need system users and
4800 groups created during early boot. systemd now also ships
4801 with two default sysusers.d/ files for the most basic
4802 users and groups systemd and the core operating system
4805 * A new tmpfiles snippet has been added that rebuilds the
4806 essential files in /etc on boot, should they be missing.
4808 * A directive for ensuring automatic clean-up of
4809 /var/cache/man/ has been removed from the default
4810 configuration. This line should now be shipped by the man
4811 implementation. The necessary change has been made to the
4812 man-db implementation. Note that you need to update your man
4813 implementation to one that ships this line, otherwise no
4814 automatic clean-up of /var/cache/man will take place.
4816 * A new condition ConditionNeedsUpdate= has been added that
4817 may conditionalize services to only run when /etc or /var
4818 are "older" than the vendor operating system resources in
4819 /usr. This is useful for reconstructing or updating /etc
4820 after an offline update of /usr or a factory reset, on the
4821 next reboot. Services that want to run once after such an
4822 update or reset should use this condition and order
4823 themselves before the new systemd-update-done.service, which
4824 will mark the two directories as fully updated. A number of
4825 service files have been added making use of this, to rebuild
4826 the udev hardware database, the journald message catalog and
4827 dynamic loader cache (ldconfig). The systemd-sysusers tool
4828 described above also makes use of this now. With this in
4829 place it is now possible to start up a minimal operating
4830 system with /etc empty cleanly. For more information on the
4831 concepts involved see this recent blog story:
4833 http://0pointer.de/blog/projects/stateless.html
4835 * A new system group "input" has been introduced, and all
4836 input device nodes get this group assigned. This is useful
4837 for system-level software to get access to input devices. It
4838 complements what is already done for "audio" and "video".
4840 * systemd-networkd learnt minimal DHCPv4 server support in
4841 addition to the existing DHCPv4 client support. It also
4842 learnt DHCPv6 client and IPv6 Router Solicitation client
4843 support. The DHCPv4 client gained support for static routes
4844 passed in from the server. Note that the [DHCPv4] section
4845 known in older systemd-networkd versions has been renamed to
4846 [DHCP] and is now also used by the DHCPv6 client. Existing
4847 .network files using settings of this section should be
4848 updated, though compatibility is maintained. Optionally, the
4849 client hostname may now be sent to the DHCP server.
4851 * networkd gained support for vxlan virtual networks as well
4852 as tun/tap and dummy devices.
4854 * networkd gained support for automatic allocation of address
4855 ranges for interfaces from a system-wide pool of
4856 addresses. This is useful for dynamically managing a large
4857 number of interfaces with a single network configuration
4858 file. In particular this is useful to easily assign
4859 appropriate IP addresses to the veth links of a large number
4860 of nspawn instances.
4862 * RPM macros for processing sysusers, sysctl and binfmt
4863 drop-in snippets at package installation time have been
4866 * The /etc/os-release file should now be placed in
4867 /usr/lib/os-release. The old location is automatically
4868 created as symlink. /usr/lib is the more appropriate
4869 location of this file, since it shall actually describe the
4870 vendor operating system shipped in /usr, and not the
4871 configuration stored in /etc.
4873 * .mount units gained a new boolean SloppyOptions= setting
4874 that maps to mount(8)'s -s option which enables permissive
4875 parsing of unknown mount options.
4877 * tmpfiles learnt a new "L+" directive which creates a symlink
4878 but (unlike "L") deletes a pre-existing file first, should
4879 it already exist and not already be the correct
4880 symlink. Similarly, "b+", "c+" and "p+" directives have been
4881 added as well, which create block and character devices, as
4882 well as fifos in the filesystem, possibly removing any
4883 pre-existing files of different types.
4885 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
4886 'argument' field (which so far specified the source to
4887 symlink/copy the files from) is now optional. If omitted the
4888 same file os copied from /usr/share/factory/ suffixed by the
4889 full destination path. This is useful for populating /etc
4890 with essential files, by copying them from vendor defaults
4891 shipped in /usr/share/factory/etc.
4893 * A new command "systemctl preset-all" has been added that
4894 applies the service preset settings to all installed unit
4895 files. A new switch --preset-mode= has been added that
4896 controls whether only enable or only disable operations
4899 * A new command "systemctl is-system-running" has been added
4900 that allows checking the overall state of the system, for
4901 example whether it is fully up and running.
4903 * When the system boots up with an empty /etc, the equivalent
4904 to "systemctl preset-all" is executed during early boot, to
4905 make sure all default services are enabled after a factory
4908 * systemd now contains a minimal preset file that enables the
4909 most basic services systemd ships by default.
4911 * Unit files' [Install] section gained a new DefaultInstance=
4912 field for defining the default instance to create if a
4913 template unit is enabled with no instance specified.
4915 * A new passive target cryptsetup-pre.target has been added
4916 that may be used by services that need to make they run and
4917 finish before the first LUKS cryptographic device is set up.
4919 * The /dev/loop-control and /dev/btrfs-control device nodes
4920 are now owned by the "disk" group by default, opening up
4921 access to this group.
4923 * systemd-coredump will now automatically generate a
4924 stack trace of all core dumps taking place on the system,
4925 based on elfutils' libdw library. This stack trace is logged
4928 * systemd-coredump may now optionally store coredumps directly
4929 on disk (in /var/lib/systemd/coredump, possibly compressed),
4930 instead of storing them unconditionally in the journal. This
4931 mode is the new default. A new configuration file
4932 /etc/systemd/coredump.conf has been added to configure this
4933 and other parameters of systemd-coredump.
4935 * coredumpctl gained a new "info" verb to show details about a
4936 specific coredump. A new switch "-1" has also been added
4937 that makes sure to only show information about the most
4938 recent entry instead of all entries. Also, as the tool is
4939 generally useful now the "systemd-" prefix of the binary
4940 name has been removed. Distributions that want to maintain
4941 compatibility with the old name should add a symlink from
4942 the old name to the new name.
4944 * journald's SplitMode= now defaults to "uid". This makes sure
4945 that unprivileged users can access their own coredumps with
4946 coredumpctl without restrictions.
4948 * New kernel command line options "systemd.wants=" (for
4949 pulling an additional unit during boot), "systemd.mask="
4950 (for masking a specific unit for the boot), and
4951 "systemd.debug-shell" (for enabling the debug shell on tty9)
4952 have been added. This is implemented in the new generator
4953 "systemd-debug-generator".
4955 * systemd-nspawn will now by default filter a couple of
4956 syscalls for containers, among them those required for
4957 kernel module loading, direct x86 IO port access, swap
4958 management, and kexec. Most importantly though
4959 open_by_handle_at() is now prohibited for containers,
4960 closing a hole similar to a recently discussed vulnerability
4961 in docker regarding access to files on file hierarchies the
4962 container should normally not have access to. Note that, for
4963 nspawn, we generally make no security claims anyway (and
4964 this is explicitly documented in the man page), so this is
4965 just a fix for one of the most obvious problems.
4967 * A new man page file-hierarchy(7) has been added that
4968 contains a minimized, modernized version of the file system
4969 layout systemd expects, similar in style to the FHS
4970 specification or hier(5). A new tool systemd-path(1) has
4971 been added to query many of these paths for the local
4974 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
4975 longer done. Since the directory now has a per-user size
4976 limit, and is cleaned on logout this appears unnecessary,
4977 in particular since this now brings the lifecycle of this
4978 directory closer in line with how IPC objects are handled.
4980 * systemd.pc now exports a number of additional directories,
4981 including $libdir (which is useful to identify the library
4982 path for the primary architecture of the system), and a
4983 couple of drop-in directories.
4985 * udev's predictable network interface names now use the dev_port
4986 sysfs attribute, introduced in linux 3.15 instead of dev_id to
4987 distinguish between ports of the same PCI function. dev_id should
4988 only be used for ports using the same HW address, hence the need
4991 * machined has been updated to export the OS version of a
4992 container (read from /etc/os-release and
4993 /usr/lib/os-release) on the bus. This is now shown in
4994 "machinectl status" for a machine.
4996 * A new service setting RestartForceExitStatus= has been
4997 added. If configured to a set of exit signals or process
4998 return values, the service will be restarted when the main
4999 daemon process exits with any of them, regardless of the
5002 * systemctl's -H switch for connecting to remote systemd
5003 machines has been extended so that it may be used to
5004 directly connect to a specific container on the
5005 host. "systemctl -H root@foobar:waldi" will now connect as
5006 user "root" to host "foobar", and then proceed directly to
5007 the container named "waldi". Note that currently you have to
5008 authenticate as user "root" for this to work, as entering
5009 containers is a privileged operation.
5011 Contributions from: Andreas Henriksson, Benjamin Steinwender,
5012 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
5013 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
5014 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
5015 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
5016 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
5017 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
5018 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
5019 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
5020 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
5021 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
5022 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
5024 — Berlin, 2014-07-03
5028 * As an experimental feature, udev now tries to lock the
5029 disk device node (flock(LOCK_SH|LOCK_NB)) while it
5030 executes events for the disk or any of its partitions.
5031 Applications like partitioning programs can lock the
5032 disk device node (flock(LOCK_EX)) and claim temporary
5033 device ownership that way; udev will entirely skip all event
5034 handling for this disk and its partitions. If the disk
5035 was opened for writing, the close will trigger a partition
5036 table rescan in udev's "watch" facility, and if needed
5037 synthesize "change" events for the disk and all its partitions.
5038 This is now unconditionally enabled, and if it turns out to
5039 cause major problems, we might turn it on only for specific
5040 devices, or might need to disable it entirely. Device Mapper
5041 devices are excluded from this logic.
5043 * We temporarily dropped the "-l" switch for fsck invocations,
5044 since they collide with the flock() logic above. util-linux
5045 upstream has been changed already to avoid this conflict,
5046 and we will readd "-l" as soon as util-linux with this
5047 change has been released.
5049 * The dependency on libattr has been removed. Since a long
5050 time, the extended attribute calls have moved to glibc, and
5051 libattr is thus unnecessary.
5053 * Virtualization detection works without privileges now. This
5054 means the systemd-detect-virt binary no longer requires
5055 CAP_SYS_PTRACE file capabilities, and our daemons can run
5056 with fewer privileges.
5058 * systemd-networkd now runs under its own "systemd-network"
5059 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
5060 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
5061 loses the ability to write to files owned by root this way.
5063 * Similarly, systemd-resolved now runs under its own
5064 "systemd-resolve" user with no capabilities remaining.
5066 * Similarly, systemd-bus-proxyd now runs under its own
5067 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
5069 * systemd-networkd gained support for setting up "veth"
5070 virtual Ethernet devices for container connectivity, as well
5071 as GRE and VTI tunnels.
5073 * systemd-networkd will no longer automatically attempt to
5074 manually load kernel modules necessary for certain tunnel
5075 transports. Instead, it is assumed the kernel loads them
5076 automatically when required. This only works correctly on
5077 very new kernels. On older kernels, please consider adding
5078 the kernel modules to /etc/modules-load.d/ as a work-around.
5080 * The resolv.conf file systemd-resolved generates has been
5081 moved to /run/systemd/resolve/. If you have a symlink from
5082 /etc/resolv.conf, it might be necessary to correct it.
5084 * Two new service settings, ProtectHome= and ProtectSystem=,
5085 have been added. When enabled, they will make the user data
5086 (such as /home) inaccessible or read-only and the system
5087 (such as /usr) read-only, for specific services. This allows
5088 very light-weight per-service sandboxing to avoid
5089 modifications of user data or system files from
5090 services. These two new switches have been enabled for all
5091 of systemd's long-running services, where appropriate.
5093 * Socket units gained new SocketUser= and SocketGroup=
5094 settings to set the owner user and group of AF_UNIX sockets
5095 and FIFOs in the file system.
5097 * Socket units gained a new RemoveOnStop= setting. If enabled,
5098 all FIFOS and sockets in the file system will be removed
5099 when the specific socket unit is stopped.
5101 * Socket units gained a new Symlinks= setting. It takes a list
5102 of symlinks to create to file system sockets or FIFOs
5103 created by the specific Unix sockets. This is useful to
5104 manage symlinks to socket nodes with the same life-cycle as
5107 * The /dev/log socket and /dev/initctl FIFO have been moved to
5108 /run, and have been replaced by symlinks. This allows
5109 connecting to these facilities even if PrivateDevices=yes is
5110 used for a service (which makes /dev/log itself unavailable,
5111 but /run is left). This also has the benefit of ensuring
5112 that /dev only contains device nodes, directories and
5113 symlinks, and nothing else.
5115 * sd-daemon gained two new calls sd_pid_notify() and
5116 sd_pid_notifyf(). They are similar to sd_notify() and
5117 sd_notifyf(), but allow overriding of the source PID of
5118 notification messages if permissions permit this. This is
5119 useful to send notify messages on behalf of a different
5120 process (for example, the parent process). The
5121 systemd-notify tool has been updated to make use of this
5122 when sending messages (so that notification messages now
5123 originate from the shell script invoking systemd-notify and
5124 not the systemd-notify process itself. This should minimize
5125 a race where systemd fails to associate notification
5126 messages to services when the originating process already
5129 * A new "on-abnormal" setting for Restart= has been added. If
5130 set, it will result in automatic restarts on all "abnormal"
5131 reasons for a process to exit, which includes unclean
5132 signals, core dumps, timeouts and watchdog timeouts, but
5133 does not include clean and unclean exit codes or clean
5134 signals. Restart=on-abnormal is an alternative for
5135 Restart=on-failure for services that shall be able to
5136 terminate and avoid restarts on certain errors, by
5137 indicating so with an unclean exit code. Restart=on-failure
5138 or Restart=on-abnormal is now the recommended setting for
5139 all long-running services.
5141 * If the InaccessibleDirectories= service setting points to a
5142 mount point (or if there are any submounts contained within
5143 it), it is now attempted to completely unmount it, to make
5144 the file systems truly unavailable for the respective
5147 * The ReadOnlyDirectories= service setting and
5148 systemd-nspawn's --read-only parameter are now recursively
5149 applied to all submounts, too.
5151 * Mount units may now be created transiently via the bus APIs.
5153 * The support for SysV and LSB init scripts has been removed
5154 from the systemd daemon itself. Instead, it is now
5155 implemented as a generator that creates native systemd units
5156 from these scripts when needed. This enables us to remove a
5157 substantial amount of legacy code from PID 1, following the
5158 fact that many distributions only ship a very small number
5159 of LSB/SysV init scripts nowadays.
5161 * Privileged Xen (dom0) domains are not considered
5162 virtualization anymore by the virtualization detection
5163 logic. After all, they generally have unrestricted access to
5164 the hardware and usually are used to manage the unprivileged
5167 * systemd-tmpfiles gained a new "C" line type, for copying
5168 files or entire directories.
5170 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
5171 lines. So far, they have been non-globbing versions of the
5172 latter, and have thus been redundant. In future, it is
5173 recommended to only use "z". "m" has hence been removed
5174 from the documentation, even though it stays supported.
5176 * A tmpfiles snippet to recreate the most basic structure in
5177 /var has been added. This is enough to create the /var/run →
5178 /run symlink and create a couple of structural
5179 directories. This allows systems to boot up with an empty or
5180 volatile /var. Of course, while with this change, the core OS
5181 now is capable with dealing with a volatile /var, not all
5182 user services are ready for it. However, we hope that sooner
5183 or later, many service daemons will be changed upstream so
5184 that they are able to automatically create their necessary
5185 directories in /var at boot, should they be missing. This is
5186 the first step to allow state-less systems that only require
5187 the vendor image for /usr to boot.
5189 * systemd-nspawn has gained a new --tmpfs= switch to mount an
5190 empty tmpfs instance to a specific directory. This is
5191 particularly useful for making use of the automatic
5192 reconstruction of /var (see above), by passing --tmpfs=/var.
5194 * Access modes specified in tmpfiles snippets may now be
5195 prefixed with "~", which indicates that they shall be masked
5196 by whether the existing file or directory is currently
5197 writable, readable or executable at all. Also, if specified,
5198 the sgid/suid/sticky bits will be masked for all
5201 * A new passive target unit "network-pre.target" has been
5202 added which is useful for services that shall run before any
5203 network is configured, for example firewall scripts.
