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 two 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.
153 * The ExecStart= lines of unit files are no longer required to
154 reference absolute paths. If non-absolute paths are specified the
155 specified binary name is searched within the service manager's
156 built-in $PATH, which may be queried with 'elogind-path
157 search-binaries-default'. It's generally recommended to continue to
158 use absolute paths for all binaries specified in unit files.
160 * Units gained a new load state "bad-setting", which is used when a
161 unit file was loaded, but contained fatal errors which prevent it
162 from being started (for example, a service unit has been defined
163 lacking both ExecStart= and ExecStop= lines).
165 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
166 support alternative debuggers, for example lldb. The old name
167 continues to be available however, for compatibility reasons. Use the
168 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
169 to pick an alternative debugger instead of the default gdb.
171 * systemctl and the other tools will now output escape sequences that
172 generate proper clickable hyperlinks in various terminal emulators
173 where useful (for example, in the "systemctl status" output you can
174 now click on the unit file name to quickly open it in the
175 editor/viewer of your choice). Note that not all terminal emulators
176 support this functionality yet, but many do. Unfortunately, the
177 "less" pager doesn't support this yet, hence this functionality is
178 currently automatically turned off when a pager is started (which
179 happens quite often due to auto-paging). We hope to remove this
180 limitation as soon as "less" learns these escape sequences. This new
181 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
182 environment variable. For details on these escape sequences see:
183 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
185 * networkd's .network files now support a new IPv6MTUBytes= option for
186 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
187 option in the [Route] section to configure the MTU to use for
188 specific routes. It also gained support for configuration of the DHCP
189 "UserClass" option through the new UserClass= setting. It gained
190 three new options in the new [CAN] section for configuring CAN
191 networks. The MULTICAST and ALLMULTI interface flags may now be
192 controlled explicitly with the new Multicast= and AllMulticast=
195 * networkd will now automatically make use of the kernel's route
196 expiration feature, if it is available.
198 * udevd's .link files now support setting the number of receive and
199 transmit channels, using the RxChannels=, TxChannels=,
200 OtherChannels=, CombinedChannels= settings.
202 * Support for UDPSegmentationOffload= has been removed, given its
203 limited support in hardware, and waning software support.
205 * networkd's .netdev files now support creating "netdevsim" interfaces.
207 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
208 to query the unit belonging to a specific kernel control group.
210 * elogind-analyze gained a new verb "cat-config", which may be used to
211 dump the contents of any configuration file, with all its matching
212 drop-in files added in, and honouring the usual search and masking
213 logic applied to elogind configuration files. For example use
214 "elogind-analyze cat-config elogind/system.conf" to get the complete
215 system configuration file of elogind how it would be loaded by PID 1
216 itself. Similar to this, various tools such as elogind-tmpfiles or
217 elogind-sysusers, gained a new option "--cat-config", which does the
218 corresponding operation for their own configuration settings. For
219 example, "elogind-tmpfiles --cat-config" will now output the full
220 list of tmpfiles.d/ lines in place.
222 * timedatectl gained two new verbs "timesync-status" (to show the
223 current NTP synchronization state of elogind-timesyncd) and
224 "show-timesync" (to show bus properties of elogind-timesyncd).
226 * elogind-timesyncd gained a bus interface on which it exposes details
229 * elogind-nspawn gained a new --rlimit= switch for setting initial
230 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
231 understood by elogind-timedated. It takes a colon-separated list of
232 unit names of NTP client services. The list is used by
233 "timedatectl set-ntp".
235 resource limits for the container payload. There's a new switch
236 --hostname= to explicitly override the container's hostname. A new
237 --no-new-privileges= switch may be used to control the
238 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
239 --oom-score-adjust= switch controls the OOM scoring adjustment value
240 for the payload. The new --cpu-affinity= switch controls the CPU
241 affinity of the container payload. The new --resolv-conf= switch
242 allows more detailed control of /etc/resolv.conf handling of the
243 container. Similarly, the new --timezone= switch allows more detailed
244 control of /etc/localtime handling of the container.
246 * elogind-detect-virt gained a new --list switch, which will print a
247 list of all currently known VM and container environments.
249 * Support for "Portable Services" has been added, see
250 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
251 experimental, but this is expected to change soon. Reflecting this
252 experimental state, the "portablectl" binary is not installed into
253 /usr/bin yet. The binary has to be called with the full path
254 /usr/lib/elogind/portablectl instead.
256 * journalctl's and systemctl's -o switch now knows a new log output
257 mode "with-unit". The output it generates is very similar to the
258 regular "short" mode, but displays the unit name instead of the
259 syslog tag for each log line. Also, the date is shown with timezone
260 information. This mode is probably more useful than the classic
261 "short" output mode for most purposes, except where pixel-perfect
262 compatibility with classic /var/log/messages formatting is required.
264 * A new --dump-bus-properties switch has been added to the elogind
265 binary, which may be used to dump all supported D-Bus properties.
266 (Options which are still supported, but are deprecated, are *not*
269 * sd-bus gained a set of new calls:
270 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
271 enable/disable the "floating" state of a bus slot object,
272 i.e. whether the slot object pins the bus it is allocated for into
273 memory or if the bus slot object gets disconnected when the bus goes
274 away. sd_bus_open_with_description(),
275 sd_bus_open_user_with_description(),
276 sd_bus_open_system_with_description() may be used to allocate bus
277 objects and set their description string already during allocation.
279 * sd-event gained support for watching inotify events from the event
280 loop, in an efficient way, sharing inotify handles between multiple
281 users. For this a new function sd_event_add_inotify() has been added.
283 * sd-event and sd-bus gained support for calling special user-supplied
284 destructor functions for userdata pointers associated with
285 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
286 functions sd_bus_slot_set_destroy_callback,
287 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
288 sd_bus_track_get_destroy_callback,
289 sd_event_source_set_destroy_callback,
290 sd_event_source_get_destroy_callback have been added.
292 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
294 * PID 1 will now automatically reschedule .timer units whenever the
295 local timezone changes. (They previously got rescheduled
296 automatically when the system clock changed.)
298 * New documentation has been added to document cgroups delegation,
299 portable services and the various code quality tools we have set up:
301 https://github.com/systemd/systemd/blob/master/doc/CGROUP_DELEGATION.md
302 https://github.com/systemd/systemd/blob/master/doc/PORTABLE_SERVICES.md
303 https://github.com/systemd/systemd/blob/master/doc/CODE_QUALITY.md
305 * pam_elogind will now look for PAM userdata fields elogind.memory_max,
306 elogind.tasks_max, elogind.cpu_weight, elogind.io_weight set by
307 earlier PAM modules. The data in these fields is used to initialize
308 the session scope's resource properties. Thus external PAM modules
309 may now configure per-session limits, for example sourced from
310 external user databases.
312 * socket units with Accept=yes will now maintain a "refused" counter in
313 addition to the existing "accepted" counter, counting connections
314 refused due to the enforced limits.
316 * The "elogind-path search-binaries-default" command may now be use to
317 query the default, built-in $PATH PID 1 will pass to the services it
320 * A new unit file setting PrivateMounts= has been added. It's a boolean
321 option. If enabled the unit's processes are invoked in their own file
322 system namespace. Note that this behaviour is also implied if any
323 other file system namespacing options (such as PrivateTmp=,
324 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
325 primarily useful for services that do not use any of the other file
326 system namespacing options. One such service is elogind-udevd.service
327 wher this is now used by default.
329 * A new unit "system-update-pre.target" is added, which defines an
330 optional synchronization point for offline system updates, as
331 implemented by the pre-existing "system-update.target" unit. It
332 allows ordering services before the service that executes the actual
333 update process in a generic way.
335 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
336 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
337 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
338 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
339 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
340 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
341 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
342 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
343 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
344 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
345 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
346 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
347 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
348 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
349 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
350 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
351 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
352 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
353 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
354 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
355 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
356 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
357 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
358 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
359 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
360 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
361 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
362 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
363 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
369 * The MemoryAccounting= unit property now defaults to on. After
370 discussions with the upstream control group maintainers we learnt
371 that the negative impact of cgroup memory accounting on current
372 kernels is finally relatively minimal, so that it should be safe to
373 enable this by default without affecting system performance. Besides
374 memory accounting only task accounting is turned on by default, all
375 other forms of resource accounting (CPU, IO, IP) remain off for now,
376 because it's not clear yet that their impact is small enough to move
377 from opt-in to opt-out. We recommend downstreams to leave memory
378 accounting on by default if kernel 4.14 or higher is primarily
379 used. On very resource constrained systems or when support for old
380 kernels is a necessity, -Dmemory-accounting-default=false can be used
381 to revert this change.
383 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
384 %udev_rules_update) and the journal catalog (%journal_catalog_update)
385 from the upgrade scriptlets of individual packages now do nothing.
386 Transfiletriggers have been added which will perform those updates
387 once at the end of the transaction.
389 Similar transfiletriggers have been added to execute any sysctl.d
390 and binfmt.d rules. Thus, it should be unnecessary to provide any
391 scriptlets to execute this configuration from package installation
394 * elogind-sysusers gained a mode where the configuration to execute is
395 specified on the command line, but this configuration is not executed
396 directly, but instead it is merged with the configuration on disk,
397 and the result is executed. This is useful for package installation
398 scripts which want to create the user before installing any files on
399 disk (in case some of those files are owned by that user), while
400 still allowing local admin overrides.
402 This functionality is exposed to rpm scriptlets through a new
403 %sysusers_create_package macro. Old %sysusers_create and
404 %sysusers_create_inline macros are deprecated.
406 A transfiletrigger for sysusers.d configuration is now installed,
407 which means that it should be unnecessary to call elogind-sysusers from
408 package installation scripts, unless the package installs any files
409 owned by those newly-created users, in which case
410 %sysusers_create_package should be used.
412 * Analogous change has been done for elogind-tmpfiles: it gained a mode
413 where the command-line configuration is merged with the configuration
414 on disk. This is exposed as the new %tmpfiles_create_package macro,
415 and %tmpfiles_create is deprecated. A transfiletrigger is installed
416 for tmpfiles.d, hence it should be unnecessary to call elogind-tmpfiles
417 from package installation scripts.
419 * sysusers.d configuration for a user may now also specify the group
420 number, in addition to the user number ("u username 123:456"), or
421 without the user number ("u username -:456").
423 * Configution items for elogind-sysusers can now be specified as
424 positional arguments when the new --inline switch is used.
426 * The login shell of users created through sysusers.d may now be
427 specified (previously, it was always /bin/sh for root and
428 /sbin/nologin for other users).
430 * elogind-analyze gained a new --global switch to look at global user
431 configuration. It also gained a unit-paths verb to list the unit load
432 paths that are compiled into elogind (which can be used with
433 --elogind, --user, or --global).
435 * udevadm trigger gained a new --settle/-w option to wait for any
436 triggered events to finish (but just those, and not any other events
437 which are triggered meanwhile).
439 * The action that elogind takes when the lid is closed and the
440 machine is connected to external power can now be configured using
441 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
442 was determined by HandleLidSwitch=, and, for backwards compatibility,
443 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
445 * journalctl will periodically call sd_journal_process() to make it
446 resilient against inotify queue overruns when journal files are
447 rotated very quickly.
449 * Two new functions in libelogind — sd_bus_get_n_queued_read and
450 sd_bus_get_n_queued_write — may be used to check the number of
451 pending bus messages.
453 * elogind gained a new
454 org.freedesktop.elogind1.Manager.AttachProcessesToUnit dbus call
455 which can be used to migrate foreign processes to scope and service
456 units. The primary user for this new API is elogind itself: the
457 elogind --user instance uses this call of the elogind --system
458 instance to migrate processes if it itself gets the request to
459 migrate processes and the kernel refuses this due to access
460 restrictions. Thanks to this "elogind-run --scope --user …" works
461 again in pure cgroups v2 environments when invoked from the user
464 * A new TemporaryFileSystem= setting can be used to mask out part of
465 the real file system tree with tmpfs mounts. This may be combined
466 with BindPaths= and BindReadOnlyPaths= to hide files or directories
467 not relevant to the unit, while still allowing some paths lower in
468 the tree to be accessed.
470 ProtectHome=tmpfs may now be used to hide user home and runtime
471 directories from units, in a way that is mostly equivalent to
472 "TemporaryFileSystem=/home /run/user /root".
474 * Non-service units are now started with KeyringMode=shared by default.
475 This means that mount and swapon and other mount tools have access
476 to keys in the main keyring.
478 * /sys/fs/bpf is now mounted automatically.
480 * QNX virtualization is now detected by elogind-detect-virt and may
481 be used in ConditionVirtualization=.
483 * IPAccounting= may now be enabled also for slice units.
485 * A new -Dsplit-bin= build configuration switch may be used to specify
486 whether bin and sbin directories are merged, or if they should be
487 included separately in $PATH and various listings of executable
488 directories. The build configuration scripts will try to autodetect
489 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
490 system, but distributions are encouraged to configure this
493 * A new -Dok-color= build configuration switch may be used to change
494 the colour of "OK" status messages.
496 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
497 PrivateNetwork=yes was buggy in previous versions of elogind. This
498 means that after the upgrade and daemon-reexec, any such units must
501 * INCOMPATIBILITY: as announced in the NEWS for 237, elogind-tmpfiles
502 will not exclude read-only files owned by root from cleanup.
504 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
505 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
506 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
507 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
508 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
509 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
510 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
511 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
512 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
513 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
514 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
515 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
516 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
517 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
518 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
519 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
525 * Some keyboards come with a zoom see-saw or rocker which until now got
526 mapped to the Linux "zoomin/out" keys in hwdb. However, these
527 keycodes are not recognized by any major desktop. They now produce
528 Up/Down key events so that they can be used for scrolling.
530 * INCOMPATIBILITY: elogind-tmpfiles' "f" lines changed behaviour
531 slightly: previously, if an argument was specified for lines of this
532 type (i.e. the right-most column was set) this string was appended to
533 existing files each time elogind-tmpfiles was run. This behaviour was
534 different from what the documentation said, and not particularly
535 useful, as repeated elogind-tmpfiles invocations would not be
536 idempotent and grow such files without bounds. With this release
537 behaviour has been altered slightly, to match what the documentation
538 says: lines of this type only have an effect if the indicated files
539 don't exist yet, and only then the argument string is written to the
542 * FUTURE INCOMPATIBILITY: In elogind v238 we intend to slightly change
543 elogind-tmpfiles behaviour: previously, read-only files owned by root
544 were always excluded from the file "aging" algorithm (i.e. the
545 automatic clean-up of directories like /tmp based on
546 atime/mtime/ctime). We intend to drop this restriction, and age files
547 by default even when owned by root and read-only. This behaviour was
548 inherited from older tools, but there have been requests to remove
549 it, and it's not obvious why this restriction was made in the first
550 place. Please speak up now, if you are aware of software that reqires
551 this behaviour, otherwise we'll remove the restriction in v238.
553 * A new environment variable $SYSTEMD_OFFLINE is now understood by
554 systemctl. It takes a boolean argument. If on, systemctl assumes it
555 operates on an "offline" OS tree, and will not attempt to talk to the
556 service manager. Previously, this mode was implicitly enabled if a
557 chroot() environment was detected, and this new environment variable
558 now provides explicit control.
560 * .path and .socket units may now be created transiently, too.
561 Previously only service, mount, automount and timer units were
562 supported as transient units. The elogind-run tool has been updated
563 to expose this new functionality, you may hence use it now to bind
564 arbitrary commands to path or socket activation on-the-fly from the
565 command line. Moreover, almost all properties are now exposed for the
566 unit types that already supported transient operation.
568 * The elogind-mount command gained support for a new --owner= parameter
569 which takes a user name, which is then resolved and included in uid=
570 and gid= mount options string of the file system to mount.
572 * A new unit condition ConditionControlGroupController= has been added
573 that checks whether a specific cgroup controller is available.
575 * Unit files, udev's .link files, and elogind-networkd's .netdev and
576 .network files all gained support for a new condition
577 ConditionKernelVersion= for checking against specific kernel
580 * In elogind-networkd, the [IPVLAN] section in .netdev files gained
581 support for configuring device flags in the Flags= setting. In the
582 same files, the [Tunnel] section gained support for configuring
583 AllowLocalRemote=. The [Route] section in .network files gained
584 support for configuring InitialCongestionWindow=,
585 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
586 understands RapidCommit=.
588 * elogind-networkd's DHCPv6 support gained support for Prefix
591 * sd-bus gained support for a new "watch-bind" feature. When this
592 feature is enabled, an sd_bus connection may be set up to connect to
593 an AF_UNIX socket in the file system as soon as it is created. This
594 functionality is useful for writing early-boot services that
595 automatically connect to the system bus as soon as it is started,
596 without ugly time-based polling. elogind-networkd and
597 elogind-resolved have been updated to make use of this
598 functionality. busctl exposes this functionality in a new
599 --watch-bind= command line switch.
601 * sd-bus will now optionally synthesize a local "Connected" signal as
602 soon as a D-Bus connection is set up fully. This message mirrors the
603 already existing "Disconnected" signal which is synthesized when the
604 connection is terminated. This signal is generally useful but
605 particularly handy in combination with the "watch-bind" feature
606 described above. Synthesizing of this message has to be requested
607 explicitly through the new API call sd_bus_set_connected_signal(). In
608 addition a new call sd_bus_is_ready() has been added that checks
609 whether a connection is fully set up (i.e. between the "Connected" and
610 "Disconnected" signals).