5205 * The "floppy" group that previously owned the /dev/fd*
5206 devices is no longer used. The "disk" group is now used
5207 instead. Distributions should probably deprecate usage of
5210 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
5211 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
5212 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
5213 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
5214 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
5215 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
5218 — Berlin, 2014-06-11
5222 * A new "systemd-timesyncd" daemon has been added for
5223 synchronizing the system clock across the network. It
5224 implements an SNTP client. In contrast to NTP
5225 implementations such as chrony or the NTP reference server,
5226 this only implements a client side, and does not bother with
5227 the full NTP complexity, focusing only on querying time from
5228 one remote server and synchronizing the local clock to
5229 it. Unless you intend to serve NTP to networked clients or
5230 want to connect to local hardware clocks, this simple NTP
5231 client should be more than appropriate for most
5232 installations. The daemon runs with minimal privileges, and
5233 has been hooked up with networkd to only operate when
5234 network connectivity is available. The daemon saves the
5235 current clock to disk every time a new NTP sync has been
5236 acquired, and uses this to possibly correct the system clock
5237 early at bootup, in order to accommodate for systems that
5238 lack an RTC such as the Raspberry Pi and embedded devices,
5239 and to make sure that time monotonically progresses on these
5240 systems, even if it is not always correct. To make use of
5241 this daemon, a new system user and group "systemd-timesync"
5242 needs to be created on installation of systemd.
5244 * The queue "seqnum" interface of libudev has been disabled, as
5245 it was generally incompatible with device namespacing as
5246 sequence numbers of devices go "missing" if the devices are
5247 part of a different namespace.
5249 * "systemctl list-timers" and "systemctl list-sockets" gained
5250 a --recursive switch for showing units of these types also
5251 for all local containers, similar in style to the already
5252 supported --recursive switch for "systemctl list-units".
5254 * A new RebootArgument= setting has been added for service
5255 units, which may be used to specify a kernel reboot argument
5256 to use when triggering reboots with StartLimitAction=.
5258 * A new FailureAction= setting has been added for service
5259 units which may be used to specify an operation to trigger
5260 when a service fails. This works similarly to
5261 StartLimitAction=, but unlike it, controls what is done
5262 immediately rather than only after several attempts to
5263 restart the service in question.
5265 * hostnamed got updated to also expose the kernel name,
5266 release, and version on the bus. This is useful for
5267 executing commands like hostnamectl with the -H switch.
5268 systemd-analyze makes use of this to properly display
5269 details when running non-locally.
5271 * The bootchart tool can now show cgroup information in the
5272 graphs it generates.
5274 * The CFS CPU quota cgroup attribute is now exposed for
5275 services. The new CPUQuota= switch has been added for this
5276 which takes a percentage value. Setting this will have the
5277 result that a service may never get more CPU time than the
5278 specified percentage, even if the machine is otherwise idle.
5280 * systemd-networkd learned IPIP and SIT tunnel support.
5282 * LSB init scripts exposing a dependency on $network will now
5283 get a dependency on network-online.target rather than simply
5284 network.target. This should bring LSB handling closer to
5285 what it was on SysV systems.
5287 * A new fsck.repair= kernel option has been added to control
5288 how fsck shall deal with unclean file systems at boot.
5290 * The (.ini) configuration file parser will now silently
5291 ignore sections whose name begins with "X-". This may be
5292 used to maintain application-specific extension sections in unit
5295 * machined gained a new API to query the IP addresses of
5296 registered containers. "machinectl status" has been updated
5297 to show these addresses in its output.
5299 * A new call sd_uid_get_display() has been added to the
5300 sd-login APIs for querying the "primary" session of a
5301 user. The "primary" session of the user is elected from the
5302 user's sessions and generally a graphical session is
5303 preferred over a text one.
5305 * A minimal systemd-resolved daemon has been added. It
5306 currently simply acts as a companion to systemd-networkd and
5307 manages resolv.conf based on per-interface DNS
5308 configuration, possibly supplied via DHCP. In the long run
5309 we hope to extend this into a local DNSSEC enabled DNS and
5312 * The systemd-networkd-wait-online tool is now enabled by
5313 default. It will delay network-online.target until a network
5314 connection has been configured. The tool primarily integrates
5315 with networkd, but will also make a best effort to make sense
5316 of network configuration performed in some other way.
5318 * Two new service options StartupCPUShares= and
5319 StartupBlockIOWeight= have been added that work similarly to
5320 CPUShares= and BlockIOWeight= however only apply during
5321 system startup. This is useful to prioritize certain services
5322 differently during bootup than during normal runtime.
5324 * hostnamed has been changed to prefer the statically
5325 configured hostname in /etc/hostname (unless set to
5326 'localhost' or empty) over any dynamic one supplied by
5327 dhcp. With this change, the rules for picking the hostname
5328 match more closely the rules of other configuration settings
5329 where the local administrator's configuration in /etc always
5330 overrides any other settings.
5332 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
5333 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
5334 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
5335 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
5336 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
5337 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
5338 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
5339 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
5340 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
5341 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
5342 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
5343 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
5344 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
5345 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
5346 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
5347 Lindskog, WaLyong Cho, Will Woods, Zbigniew
5350 — Beijing, 2014-05-28
5354 * When restoring the screen brightness at boot, stay away from
5355 the darkest setting or from the lowest 5% of the available
5356 range, depending on which is the larger value of both. This
5357 should effectively protect the user from rebooting into a
5358 black screen, should the brightness have been set to minimum
5361 * sd-login gained a new sd_machine_get_class() call to
5362 determine the class ("vm" or "container") of a machine
5363 registered with machined.
5365 * sd-login gained new calls
5366 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
5367 to query the identity of the peer of a local AF_UNIX
5368 connection. They operate similarly to their sd_pid_get_xyz()
5371 * PID 1 will now maintain a system-wide system state engine
5372 with the states "starting", "running", "degraded",
5373 "maintenance", "stopping". These states are bound to system
5374 startup, normal runtime, runtime with at least one failed
5375 service, rescue/emergency mode and system shutdown. This
5376 state is shown in the "systemctl status" output when no unit
5377 name is passed. It is useful to determine system state, in
5378 particularly when doing so for many systems or containers at
5381 * A new command "list-machines" has been added to "systemctl"
5382 that lists all local OS containers and shows their system
5383 state (see above), if systemd runs inside of them.
5385 * systemctl gained a new "-r" switch to recursively enumerate
5386 units on all local containers, when used with the
5387 "list-unit" command (which is the default one that is
5388 executed when no parameters are specified).
5390 * The GPT automatic partition discovery logic will now honour
5391 two GPT partition flags: one may be set on a partition to
5392 cause it to be mounted read-only, and the other may be set
5393 on a partition to ignore it during automatic discovery.
5395 * Two new GPT type UUIDs have been added for automatic root
5396 partition discovery, for 32-bit and 64-bit ARM. This is not
5397 particularly useful for discovering the root directory on
5398 these architectures during bare-metal boots (since UEFI is
5399 not common there), but still very useful to allow booting of
5400 ARM disk images in nspawn with the -i option.
5402 * MAC addresses of interfaces created with nspawn's
5403 --network-interface= switch will now be generated from the
5404 machine name, and thus be stable between multiple invocations
5407 * logind will now automatically remove all IPC objects owned
5408 by a user if she or he fully logs out. This makes sure that
5409 users who are logged out cannot continue to consume IPC
5410 resources. This covers SysV memory, semaphores and message
5411 queues as well as POSIX shared memory and message
5412 queues. Traditionally, SysV and POSIX IPC had no life-cycle
5413 limits. With this functionality, that is corrected. This may
5414 be turned off by using the RemoveIPC= switch of logind.conf.
5416 * The systemd-machine-id-setup and tmpfiles tools gained a
5417 --root= switch to operate on a specific root directory,
5420 * journald can now forward logged messages to the TTYs of all
5421 logged in users ("wall"). This is the default for all
5422 emergency messages now.
5424 * A new tool systemd-journal-remote has been added to stream
5425 journal log messages across the network.
5427 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
5428 controller trees are mounted into it. Note that the
5429 directories mounted beneath it are not read-only. This is a
5430 security measure and is particularly useful because glibc
5431 actually includes a search logic to pick any tmpfs it can
5432 find to implement shm_open() if /dev/shm is not available
5433 (which it might very well be in namespaced setups).
5435 * machinectl gained a new "poweroff" command to cleanly power
5436 down a local OS container.
5438 * The PrivateDevices= unit file setting will now also drop the
5439 CAP_MKNOD capability from the capability bound set, and
5440 imply DevicePolicy=closed.
5442 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
5443 comprehensively on all long-running systemd services where
5444 this is appropriate.
5446 * systemd-udevd will now run in a disassociated mount
5447 namespace. To mount directories from udev rules, make sure to
5448 pull in mount units via SYSTEMD_WANTS properties.
5450 * The kdbus support gained support for uploading policy into
5451 the kernel. sd-bus gained support for creating "monitoring"
5452 connections that can eavesdrop into all bus communication
5453 for debugging purposes.
5455 * Timestamps may now be specified in seconds since the UNIX
5456 epoch Jan 1st, 1970 by specifying "@" followed by the value
5459 * Native tcpwrap support in systemd has been removed. tcpwrap
5460 is old code, not really maintained anymore and has serious
5461 shortcomings, and better options such as firewalls
5462 exist. For setups that require tcpwrap usage, please
5463 consider invoking your socket-activated service via tcpd,
5464 like on traditional inetd.
5466 * A new system.conf configuration option
5467 DefaultTimerAccuracySec= has been added that controls the
5468 default AccuracySec= setting of .timer units.
5470 * Timer units gained a new WakeSystem= switch. If enabled,
5471 timers configured this way will cause the system to resume
5472 from system suspend (if the system supports that, which most
5475 * Timer units gained a new Persistent= switch. If enabled,
5476 timers configured this way will save to disk when they have
5477 been last triggered. This information is then used on next
5478 reboot to possible execute overdue timer events, that
5479 could not take place because the system was powered off.
5480 This enables simple anacron-like behaviour for timer units.
5482 * systemctl's "list-timers" will now also list the time a
5483 timer unit was last triggered in addition to the next time
5484 it will be triggered.
5486 * systemd-networkd will now assign predictable IPv4LL
5487 addresses to its local interfaces.
5489 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
5490 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
5491 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
5492 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
5493 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
5494 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
5495 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
5496 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
5499 — Berlin, 2014-03-25
5503 * A new unit file setting RestrictAddressFamilies= has been
5504 added to restrict which socket address families unit
5505 processes gain access to. This takes address family names
5506 like "AF_INET" or "AF_UNIX", and is useful to minimize the
5507 attack surface of services via exotic protocol stacks. This
5508 is built on seccomp system call filters.
5510 * Two new unit file settings RuntimeDirectory= and
5511 RuntimeDirectoryMode= have been added that may be used to
5512 manage a per-daemon runtime directories below /run. This is
5513 an alternative for setting up directory permissions with
5514 tmpfiles snippets, and has the advantage that the runtime
5515 directory's lifetime is bound to the daemon runtime and that
5516 the daemon starts up with an empty directory each time. This
5517 is particularly useful when writing services that drop
5518 privileges using the User= or Group= setting.
5520 * The DeviceAllow= unit setting now supports globbing for
5521 matching against device group names.
5523 * The systemd configuration file system.conf gained new
5524 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
5525 DefaultMemoryAccounting= to globally turn on/off accounting
5526 for specific resources (cgroups) for all units. These
5527 settings may still be overridden individually in each unit
5530 * systemd-gpt-auto-generator is now able to discover /srv and
5531 root partitions in addition to /home and swap partitions. It
5532 also supports LUKS-encrypted partitions now. With this in
5533 place, automatic discovery of partitions to mount following
5534 the Discoverable Partitions Specification
5535 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
5536 is now a lot more complete. This allows booting without
5537 /etc/fstab and without root= on the kernel command line on
5538 systems prepared appropriately.
5540 * systemd-nspawn gained a new --image= switch which allows
5541 booting up disk images and Linux installations on any block
5542 device that follow the Discoverable Partitions Specification
5543 (see above). This means that installations made with
5544 appropriately updated installers may now be started and
5545 deployed using container managers, completely
5546 unmodified. (We hope that libvirt-lxc will add support for
5547 this feature soon, too.)
5549 * systemd-nspawn gained a new --network-macvlan= setting to
5550 set up a private macvlan interface for the
5551 container. Similarly, systemd-networkd gained a new
5552 Kind=macvlan setting in .netdev files.
5554 * systemd-networkd now supports configuring local addresses
5557 * A new tool systemd-network-wait-online has been added to
5558 synchronously wait for network connectivity using
5561 * The sd-bus.h bus API gained a new sd_bus_track object for
5562 tracking the life-cycle of bus peers. Note that sd-bus.h is
5563 still not a public API though (unless you specify
5564 --enable-kdbus on the configure command line, which however
5565 voids your warranty and you get no API stability guarantee).
5567 * The $XDG_RUNTIME_DIR runtime directories for each user are
5568 now individual tmpfs instances, which has the benefit of
5569 introducing separate pools for each user, with individual
5570 size limits, and thus making sure that unprivileged clients
5571 can no longer negatively impact the system or other users by
5572 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
5573 RuntimeDirectorySize= has been introduced that allows
5574 controlling the default size limit for all users. It
5575 defaults to 10% of the available physical memory. This is no
5576 replacement for quotas on tmpfs though (which the kernel
5577 still does not support), as /dev/shm and /tmp are still
5578 shared resources used by both the system and unprivileged
5581 * logind will now automatically turn off automatic suspending
5582 on laptop lid close when more than one display is
5583 connected. This was previously expected to be implemented
5584 individually in desktop environments (such as GNOME),
5585 however has been added to logind now, in order to fix a
5586 boot-time race where a desktop environment might not have
5587 been started yet and thus not been able to take an inhibitor
5588 lock at the time where logind already suspends the system
5589 due to a closed lid.
5591 * logind will now wait at least 30s after each system
5592 suspend/resume cycle, and 3min after system boot before
5593 suspending the system due to a closed laptop lid. This
5594 should give USB docking stations and similar enough time to
5595 be probed and configured after system resume and boot in
5596 order to then act as suspend blocker.
5598 * systemd-run gained a new --property= setting which allows
5599 initialization of resource control properties (and others)
5600 for the created scope or service unit. Example: "systemd-run
5601 --property=BlockIOWeight=10 updatedb" may be used to run
5602 updatedb at a low block IO scheduling weight.
5604 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
5605 now also work in --scope mode.
5607 * When systemd is compiled with kdbus support, basic support
5608 for enforced policies is now in place. (Note that enabling
5609 kdbus still voids your warranty and no API compatibility
5612 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
5613 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5614 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
5615 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
5616 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
5617 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
5618 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
5619 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
5620 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
5621 Zbigniew Jędrzejewski-Szmek
5623 — Berlin, 2014-03-12
5627 * systemd will now relabel /dev after loading the SMACK policy
5628 according to SMACK rules.
5630 * A new unit file option AppArmorProfile= has been added to
5631 set the AppArmor profile for the processes of a unit.
5633 * A new condition check ConditionArchitecture= has been added
5634 to conditionalize units based on the system architecture, as
5635 reported by uname()'s "machine" field.
5637 * systemd-networkd now supports matching on the system
5638 virtualization, architecture, kernel command line, host name
5641 * logind is now a lot more aggressive when suspending the
5642 machine due to a closed laptop lid. Instead of acting only
5643 on the lid close action, it will continuously watch the lid
5644 status and act on it. This is useful for laptops where the
5645 power button is on the outside of the chassis so that it can
5646 be reached without opening the lid (such as the Lenovo
5647 Yoga). On those machines, logind will now immediately
5648 re-suspend the machine if the power button has been
5649 accidentally pressed while the laptop was suspended and in a
5650 backpack or similar.
5652 * logind will now watch SW_DOCK switches and inhibit reaction
5653 to the lid switch if it is pressed. This means that logind
5654 will not suspend the machine anymore if the lid is closed
5655 and the system is docked, if the laptop supports SW_DOCK
5656 notifications via the input layer. Note that ACPI docking
5657 stations do not generate this currently. Also note that this
5658 logic is usually not fully sufficient and Desktop
5659 Environments should take a lid switch inhibitor lock when an
5660 external display is connected, as systemd will not watch
5663 * nspawn will now make use of the devices cgroup controller by
5664 default, and only permit creation of and access to the usual
5665 API device nodes like /dev/null or /dev/random, as well as
5666 access to (but not creation of) the pty devices.
5668 * We will now ship a default .network file for
5669 systemd-networkd that automatically configures DHCP for
5670 network interfaces created by nspawn's --network-veth or
5671 --network-bridge= switches.
5673 * systemd will now understand the usual M, K, G, T suffixes
5674 according to SI conventions (i.e. to the base 1000) when
5675 referring to throughput and hardware metrics. It will stay
5676 with IEC conventions (i.e. to the base 1024) for software
5677 metrics, according to what is customary according to
5678 Wikipedia. We explicitly document which base applies for
5679 each configuration option.
5681 * The DeviceAllow= setting in unit files now supports a syntax
5682 to whitelist an entire group of devices node majors at once,
5683 based on the /proc/devices listing. For example, with the
5684 string "char-pts", it is now possible to whitelist all
5685 current and future pseudo-TTYs at once.
5687 * sd-event learned a new "post" event source. Event sources of
5688 this type are triggered by the dispatching of any event
5689 source of a type that is not "post". This is useful for
5690 implementing clean-up and check event sources that are
5691 triggered by other work being done in the program.
5693 * systemd-networkd is no longer statically enabled, but uses
5694 the usual [Install] sections so that it can be
5695 enabled/disabled using systemctl. It still is enabled by
5698 * When creating a veth interface pair with systemd-nspawn, the
5699 host side will now be prefixed with "vb-" if
5700 --network-bridge= is used, and with "ve-" if --network-veth
5701 is used. This way, it is easy to distinguish these cases on
5702 the host, for example to apply different configuration to
5703 them with systemd-networkd.
5705 * The compatibility libraries for libsystemd-journal.so,
5706 libsystem-id128.so, libsystemd-login.so and
5707 libsystemd-daemon.so do not make use of IFUNC
5708 anymore. Instead, we now build libsystemd.so multiple times
5709 under these alternative names. This means that the footprint
5710 is drastically increased, but given that these are
5711 transitional compatibility libraries, this should not matter
5712 much. This change has been made necessary to support the ARM
5713 platform for these compatibility libraries, as the ARM
5714 toolchain is not really at the same level as the toolchain
5715 for other architectures like x86 and does not support
5716 IFUNC. Please make sure to use --enable-compat-libs only
5717 during a transitional period!
5719 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
5720 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5721 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
5722 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
5723 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
5724 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
5725 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
5726 Zbigniew Jędrzejewski-Szmek
5728 — Berlin, 2014-02-24
5732 * A new component "systemd-networkd" has been added that can
5733 be used to configure local network interfaces statically or
5734 via DHCP. It is capable of bringing up bridges, VLANs, and
5735 bonding. Currently, no hook-ups for interactive network
5736 configuration are provided. Use this for your initrd,
5737 container, embedded, or server setup if you need a simple,
5738 yet powerful, network configuration solution. This
5739 configuration subsystem is quite nifty, as it allows wildcard
5740 hotplug matching in interfaces. For example, with a single
5741 configuration snippet, you can configure that all Ethernet
5742 interfaces showing up are automatically added to a bridge,
5743 or similar. It supports link-sensing and more.
5745 * A new tool "systemd-socket-proxyd" has been added which can
5746 act as a bidirectional proxy for TCP sockets. This is
5747 useful for adding socket activation support to services that
5748 do not actually support socket activation, including virtual
5749 machines and the like.
5751 * Add a new tool to save/restore rfkill state on
5754 * Save/restore state of keyboard backlights in addition to
5755 display backlights on shutdown/boot.
5757 * udev learned a new SECLABEL{} construct to label device
5758 nodes with a specific security label when they appear. For
5759 now, only SECLABEL{selinux} is supported, but the syntax is
5760 prepared for additional security frameworks.
5762 * udev gained a new scheme to configure link-level attributes
5763 from files in /etc/systemd/network/*.link. These files can
5764 match against MAC address, device path, driver name and type,
5765 and will apply attributes like the naming policy, link speed,
5766 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
5767 address assignment policy (randomized, ...).
5769 * The configuration of network interface naming rules for
5770 "permanent interface names" has changed: a new NamePolicy=
5771 setting in the [Link] section of .link files determines the
5772 priority of possible naming schemes (onboard, slot, MAC,
5773 path). The default value of this setting is determined by
5774 /usr/lib/net/links/99-default.link. Old
5775 80-net-name-slot.rules udev configuration file has been
5776 removed, so local configuration overriding this file should
5777 be adapted to override 99-default.link instead.
5779 * When the User= switch is used in a unit file, also
5780 initialize $SHELL= based on the user database entry.
5782 * systemd no longer depends on libdbus. All communication is
5783 now done with sd-bus, systemd's low-level bus library
5786 * kdbus support has been added to PID 1 itself. When kdbus is
5787 enabled, this causes PID 1 to set up the system bus and
5788 enable support for a new ".busname" unit type that
5789 encapsulates bus name activation on kdbus. It works a little
5790 bit like ".socket" units, except for bus names. A new
5791 generator has been added that converts classic dbus1 service
5792 activation files automatically into native systemd .busname
5795 * sd-bus: add a light-weight vtable implementation that allows
5796 defining objects on the bus with a simple static const
5797 vtable array of its methods, signals and properties.
5799 * systemd will not generate or install static dbus
5800 introspection data anymore to /usr/share/dbus-1/interfaces,
5801 as the precise format of these files is unclear, and
5802 nothing makes use of it.
5804 * A proxy daemon is now provided to proxy clients connecting
5805 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
5806 compatibility with classic D-Bus.
5808 * A bus driver implementation has been added that supports the
5809 classic D-Bus bus driver calls on kdbus, also for
5810 compatibility purposes.
5812 * A new API "sd-event.h" has been added that implements a
5813 minimal event loop API built around epoll. It provides a
5814 couple of features that direct epoll usage is lacking:
5815 prioritization of events, scales to large numbers of timer
5816 events, per-event timer slack (accuracy), system-wide
5817 coalescing of timer events, exit handlers, watchdog
5818 supervision support using systemd's sd_notify() API, child
5821 * A new API "sd-rntl.h" has been added that provides an API
5822 around the route netlink interface of the kernel, similar in
5823 style to "sd-bus.h".
5825 * A new API "sd-dhcp-client.h" has been added that provides a
5826 small DHCPv4 client-side implementation. This is used by
5829 * There is a new kernel command line option
5830 "systemd.restore_state=0|1". When set to "0", none of the
5831 systemd tools will restore saved runtime state to hardware
5832 devices. More specifically, the rfkill and backlight states
5835 * The FsckPassNo= compatibility option in mount/service units
5836 has been removed. The fstab generator will now add the
5837 necessary dependencies automatically, and does not require
5838 PID1's support for that anymore.
5840 * journalctl gained a new switch, --list-boots, that lists
5841 recent boots with their times and boot IDs.
5843 * The various tools like systemctl, loginctl, timedatectl,
5844 busctl, systemd-run, ... have gained a new switch "-M" to
5845 connect to a specific, local OS container (as direct
5846 connection, without requiring SSH). This works on any
5847 container that is registered with machined, such as those
5848 created by libvirt-lxc or nspawn.
5850 * systemd-run and systemd-analyze also gained support for "-H"
5851 to connect to remote hosts via SSH. This is particularly
5852 useful for systemd-run because it enables queuing of jobs
5853 onto remote systems.
5855 * machinectl gained a new command "login" to open a getty
5856 login in any local container. This works with any container
5857 that is registered with machined (such as those created by
5858 libvirt-lxc or nspawn), and which runs systemd inside.
5860 * machinectl gained a new "reboot" command that may be used to
5861 trigger a reboot on a specific container that is registered
5862 with machined. This works on any container that runs an init
5863 system of some kind.
5865 * systemctl gained a new "list-timers" command to print a nice
5866 listing of installed timer units with the times they elapse
5869 * Alternative reboot() parameters may now be specified on the
5870 "systemctl reboot" command line and are passed to the
5871 reboot() system call.
5873 * systemctl gained a new --job-mode= switch to configure the
5874 mode to queue a job with. This is a more generic version of
5875 --fail, --irreversible, and --ignore-dependencies, which are
5876 still available but not advertised anymore.
5878 * /etc/systemd/system.conf gained new settings to configure
5879 various default timeouts of units, as well as the default
5880 start limit interval and burst. These may still be overridden
5883 * PID1 will now export on the bus profile data of the security
5884 policy upload process (such as the SELinux policy upload to
5887 * journald: when forwarding logs to the console, include
5888 timestamps (following the setting in
5889 /sys/module/printk/parameters/time).
5891 * OnCalendar= in timer units now understands the special
5892 strings "yearly" and "annually". (Both are equivalent)
5894 * The accuracy of timer units is now configurable with the new
5895 AccuracySec= setting. It defaults to 1min.
5897 * A new dependency type JoinsNamespaceOf= has been added that
5898 allows running two services within the same /tmp and network
5899 namespace, if PrivateNetwork= or PrivateTmp= are used.
5901 * A new command "cat" has been added to systemctl. It outputs
5902 the original unit file of a unit, and concatenates the
5903 contents of additional "drop-in" unit file snippets, so that
5904 the full configuration is shown.
5906 * systemctl now supports globbing on the various "list-xyz"
5907 commands, like "list-units" or "list-sockets", as well as on
5908 those commands which take multiple unit names.
5910 * journalctl's --unit= switch gained support for globbing.
5912 * All systemd daemons now make use of the watchdog logic so
5913 that systemd automatically notices when they hang.
5915 * If the $container_ttys environment variable is set,
5916 getty-generator will automatically spawn a getty for each
5917 listed tty. This is useful for container managers to request
5918 login gettys to be spawned on as many ttys as needed.
5920 * %h, %s, %U specifier support is not available anymore when
5921 used in unit files for PID 1. This is because NSS calls are
5922 not safe from PID 1. They stay available for --user
5923 instances of systemd, and as special case for the root user.
5925 * loginctl gained a new "--no-legend" switch to turn off output
5928 * The "sd-login.h" API gained three new calls:
5929 sd_session_is_remote(), sd_session_get_remote_user(),
5930 sd_session_get_remote_host() to query information about
5933 * The udev hardware database now also carries vendor/product
5934 information of SDIO devices.
5936 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
5937 determine whether watchdog notifications are requested by
5940 * Socket-activated per-connection services now include a
5941 short description of the connection parameters in the
5944 * tmpfiles gained a new "--boot" option. When this is not used,
5945 only lines where the command character is not suffixed with
5946 "!" are executed. When this option is specified, those
5947 options are executed too. This partitions tmpfiles
5948 directives into those that can be safely executed at any
5949 time, and those which should be run only at boot (for
5950 example, a line that creates /run/nologin).
5952 * A new API "sd-resolve.h" has been added which provides a simple
5953 asynchronous wrapper around glibc NSS host name resolution
5954 calls, such as getaddrinfo(). In contrast to glibc's
5955 getaddrinfo_a(), it does not use signals. In contrast to most
5956 other asynchronous name resolution libraries, this one does
5957 not reimplement DNS, but reuses NSS, so that alternate
5958 host name resolution systems continue to work, such as mDNS,
5959 LDAP, etc. This API is based on libasyncns, but it has been
5960 cleaned up for inclusion in systemd.
5962 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
5963 "sd-daemon.h" are no longer found in individual libraries
5964 libsystemd-journal.so, libsystemd-login.so,
5965 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
5966 merged them into a single library, libsystemd.so, which
5967 provides all symbols. The reason for this is cyclic
5968 dependencies, as these libraries tend to use each other's
5969 symbols. So far, we have managed to workaround that by linking
5970 a copy of a good part of our code into each of these
5971 libraries again and again, which, however, makes certain
5972 things hard to do, like sharing static variables. Also, it
5973 substantially increases footprint. With this change, there
5974 is only one library for the basic APIs systemd
5975 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
5976 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
5977 library as well, however are subject to the --enable-kdbus
5978 switch (see below). Note that "sd-dhcp-client.h" is not part
5979 of this library (this is because it only consumes, never
5980 provides, services of/to other APIs). To make the transition
5981 easy from the separate libraries to the unified one, we
5982 provide the --enable-compat-libs compile-time switch which
5983 will generate stub libraries that are compatible with the
5984 old ones but redirect all calls to the new one.
5986 * All of the kdbus logic and the new APIs "sd-bus.h",
5987 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
5988 and "sd-utf8.h" are compile-time optional via the
5989 "--enable-kdbus" switch, and they are not compiled in by
5990 default. To make use of kdbus, you have to explicitly enable
5991 the switch. Note however, that neither the kernel nor the
5992 userspace API for all of this is considered stable yet. We
5993 want to maintain the freedom to still change the APIs for
5994 now. By specifying this build-time switch, you acknowledge
5995 that you are aware of the instability of the current
5998 * Also, note that while kdbus is pretty much complete,
5999 it lacks one thing: proper policy support. This means you
6000 can build a fully working system with all features; however,
6001 it will be highly insecure. Policy support will be added in
6002 one of the next releases, at the same time that we will
6003 declare the APIs stable.
6005 * When the kernel command line argument "kdbus" is specified,
6006 systemd will automatically load the kdbus.ko kernel module. At
6007 this stage of development, it is only useful for testing kdbus
6008 and should not be used in production. Note: if "--enable-kdbus"
6009 is specified, and the kdbus.ko kernel module is available, and
6010 "kdbus" is added to the kernel command line, the entire system
6011 runs with kdbus instead of dbus-daemon, with the above mentioned
6012 problem of missing the system policy enforcement. Also a future
6013 version of kdbus.ko or a newer systemd will not be compatible with
6014 each other, and will unlikely be able to boot the machine if only
6015 one of them is updated.
6017 * systemctl gained a new "import-environment" command which
6018 uploads the caller's environment (or parts thereof) into the
6019 service manager so that it is inherited by services started
6020 by the manager. This is useful to upload variables like
6021 $DISPLAY into the user service manager.
6023 * A new PrivateDevices= switch has been added to service units
6024 which allows running a service with a namespaced /dev
6025 directory that does not contain any device nodes for
6026 physical devices. More specifically, it only includes devices
6027 such as /dev/null, /dev/urandom, and /dev/zero which are API
6030 * logind has been extended to support behaviour like VT
6031 switching on seats that do not support a VT. This makes
6032 multi-session available on seats that are not the first seat
6033 (seat0), and on systems where kernel support for VTs has
6034 been disabled at compile-time.
6036 * If a process holds a delay lock for system sleep or shutdown
6037 and fails to release it in time, we will now log its
6038 identity. This makes it easier to identify processes that
6039 cause slow suspends or power-offs.
6041 * When parsing /etc/crypttab, support for a new key-slot=
6042 option as supported by Debian is added. It allows indicating
6043 which LUKS slot to use on disk, speeding up key loading.
6045 * The sd_journal_sendv() API call has been checked and
6046 officially declared to be async-signal-safe so that it may
6047 be invoked from signal handlers for logging purposes.
6049 * Boot-time status output is now enabled automatically after a
6050 short timeout if boot does not progress, in order to give
6051 the user an indication what she or he is waiting for.
6053 * The boot-time output has been improved to show how much time
6054 remains until jobs expire.
6056 * The KillMode= switch in service units gained a new possible
6057 value "mixed". If set, and the unit is shut down, then the
6058 initial SIGTERM signal is sent only to the main daemon
6059 process, while the following SIGKILL signal is sent to
6060 all remaining processes of the service.
6062 * When a scope unit is registered, a new property "Controller"
6063 may be set. If set to a valid bus name, systemd will send a
6064 RequestStop() signal to this name when it would like to shut
6065 down the scope. This may be used to hook manager logic into
6066 the shutdown logic of scope units. Also, scope units may now
6067 be put in a special "abandoned" state, in which case the
6068 manager process which created them takes no further
6069 responsibilities for it.
6071 * When reading unit files, systemd will now verify
6072 the access mode of these files, and warn about certain
6073 suspicious combinations. This has been added to make it
6074 easier to track down packaging bugs where unit files are
6075 marked executable or world-writable.
6077 * systemd-nspawn gained a new "--setenv=" switch to set
6078 container-wide environment variables. The similar option in
6079 systemd-activate was renamed from "--environment=" to
6080 "--setenv=" for consistency.
6082 * systemd-nspawn has been updated to create a new kdbus domain
6083 for each container that is invoked, thus allowing each
6084 container to have its own set of system and user buses,
6085 independent of the host.
6087 * systemd-nspawn gained a new --drop-capability= switch to run
6088 the container with less capabilities than the default. Both
6089 --drop-capability= and --capability= now take the special
6090 string "all" for dropping or keeping all capabilities.
6092 * systemd-nspawn gained new switches for executing containers
6093 with specific SELinux labels set.