612 * sd-bus gained two new calls sd_bus_request_name_async() and
613 sd_bus_release_name_async() for asynchronously registering bus
614 names. Similar, there is now sd_bus_add_match_async() for installing
615 a signal match asynchronously. All of elogind's own services have
616 been updated to make use of these calls. Doing these operations
617 asynchronously has two benefits: it reduces the risk of deadlocks in
618 case of cyclic dependencies between bus services, and it speeds up
619 service initialization since synchronization points for bus
620 round-trips are removed.
622 * sd-bus gained two new calls sd_bus_match_signal() and
623 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
624 and sd_bus_add_match_async() but instead of taking a D-Bus match
625 string take match fields as normal function parameters.
627 * sd-bus gained two new calls sd_bus_set_sender() and
628 sd_bus_message_set_sender() for setting the sender name of outgoing
629 messages (either for all outgoing messages or for just one specific
630 one). These calls are only useful in direct connections as on
631 brokered connections the broker fills in the sender anyway,
632 overwriting whatever the client filled in.
634 * sd-event gained a new pseudo-handle that may be specified on all API
635 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
636 used this refers to the default event loop object of the calling
637 thread. Note however that this does not implicitly allocate one —
638 which has to be done prior by using sd_event_default(). Similarly
639 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
640 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
641 to the default bus of the specified type of the calling thread. Here
642 too this does not implicitly allocate bus connection objects, this
643 has to be done prior with sd_bus_default() and friends.
645 * sd-event gained a new call pair
646 sd_event_source_{get|set}_io_fd_own(). This may be used to request
647 automatic closure of the file descriptor an IO event source watches
648 when the event source is destroyed.
650 * elogind-networkd gained support for natively configuring WireGuard
653 * In previous versions elogind synthesized user records both for the
654 "nobody" (UID 65534) and "root" (UID 0) users in nss-elogind and
655 internally. In order to simplify distribution-wide renames of the
656 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
657 new transitional flag file has been added: if
658 /etc/elogind/dont-synthesize-nobody exists synthesizing of the 65534
659 user and group record within the elogind codebase is disabled.
661 * elogind-notify gained a new --uid= option for selecting the source
662 user/UID to use for notification messages sent to the service
665 * journalctl gained a new --grep= option to list only entries in which
666 the message matches a certain pattern. By default matching is case
667 insensitive if the pattern is lowercase, and case sensitive
668 otherwise. Option --case-sensitive=yes|no can be used to override
669 this an specify case sensitivity or case insensitivity.
671 * There's now a "elogind-analyze service-watchdogs" command for printing
672 the current state of the service runtime watchdog, and optionally
673 enabling or disabling the per-service watchdogs system-wide if given a
674 boolean argument (i.e. the concept you configure in WatchdogSec=), for
675 debugging purposes. There's also a kernel command line option
676 elogind.service_watchdogs= for controlling the same.
678 * Two new "log-level" and "log-target" options for elogind-analyze were
679 added that merge the now deprecated get-log-level, set-log-level and
680 get-log-target, set-log-target pairs. The deprecated options are still
681 understood for backwards compatibility. The two new options print the
682 current value when no arguments are given, and set them when a
683 level/target is given as an argument.
685 * sysusers.d's "u" lines now optionally accept both a UID and a GID
686 specification, separated by a ":" character, in order to create users
687 where UID and GID do not match.
689 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
690 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
691 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
692 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
693 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
694 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
695 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
696 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
697 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
698 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
699 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
700 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
701 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
702 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
703 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
704 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
705 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
706 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
707 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
708 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
715 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
716 in v235 has been extended to also set the dummy.ko module option
717 numdummies=0, preventing the kernel from automatically creating
718 dummy0. All dummy interfaces must now be explicitly created.
720 * Unknown '%' specifiers in configuration files are now rejected. This
721 applies to units and tmpfiles.d configuration. Any percent characters
722 that are followed by a letter or digit that are not supposed to be
723 interpreted as the beginning of a specifier should be escaped by
724 doubling ("%%"). (So "size=5%" is still accepted, as well as
725 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
726 valid specifiers today.)
728 * systemd-resolved now maintains a new dynamic
729 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
730 recommended to make /etc/resolv.conf a symlink to it. This file
731 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
732 includes dynamically acquired search domains, achieving more correct
733 DNS resolution by software that bypasses local DNS APIs such as NSS.
735 * The "uaccess" udev tag has been dropped from /dev/kvm and
736 /dev/dri/renderD*. These devices now have the 0666 permissions by
737 default (but this may be changed at build-time). /dev/dri/renderD*
738 will now be owned by the "render" group along with /dev/kfd.
740 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
741 systemd-journal-gatewayd.service and
742 systemd-journal-upload.service. This means "nss-systemd" must be
743 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
744 services are resolved properly.
746 * In /etc/fstab two new mount options are now understood:
747 x-systemd.makefs and x-systemd.growfs. The former has the effect that
748 the configured file system is formatted before it is mounted, the
749 latter that the file system is resized to the full block device size
750 after it is mounted (i.e. if the file system is smaller than the
751 partition it resides on, it's grown). This is similar to the fsck
752 logic in /etc/fstab, and pulls in systemd-makefs@.service and
753 systemd-growfs@.service as necessary, similar to
754 systemd-fsck@.service. Resizing is currently only supported on ext4
757 * In systemd-networkd, the IPv6 RA logic now optionally may announce
758 DNS server and domain information.
760 * Support for the LUKS2 on-disk format for encrypted partitions has
761 been added. This requires libcryptsetup2 during compilation and
764 * The systemd --user instance will now signal "readiness" when its
765 basic.target unit has been reached, instead of when the run queue ran
766 empty for the first time.
768 * Tmpfiles.d with user configuration are now also supported.
769 systemd-tmpfiles gained a new --user switch, and snippets placed in
770 ~/.config/user-tmpfiles.d/ and corresponding directories will be
771 executed by systemd-tmpfiles --user running in the new
772 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
773 running in the user session.
775 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
776 %S resolves to the top-level state directory (/var/lib for the system
777 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
778 top-level cache directory (/var/cache for the system instance,
779 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
780 logs directory (/var/log for the system instance,
781 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
782 existing %t specifier, that resolves to the top-level runtime
783 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
786 * journalctl learnt a new parameter --output-fields= for limiting the
787 set of journal fields to output in verbose and JSON output modes.
789 * systemd-timesyncd's configuration file gained a new option
790 RootDistanceMaxSec= for setting the maximum root distance of servers
791 it'll use, as well as the new options PollIntervalMinSec= and
792 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
794 * bootctl gained a new command "list" for listing all available boot
795 menu items on systems that follow the boot loader specification.
797 * systemctl gained a new --dry-run switch that shows what would be done
798 instead of doing it, and is currently supported by the shutdown and
801 * ConditionSecurity= can now detect the TOMOYO security module.
803 * Unit file [Install] sections are now also respected in unit drop-in
804 files. This is intended to be used by drop-ins under /usr/lib/.
806 * systemd-firstboot may now also set the initial keyboard mapping.
808 * Udev "changed" events for devices which are exposed as systemd
809 .device units are now propagated to units specified in
810 ReloadPropagatedFrom= as reload requests.
812 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
813 unit template name (i.e. a name in the form of 'foobar@.service',
814 without the instance component between the '@' and - the '.'), then
815 the escaped sysfs path of the device is automatically used as the
818 * SystemCallFilter= in unit files has been extended so that an "errno"
819 can be specified individually for each system call. Example:
820 SystemCallFilter=~uname:EILSEQ.
822 * The cgroup delegation logic has been substantially updated. Delegate=
823 now optionally takes a list of controllers (instead of a boolean, as
824 before), which lists the controllers to delegate at least.
826 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
828 * A new LogLevelMax= setting configures the maximum log level any
829 process of the service may log at (i.e. anything with a lesser
830 priority than what is specified is automatically dropped). A new
831 LogExtraFields= setting allows configuration of additional journal
832 fields to attach to all log records generated by any of the unit's
835 * New StandardInputData= and StandardInputText= settings along with the
836 new option StandardInput=data may be used to configure textual or
837 binary data that shall be passed to the executed service process via
838 standard input, encoded in-line in the unit file.
840 * StandardInput=, StandardOutput= and StandardError= may now be used to
841 connect stdin/stdout/stderr of executed processes directly with a
842 file or AF_UNIX socket in the file system, using the new "file:" option.
844 * A new unit file option CollectMode= has been added, that allows
845 tweaking the garbage collection logic for units. It may be used to
846 tell systemd to garbage collect units that have failed automatically
847 (normally it only GCs units that exited successfully). systemd-run
848 and systemd-mount expose this new functionality with a new -G option.
850 * "machinectl bind" may now be used to bind mount non-directories
851 (i.e. regularfiles, devices, fifos, sockets).
853 * systemd-analyze gained a new verb "calendar" for validating and
854 testing calendar time specifications to use for OnCalendar= in timer
855 units. Besides validating the expression it will calculate the next
856 time the specified expression would elapse.
858 * In addition to the pre-existing FailureAction= unit file setting
859 there's now SuccessAction=, for configuring a shutdown action to
860 execute when a unit completes successfully. This is useful in
861 particular inside containers that shall terminate after some workload
862 has been completed. Also, both options are now supported for all unit
863 types, not just services.
865 * networkds's IP rule support gained two new options
866 IncomingInterface= and OutgoingInterface= for configuring the incoming
867 and outgoing interfaces of configured rules. systemd-networkd also
868 gained support for "vxcan" network devices.
870 * networkd gained a new setting RequiredForOnline=, taking a
871 boolean. If set, systemd-wait-online will take it into consideration
872 when determining that the system is up, otherwise it will ignore the
873 interface for this purpose.
875 * The sd_notify() protocol gained support for a new operation: with
876 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
877 store again, ahead of POLLHUP or POLLERR when they are removed
880 * A new document doc/UIDS-GIDS.md has been added to the source tree,
881 that documents the UID/GID range and assignment assumptions and
882 requirements of systemd.
884 * The watchdog device PID 1 will ping may now be configured through the
885 WatchdogDevice= configuration file setting, or by setting the
886 systemd.watchdog_service= kernel commandline option.
888 * systemd-resolved's gained support for registering DNS-SD services on
889 the local network using MulticastDNS. Services may either be
890 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
891 the same dir below /run, /usr/lib), or through its D-Bus API.
893 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
894 extend the effective start, runtime, and stop time. The service must
895 continue to send EXTEND_TIMEOUT_USEC within the period specified to
896 prevent the service manager from making the service as timedout.
898 * elogind-resolved's DNSSEC support gained support for RFC 8080
899 (Ed25519 keys and signatures).
901 * The elogind-resolve command line tool gained a new set of options
902 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
903 --set-nta= and --revert to configure per-interface DNS configuration
904 dynamically during runtime. It's useful for pushing DNS information
905 into elogind-resolved from DNS hook scripts that various interface
906 managing software supports (such as pppd).
908 * elogind-nspawn gained a new --network-namespace-path= command line
909 option, which may be used to make a container join an existing
910 network namespace, by specifying a path to a "netns" file.
912 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
913 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
914 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
915 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
916 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
917 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
918 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
919 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
920 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
921 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
922 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
923 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
924 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
925 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
926 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
927 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
928 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
929 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
930 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
931 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
932 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
933 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
934 Jędrzejewski-Szmek, Zeal Jagannatha
940 * INCOMPATIBILITY: systemd-logind.service and other long-running
941 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
942 communication with the outside. This generally improves security of
943 the system, and is in almost all cases a safe and good choice, as
944 these services do not and should not provide any network-facing
945 functionality. However, systemd-logind uses the glibc NSS API to
946 query the user database. This creates problems on systems where NSS
947 is set up to directly consult network services for user database
948 lookups. In particular, this creates incompatibilities with the
949 "nss-nis" module, which attempts to directly contact the NIS/YP
950 network servers it is configured for, and will now consistently
951 fail. In such cases, it is possible to turn off IP sandboxing for
952 systemd-logind.service (set IPAddressDeny= in its [Service] section
953 to the empty string, via a .d/ unit file drop-in). Downstream
954 distributions might want to update their nss-nis packaging to include
955 such a drop-in snippet, accordingly, to hide this incompatibility
956 from the user. Another option is to make use of glibc's nscd service
957 to proxy such network requests through a privilege-separated, minimal
958 local caching daemon, or to switch to more modern technologies such
959 sssd, whose NSS hook-ups generally do not involve direct network
960 access. In general, we think it's definitely time to question the
961 implementation choices of nss-nis, i.e. whether it's a good idea
962 today to embed a network-facing loadable module into all local
963 processes that need to query the user database, including the most
964 trivial and benign ones, such as "ls". For more details about
965 IPAddressDeny= see below.
967 * A new modprobe.d drop-in is now shipped by default that sets the
968 bonding module option max_bonds=0. This overrides the kernel default,
969 to avoid conflicts and ambiguity as to whether or not bond0 should be
970 managed by systemd-networkd or not. This resolves multiple issues
971 with bond0 properties not being applied, when bond0 is configured
972 with systemd-networkd. Distributors may choose to not package this,
973 however in that case users will be prevented from correctly managing
974 bond0 interface using systemd-networkd.
976 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
977 which print the logging level and target of the system manager. They
978 complement the existing "set-log-level" and "set-log-target" verbs
979 used to change those values.
981 * journald.conf gained a new boolean setting ReadKMsg= which defaults
982 to on. If turned off kernel log messages will not be read by
983 systemd-journald or included in the logs. It also gained a new
984 setting LineMax= for configuring the maximum line length in
985 STDOUT/STDERR log streams. The new default for this value is 48K, up
986 from the previous hardcoded 2048.
988 * A new unit setting RuntimeDirectoryPreserve= has been added, which
989 allows more detailed control of what to do with a runtime directory
990 configured with RuntimeDirectory= (i.e. a directory below /run or
991 $XDG_RUNTIME_DIR) after a unit is stopped.
993 * The RuntimeDirectory= setting for units gained support for creating
994 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
995 one top-level directory.
997 * Units gained new options StateDirectory=, CacheDirectory=,
998 LogsDirectory= and ConfigurationDirectory= which are closely related
999 to RuntimeDirectory= but manage per-service directories below
1000 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
1001 possible to write unit files which when activated automatically gain
1002 properly owned service specific directories in these locations, thus
1003 making unit files self-contained and increasing compatibility with
1004 stateless systems and factory reset where /etc or /var are
1005 unpopulated at boot. Matching these new settings there's also
1006 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
1007 ConfigurationDirectoryMode= for configuring the access mode of these
1008 directories. These settings are particularly useful in combination
1009 with DynamicUser=yes as they provide secure, properly-owned,
1010 writable, and stateful locations for storage, excluded from the
1011 sandbox that such services live in otherwise.
1013 * Automake support has been removed from this release. systemd is now
1016 * systemd-journald will now aggressively cache client metadata during
1017 runtime, speeding up log write performance under pressure. This comes
1018 at a small price though: as much of the metadata is read
1019 asynchronously from /proc/ (and isn't implicitly attached to log
1020 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
1021 metadata stored alongside a log entry might be slightly
1022 out-of-date. Previously it could only be slightly newer than the log
1023 message. The time window is small however, and given that the kernel
1024 is unlikely to be improved anytime soon in this regard, this appears
1027 * nss-myhostname/systemd-resolved will now by default synthesize an
1028 A/AAAA resource record for the "_gateway" hostname, pointing to the
1029 current default IP gateway. Previously it did that for the "gateway"
1030 name, hampering adoption, as some distributions wanted to leave that
1031 host name open for local use. The old behaviour may still be
1032 requested at build time.
1034 * systemd-networkd's [Address] section in .network files gained a new
1035 Scope= setting for configuring the IP address scope. The [Network]
1036 section gained a new boolean setting ConfigureWithoutCarrier= that
1037 tells systemd-networkd to ignore link sensing when configuring the
1038 device. The [DHCP] section gained a new Anonymize= boolean option for
1039 turning on a number of options suggested in RFC 7844. A new
1040 [RoutingPolicyRule] section has been added for configuring the IP
1041 routing policy. The [Route] section has gained support for a new
1042 Type= setting which permits configuring
1043 blackhole/unreachable/prohibit routes.
1045 * The [VRF] section in .netdev files gained a new Table= setting for
1046 configuring the routing table to use. The [Tunnel] section gained a
1047 new Independent= boolean field for configuring tunnels independent of
1048 an underlying network interface. The [Bridge] section gained a new
1049 GroupForwardMask= option for configuration of propagation of link
1050 local frames between bridge ports.
1052 * The WakeOnLan= setting in .link files gained support for a number of
1053 new modes. A new TCP6SegmentationOffload= setting has been added for
1054 configuring TCP/IPv6 hardware segmentation offload.
1056 * The IPv6 RA sender implementation may now optionally send out RDNSS
1057 and RDNSSL records to supply DNS configuration to peers.
1059 * systemd-nspawn gained support for a new --system-call-filter= command
1060 line option for adding and removing entries in the default system
1061 call filter it applies. Moreover systemd-nspawn has been changed to
1062 implement a system call whitelist instead of a blacklist.