6095 * systemd-nspawn gained a new --quiet switch to not generate
6096 any additional output but the container's own console
6099 * systemd-nspawn gained a new --share-system switch to run a
6100 container without PID namespacing enabled.
6102 * systemd-nspawn gained a new --register= switch to control
6103 whether the container is registered with systemd-machined or
6104 not. This is useful for containers that do not run full
6105 OS images, but only specific apps.
6107 * systemd-nspawn gained a new --keep-unit which may be used
6108 when invoked as the only program from a service unit, and
6109 results in registration of the unit service itself in
6110 systemd-machined, instead of a newly opened scope unit.
6112 * systemd-nspawn gained a new --network-interface= switch for
6113 moving arbitrary interfaces to the container. The new
6114 --network-veth switch creates a virtual Ethernet connection
6115 between host and container. The new --network-bridge=
6116 switch then allows assigning the host side of this virtual
6117 Ethernet connection to a bridge device.
6119 * systemd-nspawn gained a new --personality= switch for
6120 setting the kernel personality for the container. This is
6121 useful when running a 32-bit container on a 64-bit host. A
6122 similar option Personality= is now also available for service
6125 * logind will now also track a "Desktop" identifier for each
6126 session which encodes the desktop environment of it. This is
6127 useful for desktop environments that want to identify
6128 multiple running sessions of itself easily.
6130 * A new SELinuxContext= setting for service units has been
6131 added that allows setting a specific SELinux execution
6132 context for a service.
6134 * Most systemd client tools will now honour $SYSTEMD_LESS for
6135 settings of the "less" pager. By default, these tools will
6136 override $LESS to allow certain operations to work, such as
6137 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
6138 influence this logic.
6140 * systemd's "seccomp" hook-up has been changed to make use of
6141 the libseccomp library instead of using its own
6142 implementation. This has benefits for portability among
6145 * For usage together with SystemCallFilter=, a new
6146 SystemCallErrorNumber= setting has been introduced that
6147 allows configuration of a system error number to be returned
6148 on filtered system calls, instead of immediately killing the
6149 process. Also, SystemCallArchitectures= has been added to
6150 limit access to system calls of a particular architecture
6151 (in order to turn off support for unused secondary
6152 architectures). There is also a global
6153 SystemCallArchitectures= setting in system.conf now to turn
6154 off support for non-native system calls system-wide.
6156 * systemd requires a kernel with a working name_to_handle_at(),
6157 please see the kernel config requirements in the README file.
6159 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
6160 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
6161 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
6162 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
6163 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
6164 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
6165 Elia Pinto, Florian Weimer, George McCollister, Goffredo
6166 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
6167 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
6168 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
6169 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
6170 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
6171 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
6172 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
6173 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
6174 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
6175 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
6176 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
6177 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
6178 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
6179 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
6180 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
6181 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
6182 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
6184 — Berlin, 2014-02-20
6188 * logind has gained support for facilitating privileged input
6189 and drm device access for unprivileged clients. This work is
6190 useful to allow Wayland display servers (and similar
6191 programs, such as kmscon) to run under the user's ID and
6192 access input and drm devices which are normally
6193 protected. When this is used (and the kernel is new enough)
6194 logind will "mute" IO on the file descriptors passed to
6195 Wayland as long as it is in the background and "unmute" it
6196 if it returns into the foreground. This allows secure
6197 session switching without allowing background sessions to
6198 eavesdrop on input and display data. This also introduces
6199 session switching support if VT support is turned off in the
6200 kernel, and on seats that are not seat0.
6202 * A new kernel command line option luks.options= is understood
6203 now which allows specifying LUKS options for usage for LUKS
6204 encrypted partitions specified with luks.uuid=.
6206 * tmpfiles.d(5) snippets may now use specifier expansion in
6207 path names. More specifically %m, %b, %H, %v, are now
6208 replaced by the local machine id, boot id, hostname, and
6209 kernel version number.
6211 * A new tmpfiles.d(5) command "m" has been introduced which
6212 may be used to change the owner/group/access mode of a file
6213 or directory if it exists, but do nothing if it does not.
6215 * This release removes high-level support for the
6216 MemorySoftLimit= cgroup setting. The underlying kernel
6217 cgroup attribute memory.soft_limit= is currently badly
6218 designed and likely to be removed from the kernel API in its
6219 current form, hence we should not expose it for now.
6221 * The memory.use_hierarchy cgroup attribute is now enabled for
6222 all cgroups systemd creates in the memory cgroup
6223 hierarchy. This option is likely to be come the built-in
6224 default in the kernel anyway, and the non-hierarchical mode
6225 never made much sense in the intrinsically hierarchical
6228 * A new field _SYSTEMD_SLICE= is logged along with all journal
6229 messages containing the slice a message was generated
6230 from. This is useful to allow easy per-customer filtering of
6231 logs among other things.
6233 * systemd-journald will no longer adjust the group of journal
6234 files it creates to the "systemd-journal" group. Instead we
6235 rely on the journal directory to be owned by the
6236 "systemd-journal" group, and its setgid bit set, so that the
6237 kernel file system layer will automatically enforce that
6238 journal files inherit this group assignment. The reason for
6239 this change is that we cannot allow NSS look-ups from
6240 journald which would be necessary to resolve
6241 "systemd-journal" to a numeric GID, because this might
6242 create deadlocks if NSS involves synchronous queries to
6243 other daemons (such as nscd, or sssd) which in turn are
6244 logging clients of journald and might block on it, which
6245 would then dead lock. A tmpfiles.d(5) snippet included in
6246 systemd will make sure the setgid bit and group are
6247 properly set on the journal directory if it exists on every
6248 boot. However, we recommend adjusting it manually after
6249 upgrades too (or from RPM scriptlets), so that the change is
6250 not delayed until next reboot.
6252 * Backlight and random seed files in /var/lib/ have moved into
6253 the /var/lib/systemd/ directory, in order to centralize all
6254 systemd generated files in one directory.
6256 * Boot time performance measurements (as displayed by
6257 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
6258 performance information if that's available to determine how
6259 much time BIOS and boot loader initialization required. With
6260 a sufficiently new BIOS you hence no longer need to boot
6261 with Gummiboot to get access to such information.
6263 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
6264 Cristian Rodríguez, Dave Reisner, David Herrmann, David
6265 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
6266 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
6267 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
6268 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
6269 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6271 — Berlin, 2013-10-02
6275 * The Restart= option for services now understands a new
6276 on-watchdog setting, which will restart the service
6277 automatically if the service stops sending out watchdog keep
6278 alive messages (as configured with WatchdogSec=).
6280 * The getty generator (which is responsible for bringing up a
6281 getty on configured serial consoles) will no longer only
6282 start a getty on the primary kernel console but on all
6283 others, too. This makes the order in which console= is
6284 specified on the kernel command line less important.
6286 * libsystemd-logind gained a new sd_session_get_vt() call to
6287 retrieve the VT number of a session.
6289 * If the option "tries=0" is set for an entry of /etc/crypttab
6290 its passphrase is queried indefinitely instead of any
6291 maximum number of tries.
6293 * If a service with a configure PID file terminates its PID
6294 file will now be removed automatically if it still exists
6295 afterwards. This should put an end to stale PID files.
6297 * systemd-run will now also take relative binary path names
6298 for execution and no longer insists on absolute paths.
6300 * InaccessibleDirectories= and ReadOnlyDirectories= now take
6301 paths that are optionally prefixed with "-" to indicate that
6302 it should not be considered a failure if they do not exist.
6304 * journalctl -o (and similar commands) now understands a new
6305 output mode "short-precise", it is similar to "short" but
6306 shows timestamps with usec accuracy.
6308 * The option "discard" (as known from Debian) is now
6309 synonymous to "allow-discards" in /etc/crypttab. In fact,
6310 "discard" is preferred now (since it is easier to remember
6313 * Some licensing clean-ups were made, so that more code is now
6314 LGPL-2.1 licensed than before.
6316 * A minimal tool to save/restore the display backlight
6317 brightness across reboots has been added. It will store the
6318 backlight setting as late as possible at shutdown, and
6319 restore it as early as possible during reboot.
6321 * A logic to automatically discover and enable home and swap
6322 partitions on GPT disks has been added. With this in place
6323 /etc/fstab becomes optional for many setups as systemd can
6324 discover certain partitions located on the root disk
6325 automatically. Home partitions are recognized under their
6326 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
6327 partitions are recognized under their GPT type ID
6328 0657fd6da4ab43c484e50933c84b4f4f.
6330 * systemd will no longer pass any environment from the kernel
6331 or initrd to system services. If you want to set an
6332 environment for all services, do so via the kernel command
6333 line systemd.setenv= assignment.
6335 * The systemd-sysctl tool no longer natively reads the file
6336 /etc/sysctl.conf. If desired, the file should be symlinked
6337 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
6338 legacy support by a symlink rather than built-in code, it
6339 also makes the otherwise hidden order of application of the
6340 different files visible. (Note that this partly reverts to a
6341 pre-198 application order of sysctl knobs!)
6343 * The "systemctl set-log-level" and "systemctl dump" commands
6344 have been moved to systemd-analyze.
6346 * systemd-run learned the new --remain-after-exit switch,
6347 which causes the scope unit not to be cleaned up
6348 automatically after the process terminated.
6350 * tmpfiles learned a new --exclude-prefix= switch to exclude
6351 certain paths from operation.
6353 * journald will now automatically flush all messages to disk
6354 as soon as a message at the log level CRIT, ALERT or EMERG
6357 Contributions from: Andrew Cook, Brandon Philips, Christian
6358 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
6359 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
6360 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
6361 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
6362 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
6363 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
6364 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
6365 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
6366 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
6367 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
6368 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
6369 William Giokas, Zbigniew Jędrzejewski-Szmek
6371 — Berlin, 2013-09-13
6375 * The documentation has been updated to cover the various new
6376 concepts introduced with 205.
6378 * Unit files now understand the new %v specifier which
6379 resolves to the kernel version string as returned by "uname
6382 * systemctl now supports filtering the unit list output by
6383 load state, active state and sub state, using the new
6386 * "systemctl status" will now show the results of the
6387 condition checks (like ConditionPathExists= and similar) of
6388 the last start attempts of the unit. They are also logged to
6391 * "journalctl -b" may now be used to look for boot output of a
6392 specific boot. Try "journalctl -b -1" for the previous boot,
6393 but the syntax is substantially more powerful.
6395 * "journalctl --show-cursor" has been added which prints the
6396 cursor string the last shown log line. This may then be used
6397 with the new "journalctl --after-cursor=" switch to continue
6398 browsing logs from that point on.
6400 * "journalctl --force" may now be used to force regeneration
6403 * Creation of "dead" device nodes has been moved from udev
6404 into kmod and tmpfiles. Previously, udev would read the kmod
6405 databases to pre-generate dead device nodes based on meta
6406 information contained in kernel modules, so that these would
6407 be auto-loaded on access rather then at boot. As this
6408 does not really have much to do with the exposing actual
6409 kernel devices to userspace this has always been slightly
6410 alien in the udev codebase. Following the new scheme kmod
6411 will now generate a runtime snippet for tmpfiles from the
6412 module meta information and it now is tmpfiles' job to the
6413 create the nodes. This also allows overriding access and
6414 other parameters for the nodes using the usual tmpfiles
6415 facilities. As side effect this allows us to remove the
6416 CAP_SYS_MKNOD capability bit from udevd entirely.
6418 * logind's device ACLs may now be applied to these "dead"
6419 devices nodes too, thus finally allowing managed access to
6420 devices such as /dev/snd/sequencer without loading the
6421 backing module right-away.
6423 * A new RPM macro has been added that may be used to apply
6424 tmpfiles configuration during package installation.
6426 * systemd-detect-virt and ConditionVirtualization= now can
6427 detect User-Mode-Linux machines (UML).
6429 * journald will now implicitly log the effective capabilities
6430 set of processes in the message metadata.
6432 * systemd-cryptsetup has gained support for TrueCrypt volumes.
6434 * The initrd interface has been simplified (more specifically,
6435 support for passing performance data via environment
6436 variables and fsck results via files in /run has been
6437 removed). These features were non-essential, and are
6438 nowadays available in a much nicer way by having systemd in
6439 the initrd serialize its state and have the hosts systemd
6440 deserialize it again.
6442 * The udev "keymap" data files and tools to apply keyboard
6443 specific mappings of scan to key codes, and force-release
6444 scan code lists have been entirely replaced by a udev
6445 "keyboard" builtin and a hwdb data file.
6447 * systemd will now honour the kernel's "quiet" command line
6448 argument also during late shutdown, resulting in a
6449 completely silent shutdown when used.
6451 * There's now an option to control the SO_REUSEPORT socket
6452 option in .socket units.
6454 * Instance units will now automatically get a per-template
6455 subslice of system.slice unless something else is explicitly
6456 configured. For example, instances of sshd@.service will now
6457 implicitly be placed in system-sshd.slice rather than
6458 system.slice as before.
6460 * Test coverage support may now be enabled at build time.
6462 Contributions from: Dave Reisner, Frederic Crozat, Harald
6463 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
6464 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
6465 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
6466 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
6467 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
6468 Giokas, Zbigniew Jędrzejewski-Szmek
6470 — Berlin, 2013-07-23
6474 * Two new unit types have been introduced:
6476 Scope units are very similar to service units, however, are
6477 created out of pre-existing processes — instead of PID 1
6478 forking off the processes. By using scope units it is
6479 possible for system services and applications to group their
6480 own child processes (worker processes) in a powerful way
6481 which then maybe used to organize them, or kill them
6482 together, or apply resource limits on them.
6484 Slice units may be used to partition system resources in an
6485 hierarchical fashion and then assign other units to them. By
6486 default there are now three slices: system.slice (for all
6487 system services), user.slice (for all user sessions),
6488 machine.slice (for VMs and containers).
6490 Slices and scopes have been introduced primarily in
6491 context of the work to move cgroup handling to a
6492 single-writer scheme, where only PID 1
6493 creates/removes/manages cgroups.
6495 * There's a new concept of "transient" units. In contrast to
6496 normal units these units are created via an API at runtime,
6497 not from configuration from disk. More specifically this
6498 means it is now possible to run arbitrary programs as
6499 independent services, with all execution parameters passed
6500 in via bus APIs rather than read from disk. Transient units
6501 make systemd substantially more dynamic then it ever was,
6502 and useful as a general batch manager.
6504 * logind has been updated to make use of scope and slice units
6505 for managing user sessions. As a user logs in he will get
6506 his own private slice unit, to which all sessions are added
6507 as scope units. We also added support for automatically
6508 adding an instance of user@.service for the user into the
6509 slice. Effectively logind will no longer create cgroup
6510 hierarchies on its own now, it will defer entirely to PID 1
6511 for this by means of scope, service and slice units. Since
6512 user sessions this way become entities managed by PID 1
6513 the output of "systemctl" is now a lot more comprehensive.
6515 * A new mini-daemon "systemd-machined" has been added which
6516 may be used by virtualization managers to register local
6517 VMs/containers. nspawn has been updated accordingly, and
6518 libvirt will be updated shortly. machined will collect a bit
6519 of meta information about the VMs/containers, and assign
6520 them their own scope unit (see above). The collected
6521 meta-data is then made available via the "machinectl" tool,
6522 and exposed in "ps" and similar tools. machined/machinectl
6523 is compile-time optional.
6525 * As discussed earlier, the low-level cgroup configuration
6526 options ControlGroup=, ControlGroupModify=,
6527 ControlGroupPersistent=, ControlGroupAttribute= have been
6528 removed. Please use high-level attribute settings instead as
6529 well as slice units.
6531 * A new bus call SetUnitProperties() has been added to alter
6532 various runtime parameters of a unit. This is primarily
6533 useful to alter cgroup parameters dynamically in a nice way,
6534 but will be extended later on to make more properties
6535 modifiable at runtime. systemctl gained a new set-properties
6536 command that wraps this call.
6538 * A new tool "systemd-run" has been added which can be used to
6539 run arbitrary command lines as transient services or scopes,
6540 while configuring a number of settings via the command
6541 line. This tool is currently very basic, however already
6542 very useful. We plan to extend this tool to even allow
6543 queuing of execution jobs with time triggers from the
6544 command line, similar in fashion to "at".
6546 * nspawn will now inform the user explicitly that kernels with
6547 audit enabled break containers, and suggest the user to turn
6550 * Support for detecting the IMA and AppArmor security
6551 frameworks with ConditionSecurity= has been added.