1064 * systemd-run gained support for a new --pipe command line option. If
1065 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
1066 are directly passed on to the activated transient service
1067 executable. This allows invoking arbitrary processes as systemd
1068 services (for example to take benefit of dependency management,
1069 accounting management, resource management or log management that is
1070 done automatically for services) — while still allowing them to be
1071 integrated in a classic UNIX shell pipeline.
1073 * When a service sends RELOAD=1 via sd_notify() and reload propagation
1074 using ReloadPropagationTo= is configured, a reload is now propagated
1075 to configured units. (Previously this was only done on explicitly
1076 requested reloads, using "systemctl reload" or an equivalent
1079 * For each service unit a restart counter is now kept: it is increased
1080 each time the service is restarted due to Restart=, and may be
1081 queried using "systemctl show -p NRestarts …".
1083 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
1084 @signal and @timer have been added, for usage with SystemCallFilter=
1085 in unit files and the new --system-call-filter= command line option
1086 of systemd-nspawn (see above).
1088 * ExecStart= lines in unit files gained two new modifiers: when a
1089 command line is prefixed with "!" the command will be executed as
1090 configured, except for the credentials applied by
1091 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
1092 "+", but does still apply namespacing options unlike "+". There's
1093 also "!!" now, which is mostly identical, but becomes a NOP on
1094 systems that support ambient capabilities. This is useful to write
1095 unit files that work with ambient capabilities where possible but
1096 automatically fall back to traditional privilege dropping mechanisms
1097 on systems where this is not supported.
1099 * ListenNetlink= settings in socket units now support RDMA netlink
1102 * A new unit file setting LockPersonality= has been added which permits
1103 locking down the chosen execution domain ("personality") of a service
1106 * A new special target "getty-pre.target" has been added, which is
1107 ordered before all text logins, and may be used to order services
1108 before textual logins acquire access to the console.
1110 * systemd will now attempt to load the virtio-rng.ko kernel module very
1111 early on if a VM environment supporting this is detected. This should
1112 improve entropy during early boot in virtualized environments.
1114 * A _netdev option is now supported in /etc/crypttab that operates in a
1115 similar way as the same option in /etc/fstab: it permits configuring
1116 encrypted devices that need to be ordered after the network is up.
1117 Following this logic, two new special targets
1118 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
1119 added that are to cryptsetup.target what remote-fs.target and
1120 remote-fs-pre.target are to local-fs.target.
1122 * Service units gained a new UnsetEnvironment= setting which permits
1123 unsetting specific environment variables for services that are
1124 normally passed to it (for example in order to mask out locale
1125 settings for specific services that can't deal with it).
1127 * Units acquired a new boolean option IPAccounting=. When turned on, IP
1128 traffic accounting (packet count as well as byte count) is done for
1129 the service, and shown as part of "systemctl status" or "systemd-run
1132 * Service units acquired two new options IPAddressAllow= and
1133 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
1134 for configuring a simple IP access control list for all sockets of
1135 the unit. These options are available also on .slice and .socket
1136 units, permitting flexible access list configuration for individual
1137 services as well as groups of services (as defined by a slice unit),
1138 including system-wide. Note that IP ACLs configured this way are
1139 enforced on every single IPv4 and IPv6 socket created by any process
1140 of the service unit, and apply to ingress as well as egress traffic.
1142 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
1143 structured log message is generated each time the unit is stopped,
1144 containing information about the consumed resources of this
1147 * A new setting KeyringMode= has been added to unit files, which may be
1148 used to control how the kernel keyring is set up for executed
1151 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
1152 "systemctl kexec" and "systemctl exit" are now always asynchronous in
1153 behaviour (that is: these commands return immediately after the
1154 operation was enqueued instead of waiting for the operation to
1155 complete). Previously, "systemctl poweroff" and "systemctl reboot"
1156 were asynchronous on systems using systemd-logind (i.e. almost
1157 always, and like they were on sysvinit), and the other three commands
1158 were unconditionally synchronous. With this release this is cleaned
1159 up, and callers will see the same asynchronous behaviour on all
1160 systems for all five operations.
1162 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
1165 * .timer units now accept calendar specifications in other timezones
1166 than UTC or the local timezone.
1168 * The tmpfiles snippet var.conf has been changed to create
1169 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
1170 the "utmp" group already, and it appears to be generally understood
1171 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
1172 databases. Previously this was implemented correctly for all these
1173 databases excepts btmp, which has been opened up like this now
1174 too. Note that while the other databases are world-readable
1175 (i.e. 0644), btmp is not and remains more restrictive.
1177 * The systemd-resolve tool gained a new --reset-server-features
1178 switch. When invoked like this systemd-resolved will forget
1179 everything it learnt about the features supported by the configured
1180 upstream DNS servers, and restarts the feature probing logic on the
1181 next resolver look-up for them at the highest feature level
1184 * The status dump systemd-resolved sends to the logs upon receiving
1185 SIGUSR1 now also includes information about all DNS servers it is
1186 configured to use, and the features levels it probed for them.
1188 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
1189 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
1190 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
1191 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
1192 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
1193 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
1194 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
1195 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
1196 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
1197 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
1198 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
1199 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
1200 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
1201 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
1202 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
1203 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
1204 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
1205 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
1206 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
1207 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1209 — Berlin, 2017-10-06
1213 * Meson is now supported as build system in addition to Automake. It is
1214 our plan to remove Automake in one of our next releases, so that
1215 Meson becomes our exclusive build system. Hence, please start using
1216 the Meson build system in your downstream packaging. There's plenty
1217 of documentation around how to use Meson, the extremely brief
1220 ./autogen.sh && ./configure && make && sudo make install
1224 meson build && ninja -C build && sudo ninja -C build install
1226 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
1227 which permits configuring a timeout on the time a job is
1228 running. This is particularly useful for setting timeouts on jobs for
1231 * Unit files gained two new options ConditionUser= and ConditionGroup=
1232 for conditionalizing units based on the identity of the user/group
1233 running a systemd user instance.
1235 * systemd-networkd now understands a new FlowLabel= setting in the
1236 [VXLAN] section of .network files, as well as a Priority= in
1237 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
1238 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
1239 gained support for configuration of GENEVE links, and IPv6 address
1240 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
1242 * .link files now understand a new Port= setting.
1244 * systemd-networkd's DHCP support gained support for DHCP option 119
1245 (domain search list).
1247 * systemd-networkd gained support for serving IPv6 address ranges using
1248 the Router Advertisement protocol. The new .network configuration
1249 section [IPv6Prefix] may be used to configure the ranges to
1250 serve. This is implemented based on a new, minimal, native server
1251 implementation of RA.
1253 * journalctl's --output= switch gained support for a new parameter
1254 "short-iso-precise" for a mode where timestamps are shown as precise
1257 * systemd-udevd's "net_id" builtin may now generate stable network
1258 interface names from IBM PowerVM VIO devices as well as ACPI platform
1261 * MulticastDNS support in systemd-resolved may now be explicitly
1262 enabled/disabled using the new MulticastDNS= configuration file
1265 * systemd-resolved may now optionally use libidn2 instead of the libidn
1266 for processing internationalized domain names. Support for libidn2
1267 should be considered experimental and should not be enabled by
1270 * "machinectl pull-tar" and related call may now do verification of
1271 downloaded images using SUSE-style .sha256 checksum files in addition
1272 to the already existing support for validating using Ubuntu-style
1275 * sd-bus gained support for a new sd_bus_message_appendv() call which
1276 is va_list equivalent of sd_bus_message_append().
1278 * sd-boot gained support for validating images using SHIM/MOK.
1280 * The SMACK code learnt support for "onlycap".
1282 * systemd-mount --umount is now much smarter in figuring out how to
1283 properly unmount a device given its mount or device path.
1285 * The code to call libnss_dns as a fallback from libnss_resolve when
1286 the communication with systemd-resolved fails was removed. This
1287 fallback was redundant and interfered with the [!UNAVAIL=return]
1288 suffix. See nss-resolve(8) for the recommended configuration.
1290 * systemd-logind may now be restarted without losing state. It stores
1291 the file descriptors for devices it manages in the system manager
1292 using the FDSTORE= mechanism. Please note that further changes in
1293 other components may be required to make use of this (for example
1294 Xorg has code to listen for stops of systemd-logind and terminate
1295 itself when logind is stopped or restarted, in order to avoid using
1296 stale file descriptors for graphical devices, which is now
1297 counterproductive and must be reverted in order for restarts of
1298 systemd-logind to be safe. See
1299 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
1301 * All kernel install plugins are called with the environment variable
1302 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
1303 /etc/machine-id. If the file is missing or empty, the variable is
1304 empty and BOOT_DIR_ABS is the path of a temporary directory which is
1305 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
1306 is empty, all plugins should not put anything in BOOT_DIR_ABS.
1308 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
1309 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
1310 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
1311 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
1312 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
1313 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
1314 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
1315 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
1316 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
1317 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
1318 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
1319 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
1320 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
1321 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
1322 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
1323 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
1324 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
1325 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
1326 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
1327 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
1328 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
1329 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
1330 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
1331 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
1332 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
1333 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
1334 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
1335 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
1336 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
1339 — Berlin, 2017-07-12
1343 * The "hybrid" control group mode has been modified to improve
1344 compatibility with "legacy" cgroups-v1 setups. Specifically, the
1345 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
1346 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
1347 cgroups-v1 hierarchy), the only externally visible change being that
1348 the cgroups-v2 hierarchy is also mounted, to
1349 /sys/fs/cgroup/unified. This should provide a large degree of
1350 compatibility with "legacy" cgroups-v1, while taking benefit of the
1351 better management capabilities of cgroups-v2.
1353 * The default control group setup mode may be selected both a boot-time
1354 via a set of kernel command line parameters (specifically:
1355 systemd.unified_cgroup_hierarchy= and
1356 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
1357 default selected on the configure command line
1358 (--with-default-hierarchy=). The upstream default is "hybrid"
1359 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
1360 this will change in a future systemd version to be "unified" (pure
1361 cgroups-v2 mode). The third option for the compile time option is
1362 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
1363 distributions to default to "hybrid" mode for release distributions,
1364 starting with v233. We recommend "unified" for development
1365 distributions (specifically: distributions such as Fedora's rawhide)
1366 as that's where things are headed in the long run. Use "legacy" for
1367 greatest stability and compatibility only.
1369 * Note one current limitation of "unified" and "hybrid" control group
1370 setup modes: the kernel currently does not permit the systemd --user
1371 instance (i.e. unprivileged code) to migrate processes between two
1372 disconnected cgroup subtrees, even if both are managed and owned by
1373 the user. This effectively means "systemd-run --user --scope" doesn't
1374 work when invoked from outside of any "systemd --user" service or
1375 scope. Specifically, it is not supported from session scopes. We are
1376 working on fixing this in a future systemd version. (See #3388 for
1377 further details about this.)
1379 * DBus policy files are now installed into /usr rather than /etc. Make
1380 sure your system has dbus >= 1.9.18 running before upgrading to this
1381 version, or override the install path with --with-dbuspolicydir= .
1383 * All python scripts shipped with systemd (specifically: the various
1384 tests written in Python) now require Python 3.
1386 * systemd unit tests can now run standalone (without the source or
1387 build directories), and can be installed into /usr/lib/systemd/tests/
1388 with 'make install-tests'.
1390 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
1391 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
1394 * Support for the %c, %r, %R specifiers in unit files has been
1395 removed. Specifiers are not supposed to be dependent on configuration
1396 in the unit file itself (so that they resolve the same regardless
1397 where used in the unit files), but these specifiers were influenced
1398 by the Slice= option.
1400 * The shell invoked by debug-shell.service now defaults to /bin/sh in
1401 all cases. If distributions want to use a different shell for this
1402 purpose (for example Fedora's /sbin/sushell) they need to specify
1403 this explicitly at configure time using --with-debug-shell=.
1405 * The confirmation spawn prompt has been reworked to offer the
1408 (c)ontinue, proceed without asking anymore
1409 (D)ump, show the state of the unit
1410 (f)ail, don't execute the command and pretend it failed
1412 (i)nfo, show a short summary of the unit
1413 (j)obs, show jobs that are in progress
1414 (s)kip, don't execute the command and pretend it succeeded
1415 (y)es, execute the command
1417 The 'n' choice for the confirmation spawn prompt has been removed,
1418 because its meaning was confusing.
1420 The prompt may now also be redirected to an alternative console by
1421 specifying the console as parameter to systemd.confirm_spawn=.
1423 * Services of Type=notify require a READY=1 notification to be sent
1424 during startup. If no such message is sent, the service now fails,
1425 even if the main process exited with a successful exit code.
1427 * Services that fail to start up correctly now always have their
1428 ExecStopPost= commands executed. Previously, they'd enter "failed"
1429 state directly, without executing these commands.
1431 * The option MulticastDNS= of network configuration files has acquired
1432 an actual implementation. With MulticastDNS=yes a host can resolve
1433 names of remote hosts and reply to mDNS A and AAAA requests.
1435 * When units are about to be started an additional check is now done to
1436 ensure that all dependencies of type BindsTo= (when used in
1437 combination with After=) have been started.
1439 * systemd-analyze gained a new verb "syscall-filter" which shows which
1440 system call groups are defined for the SystemCallFilter= unit file
1441 setting, and which system calls they contain.
1443 * A new system call filter group "@filesystem" has been added,
1444 consisting of various file system related system calls. Group
1445 "@reboot" has been added, covering reboot, kexec and shutdown related
1446 calls. Finally, group "@swap" has been added covering swap
1447 configuration related calls.
1449 * A new unit file option RestrictNamespaces= has been added that may be
1450 used to restrict access to the various process namespace types the
1451 Linux kernel provides. Specifically, it may be used to take away the
1452 right for a service unit to create additional file system, network,
1453 user, and other namespaces. This sandboxing option is particularly
1454 relevant due to the high amount of recently discovered namespacing
1455 related vulnerabilities in the kernel.
1457 * systemd-udev's .link files gained support for a new AutoNegotiation=
1458 setting for configuring Ethernet auto-negotiation.
1460 * systemd-networkd's .network files gained support for a new
1461 ListenPort= setting in the [DHCP] section to explicitly configure the
1462 UDP client port the DHCP client shall listen on.
1464 * .network files gained a new Unmanaged= boolean setting for explicitly
1465 excluding one or more interfaces from management by systemd-networkd.
1467 * The systemd-networkd ProxyARP= option has been renamed to
1468 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
1469 renamed to ReduceARPProxy=. The old names continue to be available
1472 * systemd-networkd gained support for configuring IPv6 Proxy NDP
1473 addresses via the new IPv6ProxyNDPAddress= .network file setting.
1475 * systemd-networkd's bonding device support gained support for two new
1476 configuration options ActiveSlave= and PrimarySlave=.
1478 * The various options in the [Match] section of .network files gained
1479 support for negative matching.
1481 * New systemd-specific mount options are now understood in /etc/fstab:
1483 x-systemd.mount-timeout= may be used to configure the maximum
1484 permitted runtime of the mount command.
1486 x-systemd.device-bound may be set to bind a mount point to its
1487 backing device unit, in order to automatically remove a mount point
1488 if its backing device is unplugged. This option may also be
1489 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
1490 on the block device, which is now automatically set for all CDROM
1491 drives, so that mounted CDs are automatically unmounted when they are
1492 removed from the drive.
1494 x-systemd.after= and x-systemd.before= may be used to explicitly
1495 order a mount after or before another unit or mount point.
1497 * Enqueued start jobs for device units are now automatically garbage
1498 collected if there are no jobs waiting for them anymore.
1500 * systemctl list-jobs gained two new switches: with --after, for every
1501 queued job the jobs it's waiting for are shown; with --before the
1502 jobs which it's blocking are shown.
1504 * systemd-nspawn gained support for ephemeral boots from disk images
1505 (or in other words: --ephemeral and --image= may now be
1506 combined). Moreover, ephemeral boots are now supported for normal
1507 directories, even if the backing file system is not btrfs. Of course,
1508 if the file system does not support file system snapshots or
1509 reflinks, the initial copy operation will be relatively expensive, but
1510 this should still be suitable for many use cases.
1512 * Calendar time specifications in .timer units now support
1513 specifications relative to the end of a month by using "~" instead of
1514 "-" as separator between month and day. For example, "*-02~03" means
1515 "the third last day in February". In addition a new syntax for
1516 repeated events has been added using the "/" character. For example,
1517 "9..17/2:00" means "every two hours from 9am to 5pm".
1519 * systemd-socket-proxyd gained a new parameter --connections-max= for
1520 configuring the maximum number of concurrent connections.
1522 * sd-id128 gained a new API for generating unique IDs for the host in a
1523 way that does not leak the machine ID. Specifically,
1524 sd_id128_get_machine_app_specific() derives an ID based on the
1525 machine ID a in well-defined, non-reversible, stable way. This is
1526 useful whenever an identifier for the host is needed but where the
1527 identifier shall not be useful to identify the system beyond the
1528 scope of the application itself. (Internally this uses HMAC-SHA256 as
1529 keyed hash function using the machine ID as input.)
1531 * NotifyAccess= gained a new supported value "exec". When set
1532 notifications are accepted from all processes systemd itself invoked,
1533 including all control processes.
1535 * .nspawn files gained support for defining overlay mounts using the
1536 Overlay= and OverlayReadOnly= options. Previously this functionality
1537 was only available on the systemd-nspawn command line.
1539 * systemd-nspawn's --bind= and --overlay= options gained support for
1540 bind/overlay mounts whose source lies within the container tree by
1541 prefixing the source path with "+".
1543 * systemd-nspawn's --bind= and --overlay= options gained support for
1544 automatically allocating a temporary source directory in /var/tmp
1545 that is removed when the container dies. Specifically, if the source
1546 directory is specified as empty string this mechanism is selected. An
1547 example usage is --overlay=+/var::/var, which creates an overlay
1548 mount based on the original /var contained in the image, overlayed
1549 with a temporary directory in the host's /var/tmp. This way changes
1550 to /var are automatically flushed when the container shuts down.
1552 * systemd-nspawn --image= option does now permit raw file system block
1553 devices (in addition to images containing partition tables, as
1556 * The disk image dissection logic in systemd-nspawn gained support for
1557 automatically setting up LUKS encrypted as well as Verity protected
1558 partitions. When a container is booted from an encrypted image the
1559 passphrase is queried at start-up time. When a container with Verity
1560 data is started, the root hash is search in a ".roothash" file
1561 accompanying the disk image (alternatively, pass the root hash via
1562 the new --root-hash= command line option).
1564 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
1565 be used to dissect disk images the same way as systemd-nspawn does
1566 it, following the Bootable Partition Specification. It may even be
1567 used to mount disk images with complex partition setups (including
1568 LUKS and Verity partitions) to a local host directory, in order to
1569 inspect them. This tool is not considered public API (yet), and is
1570 thus not installed into /usr/bin. Please do not rely on its
1571 existence, since it might go away or be changed in later systemd
1574 * A new generator "systemd-verity-generator" has been added, similar in
1575 style to "systemd-cryptsetup-generator", permitting automatic setup of
1576 Verity root partitions when systemd boots up. In order to make use of
1577 this your partition setup should follow the Discoverable Partitions
1578 Specification, and the GPT partition ID of the root file system
1579 partition should be identical to the upper 128bit of the Verity root
1580 hash. The GPT partition ID of the Verity partition protecting it
1581 should be the lower 128bit of the Verity root hash. If the partition
1582 image follows this model it is sufficient to specify a single
1583 "roothash=" kernel command line argument to both configure which root
1584 image and verity partition to use as well as the root hash for
1585 it. Note that systemd-nspawn's Verity support follows the same
1586 semantics, meaning that disk images with proper Verity data in place
1587 may be booted in containers with systemd-nspawn as well as on
1588 physical systems via the verity generator. Also note that the "mkosi"
1589 tool available at https://github.com/systemd/mkosi has been updated
1590 to generate Verity protected disk images following this scheme. In
1591 fact, it has been updated to generate disk images that optionally
1592 implement a complete UEFI SecureBoot trust chain, involving a signed
1593 kernel and initrd image that incorporates such a root hash as well as
1594 a Verity-enabled root partition.
1596 * The hardware database (hwdb) udev supports has been updated to carry
1597 accelerometer quirks.
1599 * All system services are now run with a fresh kernel keyring set up
1600 for them. The invocation ID is stored by default in it, thus
1601 providing a safe, non-overridable way to determine the invocation
1604 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
1605 options for bind mounting arbitrary paths in a service-specific
1606 way. When these options are used, arbitrary host or service files and
1607 directories may be mounted to arbitrary locations in the service's
1610 * Documentation has been added that lists all of systemd's low-level
1611 environment variables:
1613 https://github.com/systemd/systemd/blob/master/doc/ENVIRONMENT.md
1615 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
1616 whether a specific socket file descriptor matches a specified socket
1619 * systemd-firstboot has been updated to check for the
1620 systemd.firstboot= kernel command line option. It accepts a boolean
1621 and when set to false the first boot questions are skipped.
1623 * systemd-fstab-generator has been updated to check for the
1624 systemd.volatile= kernel command line option, which either takes an
1625 optional boolean parameter or the special value "state". If used the
1626 system may be booted in a "volatile" boot mode. Specifically,
1627 "systemd.volatile" is used, the root directory will be mounted as
1628 tmpfs, and only /usr is mounted from the actual root file system. If
1629 "systemd.volatile=state" is used, the root directory will be mounted
1630 as usual, but /var is mounted as tmpfs. This concept provides similar
1631 functionality as systemd-nspawn's --volatile= option, but provides it
1632 on physical boots. Use this option for implementing stateless
1633 systems, or testing systems with all state and/or configuration reset
1634 to the defaults. (Note though that many distributions are not
1635 prepared to boot up without a populated /etc or /var, though.)
1637 * systemd-gpt-auto-generator gained support for LUKS encrypted root
1638 partitions. Previously it only supported LUKS encrypted partitions
1639 for all other uses, except for the root partition itself.
1641 * Socket units gained support for listening on AF_VSOCK sockets for
1642 communication in virtualized QEMU environments.
1644 * The "configure" script gained a new option --with-fallback-hostname=
1645 for specifying the fallback hostname to use if none is configured in
1646 /etc/hostname. For example, by specifying
1647 --with-fallback-hostname=fedora it is possible to default to a
1648 hostname of "fedora" on pristine installations.
1650 * systemd-cgls gained support for a new --unit= switch for listing only
1651 the control groups of a specific unit. Similar --user-unit= has been
1652 added for listing only the control groups of a specific user unit.
1654 * systemd-mount gained a new --umount switch for unmounting a mount or
1655 automount point (and all mount/automount points below it).
1657 * systemd will now refuse full configuration reloads (via systemctl
1658 daemon-reload and related calls) unless at least 16MiB of free space
1659 are available in /run. This is a safety precaution in order to ensure
1660 that generators can safely operate after the reload completed.
1662 * A new unit file option RootImage= has been added, which has a similar
1663 effect as RootDirectory= but mounts the service's root directory from
1664 a disk image instead of plain directory. This logic reuses the same
1665 image dissection and mount logic that systemd-nspawn already uses,
1666 and hence supports any disk images systemd-nspawn supports, including
1667 those following the Discoverable Partition Specification, as well as
1668 Verity enabled images. This option enables systemd to run system
1669 services directly off disk images acting as resource bundles,
1670 possibly even including full integrity data.
1672 * A new MountAPIVFS= unit file option has been added, taking a boolean
1673 argument. If enabled /proc, /sys and /dev (collectively called the
1674 "API VFS") will be mounted for the service. This is only relevant if
1675 RootDirectory= or RootImage= is used for the service, as these mounts
1676 are of course in place in the host mount namespace anyway.
1678 * systemd-nspawn gained support for a new --pivot-root= switch. If
1679 specified the root directory within the container image is pivoted to
1680 the specified mount point, while the original root disk is moved to a
1681 different place. This option enables booting of ostree images
1682 directly with systemd-nspawn.
1684 * The systemd build scripts will no longer complain if the NTP server
1685 addresses are not changed from the defaults. Google now supports
1686 these NTP servers officially. We still recommend downstreams to
1687 properly register an NTP pool with the NTP pool project though.
1689 * coredumpctl gained a new "--reverse" option for printing the list
1690 of coredumps in reverse order.
1692 * coredumpctl will now show additional information about truncated and
1693 inaccessible coredumps, as well as coredumps that are still being
1694 processed. It also gained a new --quiet switch for suppressing
1695 additional informational message in its output.
1697 * coredumpctl gained support for only showing coredumps newer and/or
1698 older than specific timestamps, using the new --since= and --until=
1699 options, reminiscent of journalctl's options by the same name.
1701 * The systemd-coredump logic has been improved so that it may be reused
1702 to collect backtraces in non-compiled languages, for example in
1703 scripting languages such as Python.
1705 * machinectl will now show the UID shift of local containers, if user
1706 namespacing is enabled for them.
1708 * systemd will now optionally run "environment generator" binaries at
1709 configuration load time. They may be used to add environment
1710 variables to the environment block passed to services invoked. One
1711 user environment generator is shipped by default that sets up
1712 environment variables based on files dropped into /etc/environment.d
1713 and ~/.config/environment.d/.
1715 * systemd-resolved now includes the new, recently published 2017 DNSSEC
1718 * hostnamed has been updated to report a new chassis type of
1719 "convertible" to cover "foldable" laptops that can both act as a
1720 tablet and as a laptop, such as various Lenovo Yoga devices.
1722 Contributions from: Adrián López, Alexander Galanin, Alexander
1723 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
1724 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
1725 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
1726 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
1727 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
1728 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
1729 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
1730 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
1731 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
1732 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
1733 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
1734 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
1735 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
1736 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
1737 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
1738 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
1739 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
1740 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
1741 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
1742 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
1743 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
1744 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
1745 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
1746 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
1747 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
1748 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
1749 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
1752 — Berlin, 2017-03-01
1756 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
1757 RestrictAddressFamilies= enabled. These sandboxing options should
1758 generally be compatible with the various external udev call-out
1759 binaries we are aware of, however there may be exceptions, in
1760 particular when exotic languages for these call-outs are used. In
1761 this case, consider turning off these settings locally.
1763 * The new RemoveIPC= option can be used to remove IPC objects owned by
1764 the user or group of a service when that service exits.
1766 * The new ProtectKernelModules= option can be used to disable explicit
1767 load and unload operations of kernel modules by a service. In
1768 addition access to /usr/lib/modules is removed if this option is set.
1770 * ProtectSystem= option gained a new value "strict", which causes the
1771 whole file system tree with the exception of /dev, /proc, and /sys,
1772 to be remounted read-only for a service.
1774 * The new ProtectKernelTunables= option can be used to disable
1775 modification of configuration files in /sys and /proc by a service.
1776 Various directories and files are remounted read-only, so access is
1777 restricted even if the file permissions would allow it.
1779 * The new ProtectControlGroups= option can be used to disable write
1780 access by a service to /sys/fs/cgroup.
1782 * Various systemd services have been hardened with
1783 ProtectKernelTunables=yes, ProtectControlGroups=yes,
1784 RestrictAddressFamilies=.
1786 * Support for dynamically creating users for the lifetime of a service
1787 has been added. If DynamicUser=yes is specified, user and group IDs
1788 will be allocated from the range 61184..65519 for the lifetime of the
1789 service. They can be resolved using the new nss-systemd.so NSS
1790 module. The module must be enabled in /etc/nsswitch.conf. Services
1791 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
1792 any resources allocated by the service will be cleaned up when the
1793 service exits. They also have ProtectHome=read-only and
1794 ProtectSystem=strict enabled, so they are not able to make any
1795 permanent modifications to the system.
1797 * The nss-systemd module also always resolves root and nobody, making
1798 it possible to have no /etc/passwd or /etc/group files in minimal
1799 container or chroot environments.
1801 * Services may be started with their own user namespace using the new
1802 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
1803 under which the service is running are mapped. All other users are
1806 * Support for the cgroup namespace has been added to systemd-nspawn. If
1807 supported by kernel, the container system started by systemd-nspawn
1808 will have its own view of the cgroup hierarchy. This new behaviour
1809 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
1811 * The new MemorySwapMax= option can be used to limit the maximum swap
1812 usage under the unified cgroup hierarchy.
1814 * Support for the CPU controller in the unified cgroup hierarchy has
1815 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
1816 options. This controller requires out-of-tree patches for the kernel
1817 and the support is provisional.
1819 * Mount and automount units may now be created transiently
1820 (i.e. dynamically at runtime via the bus API, instead of requiring
1821 unit files in the file system).
1823 * systemd-mount is a new tool which may mount file systems – much like
1824 mount(8), optionally pulling in additional dependencies through
1825 transient .mount and .automount units. For example, this tool
1826 automatically runs fsck on a backing block device before mounting,
1827 and allows the automount logic to be used dynamically from the
1828 command line for establishing mount points. This tool is particularly
1829 useful when dealing with removable media, as it will ensure fsck is
1830 run – if necessary – before the first access and that the file system
1831 is quickly unmounted after each access by utilizing the automount
1832 logic. This maximizes the chance that the file system on the
1833 removable media stays in a clean state, and if it isn't in a clean
1834 state is fixed automatically.
1836 * LazyUnmount=yes option for mount units has been added to expose the
1837 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
1840 * /efi will be used as the mount point of the EFI boot partition, if
1841 the directory is present, and the mount point was not configured
1842 through other means (e.g. fstab). If /efi directory does not exist,
1843 /boot will be used as before. This makes it easier to automatically
1844 mount the EFI partition on systems where /boot is used for something
1847 * When operating on GPT disk images for containers, systemd-nspawn will
1848 now mount the ESP to /boot or /efi according to the same rules as PID
1849 1 running on a host. This allows tools like "bootctl" to operate
1850 correctly within such containers, in order to make container images
1851 bootable on physical systems.
1853 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
1855 * Two new user session targets have been added to support running
1856 graphical sessions under the systemd --user instance:
1857 graphical-session.target and graphical-session-pre.target. See
1858 systemd.special(7) for a description of how those targets should be
1861 * The vconsole initialization code has been significantly reworked to
1862 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
1863 support unicode keymaps. Font and keymap configuration will now be
1864 copied to all allocated virtual consoles.
1866 * FreeBSD's bhyve virtualization is now detected.
1868 * Information recorded in the journal for core dumps now includes the
1869 contents of /proc/mountinfo and the command line of the process at
1870 the top of the process hierarchy (which is usually the init process
1873 * systemd-journal-gatewayd learned the --directory= option to serve
1874 files from the specified location.
1876 * journalctl --root=… can be used to peruse the journal in the
1877 /var/log/ directories inside of a container tree. This is similar to
1878 the existing --machine= option, but does not require the container to
1881 * The hardware database has been extended to support
1882 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
1885 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
1886 specify the click rate for mice which include a horizontal wheel with
1887 a click rate that is different than the one for the vertical wheel.
1889 * systemd-run gained a new --wait option that makes service execution
1890 synchronous. (Specifically, the command will not return until the
1891 specified service binary exited.)
1893 * systemctl gained a new --wait option that causes the start command to
1894 wait until the units being started have terminated again.
1896 * A new journal output mode "short-full" has been added which displays
1897 timestamps with abbreviated English day names and adds a timezone
1898 suffix. Those timestamps include more information than the default
1899 "short" output mode, and can be passed directly to journalctl's
1900 --since= and --until= options.
1902 * /etc/resolv.conf will be bind-mounted into containers started by
1903 systemd-nspawn, if possible, so any changes to resolv.conf contents
1904 are automatically propagated to the container.
1906 * The number of instances for socket-activated services originating
1907 from a single IP address can be limited with
1908 MaxConnectionsPerSource=, extending the existing setting of
1911 * systemd-networkd gained support for vcan ("Virtual CAN") interface
1914 * .netdev and .network configuration can now be extended through
1917 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
1918 Segmentation Offload, Generic Receive Offload, Large Receive Offload
1919 can be enabled and disabled using the new UDPSegmentationOffload=,
1920 TCPSegmentationOffload=, GenericSegmentationOffload=,
1921 GenericReceiveOffload=, LargeReceiveOffload= options in the
1922 [Link] section of .link files.
1924 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
1925 Port VLAN ID can be configured for bridge devices using the new STP=,
1926 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
1927 section of .netdev files.
1929 * The route table to which routes received over DHCP or RA should be
1930 added can be configured with the new RouteTable= option in the [DHCP]
1931 and [IPv6AcceptRA] sections of .network files.
1933 * The Address Resolution Protocol can be disabled on links managed by
1934 systemd-networkd using the ARP=no setting in the [Link] section of
1937 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
1938 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
1939 encode information about the result and exit codes of the current
1940 service runtime cycle.
1942 * systemd-sysctl will now configure kernel parameters in the order
1943 they occur in the configuration files. This matches what sysctl
1944 has been traditionally doing.
1946 * kernel-install "plugins" that are executed to perform various
1947 tasks after a new kernel is added and before an old one is removed
1948 can now return a special value to terminate the procedure and
1949 prevent any later plugins from running.
1951 * Journald's SplitMode=login setting has been deprecated. It has been
1952 removed from documentation, and its use is discouraged. In a future
1953 release it will be completely removed, and made equivalent to current
1954 default of SplitMode=uid.
1956 * Storage=both option setting in /etc/systemd/coredump.conf has been
1957 removed. With fast LZ4 compression storing the core dump twice is not
1960 * The --share-system systemd-nspawn option has been replaced with an
1961 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
1962 this functionality is discouraged. In addition the variables
1963 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
1964 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
1965 individual namespaces.
1967 * "machinectl list" now shows the IP address of running containers in
1968 the output, as well as OS release information.
1970 * "loginctl list" now shows the TTY of each session in the output.
1972 * sd-bus gained new API calls sd_bus_track_set_recursive(),
1973 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
1974 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
1975 tracking objects in a "recursive" mode, where a single client can be
1976 counted multiple times, if it takes multiple references.
1978 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
1979 sd_bus_get_exit_on_disconnect(). They may be used to make a
1980 process using sd-bus automatically exit if the bus connection is
1983 * Bus clients of the service manager may now "pin" loaded units into
1984 memory, by taking an explicit reference on them. This is useful to
1985 ensure the client can retrieve runtime data about the service even
1986 after the service completed execution. Taking such a reference is
1987 available only for privileged clients and should be helpful to watch
1988 running services in a race-free manner, and in particular collect
1989 information about exit statuses and results.