6553 * journalctl gained a new "-k" switch for showing only kernel
6554 messages, mimicking dmesg output; in addition to "--user"
6555 and "--system" switches for showing only user's own logs
6558 * systemd-delta can now show information about drop-in
6559 snippets extending unit files.
6561 * libsystemd-bus has been substantially updated but is still
6562 not available as public API.
6564 * systemd will now look for the "debug" argument on the kernel
6565 command line and enable debug logging, similar to what
6566 "systemd.log_level=debug" already did before.
6568 * "systemctl set-default", "systemctl get-default" has been
6569 added to configure the default.target symlink, which
6570 controls what to boot into by default.
6572 * "systemctl set-log-level" has been added as a convenient
6573 way to raise and lower systemd logging threshold.
6575 * "systemd-analyze plot" will now show the time the various
6576 generators needed for execution, as well as information
6577 about the unit file loading.
6579 * libsystemd-journal gained a new sd_journal_open_files() call
6580 for opening specific journal files. journactl also gained a
6581 new switch to expose this new functionality. Previously we
6582 only supported opening all files from a directory, or all
6583 files from the system, as opening individual files only is
6584 racy due to journal file rotation.
6586 * systemd gained the new DefaultEnvironment= setting in
6587 /etc/systemd/system.conf to set environment variables for
6590 * If a privileged process logs a journal message with the
6591 OBJECT_PID= field set, then journald will automatically
6592 augment this with additional OBJECT_UID=, OBJECT_GID=,
6593 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
6594 system services want to log events about specific client
6595 processes. journactl/systemctl has been updated to make use
6596 of this information if all log messages regarding a specific
6599 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
6600 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
6601 Reisner, David Coppa, David King, David Strauss, Eelco
6602 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
6603 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
6604 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
6605 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
6606 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
6607 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
6608 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
6609 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
6610 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
6611 Łukasz Stelmach, 장동준
6615 * The Python bindings gained some minimal support for the APIs
6616 exposed by libsystemd-logind.
6618 * ConditionSecurity= gained support for detecting SMACK. Since
6619 this condition already supports SELinux and AppArmor we only
6620 miss IMA for this. Patches welcome!
6622 Contributions from: Karol Lewandowski, Lennart Poettering,
6623 Zbigniew Jędrzejewski-Szmek
6627 * systemd-nspawn will now create /etc/resolv.conf if
6628 necessary, before bind-mounting the host's file onto it.
6630 * systemd-nspawn will now store meta information about a
6631 container on the container's cgroup as extended attribute
6632 fields, including the root directory.
6634 * The cgroup hierarchy has been reworked in many ways. All
6635 objects any of the components systemd creates in the cgroup
6636 tree are now suffixed. More specifically, user sessions are
6637 now placed in cgroups suffixed with ".session", users in
6638 cgroups suffixed with ".user", and nspawn containers in
6639 cgroups suffixed with ".nspawn". Furthermore, all cgroup
6640 names are now escaped in a simple scheme to avoid collision
6641 of userspace object names with kernel filenames. This work
6642 is preparation for making these objects relocatable in the
6643 cgroup tree, in order to allow easy resource partitioning of
6644 these objects without causing naming conflicts.
6646 * systemctl list-dependencies gained the new switches
6647 --plain, --reverse, --after and --before.
6649 * systemd-inhibit now shows the process name of processes that
6650 have taken an inhibitor lock.
6652 * nss-myhostname will now also resolve "localhost"
6653 implicitly. This makes /etc/hosts an optional file and
6654 nicely handles that on IPv6 ::1 maps to both "localhost" and
6657 * libsystemd-logind.so gained a new call
6658 sd_get_machine_names() to enumerate running containers and
6659 VMs (currently only supported by very new libvirt and
6660 nspawn). sd_login_monitor can now be used to watch
6661 VMs/containers coming and going.
6663 * .include is not allowed recursively anymore, and only in
6664 unit files. Usually it is better to use drop-in snippets in
6665 .d/*.conf anyway, as introduced with systemd 198.
6667 * systemd-analyze gained a new "critical-chain" command that
6668 determines the slowest chain of units run during system
6669 boot-up. It is very useful for tracking down where
6670 optimizing boot time is the most beneficial.
6672 * systemd will no longer allow manipulating service paths in
6673 the name=systemd:/system cgroup tree using ControlGroup= in
6674 units. (But is still fine with it in all other dirs.)
6676 * There's a new systemd-nspawn@.service service file that may
6677 be used to easily run nspawn containers as system
6678 services. With the container's root directory in
6679 /var/lib/container/foobar it is now sufficient to run
6680 "systemctl start systemd-nspawn@foobar.service" to boot it.
6682 * systemd-cgls gained a new parameter "--machine" to list only
6683 the processes within a certain container.
6685 * ConditionSecurity= now can check for "apparmor". We still
6686 are lacking checks for SMACK and IMA for this condition
6687 check though. Patches welcome!
6689 * A new configuration file /etc/systemd/sleep.conf has been
6690 added that may be used to configure which kernel operation
6691 systemd is supposed to execute when "suspend", "hibernate"
6692 or "hybrid-sleep" is requested. This makes the new kernel
6693 "freeze" state accessible to the user.
6695 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
6696 the passed argument if applicable.
6698 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
6699 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
6700 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
6701 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
6702 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
6703 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
6704 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
6709 * The output of 'systemctl list-jobs' got some polishing. The
6710 '--type=' argument may now be passed more than once. A new
6711 command 'systemctl list-sockets' has been added which shows
6712 a list of kernel sockets systemd is listening on with the
6713 socket units they belong to, plus the units these socket
6716 * The experimental libsystemd-bus library got substantial
6717 updates to work in conjunction with the (also experimental)
6718 kdbus kernel project. It works well enough to exchange
6719 messages with some sophistication. Note that kdbus is not
6720 ready yet, and the library is mostly an elaborate test case
6721 for now, and not installable.
6723 * systemd gained a new unit 'systemd-static-nodes.service'
6724 that generates static device nodes earlier during boot, and
6725 can run in conjunction with udev.
6727 * libsystemd-login gained a new call sd_pid_get_user_unit()
6728 to retrieve the user systemd unit a process is running
6729 in. This is useful for systems where systemd is used as
6732 * systemd-nspawn now places all containers in the new /machine
6733 top-level cgroup directory in the name=systemd
6734 hierarchy. libvirt will soon do the same, so that we get a
6735 uniform separation of /system, /user and /machine for system
6736 services, user processes and containers/virtual
6737 machines. This new cgroup hierarchy is also useful to stick
6738 stable names to specific container instances, which can be
6739 recognized later this way (this name may be controlled
6740 via systemd-nspawn's new -M switch). libsystemd-login also
6741 gained a new call sd_pid_get_machine_name() to retrieve the
6742 name of the container/VM a specific process belongs to.
6744 * bootchart can now store its data in the journal.
6746 * libsystemd-journal gained a new call
6747 sd_journal_add_conjunction() for AND expressions to the
6748 matching logic. This can be used to express more complex
6749 logical expressions.
6751 * journactl can now take multiple --unit= and --user-unit=
6754 * The cryptsetup logic now understands the "luks.key=" kernel
6755 command line switch for specifying a file to read the
6756 decryption key from. Also, if a configured key file is not
6757 found the tool will now automatically fall back to prompting
6760 * Python systemd.journal module was updated to wrap recently
6761 added functions from libsystemd-journal. The interface was
6762 changed to bring the low level interface in s.j._Reader
6763 closer to the C API, and the high level interface in
6764 s.j.Reader was updated to wrap and convert all data about
6767 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
6768 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
6769 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
6770 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
6771 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
6772 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6776 * journalctl --update-catalog now understands a new --root=
6777 option to operate on catalogs found in a different root
6780 * During shutdown after systemd has terminated all running
6781 services a final killing loop kills all remaining left-over
6782 processes. We will now print the name of these processes
6783 when we send SIGKILL to them, since this usually indicates a
6786 * If /etc/crypttab refers to password files stored on
6787 configured mount points automatic dependencies will now be
6788 generated to ensure the specific mount is established first
6789 before the key file is attempted to be read.
6791 * 'systemctl status' will now show information about the
6792 network sockets a socket unit is listening on.
6794 * 'systemctl status' will also shown information about any
6795 drop-in configuration file for units. (Drop-In configuration
6796 files in this context are files such as
6797 /etc/systemd/systemd/foobar.service.d/*.conf)
6799 * systemd-cgtop now optionally shows summed up CPU times of
6800 cgroups. Press '%' while running cgtop to switch between
6801 percentage and absolute mode. This is useful to determine
6802 which cgroups use up the most CPU time over the entire
6803 runtime of the system. systemd-cgtop has also been updated
6804 to be 'pipeable' for processing with further shell tools.
6806 * 'hostnamectl set-hostname' will now allow setting of FQDN
6809 * The formatting and parsing of time span values has been
6810 changed. The parser now understands fractional expressions
6811 such as "5.5h". The formatter will now output fractional
6812 expressions for all time spans under 1min, i.e. "5.123456s"
6813 rather than "5s 123ms 456us". For time spans under 1s
6814 millisecond values are shown, for those under 1ms
6815 microsecond values are shown. This should greatly improve
6816 all time-related output of systemd.
6818 * libsystemd-login and libsystemd-journal gained new
6819 functions for querying the poll() events mask and poll()
6820 timeout value for integration into arbitrary event
6823 * localectl gained the ability to list available X11 keymaps
6824 (models, layouts, variants, options).
6826 * 'systemd-analyze dot' gained the ability to filter for
6827 specific units via shell-style globs, to create smaller,
6828 more useful graphs. I.e. it is now possible to create simple
6829 graphs of all the dependencies between only target units, or
6830 of all units that Avahi has dependencies with.
6832 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
6833 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
6834 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
6835 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
6836 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
6837 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
6838 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
6842 * The boot-time readahead implementation for rotating media
6843 will now read the read-ahead data in multiple passes which
6844 consist of all read requests made in equidistant time
6845 intervals. This means instead of strictly reading read-ahead
6846 data in its physical order on disk we now try to find a
6847 middle ground between physical and access time order.
6849 * /etc/os-release files gained a new BUILD_ID= field for usage
6850 on operating systems that provide continuous builds of OS
6853 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
6854 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
6855 William Douglas, Zbigniew Jędrzejewski-Szmek
6859 * systemd-python gained an API exposing libsystemd-daemon.
6861 * The SMACK setup logic gained support for uploading CIPSO
6864 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
6865 ReadOnlyDirectories= and InaccessibleDirectories= has
6866 changed. The private /tmp and /var/tmp directories are now
6867 shared by all processes of a service (which means
6868 ExecStartPre= may now leave data in /tmp that ExecStart= of
6869 the same service can still access). When a service is
6870 stopped its temporary directories are immediately deleted
6871 (normal clean-up with tmpfiles is still done in addition to
6874 * By default, systemd will now set a couple of sysctl
6875 variables in the kernel: the safe sysrq options are turned
6876 on, IP route verification is turned on, and source routing
6877 disabled. The recently added hardlink and softlink
6878 protection of the kernel is turned on. These settings should
6879 be reasonably safe, and good defaults for all new systems.
6881 * The predictable network naming logic may now be turned off
6882 with a new kernel command line switch: net.ifnames=0.
6884 * A new libsystemd-bus module has been added that implements a
6885 pretty complete D-Bus client library. For details see:
6887 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
6889 * journald will now explicitly flush the journal files to disk
6890 at the latest 5min after each write. The file will then also
6891 be marked offline until the next write. This should increase
6892 reliability in case of a crash. The synchronization delay
6893 can be configured via SyncIntervalSec= in journald.conf.
6895 * There's a new remote-fs-setup.target unit that can be used
6896 to pull in specific services when at least one remote file
6897 system is to be mounted.
6899 * There are new targets timers.target and paths.target as
6900 canonical targets to pull user timer and path units in
6901 from. This complements sockets.target with a similar
6902 purpose for socket units.
6904 * libudev gained a new call udev_device_set_attribute_value()
6905 to set sysfs attributes of a device.
6907 * The udev daemon now sets the default number of worker
6908 processes executed in parallel based on the number of available
6909 CPUs instead of the amount of available RAM. This is supposed
6910 to provide a more reliable default and limit a too aggressive
6911 parallelism for setups with 1000s of devices connected.
6913 Contributions from: Auke Kok, Colin Walters, Cristian
6914 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
6915 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
6916 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
6917 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
6918 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
6919 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
6920 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
6921 Zbigniew Jędrzejewski-Szmek
6925 * Configuration of unit files may now be extended via drop-in
6926 files without having to edit/override the unit files
6927 themselves. More specifically, if the administrator wants to
6928 change one value for a service file foobar.service he can
6929 now do so by dropping in a configuration snippet into
6930 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
6931 will load all these snippets and apply them on top of the
6932 main unit configuration file, possibly extending or
6933 overriding its settings. Using these drop-in snippets is
6934 generally nicer than the two earlier options for changing
6935 unit files locally: copying the files from
6936 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
6937 them there; or creating a new file in /etc/systemd/system/
6938 that incorporates the original one via ".include". Drop-in
6939 snippets into these .d/ directories can be placed in any
6940 directory systemd looks for units in, and the usual
6941 overriding semantics between /usr/lib, /etc and /run apply
6944 * Most unit file settings which take lists of items can now be
6945 reset by assigning the empty string to them. For example,
6946 normally, settings such as Environment=FOO=BAR append a new
6947 environment variable assignment to the environment block,
6948 each time they are used. By assigning Environment= the empty
6949 string the environment block can be reset to empty. This is
6950 particularly useful with the .d/*.conf drop-in snippets
6951 mentioned above, since this adds the ability to reset list
6952 settings from vendor unit files via these drop-ins.
6954 * systemctl gained a new "list-dependencies" command for
6955 listing the dependencies of a unit recursively.
6957 * Inhibitors are now honored and listed by "systemctl
6958 suspend", "systemctl poweroff" (and similar) too, not only
6959 GNOME. These commands will also list active sessions by
6962 * Resource limits (as exposed by the various control group
6963 controllers) can now be controlled dynamically at runtime
6964 for all units. More specifically, you can now use a command
6965 like "systemctl set-cgroup-attr foobar.service cpu.shares
6966 2000" to alter the CPU shares a specific service gets. These
6967 settings are stored persistently on disk, and thus allow the
6968 administrator to easily adjust the resource usage of
6969 services with a few simple commands. This dynamic resource
6970 management logic is also available to other programs via the
6971 bus. Almost any kernel cgroup attribute and controller is
6974 * systemd-vconsole-setup will now copy all font settings to
6975 all allocated VTs, where it previously applied them only to
6978 * libsystemd-login gained the new sd_session_get_tty() API
6981 * This release drops support for a few legacy or
6982 distribution-specific LSB facility names when parsing init
6983 scripts: $x-display-manager, $mail-transfer-agent,
6984 $mail-transport-agent, $mail-transfer-agent, $smtp,
6985 $null. Also, the mail-transfer-agent.target unit backing
6986 this has been removed. Distributions which want to retain
6987 compatibility with this should carry the burden for
6988 supporting this themselves and patch support for these back
6989 in, if they really need to. Also, the facilities $syslog and
6990 $local_fs are now ignored, since systemd does not support
6991 early-boot LSB init scripts anymore, and these facilities
6992 are implied anyway for normal services. syslog.target has
6995 * There are new bus calls on PID1's Manager object for
6996 cancelling jobs, and removing snapshot units. Previously,
6997 both calls were only available on the Job and Snapshot
7000 * systemd-journal-gatewayd gained SSL support.
7002 * The various "environment" files, such as /etc/locale.conf
7003 now support continuation lines with a backslash ("\") as
7004 last character in the line, similarly in style (but different)
7005 to how this is supported in shells.
7007 * For normal user processes the _SYSTEMD_USER_UNIT= field is
7008 now implicitly appended to every log entry logged. systemctl
7009 has been updated to filter by this field when operating on a
7010 user systemd instance.
7012 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
7013 CAP_AUDIT_CONTROL capabilities to the capabilities set for
7014 the container. This makes it easier to boot unmodified
7015 Fedora systems in a container, which however still requires
7016 audit=0 to be passed on the kernel command line. Auditing in
7017 kernel and userspace is unfortunately still too broken in
7018 context of containers, hence we recommend compiling it out
7019 of the kernel or using audit=0. Hopefully this will be fixed
7020 one day for good in the kernel.
7022 * nspawn gained the new --bind= and --bind-ro= parameters to
7023 bind mount specific directories from the host into the
7026 * nspawn will now mount its own devpts file system instance
7027 into the container, in order not to leak pty devices from
7028 the host into the container.