1991 * The nss-resolve module has been changed to strictly return UNAVAIL
1992 when communication via D-Bus with resolved failed, and NOTFOUND when
1993 a lookup completed but was negative. This means it is now possible to
1994 neatly configure fallbacks using nsswitch.conf result checking
1995 expressions. Taking benefit of this, the new recommended
1996 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
1998 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
2000 * A new setting CtrlAltDelBurstAction= has been added to
2001 /etc/systemd/system.conf which may be used to configure the precise
2002 behaviour if the user on the console presses Ctrl-Alt-Del more often
2003 than 7 times in 2s. Previously this would unconditionally result in
2004 an expedited, immediate reboot. With this new setting the precise
2005 operation may be configured in more detail, and also turned off
2008 * In .netdev files two new settings RemoteChecksumTx= and
2009 RemoteChecksumRx= are now understood that permit configuring the
2010 remote checksumming logic for VXLAN networks.
2012 * The service manager learnt a new "invocation ID" concept for invoked
2013 services. Each runtime cycle of a service will get a new invocation
2014 ID (a 128bit random UUID) assigned that identifies the current
2015 run of the service uniquely and globally. A new invocation ID
2016 is generated each time a service starts up. The journal will store
2017 the invocation ID of a service along with any logged messages, thus
2018 making the invocation ID useful for matching the online runtime of a
2019 service with the offline log data it generated in a safe way without
2020 relying on synchronized timestamps. In many ways this new service
2021 invocation ID concept is similar to the kernel's boot ID concept that
2022 uniquely and globally identifies the runtime of each boot. The
2023 invocation ID of a service is passed to the service itself via an
2024 environment variable ($INVOCATION_ID). A new bus call
2025 GetUnitByInvocationID() has been added that is similar to GetUnit()
2026 but instead of retrieving the bus path for a unit by its name
2027 retrieves it by its invocation ID. The returned path is valid only as
2028 long as the passed invocation ID is current.
2030 * systemd-resolved gained a new "DNSStubListener" setting in
2031 resolved.conf. It either takes a boolean value or the special values
2032 "udp" and "tcp", and configures whether to enable the stub DNS
2033 listener on 127.0.0.53:53.
2035 * IP addresses configured via networkd may now carry additional
2036 configuration settings supported by the kernel. New options include:
2037 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
2038 PrefixRoute=, AutoJoin=.
2040 * The PAM configuration fragment file for "user@.service" shipped with
2041 systemd (i.e. the --user instance of systemd) has been stripped to
2042 the minimum necessary to make the system boot. Previously, it
2043 contained Fedora-specific stanzas that did not apply to other
2044 distributions. It is expected that downstream distributions add
2045 additional configuration lines, matching their needs to this file,
2046 using it only as rough template of what systemd itself needs. Note
2047 that this reduced fragment does not even include an invocation of
2048 pam_limits which most distributions probably want to add, even though
2049 systemd itself does not need it. (There's also the new build time
2050 option --with-pamconfdir=no to disable installation of the PAM
2053 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
2054 capability is now also dropped from its set (in addition to
2055 CAP_SYS_MKNOD as before).
2057 * In service unit files it is now possible to connect a specific named
2058 file descriptor with stdin/stdout/stdout of an executed service. The
2059 name may be specified in matching .socket units using the
2060 FileDescriptorName= setting.
2062 * A number of journal settings may now be configured on the kernel
2063 command line. Specifically, the following options are now understood:
2064 systemd.journald.max_level_console=,
2065 systemd.journald.max_level_store=,
2066 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
2067 systemd.journald.max_level_wall=.
2069 * "systemctl is-enabled --full" will now show by which symlinks a unit
2070 file is enabled in the unit dependency tree.
2072 * Support for VeraCrypt encrypted partitions has been added to the
2073 "cryptsetup" logic and /etc/crypttab.
2075 * systemd-detect-virt gained support for a new --private-users switch
2076 that checks whether the invoking processes are running inside a user
2077 namespace. Similar, a new special value "private-users" for the
2078 existing ConditionVirtualization= setting has been added, permitting
2079 skipping of specific units in user namespace environments.
2081 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
2082 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
2083 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
2084 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
2085 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
2086 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
2087 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
2088 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
2089 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
2090 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
2091 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
2092 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
2093 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
2094 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
2095 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
2096 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
2097 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
2098 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
2099 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
2100 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
2101 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
2102 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
2103 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
2104 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
2105 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
2106 Jędrzejewski-Szmek, Zeal Jagannatha
2108 — Santa Fe, 2016-11-03
2112 * In service units the various ExecXYZ= settings have been extended
2113 with an additional special character as first argument of the
2114 assigned value: if the character '+' is used the specified command
2115 line it will be run with full privileges, regardless of User=,
2116 Group=, CapabilityBoundingSet= and similar options. The effect is
2117 similar to the existing PermissionsStartOnly= option, but allows
2118 configuration of this concept for each executed command line
2121 * Services may now alter the service watchdog timeout at runtime by
2122 sending a WATCHDOG_USEC= message via sd_notify().
2124 * MemoryLimit= and related unit settings now optionally take percentage
2125 specifications. The percentage is taken relative to the amount of
2126 physical memory in the system (or in case of containers, the assigned
2127 amount of memory). This allows scaling service resources neatly with
2128 the amount of RAM available on the system. Similarly, systemd-logind's
2129 RuntimeDirectorySize= option now also optionally takes percentage
2132 * In similar fashion TasksMax= takes percentage values now, too. The
2133 value is taken relative to the configured maximum number of processes
2134 on the system. The per-service task maximum has been changed to 15%
2135 using this functionality. (Effectively this is an increase of 512 →
2136 4915 for service units, given the kernel's default pid_max setting.)
2138 * Calendar time specifications in .timer units now understand a ".."
2139 syntax for time ranges. Example: "4..7:10" may now be used for
2140 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
2143 * The InaccessableDirectories=, ReadOnlyDirectories= and
2144 ReadWriteDirectories= unit file settings have been renamed to
2145 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
2146 applied to all kinds of file nodes, and not just directories, with
2147 the exception of symlinks. Specifically these settings may now be
2148 used on block and character device nodes, UNIX sockets and FIFOS as
2149 well as regular files. The old names of these settings remain
2150 available for compatibility.
2152 * systemd will now log about all service processes it kills forcibly
2153 (using SIGKILL) because they remained after the clean shutdown phase
2154 of the service completed. This should help identifying services that
2155 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
2156 systemd-logind's configuration a similar log message is generated for
2157 processes killed at the end of each session due to this setting.
2159 * systemd will now set the $JOURNAL_STREAM environment variable for all
2160 services whose stdout/stderr are connected to the Journal (which
2161 effectively means by default: all services). The variable contains
2162 the device and inode number of the file descriptor used for
2163 stdout/stderr. This may be used by invoked programs to detect whether
2164 their stdout/stderr is connected to the Journal, in which case they
2165 can switch over to direct Journal communication, thus being able to
2166 pass extended, structured metadata along with their log messages. As
2167 one example, this is now used by glib's logging primitives.
2169 * When using systemd's default tmp.mount unit for /tmp, the mount point
2170 will now be established with the "nosuid" and "nodev" options. This
2171 avoids privilege escalation attacks that put traps and exploits into
2172 /tmp. However, this might cause problems if you e. g. put container
2173 images or overlays into /tmp; if you need this, override tmp.mount's
2174 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
2177 * systemd now supports the "memory" cgroup controller also on
2180 * The systemd-cgtop tool now optionally takes a control group path as
2181 command line argument. If specified, the control group list shown is
2182 limited to subgroups of that group.
2184 * The SystemCallFilter= unit file setting gained support for
2185 pre-defined, named system call filter sets. For example
2186 SystemCallFilter=@clock is now an effective way to make all clock
2187 changing-related system calls unavailable to a service. A number of
2188 similar pre-defined groups are defined. Writing system call filters
2189 for system services is simplified substantially with this new
2190 concept. Accordingly, all of systemd's own, long-running services now
2191 enable system call filtering based on this, by default.
2193 * A new service setting MemoryDenyWriteExecute= has been added, taking
2194 a boolean value. If turned on, a service may no longer create memory
2195 mappings that are writable and executable at the same time. This
2196 enhances security for services where this is enabled as it becomes
2197 harder to dynamically write and then execute memory in exploited
2198 service processes. This option has been enabled for all of systemd's
2199 own long-running services.
2201 * A new RestrictRealtime= service setting has been added, taking a
2202 boolean argument. If set the service's processes may no longer
2203 acquire realtime scheduling. This improves security as realtime
2204 scheduling may otherwise be used to easily freeze the system.
2206 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
2207 value. This may be used for requesting that the system manager inside
2208 of the container reports start-up completion to nspawn which then
2209 propagates this notification further to the service manager
2210 supervising nspawn itself. A related option NotifyReady= in .nspawn
2211 files has been added too. This functionality allows ordering of the
2212 start-up of multiple containers using the usual systemd ordering
2215 * machinectl gained a new command "stop" that is an alias for
2218 * systemd-resolved gained support for contacting DNS servers on
2219 link-local IPv6 addresses.
2221 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
2222 its caches. A method call for requesting the same operation has been
2223 added to the bus API too, and is made available via "systemd-resolve
2226 * systemd-resolve gained a new --status switch. If passed a brief
2227 summary of the used DNS configuration with per-interface information
2230 * resolved.conf gained a new Cache= boolean option, defaulting to
2231 on. If turned off local DNS caching is disabled. This comes with a
2232 performance penalty in particular when DNSSEC is enabled. Note that
2233 resolved disables its internal caching implicitly anyway, when the
2234 configured DNS server is on a host-local IP address such as ::1 or
2235 127.0.0.1, thus automatically avoiding double local caching.
2237 * systemd-resolved now listens on the local IP address 127.0.0.53:53
2238 for DNS requests. This improves compatibility with local programs
2239 that do not use the libc NSS or systemd-resolved's bus APIs for name
2240 resolution. This minimal DNS service is only available to local
2241 programs and does not implement the full DNS protocol, but enough to
2242 cover local DNS clients. A new, static resolv.conf file, listing just
2243 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
2244 now recommended to make /etc/resolv.conf a symlink to this file in
2245 order to route all DNS lookups to systemd-resolved, regardless if
2246 done via NSS, the bus API or raw DNS packets. Note that this local
2247 DNS service is not as fully featured as the libc NSS or
2248 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
2249 used to deliver link-local address information (as this implies
2250 sending a local interface index along), LLMNR/mDNS support via this
2251 interface is severely restricted. It is thus strongly recommended for
2252 all applications to use the libc NSS API or native systemd-resolved
2255 * systemd-networkd's bridge support learned a new setting
2256 VLANFiltering= for controlling VLAN filtering. Moreover a new section
2257 in .network files has been added for configuring VLAN bridging in
2258 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
2260 * systemd-networkd's IPv6 Router Advertisement code now makes use of
2261 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
2262 now be acquired without relying on DHCPv6. Two new options
2263 UseDomains= and UseDNS= have been added to configure this behaviour.
2265 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
2266 renamed IPv6AcceptRA=, without altering its behaviour. The old
2267 setting name remains available for compatibility reasons.
2269 * The systemd-networkd VTI/VTI6 tunneling support gained new options
2270 Key=, InputKey= and OutputKey=.
2272 * systemd-networkd gained support for VRF ("Virtual Routing Function")
2273 interface configuration.
2275 * "systemctl edit" may now be used to create new unit files by
2276 specifying the --force switch.
2278 * sd-event gained a new function sd_event_get_iteration() for
2279 requesting the current iteration counter of the event loop. It starts
2280 at zero and is increased by one with each event loop iteration.
2282 * A new rpm macro %systemd_ordering is provided by the macros.systemd
2283 file. It can be used in lieu of %systemd_requires in packages which
2284 don't use any systemd functionality and are intended to be installed
2285 in minimal containers without systemd present. This macro provides
2286 ordering dependencies to ensure that if the package is installed in
2287 the same rpm transaction as systemd, systemd will be installed before
2288 the scriptlets for the package are executed, allowing unit presets
2291 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
2292 been added to simplify packaging of generators.
2294 * The os-release file gained VERSION_CODENAME field for the
2295 distribution nickname (e.g. VERSION_CODENAME=woody).
2297 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
2298 can be set to disable parsing of metadata and the creation
2299 of persistent symlinks for that device.
2301 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
2302 to make them available to logged-in users has been reverted.
2304 * Much of the common code of the various systemd components is now
2305 built into an internal shared library libsystemd-shared-231.so
2306 (incorporating the systemd version number in the name, to be updated
2307 with future releases) that the components link to. This should
2308 decrease systemd footprint both in memory during runtime and on
2309 disk. Note that the shared library is not for public use, and is
2310 neither API nor ABI stable, but is likely to change with every new
2311 released update. Packagers need to make sure that binaries
2312 linking to libsystemd-shared.so are updated in step with the
2315 * Configuration for "mkosi" is now part of the systemd
2316 repository. mkosi is a tool to easily build legacy-free OS images,
2317 and is available on github: https://github.com/systemd/mkosi. If
2318 "mkosi" is invoked in the build tree a new raw OS image is generated
2319 incorporating the systemd sources currently being worked on and a
2320 clean, fresh distribution installation. The generated OS image may be
2321 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
2322 UEFI PC. This functionality is particularly useful to easily test
2323 local changes made to systemd in a pristine, defined environment. See
2324 doc/HACKING for details.
2326 * configure learned the --with-support-url= option to specify the
2327 distribution's bugtracker.
2329 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
2330 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
2331 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
2332 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
2333 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
2334 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
2335 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
2336 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
2337 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
2338 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
2339 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
2340 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
2341 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
2342 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
2343 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
2344 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
2345 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
2346 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
2347 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
2349 — Berlin, 2016-07-25
2353 * DNSSEC is now turned on by default in systemd-resolved (in
2354 "allow-downgrade" mode), but may be turned off during compile time by
2355 passing "--with-default-dnssec=no" to "configure" (and of course,
2356 during runtime with DNSSEC= in resolved.conf). We recommend
2357 downstreams to leave this on at least during development cycles and
2358 report any issues with the DNSSEC logic upstream. We are very
2359 interested in collecting feedback about the DNSSEC validator and its
2360 limitations in the wild. Note however, that DNSSEC support is
2361 probably nothing downstreams should turn on in stable distros just
2362 yet, as it might create incompatibilities with a few DNS servers and
2363 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
2364 automatically whenever we detect such incompatible setups, but there
2365 might be systems we do not cover yet. Hence: please help us testing
2366 the DNSSEC code, leave this on where you can, report back, but then
2367 again don't consider turning this on in your stable, LTS or
2368 production release just yet. (Note that you have to enable
2369 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
2370 and its DNSSEC mode for host name resolution from local
2373 * systemd-resolve conveniently resolves DANE records with the --tlsa
2374 option and OPENPGPKEY records with the --openpgp option. It also
2375 supports dumping raw DNS record data via the new --raw= switch.
2377 * systemd-logind will now by default terminate user processes that are
2378 part of the user session scope unit (session-XX.scope) when the user
2379 logs out. This behavior is controlled by the KillUserProcesses=
2380 setting in logind.conf, and the previous default of "no" is now
2381 changed to "yes". This means that user sessions will be properly
2382 cleaned up after, but additional steps are necessary to allow
2383 intentionally long-running processes to survive logout.
2385 While the user is logged in at least once, user@.service is running,
2386 and any service that should survive the end of any individual login
2387 session can be started at a user service or scope using systemd-run.
2388 systemd-run(1) man page has been extended with an example which shows
2389 how to run screen in a scope unit underneath user@.service. The same
2390 command works for tmux.
2392 After the user logs out of all sessions, user@.service will be
2393 terminated too, by default, unless the user has "lingering" enabled.
2394 To effectively allow users to run long-term tasks even if they are
2395 logged out, lingering must be enabled for them. See loginctl(1) for
2396 details. The default polkit policy was modified to allow users to
2397 set lingering for themselves without authentication.
2399 Previous defaults can be restored at compile time by the
2400 --without-kill-user-processes option to "configure".
2402 * systemd-logind gained new configuration settings SessionsMax= and
2403 InhibitorsMax=, both with a default of 8192. It will not register new
2404 user sessions or inhibitors above this limit.
2406 * systemd-logind will now reload configuration on SIGHUP.
2408 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
2409 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
2410 enable. Also, support for the "io" cgroup controller in the unified
2411 hierarchy has been added, so that the "memory", "pids" and "io" are
2412 now the controllers that are supported on the unified hierarchy.
2414 WARNING: it is not possible to use previous systemd versions with
2415 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
2416 is necessary to also update systemd in the initramfs if using the
2417 unified hierarchy. An updated SELinux policy is also required.
2419 * LLDP support has been extended, and both passive (receive-only) and
2420 active (sender) modes are supported. Passive mode ("routers-only") is
2421 enabled by default in systemd-networkd. Active LLDP mode is enabled
2422 by default for containers on the internal network. The "networkctl
2423 lldp" command may be used to list information gathered. "networkctl
2424 status" will also show basic LLDP information on connected peers now.
2426 * The IAID and DUID unique identifier sent in DHCP requests may now be
2427 configured for the system and each .network file managed by
2428 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
2430 * systemd-networkd gained support for configuring proxy ARP support for
2431 each interface, via the ProxyArp= setting in .network files. It also
2432 gained support for configuring the multicast querier feature of
2433 bridge devices, via the new MulticastQuerier= setting in .netdev
2434 files. Similarly, snooping on the IGMP traffic can be controlled
2435 via the new setting MulticastSnooping=.
2437 A new setting PreferredLifetime= has been added for addresses
2438 configured in .network file to configure the lifetime intended for an
2441 The systemd-networkd DHCP server gained the option EmitRouter=, which
2442 defaults to yes, to configure whether the DHCP Option 3 (Router)
2445 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
2446 systemd-socket-activate and installed into /usr/bin. It is now fully
2449 * systemd-journald now uses separate threads to flush changes to disk
2450 when closing journal files, thus reducing impact of slow disk I/O on
2451 logging performance.
2453 * The sd-journal API gained two new calls
2454 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
2455 can be used to open journal files using file descriptors instead of
2456 file or directory paths. sd_journal_open_container() has been
2457 deprecated, sd_journal_open_directory_fd() should be used instead
2458 with the flag SD_JOURNAL_OS_ROOT.
2460 * journalctl learned a new output mode "-o short-unix" that outputs log
2461 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
2462 UTC). It also gained support for a new --no-hostname setting to
2463 suppress the hostname column in the family of "short" output modes.
2465 * systemd-ask-password now optionally skips printing of the password to
2466 stdout with --no-output which can be useful in scripts.
2468 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
2469 (devices tagged with ID_MAKER_TOOL) are now tagged with
2470 "uaccess" and are available to logged in users.
2472 * The DeviceAllow= unit setting now supports specifiers (with "%").
2474 * "systemctl show" gained a new --value switch, which allows print a
2475 only the contents of a specific unit property, without also printing
2476 the property's name. Similar support was added to "show*" verbs
2477 of loginctl and machinectl that output "key=value" lists.
2479 * A new unit type "generated" was added for files dynamically generated
2480 by generator tools. Similarly, a new unit type "transient" is used
2481 for unit files created using the runtime API. "systemctl enable" will
2482 refuse to operate on such files.
2484 * A new command "systemctl revert" has been added that may be used to
2485 revert to the vendor version of a unit file, in case local changes
2486 have been made by adding drop-ins or overriding the unit file.
2488 * "machinectl clean" gained a new verb to automatically remove all or
2489 just hidden container images.
2491 * systemd-tmpfiles gained support for a new line type "e" for emptying
2492 directories, if they exist, without creating them if they don't.
2494 * systemd-nspawn gained support for automatically patching the UID/GIDs
2495 of the owners and the ACLs of all files and directories in a
2496 container tree to match the UID/GID user namespacing range selected
2497 for the container invocation. This mode is enabled via the new
2498 --private-users-chown switch. It also gained support for
2499 automatically choosing a free, previously unused UID/GID range when
2500 starting a container, via the new --private-users=pick setting (which
2501 implies --private-users-chown). Together, these options for the first
2502 time make user namespacing for nspawn containers fully automatic and
2503 thus deployable. The systemd-nspawn@.service template unit file has
2504 been changed to use this functionality by default.
2506 * systemd-nspawn gained a new --network-zone= switch, that allows
2507 creating ad-hoc virtual Ethernet links between multiple containers,
2508 that only exist as long as at least one container referencing them is
2509 running. This allows easy connecting of multiple containers with a
2510 common link that implements an Ethernet broadcast domain. Each of
2511 these network "zones" may be named relatively freely by the user, and
2512 may be referenced by any number of containers, but each container may
2513 only reference one of these "zones". On the lower level, this is
2514 implemented by an automatically managed bridge network interface for
2515 each zone, that is created when the first container referencing its
2516 zone is created and removed when the last one referencing its zone
2519 * The default start timeout may now be configured on the kernel command
2520 line via systemd.default_timeout_start_sec=. It was already
2521 configurable via the DefaultTimeoutStartSec= option in
2522 /etc/systemd/system.conf.
2524 * Socket units gained a new TriggerLimitIntervalSec= and
2525 TriggerLimitBurst= setting to configure a limit on the activation
2526 rate of the socket unit.
2528 * The LimitNICE= setting now optionally takes normal UNIX nice values
2529 in addition to the raw integer limit value. If the specified
2530 parameter is prefixed with "+" or "-" and is in the range -20..19 the
2531 value is understood as UNIX nice value. If not prefixed like this it
2532 is understood as raw RLIMIT_NICE limit.
2534 * Note that the effect of the PrivateDevices= unit file setting changed
2535 slightly with this release: the per-device /dev file system will be
2536 mounted read-only from this version on, and will have "noexec"
2537 set. This (minor) change of behavior might cause some (exceptional)
2538 legacy software to break, when PrivateDevices=yes is set for its
2539 service. Please leave PrivateDevices= off if you run into problems
2542 * systemd-bootchart has been split out to a separate repository:
2543 https://github.com/systemd/systemd-bootchart
2545 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
2546 merged into the kernel in its current form.
2548 * The compatibility libraries libsystemd-daemon.so,
2549 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
2550 which have been deprecated since systemd-209 have been removed along
2551 with the corresponding pkg-config files. All symbols provided by
2552 those libraries are provided by libsystemd.so.
2554 * The Capabilities= unit file setting has been removed (it is ignored
2555 for backwards compatibility). AmbientCapabilities= and
2556 CapabilityBoundingSet= should be used instead.
2558 * A new special target has been added, initrd-root-device.target,
2559 which creates a synchronization point for dependencies of the root
2560 device in early userspace. Initramfs builders must ensure that this
2561 target is now included in early userspace.
2563 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
2564 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
2565 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
2566 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
2567 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
2568 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
2569 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
2570 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
2571 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
2572 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
2573 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
2574 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
2575 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
2576 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
2577 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
2578 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
2579 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
2580 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
2581 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
2582 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
2583 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
2584 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
2585 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
2586 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
2587 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
2590 — Fairfax, 2016-05-21
2594 * The systemd-resolved DNS resolver service has gained a substantial
2595 set of new features, most prominently it may now act as a DNSSEC
2596 validating stub resolver. DNSSEC mode is currently turned off by
2597 default, but is expected to be turned on by default in one of the
2598 next releases. For now, we invite everybody to test the DNSSEC logic
2599 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
2600 service also gained a full set of D-Bus interfaces, including calls
2601 to configure DNS and DNSSEC settings per link (for use by external
2602 network management software). systemd-resolved and systemd-networkd
2603 now distinguish between "search" and "routing" domains. The former
2604 are used to qualify single-label names, the latter are used purely
2605 for routing lookups within certain domains to specific links.
2606 resolved now also synthesizes RRs for all entries from /etc/hosts.
2608 * The systemd-resolve tool (which is a client utility for
2609 systemd-resolved) has been improved considerably and is now fully
2610 supported and documented. Hence it has moved from /usr/lib/systemd to
2613 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
2616 * The coredump collection logic has been reworked: when a coredump is
2617 collected it is now written to disk, compressed and processed
2618 (including stacktrace extraction) from a new instantiated service
2619 systemd-coredump@.service, instead of directly from the
2620 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
2621 processing large coredumps can take up a substantial amount of
2622 resources and time, and this previously happened entirely outside of
2623 systemd's service supervision. With the new logic the core_pattern
2624 hook only does minimal metadata collection before passing off control
2625 to the new instantiated service, which is configured with a time
2626 limit, a nice level and other settings to minimize negative impact on
2627 the rest of the system. Also note that the new logic will honour the
2628 RLIMIT_CORE setting of the crashed process, which now allows users
2629 and processes to turn off coredumping for their processes by setting
2632 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
2633 and all forked processes by default. Previously, PID 1 would leave
2634 the setting at "0" for all processes, as set by the kernel. Note that
2635 the resource limit traditionally has no effect on the generated
2636 coredumps on the system if the /proc/sys/kernel/core_pattern hook
2637 logic is used. Since the limit is now honoured (see above) its
2638 default has been changed so that the coredumping logic is enabled by
2639 default for all processes, while allowing specific opt-out.
2641 * When the stacktrace is extracted from processes of system users, this
2642 is now done as "systemd-coredump" user, in order to sandbox this
2643 potentially security sensitive parsing operation. (Note that when
2644 processing coredumps of normal users this is done under the user ID
2645 of process that crashed, as before.) Packagers should take notice
2646 that it is now necessary to create the "systemd-coredump" system user
2647 and group at package installation time.
2649 * The systemd-activate socket activation testing tool gained support
2650 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
2651 and --seqpacket switches. It also has been extended to support both
2652 new-style and inetd-style file descriptor passing. Use the new
2653 --inetd switch to request inetd-style file descriptor passing.
2655 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
2656 variable, which takes a boolean value. If set to false, ANSI color
2657 output is disabled in the tools even when run on a terminal that
2660 * The VXLAN support in networkd now supports two new settings
2661 DestinationPort= and PortRange=.
2663 * A new systemd.machine_id= kernel command line switch has been added,
2664 that may be used to set the machine ID in /etc/machine-id if it is
2665 not initialized yet. This command line option has no effect if the
2666 file is already initialized.
2668 * systemd-nspawn gained a new --as-pid2 switch that invokes any
2669 specified command line as PID 2 rather than PID 1 in the
2670 container. In this mode PID 1 is a minimal stub init process that
2671 implements the special POSIX and Linux semantics of PID 1 regarding
2672 signal and child process management. Note that this stub init process
2673 is implemented in nspawn itself and requires no support from the
2674 container image. This new logic is useful to support running
2675 arbitrary commands in the container, as normal processes are
2676 generally not prepared to run as PID 1.
2678 * systemd-nspawn gained a new --chdir= switch for setting the current
2679 working directory for the process started in the container.
2681 * "journalctl /dev/sda" will now output all kernel log messages for
2682 specified device from the current boot, in addition to all devices
2683 that are parents of it. This should make log output about devices
2684 pretty useful, as long as kernel drivers attach enough metadata to
2685 the log messages. (The usual SATA drivers do.)
2687 * The sd-journal API gained two new calls
2688 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
2689 that report whether log data from /run or /var has been found.
2691 * journalctl gained a new switch "--fields" that prints all journal
2692 record field names currently in use in the journal. This is backed
2693 by two new sd-journal API calls sd_journal_enumerate_fields() and
2694 sd_journal_restart_fields().
2696 * Most configurable timeouts in systemd now expect an argument of
2697 "infinity" to turn them off, instead of "0" as before. The semantics
2698 from now on is that a timeout of "0" means "now", and "infinity"
2699 means "never". To maintain backwards compatibility, "0" continues to
2700 turn off previously existing timeout settings.
2702 * "systemctl reload-or-try-restart" has been renamed to "systemctl
2703 try-reload-or-restart" to clarify what it actually does: the "try"
2704 logic applies to both reloading and restarting, not just restarting.
2705 The old name continues to be accepted for compatibility.
2707 * On boot-up, when PID 1 detects that the system clock is behind the
2708 release date of the systemd version in use, the clock is now set
2709 to the latter. Previously, this was already done in timesyncd, in order
2710 to avoid running with clocks set to the various clock epochs such as
2711 1902, 1938 or 1970. With this change the logic is now done in PID 1
2712 in addition to timesyncd during early boot-up, so that it is enforced
2713 before the first process is spawned by systemd. Note that the logic
2714 in timesyncd remains, as it is more comprehensive and ensures
2715 clock monotonicity by maintaining a persistent timestamp file in
2716 /var. Since /var is generally not available in earliest boot or the
2717 initrd, this part of the logic remains in timesyncd, and is not done
2720 * Support for tweaking details in net_cls.class_id through the
2721 NetClass= configuration directive has been removed, as the kernel
2722 people have decided to deprecate that controller in cgroup v2.
2723 Userspace tools such as nftables are moving over to setting rules
2724 that are specific to the full cgroup path of a task, which obsoletes
2725 these controllers anyway. The NetClass= directive is kept around for
2726 legacy compatibility reasons. For a more in-depth description of the
2727 kernel change, please refer to the respective upstream commit:
2729 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
2731 * A new service setting RuntimeMaxSec= has been added that may be used
2732 to specify a maximum runtime for a service. If the timeout is hit, the
2733 service is terminated and put into a failure state.
2735 * A new service setting AmbientCapabilities= has been added. It allows
2736 configuration of additional Linux process capabilities that are
2737 passed to the activated processes. This is only available on very
2740 * The process resource limit settings in service units may now be used
2741 to configure hard and soft limits individually.
2743 * The various libsystemd APIs such as sd-bus or sd-event now publicly
2744 expose support for gcc's __attribute__((cleanup())) C extension.
2745 Specifically, for many object destructor functions alternative
2746 versions have been added that have names suffixed with "p" and take a
2747 pointer to a pointer to the object to destroy, instead of just a
2748 pointer to the object itself. This is useful because these destructor
2749 functions may be used directly as parameters to the cleanup
2750 construct. Internally, systemd has been a heavy user of this GCC
2751 extension for a long time, and with this change similar support is
2752 now available to consumers of the library outside of systemd. Note
2753 that by using this extension in your sources compatibility with old
2754 and strictly ANSI compatible C compilers is lost. However, all gcc or
2755 LLVM versions of recent years support this extension.
2757 * Timer units gained support for a new setting RandomizedDelaySec= that
2758 allows configuring some additional randomized delay to the configured
2759 time. This is useful to spread out timer events to avoid load peaks in
2760 clusters or larger setups.
2762 * Calendar time specifications now support sub-second accuracy.
2764 * Socket units now support listening on SCTP and UDP-lite protocol
2767 * The sd-event API now comes with a full set of man pages.
2769 * Older versions of systemd contained experimental support for
2770 compressing journal files and coredumps with the LZ4 compressor that
2771 was not compatible with the lz4 binary (due to API limitations of the
2772 lz4 library). This support has been removed; only support for files
2773 compatible with the lz4 binary remains. This LZ4 logic is now
2774 officially supported and no longer considered experimental.
2776 * The dkr image import logic has been removed again from importd. dkr's
2777 micro-services focus doesn't fit into the machine image focus of
2778 importd, and quickly got out of date with the upstream dkr API.
2780 * Creation of the /run/lock/lockdev/ directory was dropped from
2781 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
2782 been available for many years. If you still need this, you need to
2783 create your own tmpfiles.d config file with:
2785 d /run/lock/lockdev 0775 root lock -
2787 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
2788 and RebootArgument= have been moved from the [Service] section of
2789 unit files to [Unit], and they are now supported on all unit types,
2790 not just service units. Of course, elogind will continue to
2791 understand these settings also at the old location, in order to
2792 maintain compatibility.
2794 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
2795 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
2796 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
2797 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
2798 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
2799 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
2800 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
2801 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
2802 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
2803 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
2804 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
2805 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
2806 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
2807 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
2808 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
2809 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
2810 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
2811 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
2812 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2814 — Berlin, 2016-02-11
2818 * A number of properties previously only settable in unit
2819 files are now also available as properties to set when
2820 creating transient units programmatically via the bus, as it
2821 is exposed with systemd-run's --property=
2822 setting. Specifically, these are: SyslogIdentifier=,
2823 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
2824 EnvironmentFile=, ReadWriteDirectories=,
2825 ReadOnlyDirectories=, InaccessibleDirectories=,
2826 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
2828 * When creating transient services via the bus API it is now
2829 possible to pass in a set of file descriptors to use as
2830 STDIN/STDOUT/STDERR for the invoked process.
2832 * Slice units may now be created transiently via the bus APIs,
2833 similar to the way service and scope units may already be
2834 created transiently.
2836 * Wherever systemd expects a calendar timestamp specification
2837 (like in journalctl's --since= and --until= switches) UTC
2838 timestamps are now supported. Timestamps suffixed with "UTC"
2839 are now considered to be in Universal Time Coordinated
2840 instead of the local timezone. Also, timestamps may now
2841 optionally be specified with sub-second accuracy. Both of
2842 these additions also apply to recurring calendar event
2843 specification, such as OnCalendar= in timer units.
2845 * journalctl gained a new "--sync" switch that asks the
2846 journal daemon to write all so far unwritten log messages to
2847 disk and sync the files, before returning.
2849 * systemd-tmpfiles learned two new line types "q" and "Q" that
2850 operate like "v", but also set up a basic btrfs quota
2851 hierarchy when used on a btrfs file system with quota
2854 * tmpfiles' "v", "q" and "Q" will now create a plain directory
2855 instead of a subvolume (even on a btrfs file system) if the
2856 root directory is a plain directory, and not a
2857 subvolume. This should simplify things with certain chroot()
2858 environments which are not aware of the concept of btrfs
2861 * systemd-detect-virt gained a new --chroot switch to detect
2862 whether execution takes place in a chroot() environment.
2864 * CPUAffinity= now takes CPU index ranges in addition to
2867 * The various memory-related resource limit settings (such as
2868 LimitAS=) now understand the usual K, M, G, ... suffixes to
2869 the base of 1024 (IEC). Similar, the time-related resource
2870 limit settings understand the usual min, h, day, ...