7030 * systemd will now read the firmware boot time performance
7031 information from the EFI variables, if the used boot loader
7032 supports this, and takes it into account for boot performance
7033 analysis via "systemd-analyze". This is currently supported
7034 only in conjunction with Gummiboot, but could be supported
7035 by other boot loaders too. For details see:
7037 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
7039 * A new generator has been added that automatically mounts the
7040 EFI System Partition (ESP) to /boot, if that directory
7041 exists, is empty, and no other file system has been
7042 configured to be mounted there.
7044 * logind will now send out PrepareForSleep(false) out
7045 unconditionally, after coming back from suspend. This may be
7046 used by applications as asynchronous notification for
7047 system resume events.
7049 * "systemctl unlock-sessions" has been added, that allows
7050 unlocking the screens of all user sessions at once, similar
7051 to how "systemctl lock-sessions" already locked all users
7052 sessions. This is backed by a new D-Bus call UnlockSessions().
7054 * "loginctl seat-status" will now show the master device of a
7055 seat. (i.e. the device of a seat that needs to be around for
7056 the seat to be considered available, usually the graphics
7059 * tmpfiles gained a new "X" line type, that allows
7060 configuration of files and directories (with wildcards) that
7061 shall be excluded from automatic cleanup ("aging").
7063 * udev default rules set the device node permissions now only
7064 at "add" events, and do not change them any longer with a
7065 later "change" event.
7067 * The log messages for lid events and power/sleep keypresses
7068 now carry a message ID.
7070 * We now have a substantially larger unit test suite, but this
7071 continues to be work in progress.
7073 * udevadm hwdb gained a new --root= parameter to change the
7074 root directory to operate relative to.
7076 * logind will now issue a background sync() request to the kernel
7077 early at shutdown, so that dirty buffers are flushed to disk early
7078 instead of at the last moment, in order to optimize shutdown
7081 * A new bootctl tool has been added that is an interface for
7082 certain boot loader operations. This is currently a preview
7083 and is likely to be extended into a small mechanism daemon
7084 like timedated, localed, hostnamed, and can be used by
7085 graphical UIs to enumerate available boot options, and
7086 request boot into firmware operations.
7088 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
7089 the rest of the package. It also has been updated to work
7090 correctly in initrds.
7092 * Policykit previously has been runtime optional, and is now
7093 also compile time optional via a configure switch.
7095 * systemd-analyze has been reimplemented in C. Also "systemctl
7096 dot" has moved into systemd-analyze.
7098 * "systemctl status" with no further parameters will now print
7099 the status of all active or failed units.
7101 * Operations such as "systemctl start" can now be executed
7102 with a new mode "--irreversible" which may be used to queue
7103 operations that cannot accidentally be reversed by a later
7104 job queuing. This is by default used to make shutdown
7105 requests more robust.
7107 * The Python API of systemd now gained a new module for
7108 reading journal files.
7110 * A new tool kernel-install has been added that can install
7111 kernel images according to the Boot Loader Specification:
7113 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
7115 * Boot time console output has been improved to provide
7116 animated boot time output for hanging jobs.
7118 * A new tool systemd-activate has been added which can be used
7119 to test socket activation with, directly from the command
7120 line. This should make it much easier to test and debug
7121 socket activation in daemons.
7123 * journalctl gained a new "--reverse" (or -r) option to show
7124 journal output in reverse order (i.e. newest line first).
7126 * journalctl gained a new "--pager-end" (or -e) option to jump
7127 to immediately jump to the end of the journal in the
7128 pager. This is only supported in conjunction with "less".
7130 * journalctl gained a new "--user-unit=" option, that works
7131 similarly to "--unit=" but filters for user units rather than
7134 * A number of unit files to ease adoption of systemd in
7135 initrds has been added. This moves some minimal logic from
7136 the various initrd implementations into systemd proper.
7138 * The journal files are now owned by a new group
7139 "systemd-journal", which exists specifically to allow access
7140 to the journal, and nothing else. Previously, we used the
7141 "adm" group for that, which however possibly covers more
7142 than just journal/log file access. This new group is now
7143 already used by systemd-journal-gatewayd to ensure this
7144 daemon gets access to the journal files and as little else
7145 as possible. Note that "make install" will also set FS ACLs
7146 up for /var/log/journal to give "adm" and "wheel" read
7147 access to it, in addition to "systemd-journal" which owns
7148 the journal files. We recommend that packaging scripts also
7149 add read access to "adm" + "wheel" to /var/log/journal, and
7150 all existing/future journal files. To normal users and
7151 administrators little changes, however packagers need to
7152 ensure to create the "systemd-journal" system group at
7153 package installation time.
7155 * The systemd-journal-gatewayd now runs as unprivileged user
7156 systemd-journal-gateway:systemd-journal-gateway. Packaging
7157 scripts need to create these system user/group at
7160 * timedated now exposes a new boolean property CanNTP that
7161 indicates whether a local NTP service is available or not.
7163 * systemd-detect-virt will now also detect xen PVs
7165 * The pstore file system is now mounted by default, if it is
7168 * In addition to the SELinux and IMA policies we will now also
7169 load SMACK policies at early boot.
7171 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
7172 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
7173 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
7174 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
7175 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
7176 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
7177 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
7178 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
7179 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
7180 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
7181 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
7182 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
7183 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
7184 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
7188 * Timer units now support calendar time events in addition to
7189 monotonic time events. That means you can now trigger a unit
7190 based on a calendar time specification such as "Thu,Fri
7191 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
7192 or fifth day of any month of the year 2013, given that it is
7193 a thursday or friday. This brings timer event support
7194 considerably closer to cron's capabilities. For details on
7195 the supported calendar time specification language see
7198 * udev now supports a number of different naming policies for
7199 network interfaces for predictable names, and a combination
7200 of these policies is now the default. Please see this wiki
7201 document for details:
7203 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
7205 * Auke Kok's bootchart implementation has been added to the
7206 systemd tree. It is an optional component that can graph the
7207 boot in quite some detail. It is one of the best bootchart
7208 implementations around and minimal in its code and
7211 * nss-myhostname has been integrated into the systemd source
7212 tree. nss-myhostname guarantees that the local hostname
7213 always stays resolvable via NSS. It has been a weak
7214 requirement of systemd-hostnamed since a long time, and
7215 since its code is actually trivial we decided to just
7216 include it in systemd's source tree. It can be turned off
7217 with a configure switch.
7219 * The read-ahead logic is now capable of properly detecting
7220 whether a btrfs file system is on SSD or rotating media, in
7221 order to optimize the read-ahead scheme. Previously, it was
7222 only capable of detecting this on traditional file systems
7225 * In udev, additional device properties are now read from the
7226 IAB in addition to the OUI database. Also, Bluetooth company
7227 identities are attached to the devices as well.
7229 * In service files %U may be used as specifier that is
7230 replaced by the configured user name of the service.
7232 * nspawn may now be invoked without a controlling TTY. This
7233 makes it suitable for invocation as its own service. This
7234 may be used to set up a simple containerized server system
7235 using only core OS tools.
7237 * systemd and nspawn can now accept socket file descriptors
7238 when they are started for socket activation. This enables
7239 implementation of socket activated nspawn
7240 containers. i.e. think about autospawning an entire OS image
7241 when the first SSH or HTTP connection is received. We expect
7242 that similar functionality will also be added to libvirt-lxc
7245 * journalctl will now suppress ANSI color codes when
7246 presenting log data.
7248 * systemctl will no longer show control group information for
7249 a unit if the control group is empty anyway.
7251 * logind can now automatically suspend/hibernate/shutdown the
7254 * /etc/machine-info and hostnamed now also expose the chassis
7255 type of the system. This can be used to determine whether
7256 the local system is a laptop, desktop, handset or
7257 tablet. This information may either be configured by the
7258 user/vendor or is automatically determined from ACPI and DMI
7259 information if possible.
7261 * A number of PolicyKit actions are now bound together with
7262 "imply" rules. This should simplify creating UIs because
7263 many actions will now authenticate similar ones as well.
7265 * Unit files learnt a new condition ConditionACPower= which
7266 may be used to conditionalize a unit depending on whether an
7267 AC power source is connected or not, of whether the system
7268 is running on battery power.
7270 * systemctl gained a new "is-failed" verb that may be used in
7271 shell scripts and suchlike to check whether a specific unit
7272 is in the "failed" state.
7274 * The EnvironmentFile= setting in unit files now supports file
7275 globbing, and can hence be used to easily read a number of
7276 environment files at once.
7278 * systemd will no longer detect and recognize specific
7279 distributions. All distribution-specific #ifdeffery has been
7280 removed, systemd is now fully generic and
7281 distribution-agnostic. Effectively, not too much is lost as
7282 a lot of the code is still accessible via explicit configure
7283 switches. However, support for some distribution specific
7284 legacy configuration file formats has been dropped. We
7285 recommend distributions to simply adopt the configuration
7286 files everybody else uses now and convert the old
7287 configuration from packaging scripts. Most distributions
7288 already did that. If that's not possible or desirable,
7289 distributions are welcome to forward port the specific
7290 pieces of code locally from the git history.
7292 * When logging a message about a unit systemd will now always
7293 log the unit name in the message meta data.
7295 * localectl will now also discover system locale data that is
7296 not stored in locale archives, but directly unpacked.
7298 * logind will no longer unconditionally use framebuffer
7299 devices as seat masters, i.e. as devices that are required
7300 to be existing before a seat is considered preset. Instead,
7301 it will now look for all devices that are tagged as
7302 "seat-master" in udev. By default, framebuffer devices will
7303 be marked as such, but depending on local systems, other
7304 devices might be marked as well. This may be used to
7305 integrate graphics cards using closed source drivers (such
7306 as NVidia ones) more nicely into logind. Note however, that
7307 we recommend using the open source NVidia drivers instead,
7308 and no udev rules for the closed-source drivers will be
7309 shipped from us upstream.
7311 Contributions from: Adam Williamson, Alessandro Crismani, Auke
7312 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
7313 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
7314 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
7315 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
7316 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
7317 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
7318 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
7319 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
7320 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
7321 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
7322 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
7327 * udev gained support for loading additional device properties
7328 from an indexed database that is keyed by vendor/product IDs
7329 and similar device identifiers. For the beginning this
7330 "hwdb" is populated with data from the well-known PCI and
7331 USB database, but also includes PNP, ACPI and OID data. In
7332 the longer run this indexed database shall grow into
7333 becoming the one central database for non-essential
7334 userspace device metadata. Previously, data from the PCI/USB
7335 database was only attached to select devices, since the
7336 lookup was a relatively expensive operation due to O(n) time
7337 complexity (with n being the number of entries in the
7338 database). Since this is now O(1), we decided to add in this
7339 data for all devices where this is available, by
7340 default. Note that the indexed database needs to be rebuilt
7341 when new data files are installed. To achieve this you need
7342 to update your packaging scripts to invoke "udevadm hwdb
7343 --update" after installation of hwdb data files. For
7344 RPM-based distributions we introduced the new
7345 %udev_hwdb_update macro for this purpose.
7347 * The Journal gained support for the "Message Catalog", an
7348 indexed database to link up additional information with
7349 journal entries. For further details please check:
7351 https://www.freedesktop.org/wiki/Software/systemd/catalog
7353 The indexed message catalog database also needs to be
7354 rebuilt after installation of message catalog files. Use
7355 "journalctl --update-catalog" for this. For RPM-based
7356 distributions we introduced the %journal_catalog_update
7357 macro for this purpose.
7359 * The Python Journal bindings gained support for the standard
7360 Python logging framework.
7362 * The Journal API gained new functions for checking whether
7363 the underlying file system of a journal file is capable of
7364 properly reporting file change notifications, or whether
7365 applications that want to reflect journal changes "live"
7366 need to recheck journal files continuously in appropriate
7369 * It is now possible to set the "age" field for tmpfiles
7370 entries to 0, indicating that files matching this entry
7371 shall always be removed when the directories are cleaned up.
7373 * coredumpctl gained a new "gdb" verb which invokes gdb
7374 right-away on the selected coredump.
7376 * There's now support for "hybrid sleep" on kernels that
7377 support this, in addition to "suspend" and "hibernate". Use
7378 "systemctl hybrid-sleep" to make use of this.
7380 * logind's HandleSuspendKey= setting (and related settings)
7381 now gained support for a new "lock" setting to simply
7382 request the screen lock on all local sessions, instead of
7383 actually executing a suspend or hibernation.
7385 * systemd will now mount the EFI variables file system by
7388 * Socket units now gained support for configuration of the
7389 SMACK security label.
7391 * timedatectl will now output the time of the last and next
7392 daylight saving change.
7394 * We dropped support for various legacy and distro-specific
7395 concepts, such as insserv, early-boot SysV services
7396 (i.e. those for non-standard runlevels such as 'b' or 'S')
7397 or ArchLinux /etc/rc.conf support. We recommend the
7398 distributions who still need support this to either continue
7399 to maintain the necessary patches downstream, or find a
7400 different solution. (Talk to us if you have questions!)
7402 * Various systemd components will now bypass PolicyKit checks
7403 for root and otherwise handle properly if PolicyKit is not
7404 found to be around. This should fix most issues for
7405 PolicyKit-less systems. Quite frankly this should have been
7406 this way since day one. It is absolutely our intention to
7407 make systemd work fine on PolicyKit-less systems, and we
7408 consider it a bug if something does not work as it should if
7409 PolicyKit is not around.
7411 * For embedded systems it is now possible to build udev and
7412 systemd without blkid and/or kmod support.
7414 * "systemctl switch-root" is now capable of switching root
7415 more than once. I.e. in addition to transitions from the
7416 initrd to the host OS it is now possible to transition to
7417 further OS images from the host. This is useful to implement
7418 offline updating tools.
7420 * Various other additions have been made to the RPM macros
7421 shipped with systemd. Use %udev_rules_update() after
7422 installing new udev rules files. %_udevhwdbdir,
7423 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
7424 %_sysctldir are now available which resolve to the right
7425 directories for packages to place various data files in.
7427 * journalctl gained the new --full switch (in addition to
7428 --all, to disable ellipsation for long messages.
7430 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
7431 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
7432 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
7433 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
7434 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
7435 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
7436 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
7437 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
7438 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
7442 * journalctl gained new --since= and --until= switches to
7443 filter by time. It also now supports nice filtering for
7444 units via --unit=/-u.
7446 * Type=oneshot services may use ExecReload= and do the
7449 * The journal daemon now supports time-based rotation and
7450 vacuuming, in addition to the usual disk-space based
7453 * The journal will now index the available field values for
7454 each field name. This enables clients to show pretty drop
7455 downs of available match values when filtering. The bash
7456 completion of journalctl has been updated
7457 accordingly. journalctl gained a new switch -F to list all
7458 values a certain field takes in the journal database.
7460 * More service events are now written as structured messages
7461 to the journal, and made recognizable via message IDs.
7463 * The timedated, localed and hostnamed mini-services which
7464 previously only provided support for changing time, locale
7465 and hostname settings from graphical DEs such as GNOME now
7466 also have a minimal (but very useful) text-based client
7467 utility each. This is probably the nicest way to changing
7468 these settings from the command line now, especially since
7469 it lists available options and is fully integrated with bash
7472 * There's now a new tool "systemd-coredumpctl" to list and
7473 extract coredumps from the journal.
7475 * We now install a README each in /var/log/ and
7476 /etc/rc.d/init.d explaining where the system logs and init
7477 scripts went. This hopefully should help folks who go to
7478 that dirs and look into the otherwise now empty void and
7479 scratch their heads.
7481 * When user-services are invoked (by systemd --user) the
7482 $MANAGERPID env var is set to the PID of systemd.
7484 * SIGRTMIN+24 when sent to a --user instance will now result
7485 in immediate termination of systemd.
7487 * gatewayd received numerous feature additions such as a
7488 "follow" mode, for live syncing and filtering.
7490 * browse.html now allows filtering and showing detailed
7491 information on specific entries. Keyboard navigation and
7492 mouse screen support has been added.