2873 * There's a new system.conf setting DefaultTasksMax= to
2874 control the default TasksMax= setting for services and
2875 scopes running on the system. (TasksMax= is the primary
2876 setting that exposes the "pids" cgroup controller on systemd
2877 and was introduced in the previous systemd release.) The
2878 setting now defaults to 512, which means services that are
2879 not explicitly configured otherwise will only be able to
2880 create 512 processes or threads at maximum, from this
2881 version on. Note that this means that thread- or
2882 process-heavy services might need to be reconfigured to set
2883 TasksMax= to a higher value. It is sufficient to set
2884 TasksMax= in these specific unit files to a higher value, or
2885 even "infinity". Similar, there's now a logind.conf setting
2886 UserTasksMax= that defaults to 4096 and limits the total
2887 number of processes or tasks each user may own
2888 concurrently. nspawn containers also have the TasksMax=
2889 value set by default now, to 8192. Note that all of this
2890 only has an effect if the "pids" cgroup controller is
2891 enabled in the kernel. The general benefit of these changes
2892 should be a more robust and safer system, that provides a
2893 certain amount of per-service fork() bomb protection.
2895 * systemd-nspawn gained the new --network-veth-extra= switch
2896 to define additional and arbitrarily-named virtual Ethernet
2897 links between the host and the container.
2899 * A new service execution setting PassEnvironment= has been
2900 added that allows importing select environment variables
2901 from PID1's environment block into the environment block of
2904 * Timer units gained support for a new RemainAfterElapse=
2905 setting which takes a boolean argument. It defaults to on,
2906 exposing behaviour unchanged to previous releases. If set to
2907 off, timer units are unloaded after they elapsed if they
2908 cannot elapse again. This is particularly useful for
2909 transient timer units, which shall not stay around longer
2910 than until they first elapse.
2912 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
2913 default now (the kernel default is 16). This is beneficial
2914 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
2915 allows substantially larger numbers of queued
2916 datagrams. This should increase the capability of systemd to
2917 parallelize boot-up, as logging and sd_notify() are unlikely
2918 to stall execution anymore. If you need to change the value
2919 from the new defaults, use the usual sysctl.d/ snippets.
2921 * The compression framing format used by the journal or
2922 coredump processing has changed to be in line with what the
2923 official LZ4 tools generate. LZ4 compression support in
2924 systemd was considered unsupported previously, as the format
2925 was not compatible with the normal tools. With this release
2926 this has changed now, and it is hence safe for downstream
2927 distributions to turn it on. While not compressing as well
2928 as the XZ, LZ4 is substantially faster, which makes
2929 it a good default choice for the compression logic in the
2930 journal and in coredump handling.
2932 * Any reference to /etc/mtab has been dropped from
2933 systemd. The file has been obsolete since a while, but
2934 systemd refused to work on systems where it was incorrectly
2935 set up (it should be a symlink or non-existent). Please make
2936 sure to update to util-linux 2.27.1 or newer in conjunction
2937 with this systemd release, which also drops any reference to
2938 /etc/mtab. If you maintain a distribution make sure that no
2939 software you package still references it, as this is a
2940 likely source of bugs. There's also a glibc bug pending,
2941 asking for removal of any reference to this obsolete file:
2943 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
2945 Note that only util-linux versions built with
2946 --enable-libmount-force-mountinfo are supported.
2948 * Support for the ".snapshot" unit type has been removed. This
2949 feature turned out to be little useful and little used, and
2950 has now been removed from the core and from systemctl.
2952 * The dependency types RequiresOverridable= and
2953 RequisiteOverridable= have been removed from systemd. They
2954 have been used only very sparingly to our knowledge and
2955 other options that provide a similar effect (such as
2956 systemctl --mode=ignore-dependencies) are much more useful
2957 and commonly used. Moreover, they were only half-way
2958 implemented as the option to control behaviour regarding
2959 these dependencies was never added to systemctl. By removing
2960 these dependency types the execution engine becomes a bit
2961 simpler. Unit files that use these dependencies should be
2962 changed to use the non-Overridable dependency types
2963 instead. In fact, when parsing unit files with these
2964 options, that's what systemd will automatically convert them
2965 too, but it will also warn, asking users to fix the unit
2966 files accordingly. Removal of these dependency types should
2967 only affect a negligible number of unit files in the wild.
2969 * Behaviour of networkd's IPForward= option changed
2970 (again). It will no longer maintain a per-interface setting,
2971 but propagate one way from interfaces where this is enabled
2972 to the global kernel setting. The global setting will be
2973 enabled when requested by a network that is set up, but
2974 never be disabled again. This change was made to make sure
2975 IPv4 and IPv6 behaviour regarding packet forwarding is
2976 similar (as the Linux IPv6 stack does not support
2977 per-interface control of this setting) and to minimize
2980 * In unit files the behaviour of %u, %U, %h, %s has
2981 changed. These specifiers will now unconditionally resolve
2982 to the various user database fields of the user that the
2983 systemd instance is running as, instead of the user
2984 configured in the specific unit via User=. Note that this
2985 effectively doesn't change much, as resolving of these
2986 specifiers was already turned off in the --system instance
2987 of systemd, as we cannot do NSS lookups from PID 1. In the
2988 --user instance of systemd these specifiers where correctly
2989 resolved, but hardly made any sense, since the user instance
2990 lacks privileges to do user switches anyway, and User= is
2991 hence useless. Moreover, even in the --user instance of
2992 systemd behaviour was awkward as it would only take settings
2993 from User= assignment placed before the specifier into
2994 account. In order to unify and simplify the logic around
2995 this the specifiers will now always resolve to the
2996 credentials of the user invoking the manager (which in case
2997 of PID 1 is the root user).
2999 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
3000 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
3001 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
3002 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
3003 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
3004 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
3005 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
3006 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
3007 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
3008 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
3009 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
3010 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
3011 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
3012 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
3015 — Berlin, 2015-11-18
3019 * systemd now depends on util-linux v2.27. More specifically,
3020 the newly added mount monitor feature in libmount now
3021 replaces systemd's former own implementation.
3023 * libmount mandates /etc/mtab not to be regular file, and
3024 systemd now enforces this condition at early boot.
3025 /etc/mtab has been deprecated and warned about for a very
3026 long time, so systems running systemd should already have
3027 stopped having this file around as anything else than a
3028 symlink to /proc/self/mounts.
3030 * Support for the "pids" cgroup controller has been added. It
3031 allows accounting the number of tasks in a cgroup and
3032 enforcing limits on it. This adds two new setting
3033 TasksAccounting= and TasksMax= to each unit, as well as a
3034 global option DefaultTasksAccounting=.
3036 * Support for the "net_cls" cgroup controller has been added.
3037 It allows assigning a net class ID to each task in the
3038 cgroup, which can then be used in firewall rules and traffic
3039 shaping configurations. Note that the kernel netfilter net
3040 class code does not currently work reliably for ingress
3041 packets on unestablished sockets.
3043 This adds a new config directive called NetClass= to CGroup
3044 enabled units. Allowed values are positive numbers for fixed
3045 assignments and "auto" for picking a free value
3048 * 'systemctl is-system-running' now returns 'offline' if the
3049 system is not booted with systemd. This command can now be
3050 used as a substitute for 'systemd-notify --booted'.
3052 * Watchdog timeouts have been increased to 3 minutes for all
3053 in-tree service files. Apparently, disk IO issues are more
3054 frequent than we hoped, and user reported >1 minute waiting
3057 * 'machine-id-commit' functionality has been merged into
3058 'machine-id-setup --commit'. The separate binary has been
3061 * The WorkingDirectory= directive in unit files may now be set
3062 to the special value '~'. In this case, the working
3063 directory is set to the home directory of the user
3064 configured in User=.
3066 * "machinectl shell" will now open the shell in the home
3067 directory of the selected user by default.
3069 * The CrashChVT= configuration file setting is renamed to
3070 CrashChangeVT=, following our usual logic of not
3071 abbreviating unnecessarily. The old directive is still
3072 supported for compat reasons. Also, this directive now takes
3073 an integer value between 1 and 63, or a boolean value. The
3074 formerly supported '-1' value for disabling stays around for
3077 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
3078 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
3079 RootDirectory= properties can now be set for transient
3082 * The systemd-analyze tool gained a new "set-log-target" verb
3083 to change the logging target the system manager logs to
3084 dynamically during runtime. This is similar to how
3085 "systemd-analyze set-log-level" already changes the log
3088 * In nspawn /sys is now mounted as tmpfs, with only a selected
3089 set of subdirectories mounted in from the real sysfs. This
3090 enhances security slightly, and is useful for ensuring user
3091 namespaces work correctly.
3093 * Support for USB FunctionFS activation has been added. This
3094 allows implementation of USB gadget services that are
3095 activated as soon as they are requested, so that they don't
3096 have to run continuously, similar to classic socket
3099 * The "systemctl exit" command now optionally takes an
3100 additional parameter that sets the exit code to return from
3101 the systemd manager when exiting. This is only relevant when
3102 running the systemd user instance, or when running the
3103 system instance in a container.
3105 * sd-bus gained the new API calls sd_bus_path_encode_many()
3106 and sd_bus_path_decode_many() that allow easy encoding and
3107 decoding of multiple identifier strings inside a D-Bus
3108 object path. Another new call sd_bus_default_flush_close()
3109 has been added to flush and close per-thread default
3112 * systemd-cgtop gained support for a -M/--machine= switch to
3113 show the control groups within a certain container only.
3115 * "systemctl kill" gained support for an optional --fail
3116 switch. If specified the requested operation will fail of no
3117 processes have been killed, because the unit had no
3118 processes attached, or similar.
3120 * A new systemd.crash_reboot=1 kernel command line option has
3121 been added that triggers a reboot after crashing. This can
3122 also be set through CrashReboot= in systemd.conf.
3124 * The RuntimeDirectory= setting now understands unit
3125 specifiers like %i or %f.
3127 * A new (still internal) library API sd-ipv4acd has been added,
3128 that implements address conflict detection for IPv4. It's
3129 based on code from sd-ipv4ll, and will be useful for
3130 detecting DHCP address conflicts.
3132 * File descriptors passed during socket activation may now be
3133 named. A new API sd_listen_fds_with_names() is added to
3134 access the names. The default names may be overridden,
3135 either in the .socket file using the FileDescriptorName=
3136 parameter, or by passing FDNAME= when storing the file
3137 descriptors using sd_notify().
3139 * systemd-networkd gained support for:
3141 - Setting the IPv6 Router Advertisement settings via
3142 IPv6AcceptRouterAdvertisements= in .network files.
3144 - Configuring the HelloTimeSec=, MaxAgeSec= and
3145 ForwardDelaySec= bridge parameters in .netdev files.
3147 - Configuring PreferredSource= for static routes in
3150 * The "ask-password" framework used to query for LUKS harddisk
3151 passwords or SSL passwords during boot gained support for
3152 caching passwords in the kernel keyring, if it is
3153 available. This makes sure that the user only has to type in
3154 a passphrase once if there are multiple objects to unlock
3155 with the same one. Previously, such password caching was
3156 available only when Plymouth was used; this moves the
3157 caching logic into the systemd codebase itself. The
3158 "systemd-ask-password" utility gained a new --keyname=
3159 switch to control which kernel keyring key to use for
3160 caching a password in. This functionality is also useful for
3161 enabling display managers such as gdm to automatically
3162 unlock the user's GNOME keyring if its passphrase, the
3163 user's password and the harddisk password are the same, if
3164 gdm-autologin is used.
3166 * When downloading tar or raw images using "machinectl
3167 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
3168 file is now also downloaded, if it is available and stored
3169 next to the image file.
3171 * Units of type ".socket" gained a new boolean setting
3172 Writable= which is only useful in conjunction with
3173 ListenSpecial=. If true, enables opening the specified
3174 special file in O_RDWR mode rather than O_RDONLY mode.
3176 * systemd-rfkill has been reworked to become a singleton
3177 service that is activated through /dev/rfkill on each rfkill
3178 state change and saves the settings to disk. This way,
3179 systemd-rfkill is now compatible with devices that exist
3180 only intermittendly, and even restores state if the previous
3181 system shutdown was abrupt rather than clean.
3183 * The journal daemon gained support for vacuuming old journal
3184 files controlled by the number of files that shall remain,
3185 in addition to the already existing control by size and by
3186 date. This is useful as journal interleaving performance
3187 degrades with too many separate journal files, and allows
3188 putting an effective limit on them. The new setting defaults
3189 to 100, but this may be changed by setting SystemMaxFiles=
3190 and RuntimeMaxFiles= in journald.conf. Also, the
3191 "journalctl" tool gained the new --vacuum-files= switch to
3192 manually vacuum journal files to leave only the specified
3193 number of files in place.
3195 * udev will now create /dev/disk/by-path links for ATA devices
3196 on kernels where that is supported.
3198 * Galician, Serbian, Turkish and Korean translations were added.
3200 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
3201 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
3202 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
3203 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
3204 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
3205 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
3206 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
3207 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
3208 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
3209 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
3210 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
3211 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
3212 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
3213 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
3214 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
3215 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
3216 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
3217 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
3219 — Berlin, 2015-10-07
3223 * The DHCP implementation of systemd-networkd gained a set of
3226 - The DHCP server now supports emitting DNS and NTP
3227 information. It may be enabled and configured via
3228 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
3229 and NTP information is enabled, but no servers are
3230 configured, the corresponding uplink information (if there
3231 is any) is propagated.
3233 - Server and client now support transmission and reception
3234 of timezone information. It can be configured via the
3235 newly introduced network options UseTimezone=,
3236 EmitTimezone=, and Timezone=. Transmission of timezone
3237 information is enabled between host and containers by
3238 default now: the container will change its local timezone
3239 to what the host has set.
3241 - Lease timeouts can now be configured via
3242 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
3244 - The DHCP server improved on the stability of
3245 leases. Clients are more likely to get the same lease
3246 information back, even if the server loses state.
3248 - The DHCP server supports two new configuration options to
3249 control the lease address pool metrics, PoolOffset= and
3252 * The encapsulation limit of tunnels in systemd-networkd may
3253 now be configured via 'EncapsulationLimit='. It allows
3254 modifying the maximum additional levels of encapsulation
3255 that are permitted to be prepended to a packet.
3257 * systemd now supports the concept of user buses replacing
3258 session buses, if used with dbus-1.10 (and enabled via dbus
3259 --enable-user-session). It previously only supported this on
3260 kdbus-enabled systems, and this release expands this to
3261 'dbus-daemon' systems.
3263 * systemd-networkd now supports predictable interface names
3266 * systemd now optionally supports the new Linux kernel
3267 "unified" control group hierarchy. If enabled via the kernel
3268 command-line option 'systemd.unified_cgroup_hierarchy=1',
3269 systemd will try to mount the unified cgroup hierarchy
3270 directly on /sys/fs/cgroup. If not enabled, or not
3271 available, systemd will fall back to the legacy cgroup
3272 hierarchy setup, as before. Host system and containers can
3273 mix and match legacy and unified hierarchies as they
3274 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
3275 environment variable to individually select the hierarchy to
3276 use for executed containers. By default, nspawn will use the
3277 unified hierarchy for the containers if the host uses the
3278 unified hierarchy, and the legacy hierarchy otherwise.
3279 Please note that at this point the unified hierarchy is an
3280 experimental kernel feature and is likely to change in one
3281 of the next kernel releases. Therefore, it should not be
3282 enabled by default in downstream distributions yet. The
3283 minimum required kernel version for the unified hierarchy to
3284 work is 4.2. Note that when the unified hierarchy is used
3285 for the first time delegated access to controllers is
3286 safe. Because of this systemd-nspawn containers will get
3287 access to controllers now, as will systemd user
3288 sessions. This means containers and user sessions may now
3289 manage their own resources, partitioning up what the system
3292 * A new special scope unit "init.scope" has been introduced
3293 that encapsulates PID 1 of the system. It may be used to
3294 determine resource usage and enforce resource limits on PID
3295 1 itself. PID 1 hence moved out of the root of the control
3298 * The cgtop tool gained support for filtering out kernel
3299 threads when counting tasks in a control group. Also, the
3300 count of processes is now recursively summed up by
3301 default. Two options -k and --recursive= have been added to
3302 revert to old behaviour. The tool has also been updated to
3303 work correctly in containers now.
3305 * systemd-nspawn's --bind= and --bind-ro= options have been
3306 extended to allow creation of non-recursive bind mounts.
3308 * libsystemd gained two new calls sd_pid_get_cgroup() and
3309 sd_peer_get_cgroup() which return the control group path of
3310 a process or peer of a connected AF_UNIX socket. This
3311 function call is particularly useful when implementing
3312 delegated subtrees support in the control group hierarchy.
3314 * The "sd-event" event loop API of libsystemd now supports
3315 correct dequeuing of real-time signals, without losing
3318 * When systemd requests a PolicyKit decision when managing
3319 units it will now add additional fields to the request,
3320 including unit name and desired operation. This enables more
3321 powerful PolicyKit policies, that make decisions depending
3322 on these parameters.
3324 * nspawn learnt support for .nspawn settings files, that may
3325 accompany the image files or directories of containers, and
3326 may contain additional settings for the container. This is
3327 an alternative to configuring container parameters via the
3328 nspawn command line.
3330 Contributions from: Cristian Rodríguez, Daniel Mack, David
3331 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
3332 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
3333 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
3334 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
3335 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
3336 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
3337 Andersen, Tom Gundersen, Torstein Husebø
3339 — Berlin, 2015-09-08
3343 * machinectl gained a new verb 'shell' which opens a fresh
3344 shell on the target container or the host. It is similar to
3345 the existing 'login' command of machinectl, but spawns the
3346 shell directly without prompting for username or
3347 password. The pseudo machine '.host' now refers to the local
3348 host and is used by default. Hence, 'machinectl shell' can
3349 be used as replacement for 'su -' which spawns a session as
3350 a fresh systemd unit in a way that is fully isolated from
3351 the originating session.
3353 * systemd-networkd learned to cope with private-zone DHCP
3354 options and allows other programs to query the values.
3356 * SELinux access control when enabling/disabling units is no
3357 longer enforced with this release. The previous
3358 implementation was incorrect, and a new corrected
3359 implementation is not yet available. As unit file operations
3360 are still protected via PolicyKit and D-Bus policy this is
3361 not a security problem. Yet, distributions which care about
3362 optimal SELinux support should probably not stabilize on
3365 * sd-bus gained support for matches of type "arg0has=", that
3366 test for membership of strings in string arrays sent in bus
3369 * systemd-resolved now dumps the contents of its DNS and LLMNR
3370 caches to the logs on reception of the SIGUSR1 signal. This
3371 is useful to debug DNS behaviour.
3373 * The coredumpctl tool gained a new --directory= option to
3374 operate on journal files in a specific directory.
3376 * "systemctl reboot" and related commands gained a new
3377 "--message=" option which may be used to set a free-text
3378 wall message when shutting down or rebooting the
3379 system. This message is also logged, which is useful for
3380 figuring out the reason for a reboot or shutdown a
3383 * The "systemd-resolve-host" tool's -i switch now takes
3384 network interface numbers as alternative to interface names.
3386 * A new unit file setting for services has been introduced:
3387 UtmpMode= allows configuration of how precisely systemd
3388 handles utmp and wtmp entries for the service if this is
3389 enabled. This allows writing services that appear similar to
3390 user sessions in the output of the "w", "who", "last" and
3393 * systemd-resolved will now locally synthesize DNS resource
3394 records for the "localhost" and "gateway" domains as well as
3395 the local hostname. This should ensure that clients querying
3396 RRs via resolved will get similar results as those going via
3397 NSS, if nss-myhostname is enabled.
3399 Contributions from: Alastair Hughes, Alex Crawford, Daniel
3400 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
3401 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
3402 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
3403 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
3404 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
3405 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
3406 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
3407 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
3408 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
3409 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
3410 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3412 — Berlin, 2015-08-27
3416 * The systemd-efi-boot-generator functionality was merged into
3417 systemd-gpt-auto-generator.
3419 * systemd-networkd now supports Group Policy for vxlan
3420 devices. It can be enabled via the new boolean configuration
3421 option called 'GroupPolicyExtension='.
3423 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
3424 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
3425 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
3427 — Berlin, 2015-07-31
3431 * The python-systemd code has been removed from the systemd repository.
3432 A new repository has been created which accommodates the code from
3433 now on, and we kindly ask distributions to create a separate package
3434 for this: https://github.com/systemd/python-systemd
3436 * The systemd daemon will now reload its main configuration
3437 (/etc/systemd/system.conf) on daemon-reload.
3439 * sd-dhcp now exposes vendor specific extensions via
3440 sd_dhcp_lease_get_vendor_specific().
3442 * systemd-networkd gained a number of new configuration options.
3444 - A new boolean configuration option for TAP devices called
3445 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
3446 device, thus allowing to send and receive GSO packets.
3448 - A new tunnel configuration option called 'CopyDSCP='.
3449 If enabled, the DSCP field of ip6 tunnels is copied into the
3450 decapsulated packet.
3452 - A set of boolean bridge configuration options were added.
3453 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
3454 and 'UnicastFlood=' are now parsed by networkd and applied to the
3455 respective bridge link device via the respective IFLA_BRPORT_*
3458 - A new string configuration option to override the hostname sent
3459 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
3460 is true, networkd will use the configured hostname instead of the
3461 system hostname when sending DHCP requests.
3463 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
3464 networkd will configure the IPv6 flow-label of the tunnel device
3465 according to RFC2460.
3467 - The 'macvtap' virtual network devices are now supported, similar to
3468 the already supported 'macvlan' devices.
3470 * systemd-resolved now implements RFC5452 to improve resilience against
3471 cache poisoning. Additionally, source port randomization is enabled
3472 by default to further protect against DNS spoofing attacks.
3474 * nss-mymachines now supports translating UIDs and GIDs of running
3475 containers with user-namespaces enabled. If a container 'foo'
3476 translates a host uid 'UID' to the container uid 'TUID', then
3477 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
3478 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
3479 mapped as 'vg-foo-TGID'.
3481 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
3482 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
3483 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
3484 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
3485 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
3486 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
3487 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
3488 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
3489 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
3490 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
3492 — Berlin, 2015-07-29
3496 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
3497 There are no known issues with current sysfs, and udev does not need
3498 or should be used to work around such bugs.
3500 * udev does no longer enable USB HID power management. Several reports
3501 indicate, that some devices cannot handle that setting.
3503 * The udev accelerometer helper was removed. The functionality
3504 is now fully included in iio-sensor-proxy. But this means,
3505 older iio-sensor-proxy versions will no longer provide
3506 accelerometer/orientation data with this systemd version.
3507 Please upgrade iio-sensor-proxy to version 1.0.
3509 * networkd gained a new configuration option IPv6PrivacyExtensions=
3510 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
3511 for Stateless Address") on selected networks.
3513 * For the sake of fewer build-time dependencies and less code in the
3514 main repository, the python bindings are about to be removed in the
3515 next release. A new repository has been created which accommodates
3516 the code from now on, and we kindly ask distributions to create a
3517 separate package for this. The removal will take place in v223.
3519 https://github.com/systemd/python-systemd
3521 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
3522 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
3523 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
3524 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
3525 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
3526 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
3527 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
3528 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
3529 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3530 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3532 — Berlin, 2015-07-07
3536 * The sd-bus.h and sd-event.h APIs have now been declared
3537 stable and have been added to the official interface of
3538 libsystemd.so. sd-bus implements an alternative D-Bus client
3539 library, that is relatively easy to use, very efficient and
3540 supports both classic D-Bus as well as kdbus as transport
3541 backend. sd-event is a generic event loop abstraction that
3542 is built around Linux epoll, but adds features such as event
3543 prioritization or efficient timer handling. Both APIs are good
3544 choices for C programs looking for a bus and/or event loop
3545 implementation that is minimal and does not have to be
3546 portable to other kernels.
3548 * kdbus support is no longer compile-time optional. It is now
3549 always built-in. However, it can still be disabled at
3550 runtime using the kdbus=0 kernel command line setting, and
3551 that setting may be changed to default to off, by specifying
3552 --disable-kdbus at build-time. Note though that the kernel
3553 command line setting has no effect if the kdbus.ko kernel
3554 module is not installed, in which case kdbus is (obviously)
3555 also disabled. We encourage all downstream distributions to
3556 begin testing kdbus by adding it to the kernel images in the
3557 development distributions, and leaving kdbus support in
3560 * The minimal required util-linux version has been bumped to
3563 * Support for chkconfig (--enable-chkconfig) was removed in
3564 favor of calling an abstraction tool
3565 /lib/systemd/systemd-sysv-install. This needs to be
3566 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
3567 in README for details.
3569 * If there's a systemd unit and a SysV init script for the
3570 same service name, and the user executes "systemctl enable"
3571 for it (or a related call), then this will now enable both
3572 (or execute the related operation on both), not just the
3575 * The libudev API documentation has been converted from gtkdoc
3578 * gudev has been removed from the systemd tree, it is now an
3581 * The systemd-cgtop tool learnt a new --raw switch to generate
3582 "raw" (machine parsable) output.
3584 * networkd's IPForwarding= .network file setting learnt the
3585 new setting "kernel", which ensures that networkd does not
3586 change the IP forwarding sysctl from the default kernel
3589 * The systemd-logind bus API now exposes a new boolean
3590 property "Docked" that reports whether logind considers the
3591 system "docked", i.e. connected to a docking station or not.
3593 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
3594 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
3595 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
3596 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
3597 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
3598 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
3599 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
3600 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
3601 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
3602 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
3603 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
3604 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
3605 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
3606 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3607 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
3608 Fink, Zbigniew Jędrzejewski-Szmek
3610 — Berlin, 2015-06-19
3614 * The gudev library has been extracted into a separate repository
3615 available at: https://git.gnome.org/browse/libgudev/
3616 It is now managed as part of the Gnome project. Distributions
3617 are recommended to pass --disable-gudev to systemd and use
3618 gudev from the Gnome project instead. gudev is still included
3619 in systemd, for now. It will be removed soon, though. Please
3620 also see the announcement-thread on systemd-devel:
3621 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
3623 * systemd now exposes a CPUUsageNSec= property for each
3624 service unit on the bus, that contains the overall consumed
3625 CPU time of a service (the sum of what each process of the
3626 service consumed). This value is only available if
3627 CPUAccounting= is turned on for a service, and is then shown
3628 in the "systemctl status" output.
3630 * Support for configuring alternative mappings of the old SysV
3631 runlevels to systemd targets has been removed. They are now
3632 hardcoded in a way that runlevels 2, 3, 4 all map to
3633 multi-user.target and 5 to graphical.target (which
3634 previously was already the default behaviour).
3636 * The auto-mounter logic gained support for mount point
3637 expiry, using a new TimeoutIdleSec= setting in .automount
3638 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
3640 * The EFI System Partition (ESP) as mounted to /boot by
3641 systemd-efi-boot-generator will now be unmounted
3642 automatically after 2 minutes of not being used. This should
3643 minimize the risk of ESP corruptions.
3645 * New /etc/fstab options x-systemd.requires= and
3646 x-systemd.requires-mounts-for= are now supported to express
3647 additional dependencies for mounts. This is useful for
3648 journalling file systems that support external journal
3649 devices or overlay file systems that require underlying file
3650 systems to be mounted.
3652 * systemd does not support direct live-upgrades (via systemctl
3653 daemon-reexec) from versions older than v44 anymore. As no
3654 distribution we are aware of shipped such old versions in a
3655 stable release this should not be problematic.
3657 * When systemd forks off a new per-connection service instance
3658 it will now set the $REMOTE_ADDR environment variable to the
3659 remote IP address, and $REMOTE_PORT environment variable to
3660 the remote IP port. This behaviour is similar to the
3661 corresponding environment variables defined by CGI.
3663 * systemd-networkd gained support for uplink failure
3664 detection. The BindCarrier= option allows binding interface
3665 configuration dynamically to the link sense of other
3666 interfaces. This is useful to achieve behaviour like in
3669 * systemd-networkd gained support for configuring the DHCP
3670 client identifier to use when requesting leases.
3672 * systemd-networkd now has a per-network UseNTP= option to
3673 configure whether NTP server information acquired via DHCP
3674 is passed on to services like systemd-timesyncd.
3676 * systemd-networkd gained support for vti6 tunnels.
3678 * Note that systemd-networkd manages the sysctl variable
3679 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
3680 it is configured for since v219. The variable controls IP
3681 forwarding, and is a per-interface alternative to the global
3682 /proc/sys/net/ipv[46]/ip_forward. This setting is
3683 configurable in the IPForward= option, which defaults to
3684 "no". This means if networkd is used for an interface it is
3685 no longer sufficient to set the global sysctl option to turn
3686 on IP forwarding! Instead, the .network file option
3687 IPForward= needs to be turned on! Note that the
3688 implementation of this behaviour was broken in v219 and has
3691 * Many bonding and vxlan options are now configurable in
3694 * systemd-nspawn gained a new --property= setting to set unit
3695 properties for the container scope. This is useful for
3696 setting resource parameters (e.g. "CPUShares=500") on
3697 containers started from the command line.
3699 * systemd-nspawn gained a new --private-users= switch to make
3700 use of user namespacing available on recent Linux kernels.
3702 * systemd-nspawn may now be called as part of a shell pipeline
3703 in which case the pipes used for stdin and stdout are passed
3704 directly to the process invoked in the container, without
3705 indirection via a pseudo tty.
3707 * systemd-nspawn gained a new switch to control the UNIX
3708 signal to use when killing the init process of the container
3711 * systemd-nspawn gained a new --overlay= switch for mounting
3712 overlay file systems into the container using the new kernel
3715 * When a container image is imported via systemd-importd and
3716 the host file system is not btrfs, a loopback block device
3717 file is created in /var/lib/machines.raw with a btrfs file
3718 system inside. It is then mounted to /var/lib/machines to
3719 enable btrfs features for container management. The loopback
3720 file and btrfs file system is grown as needed when container
3721 images are imported via systemd-importd.
3723 * systemd-machined/systemd-importd gained support for btrfs
3724 quota, to enforce container disk space limits on disk. This
3725 is exposed in "machinectl set-limit".
3727 * systemd-importd now can import containers from local .tar,
3728 .raw and .qcow2 images, and export them to .tar and .raw. It
3729 can also import dkr v2 images now from the network (on top
3732 * systemd-importd gained support for verifying downloaded
3733 images with gpg2 (previously only gpg1 was supported).
3735 * systemd-machined, systemd-logind, systemd: most bus calls
3736 are now accessible to unprivileged processes via
3737 PolicyKit. Also, systemd-logind will now allow users to kill
3738 their own sessions without further privileges or
3741 * systemd-shutdownd has been removed. This service was
3742 previously responsible for implementing scheduled shutdowns
3743 as exposed in /usr/bin/shutdown's time parameter. This
3744 functionality has now been moved into systemd-logind and is
3745 accessible via a bus interface.
3747 * "systemctl reboot" gained a new switch --firmware-setup that
3748 can be used to reboot into the EFI firmware setup, if that
3749 is available. systemd-logind now exposes an API on the bus
3750 to trigger such reboots, in case graphical desktop UIs want
3751 to cover this functionality.
3753 * "systemctl enable", "systemctl disable" and "systemctl mask"
3754 now support a new "--now" switch. If specified the units
3755 that are enabled will also be started, and the ones
3756 disabled/masked also stopped.
3758 * The Gummiboot EFI boot loader tool has been merged into
3759 systemd, and renamed to "systemd-boot". The bootctl tool has been
3760 updated to support systemd-boot.
3762 * An EFI kernel stub has been added that may be used to create
3763 kernel EFI binaries that contain not only the actual kernel,
3764 but also an initrd, boot splash, command line and OS release
3765 information. This combined binary can then be signed as a
3766 single image, so that the firmware can verify it all in one
3767 step. systemd-boot has special support for EFI binaries created
3768 like this and can extract OS release information from them
3769 and show them in the boot menu. This functionality is useful
3770 to implement cryptographically verified boot schemes.
3772 * Optional support has been added to systemd-fsck to pass
3773 fsck's progress report to an AF_UNIX socket in the file
3776 * udev will no longer create device symlinks for all block
3777 devices by default. A blacklist for excluding special block
3778 devices from this logic has been turned into a whitelist
3779 that requires picking block devices explicitly that require
3782 * A new (currently still internal) API sd-device.h has been
3783 added to libsystemd. This modernized API is supposed to
3784 replace libudev eventually. In fact, already much of libudev
3785 is now just a wrapper around sd-device.h.
3787 * A new hwdb database for storing metadata about pointing
3788 stick devices has been added.
3790 * systemd-tmpfiles gained support for setting file attributes
3791 similar to the "chattr" tool with new 'h' and 'H' lines.
3793 * systemd-journald will no longer unconditionally set the
3794 btrfs NOCOW flag on new journal files. This is instead done
3795 with tmpfiles snippet using the new 'h' line type. This
3796 allows easy disabling of this logic, by masking the
3797 journal-nocow.conf tmpfiles file.
3799 * systemd-journald will now translate audit message types to
3800 human readable identifiers when writing them to the
3801 journal. This should improve readability of audit messages.
3803 * The LUKS logic gained support for the offset= and skip=
3804 options in /etc/crypttab, as previously implemented by
3807 * /usr/lib/os-release gained a new optional field VARIANT= for
3808 distributions that support multiple variants (such as a
3809 desktop edition, a server edition, ...)
3811 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
3812 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
3813 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
3814 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
3815 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
3816 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
3817 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
3818 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
3819 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
3820 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
3821 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
3822 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
3823 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
3824 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
3825 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
3826 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
3827 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
3828 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
3829 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
3830 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
3831 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
3832 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
3833 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
3834 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
3835 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
3836 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
3837 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
3839 — Berlin, 2015-05-22
3843 * Introduce a new API "sd-hwdb.h" for querying the hardware
3844 metadata database. With this minimal interface one can query
3845 and enumerate the udev hwdb, decoupled from the old libudev
3846 library. libudev's interface for this is now only a wrapper
3847 around sd-hwdb. A new tool systemd-hwdb has been added to
3848 interface with and update the database.
3850 * When any of systemd's tools copies files (for example due to
3851 tmpfiles' C lines) a btrfs reflink will attempted first,
3852 before bytewise copying is done.
3854 * systemd-nspawn gained a new --ephemeral switch. When
3855 specified a btrfs snapshot is taken