7494 * gatewayd/journalctl now supports HTML5/JSON
7495 Server-Sent-Events as output.
7497 * The SysV init script compatibility logic will now
7498 heuristically determine whether a script supports the
7499 "reload" verb, and only then make this available as
7502 * "systemctl status --follow" has been removed, use "journalctl
7505 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
7506 have been removed since they are hardly useful to be
7509 * And I'd like to take the opportunity to specifically mention
7510 Zbigniew for his great contributions. Zbigniew, you rock!
7512 Contributions from: Andrew Eikum, Christian Hesse, Colin
7513 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
7514 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
7515 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
7516 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
7517 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
7518 Jędrzejewski-Szmek, Сковорода Никита Андреевич
7522 * If /etc/vconsole.conf is non-existent or empty we will no
7523 longer load any console font or key map at boot by
7524 default. Instead the kernel defaults will be left
7525 intact. This is definitely the right thing to do, as no
7526 configuration should mean no configuration, and hard-coding
7527 font names that are different on all archs is probably a bad
7528 idea. Also, the kernel default key map and font should be
7529 good enough for most cases anyway, and mostly identical to
7530 the userspace fonts/key maps we previously overloaded them
7531 with. If distributions want to continue to default to a
7532 non-kernel font or key map they should ship a default
7533 /etc/vconsole.conf with the appropriate contents.
7535 Contributions from: Colin Walters, Daniel J Walsh, Dave
7536 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
7537 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
7541 * journalctl gained a new --cursor= switch to show entries
7542 starting from the specified location in the journal.
7544 * We now enforce a size limit on journal entry fields exported
7545 with "-o json" in journalctl. Fields larger than 4K will be
7546 assigned null. This can be turned off with --all.
7548 * An (optional) journal gateway daemon is now available as
7549 "systemd-journal-gatewayd.service". This service provides
7550 access to the journal via HTTP and JSON. This functionality
7551 will be used to implement live log synchronization in both
7552 pull and push modes, but has various other users too, such
7553 as easy log access for debugging of embedded devices. Right
7554 now it is already useful to retrieve the journal via HTTP:
7556 # systemctl start systemd-journal-gatewayd.service
7557 # wget http://localhost:19531/entries
7559 This will download the journal contents in a
7560 /var/log/messages compatible format. The same as JSON:
7562 # curl -H"Accept: application/json" http://localhost:19531/entries
7564 This service is also accessible via a web browser where a
7565 single static HTML5 app is served that uses the JSON logic
7566 to enable the user to do some basic browsing of the
7567 journal. This will be extended later on. Here's an example
7568 screenshot of this app in its current state:
7570 http://0pointer.de/public/journal-gatewayd
7572 Contributions from: Kay Sievers, Lennart Poettering, Robert
7573 Milasan, Tom Gundersen
7577 * The bash completion logic is now available for journalctl
7580 * We do not mount the "cpuset" controller anymore together with
7581 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
7582 started if no parameters are assigned to it. "cpuset" hence
7583 broke code that assumed it could create "cpu" groups and
7586 * journalctl -f will now subscribe to terminal size changes,
7587 and line break accordingly.
7589 Contributions from: Dave Reisner, Kay Sievers, Lennart
7590 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
7594 * nspawn will now create a symlink /etc/localtime in the
7595 container environment, copying the host's timezone
7596 setting. Previously this has been done via a bind mount, but
7597 since symlinks cannot be bind mounted this has now been
7598 changed to create/update the appropriate symlink.
7600 * journalctl -n's line number argument is now optional, and
7601 will default to 10 if omitted.
7603 * journald will now log the maximum size the journal files may
7604 take up on disk. This is particularly useful if the default
7605 built-in logic of determining this parameter from the file
7606 system size is used. Use "systemctl status
7607 systemd-journald.service" to see this information.
7609 * The multi-seat X wrapper tool has been stripped down. As X
7610 is now capable of enumerating graphics devices via udev in a
7611 seat-aware way the wrapper is not strictly necessary
7612 anymore. A stripped down temporary stop-gap is still shipped
7613 until the upstream display managers have been updated to
7614 fully support the new X logic. Expect this wrapper to be
7615 removed entirely in one of the next releases.
7617 * HandleSleepKey= in logind.conf has been split up into
7618 HandleSuspendKey= and HandleHibernateKey=. The old setting
7619 is not available anymore. X11 and the kernel are
7620 distinguishing between these keys and we should too. This
7621 also means the inhibition lock for these keys has been split
7624 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
7625 Poettering, Lukas Nykryn, Václav Pavlín
7629 * Whenever a unit changes state we will now log this to the
7630 journal and show along the unit's own log output in
7633 * ConditionPathIsMountPoint= can now properly detect bind
7634 mount points too. (Previously, a bind mount of one file
7635 system to another place in the same file system could not be
7636 detected as mount, since they shared struct stat's st_dev
7639 * We will now mount the cgroup controllers cpu, cpuacct,
7640 cpuset and the controllers net_cls, net_prio together by
7643 * nspawn containers will now have a virtualized boot
7644 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
7645 over with a randomized ID at container initialization). This
7646 has the effect of making "journalctl -b" do the right thing
7649 * The JSON output journal serialization has been updated not
7650 to generate "endless" list objects anymore, but rather one
7651 JSON object per line. This is more in line how most JSON
7652 parsers expect JSON objects. The new output mode
7653 "json-pretty" has been added to provide similar output, but
7654 neatly aligned for readability by humans.
7656 * We dropped all explicit sync() invocations in the shutdown
7657 code. The kernel does this implicitly anyway in the kernel
7658 reboot() syscall. halt(8)'s -n option is now a compatibility
7661 * We now support virtualized reboot() in containers, as
7662 supported by newer kernels. We will fall back to exit() if
7663 CAP_SYS_REBOOT is not available to the container. Also,
7664 nspawn makes use of this now and will actually reboot the
7665 container if the containerized OS asks for that.
7667 * journalctl will only show local log output by default
7668 now. Use --merge (-m) to show remote log output, too.
7670 * libsystemd-journal gained the new sd_journal_get_usage()
7671 call to determine the current disk usage of all journal
7672 files. This is exposed in the new "journalctl --disk-usage"
7675 * journald gained a new configuration setting SplitMode= in
7676 journald.conf which may be used to control how user journals
7677 are split off. See journald.conf(5) for details.
7679 * A new condition type ConditionFileNotEmpty= has been added.
7681 * tmpfiles' "w" lines now support file globbing, to write
7682 multiple files at once.
7684 * We added Python bindings for the journal submission
7685 APIs. More Python APIs for a number of selected APIs will
7686 likely follow. Note that we intend to add native bindings
7687 only for the Python language, as we consider it common
7688 enough to deserve bindings shipped within systemd. There are
7689 various projects outside of systemd that provide bindings
7690 for languages such as PHP or Lua.
7692 * Many conditions will now resolve specifiers such as %i. In
7693 addition, PathChanged= and related directives of .path units
7694 now support specifiers as well.
7696 * There's now a new RPM macro definition for the system preset
7699 * journald will now warn if it ca not forward a message to the
7700 syslog daemon because its socket is full.
7702 * timedated will no longer write or process /etc/timezone,
7703 except on Debian. As we do not support late mounted /usr
7704 anymore /etc/localtime always being a symlink is now safe,
7705 and hence the information in /etc/timezone is not necessary
7708 * logind will now always reserve one VT for a text getty (VT6
7709 by default). Previously if more than 6 X sessions where
7710 started they took up all the VTs with auto-spawned gettys,
7711 so that no text gettys were available anymore.
7713 * udev will now automatically inform the btrfs kernel logic
7714 about btrfs RAID components showing up. This should make
7715 simple hotplug based btrfs RAID assembly work.
7717 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
7718 (but not for its children which will stay at the kernel
7719 default). This should allow setups with a lot more listening
7722 * systemd will now always pass the configured timezone to the
7723 kernel at boot. timedated will do the same when the timezone
7726 * logind's inhibition logic has been updated. By default,
7727 logind will now handle the lid switch, the power and sleep
7728 keys all the time, even in graphical sessions. If DEs want
7729 to handle these events on their own they should take the new
7730 handle-power-key, handle-sleep-key and handle-lid-switch
7731 inhibitors during their runtime. A simple way to achieve
7732 that is to invoke the DE wrapped in an invocation of:
7734 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
7736 * Access to unit operations is now checked via SELinux taking
7737 the unit file label and client process label into account.
7739 * systemd will now notify the administrator in the journal
7740 when he over-mounts a non-empty directory.
7742 * There are new specifiers that are resolved in unit files,
7743 for the host name (%H), the machine ID (%m) and the boot ID
7746 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
7747 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
7748 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
7749 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
7750 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
7751 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
7752 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
7756 * Support for reading structured kernel messages from
7757 /dev/kmsg has now been added and is enabled by default.
7759 * Support for reading kernel messages from /proc/kmsg has now
7760 been removed. If you want kernel messages in the journal
7761 make sure to run a recent kernel (>= 3.5) that supports
7762 reading structured messages from /dev/kmsg (see
7763 above). /proc/kmsg is now exclusive property of classic
7764 syslog daemons again.
7766 * The libudev API gained the new
7767 udev_device_new_from_device_id() call.
7769 * The logic for file system namespace (ReadOnlyDirectory=,
7770 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
7771 require pivot_root() anymore. This means fewer temporary
7772 directories are created below /tmp for this feature.
7774 * nspawn containers will now see and receive all submounts
7775 made on the host OS below the root file system of the
7778 * Forward Secure Sealing is now supported for Journal files,
7779 which provide cryptographical sealing of journal files so
7780 that attackers cannot alter log history anymore without this
7781 being detectable. Lennart will soon post a blog story about
7782 this explaining it in more detail.
7784 * There are two new service settings RestartPreventExitStatus=
7785 and SuccessExitStatus= which allow configuration of exit
7786 status (exit code or signal) which will be excepted from the
7787 restart logic, resp. consider successful.
7789 * journalctl gained the new --verify switch that can be used
7790 to check the integrity of the structure of journal files and
7791 (if Forward Secure Sealing is enabled) the contents of
7794 * nspawn containers will now be run with /dev/stdin, /dev/fd/
7795 and similar symlinks pre-created. This makes running shells
7796 as container init process a lot more fun.
7798 * The fstab support can now handle PARTUUID= and PARTLABEL=
7801 * A new ConditionHost= condition has been added to match
7802 against the hostname (with globs) and machine ID. This is
7803 useful for clusters where a single OS image is used to
7804 provision a large number of hosts which shall run slightly
7805 different sets of services.
7807 * Services which hit the restart limit will now be placed in a
7810 Contributions from: Bertram Poettering, Dave Reisner, Huang
7811 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
7812 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
7816 * When running in --user mode systemd will now become a
7817 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
7818 tree a lot more organized.
7820 * A new PartOf= unit dependency type has been introduced that
7821 may be used to group services in a natural way.
7823 * "systemctl enable" may now be used to enable instances of
7826 * journalctl now prints error log levels in red, and
7827 warning/notice log levels in bright white. It also supports
7828 filtering by log level now.
7830 * cgtop gained a new -n switch (similar to top), to configure
7831 the maximum number of iterations to run for. It also gained
7832 -b, to run in batch mode (accepting no input).
7834 * The suffix ".service" may now be omitted on most systemctl
7835 command lines involving service unit names.
7837 * There's a new bus call in logind to lock all sessions, as
7838 well as a loginctl verb for it "lock-sessions".
7840 * libsystemd-logind.so gained a new call sd_journal_perror()
7841 that works similar to libc perror() but logs to the journal
7842 and encodes structured information about the error number.
7844 * /etc/crypttab entries now understand the new keyfile-size=
7847 * shutdown(8) now can send a (configurable) wall message when
7848 a shutdown is cancelled.
7850 * The mount propagation mode for the root file system will now
7851 default to "shared", which is useful to make containers work
7852 nicely out-of-the-box so that they receive new mounts from
7853 the host. This can be undone locally by running "mount
7854 --make-rprivate /" if needed.
7856 * The prefdm.service file has been removed. Distributions
7857 should maintain this unit downstream if they intend to keep
7858 it around. However, we recommend writing normal unit files
7859 for display managers instead.
7861 * Since systemd is a crucial part of the OS we will now
7862 default to a number of compiler switches that improve
7863 security (hardening) such as read-only relocations, stack
7864 protection, and suchlike.
7866 * The TimeoutSec= setting for services is now split into
7867 TimeoutStartSec= and TimeoutStopSec= to allow configuration
7868 of individual time outs for the start and the stop phase of
7871 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
7872 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
7873 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
7874 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
7875 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
7876 Gundersen, Zbigniew Jędrzejewski-Szmek
7880 * The journal and id128 C APIs are now fully documented as man
7883 * Extra safety checks have been added when transitioning from
7884 the initial RAM disk to the main system to avoid accidental
7887 * /etc/crypttab entries now understand the new keyfile-offset=
7890 * systemctl -t can now be used to filter by unit load state.
7892 * The journal C API gained the new sd_journal_wait() call to
7893 make writing synchronous journal clients easier.
7895 * journalctl gained the new -D switch to show journals from a
7898 * journalctl now displays a special marker between log
7899 messages of two different boots.
7901 * The journal is now explicitly flushed to /var via a service
7902 systemd-journal-flush.service, rather than implicitly simply
7903 by seeing /var/log/journal to be writable.
7905 * journalctl (and the journal C APIs) can now match for much
7906 more complex expressions, with alternatives and
7909 * When transitioning from the initial RAM disk to the main
7910 system we will now kill all processes in a killing spree to
7911 ensure no processes stay around by accident.
7913 * Three new specifiers may be used in unit files: %u, %h, %s
7914 resolve to the user name, user home directory resp. user
7915 shell. This is useful for running systemd user instances.
7917 * We now automatically rotate journal files if their data
7918 object hash table gets a fill level > 75%. We also size the
7919 hash table based on the configured maximum file size. This
7920 together should lower hash collisions drastically and thus
7921 speed things up a bit.
7923 * journalctl gained the new "--header" switch to introspect
7924 header data of journal files.
7926 * A new setting SystemCallFilters= has been added to services
7927 which may be used to apply blacklists or whitelists to
7928 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
7930 * nspawn gained a new --link-journal= switch (and quicker: -j)
7931 to link the container journal with the host. This makes it
7932 very easy to centralize log viewing on the host for all
7933 guests while still keeping the journal files separated.
7935 * Many bugfixes and optimizations
7937 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
7938 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
7939 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
7944 * Several tools now understand kernel command line arguments,
7945 which are only read when run in an initial RAM disk. They
7946 usually follow closely their normal counterparts, but are
7949 * There's a new tool to analyze the readahead files that are
7950 automatically generated at boot. Use:
7952 /usr/lib/systemd/systemd-readahead analyze /.readahead
7954 * We now provide an early debug shell on tty9 if this enabled. Use:
7956 systemctl enable debug-shell.service
7958 * All plymouth related units have been moved into the Plymouth
7959 package. Please make sure to upgrade your Plymouth version
7962 * systemd-tmpfiles now supports getting passed the basename of
7963 a configuration file only, in which case it will look for it
7964 in all appropriate directories automatically.
7966 * udevadm info now takes a /dev or /sys path as argument, and
7967 does the right thing. Example:
7969 udevadm info /dev/sda
7970 udevadm info /sys/class/block/sda
7972 * systemctl now prints a warning if a unit is stopped but a
7973 unit that might trigger it continues to run. Example: a
7974 service is stopped but the socket that activates it is left
7977 * "systemctl status" will now mention if the log output was
7978 shortened due to rotation since a service has been started.
7980 * The journal API now exposes functions to determine the
7981 "cutoff" times due to rotation.
7983 * journald now understands SIGUSR1 and SIGUSR2 for triggering
7984 immediately flushing of runtime logs to /var if possible,
7985 resp. for triggering immediate rotation of the journal
7988 * It is now considered an error if a service is attempted to
7989 be stopped that is not loaded.
7991 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
7993 * systemd-analyze now supports Python 3
7995 * tmpfiles now supports cleaning up directories via aging
7996 where the first level dirs are always kept around but
7997 directories beneath it automatically aged. This is enabled
7998 by prefixing the age field with '~'.
8000 * Seat objects now expose CanGraphical, CanTTY properties
8001 which is required to deal with very fast bootups where the
8002 display manager might be running before the graphics drivers
8003 completed initialization.
8005 * Seat objects now expose a State property.
8007 * We now include RPM macros for service enabling/disabling
8008 based on the preset logic. We recommend RPM based
8009 distributions to make use of these macros if possible. This
8010 makes it simpler to reuse RPM spec files across
8013 * We now make sure that the collected systemd unit name is
8014 always valid when services log to the journal via
8017 * There's a new man page kernel-command-line(7) detailing all
8018 command line options we understand.
8020 * The fstab generator may now be disabled at boot by passing
8021 fstab=0 on the kernel command line.
8023 * A new kernel command line option modules-load= is now understood
8024 to load a specific kernel module statically, early at boot.
8026 * Unit names specified on the systemctl command line are now
8027 automatically escaped as needed. Also, if file system or
8028 device paths are specified they are automatically turned
8029 into the appropriate mount or device unit names. Example:
8031 systemctl status /home
8032 systemctl status /dev/sda
8034 * The SysVConsole= configuration option has been removed from
8035 system.conf parsing.
8037 * The SysV search path is no longer exported on the D-Bus
8040 * The Names= option has been removed from unit file parsing.
8042 * There's a new man page bootup(7) detailing the boot process.
8044 * Every unit and every generator we ship with systemd now
8045 comes with full documentation. The self-explanatory boot is
8048 * A couple of services gained "systemd-" prefixes in their
8049 name if they wrap systemd code, rather than only external
8050 code. Among them fsck@.service which is now
8051 systemd-fsck@.service.
8053 * The HaveWatchdog property has been removed from the D-Bus
8056 * systemd.confirm_spawn= on the kernel command line should now
8059 * There's a new man page crypttab(5) which details all options
8060 we actually understand.
8062 * systemd-nspawn gained a new --capability= switch to pass
8063 additional capabilities to the container.
8065 * timedated will now read known NTP implementation unit names
8066 from /usr/lib/systemd/ntp-units.d/*.list,
8067 systemd-timedated-ntp.target has been removed.
8069 * journalctl gained a new switch "-b" that lists log data of
8070 the current boot only.
8072 * The notify socket is in the abstract namespace again, in
8073 order to support daemons which chroot() at start-up.
8075 * There is a new Storage= configuration option for journald
8076 which allows configuration of where log data should go. This
8077 also provides a way to disable journal logging entirely, so
8078 that data collected is only forwarded to the console, the
8079 kernel log buffer or another syslog implementation.
8081 * Many bugfixes and optimizations
8083 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
8084 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
8085 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
8086 Shawn Landden, Tom Gundersen
8090 * "systemctl help <unit>" now shows the man page if one is
8093 * Several new man pages have been added.
8095 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
8096 MaxLevelConsole= can now be specified in
8097 journald.conf. These options allow reducing the amount of
8098 data stored on disk or forwarded by the log level.
8100 * TimerSlackNSec= can now be specified in system.conf for
8101 PID1. This allows system-wide power savings.
8103 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
8104 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
8109 * logind is now capable of (optionally) handling power and
8110 sleep keys as well as the lid switch.
8112 * journalctl now understands the syntax "journalctl
8113 /usr/bin/avahi-daemon" to get all log output of a specific
8116 * CapabilityBoundingSet= in system.conf now also influences
8117 the capability bound set of usermode helpers of the kernel.
8119 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
8120 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
8121 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
8122 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
8126 * Note that we skipped 139 releases here in order to set the
8127 new version to something that is greater than both udev's
8128 and systemd's most recent version number.
8130 * udev: all udev sources are merged into the systemd source tree now.
8131 All future udev development will happen in the systemd tree. It
8132 is still fully supported to use the udev daemon and tools without
8133 systemd running, like in initramfs or other init systems. Building
8134 udev though, will require the *build* of the systemd tree, but
8135 udev can be properly *run* without systemd.
8137 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
8138 should be used to create dead device nodes as workarounds for broken
8141 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
8142 no longer supported. udev_monitor_new_from_netlink() needs to be
8143 used to subscribe to events.
8145 * udev: when udevd is started by systemd, processes which are left
8146 behind by forking them off of udev rules, are unconditionally cleaned
8147 up and killed now after the event handling has finished. Services or
8148 daemons must be started as systemd services. Services can be
8149 pulled-in by udev to get started, but they can no longer be directly
8150 forked by udev rules.
8152 * udev: the daemon binary is called systemd-udevd now and installed
8153 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
8154 to adapt to that, create symlink, or rename the binary after building
8157 * libudev no longer provides these symbols:
8158 udev_monitor_from_socket()
8159 udev_queue_get_failed_list_entry()
8160 udev_get_{dev,sys,run}_path()
8161 The versions number was bumped and symbol versioning introduced.
8163 * systemd-loginctl and systemd-journalctl have been renamed
8164 to loginctl and journalctl to match systemctl.
8166 * The config files: /etc/systemd/systemd-logind.conf and
8167 /etc/systemd/systemd-journald.conf have been renamed to
8168 logind.conf and journald.conf. Package updates should rename
8169 the files to the new names on upgrade.
8171 * For almost all files the license is now LGPL2.1+, changed
8172 from the previous GPL2.0+. Exceptions are some minor stuff
8173 of udev (which will be changed to LGPL2.1 eventually, too),
8174 and the MIT licensed sd-daemon.[ch] library that is suitable
8175 to be used as drop-in files.
8177 * systemd and logind now handle system sleep states, in
8178 particular suspending and hibernating.
8180 * logind now implements a sleep/shutdown/idle inhibiting logic
8181 suitable for a variety of uses. Soonishly Lennart will blog
8182 about this in more detail.
8184 * var-run.mount and var-lock.mount are no longer provided
8185 (which previously bind mounted these directories to their new
8186 places). Distributions which have not converted these
8187 directories to symlinks should consider stealing these files
8188 from git history and add them downstream.
8190 * We introduced the Documentation= field for units and added
8191 this to all our shipped units. This is useful to make it
8192 easier to explore the boot and the purpose of the various
8195 * All smaller setup units (such as
8196 systemd-vconsole-setup.service) now detect properly if they
8197 are run in a container and are skipped when
8198 appropriate. This guarantees an entirely noise-free boot in
8199 Linux container environments such as systemd-nspawn.
8201 * A framework for implementing offline system updates is now
8202 integrated, for details see:
8203 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
8205 * A new service type Type=idle is available now which helps us
8206 avoiding ugly interleaving of getty output and boot status
8209 * There's now a system-wide CapabilityBoundingSet= option to
8210 globally reduce the set of capabilities for the
8211 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
8212 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
8213 even CAP_NET_ADMIN system-wide for secure systems.
8215 * There are now system-wide DefaultLimitXXX= options to
8216 globally change the defaults of the various resource limits
8217 for all units started by PID 1.
8219 * Harald Hoyer's systemd test suite has been integrated into
8220 systemd which allows easy testing of systemd builds in qemu
8221 and nspawn. (This is really awesome! Ask us for details!)
8223 * The fstab parser is now implemented as generator, not inside
8226 * systemctl will now warn you if .mount units generated from
8227 /etc/fstab are out of date due to changes in fstab that
8228 have not been read by systemd yet.
8230 * systemd is now suitable for usage in initrds. Dracut has
8231 already been updated to make use of this. With this in place
8232 initrds get a slight bit faster but primarily are much
8233 easier to introspect and debug since "systemctl status" in
8234 the host system can be used to introspect initrd services,
8235 and the journal from the initrd is kept around too.
8237 * systemd-delta has been added, a tool to explore differences
8238 between user/admin configuration and vendor defaults.
8240 * PrivateTmp= now affects both /tmp and /var/tmp.
8242 * Boot time status messages are now much prettier and feature
8243 proper english language. Booting up systemd has never been
8246 * Read-ahead pack files now include the inode number of all
8247 files to pre-cache. When the inode changes the pre-caching
8248 is not attempted. This should be nicer to deal with updated
8249 packages which might result in changes of read-ahead
8252 * We now temporaritly lower the kernel's read_ahead_kb variable
8253 when collecting read-ahead data to ensure the kernel's
8254 built-in read-ahead does not add noise to our measurements
8255 of necessary blocks to pre-cache.
8257 * There's now RequiresMountsFor= to add automatic dependencies
8258 for all mounts necessary for a specific file system path.
8260 * MountAuto= and SwapAuto= have been removed from
8261 system.conf. Mounting file systems at boot has to take place
8264 * nspawn now learned a new switch --uuid= to set the machine
8265 ID on the command line.
8267 * nspawn now learned the -b switch to automatically search
8270 * vt102 is now the default TERM for serial TTYs, upgraded from
8273 * systemd-logind now works on VT-less systems.
8275 * The build tree has been reorganized. The individual
8276 components now have directories of their own.
8278 * A new condition type ConditionPathIsReadWrite= is now available.
8280 * nspawn learned the new -C switch to create cgroups for the
8281 container in other hierarchies.
8283 * We now have support for hardware watchdogs, configurable in
8286 * The scheduled shutdown logic now has a public API.
8288 * We now mount /tmp as tmpfs by default, but this can be
8289 masked and /etc/fstab can override it.
8291 * Since udisks does not make use of /media anymore we are not
8292 mounting a tmpfs on it anymore.
8294 * journalctl gained a new --local switch to only interleave
8295 locally generated journal files.
8297 * We can now load the IMA policy at boot automatically.
8299 * The GTK tools have been split off into a systemd-ui.
8301 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
8302 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
8303 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
8304 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
8305 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
8306 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
8307 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
8308 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
8309 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
8314 * This is mostly a bugfix release
8316 * Support optional initialization of the machine ID from the
8317 KVM or container configured UUID.
8319 * Support immediate reboots with "systemctl reboot -ff"
8321 * Show /etc/os-release data in systemd-analyze output
8323 * Many bugfixes for the journal, including endianness fixes and
8324 ensuring that disk space enforcement works
8326 * sd-login.h is C++ compatible again
8328 * Extend the /etc/os-release format on request of the Debian
8331 * We now refuse non-UTF8 strings used in various configuration
8332 and unit files. This is done to ensure we do not pass invalid
8333 data over D-Bus or expose it elsewhere.
8335 * Register Mimo USB Screens as suitable for automatic seat
8338 * Read SELinux client context from journal clients in a race
8341 * Reorder configuration file lookup order. /etc now always
8342 overrides /run in order to allow the administrator to always
8343 and unconditionally override vendor-supplied or
8344 automatically generated data.
8346 * The various user visible bits of the journal now have man
8347 pages. We still lack man pages for the journal API calls
8350 * We now ship all man pages in HTML format again in the
8353 Contributions from: Dave Reisner, Dirk Eibach, Frederic
8354 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
8355 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
8360 * This is mostly a bugfix release
8362 * systems lacking /etc/os-release are no longer supported.
8364 * Various functionality updates to libsystemd-login.so
8366 * Track class of PAM logins to distinguish greeters from
8369 Contributions from: Kay Sievers, Lennart Poettering, Michael
8374 * This is an important bugfix release for v41.
8376 * Building man pages is now optional which should be useful
8377 for those building systemd from git but unwilling to install
8380 * Watchdog support for supervising services is now usable. In
8381 a future release support for hardware watchdogs
8382 (i.e. /dev/watchdog) will be added building on this.
8384 * Service start rate limiting is now configurable and can be
8385 turned off per service. When a start rate limit is hit a
8386 reboot can automatically be triggered.
8388 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
8390 Contributions from: Benjamin Franzke, Bill Nottingham,
8391 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
8392 Schmidt, Michał Górny, Piotr Drąg
8396 * The systemd binary is installed /usr/lib/systemd/systemd now;
8397 An existing /sbin/init symlink needs to be adapted with the
8400 * The code that loads kernel modules has been ported to invoke
8401 libkmod directly, instead of modprobe. This means we do not
8402 support systems with module-init-tools anymore.
8404 * Watchdog support is now already useful, but still not
8407 * A new kernel command line option systemd.setenv= is
8408 understood to set system wide environment variables
8409 dynamically at boot.
8411 * We now limit the set of capabilities of systemd-journald.
8413 * We now set SIGPIPE to ignore by default, since it only is
8414 useful in shell pipelines, and has little use in general
8415 code. This can be disabled with IgnoreSIPIPE=no in unit
8418 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
8419 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
8424 * This is mostly a bugfix release
8426 * We now expose the reason why a service failed in the
8427 "Result" D-Bus property.
8429 * Rudimentary service watchdog support (will be completed over
8430 the next few releases.)
8432 * When systemd forks off in order execute some service we will
8433 now immediately changes its argv[0] to reflect which process
8434 it will execute. This is useful to minimize the time window
8435 with a generic argv[0], which makes bootcharts more useful
8437 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
8438 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
8439 Mike Kazantsev, Ray Strode
8443 * This is mostly a test release, but incorporates many
8446 * New systemd-cgtop tool to show control groups by their
8449 * Linking against libacl for ACLs is optional again. If
8450 disabled, support tracking device access for active logins
8451 goes becomes unavailable, and so does access to the user
8452 journals by the respective users.
8454 * If a group "adm" exists, journal files are automatically
8455 owned by them, thus allow members of this group full access
8456 to the system journal as well as all user journals.
8458 * The journal now stores the SELinux context of the logging
8459 client for all entries.
8461 * Add C++ inclusion guards to all public headers
8463 * New output mode "cat" in the journal to print only text
8464 messages, without any meta data like date or time.
8466 * Include tiny X server wrapper as a temporary stop-gap to
8467 teach XOrg udev display enumeration. This is used by display
8468 managers such as gdm, and will go away as soon as XOrg
8469 learned native udev hotplugging for display devices.
8471 * Add new systemd-cat tool for executing arbitrary programs
8472 with STDERR/STDOUT connected to the journal. Can also act as
8473 BSD logger replacement, and does so by default.
8475 * Optionally store all locally generated coredumps in the
8476 journal along with meta data.
8478 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
8479 writing short strings to files (for usage for /sys), and for
8480 creating symlinks, character and block device nodes.
8482 * New unit file option ControlGroupPersistent= to make cgroups
8483 persistent, following the mechanisms outlined in
8484 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
8486 * Support multiple local RTCs in a sane way
8488 * No longer monopolize IO when replaying readahead data on
8489 rotating disks, since we might starve non-file-system IO to
8490 death, since fanotify() will not see accesses done by blkid,
8493 * Do not show kernel threads in systemd-cgls anymore, unless
8494 requested with new -k switch.
8496 Contributions from: Dan Horák, Kay Sievers, Lennart
8497 Poettering, Michal Schmidt
8501 * This is mostly a test release, but incorporates many
8504 * The git repository moved to:
8505 git://anongit.freedesktop.org/systemd/systemd
8506 ssh://git.freedesktop.org/git/systemd/systemd
8508 * First release with the journal
8509 http://0pointer.de/blog/projects/the-journal.html
8511 * The journal replaces both systemd-kmsg-syslogd and
8512 systemd-stdout-bridge.
8514 * New sd_pid_get_unit() API call in libsystemd-logind
8516 * Many systemadm clean-ups
8518 * Introduce remote-fs-pre.target which is ordered before all
8519 remote mounts and may be used to start services before all
8522 * Added Mageia support
8524 * Add bash completion for systemd-loginctl
8526 * Actively monitor PID file creation for daemons which exit in
8527 the parent process before having finished writing the PID
8528 file in the daemon process. Daemons which do this need to be
8529 fixed (i.e. PID file creation must have finished before the
8530 parent exits), but we now react a bit more gracefully to them.
8532 * Add colourful boot output, mimicking the well-known output
8533 of existing distributions.
8535 * New option PassCredentials= for socket units, for
8536 compatibility with a recent kernel ABI breakage.
8538 * /etc/rc.local is now hooked in via a generator binary, and
8539 thus will no longer act as synchronization point during
8542 * systemctl list-unit-files now supports --root=.
8544 * systemd-tmpfiles now understands two new commands: z, Z for
8545 relabelling files according to the SELinux database. This is
8546 useful to apply SELinux labels to specific files in /sys,
8549 * Output of SysV services is now forwarded to both the console
8550 and the journal by default, not only just the console.
8552 * New man pages for all APIs from libsystemd-login.
8554 * The build tree got reorganized and the build system is a
8555 lot more modular allowing embedded setups to specifically
8556 select the components of systemd they are interested in.
8558 * Support for Linux systems lacking the kernel VT subsystem is
8561 * configure's --with-rootdir= got renamed to
8562 --with-rootprefix= to follow the naming used by udev and
8565 * Unless specified otherwise we will now install to /usr instead
8566 of /usr/local by default.
8568 * Processes with '@' in argv[0][0] are now excluded from the
8569 final shut-down killing spree, following the logic explained
8571 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
8573 * All processes remaining in a service cgroup when we enter
8574 the START or START_PRE states are now killed with
8575 SIGKILL. That means it is no longer possible to spawn
8576 background processes from ExecStart= lines (which was never
8577 supported anyway, and bad style).
8579 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
8580 reloading of units together.
8582 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
8583 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
8584 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
8585 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
8586 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek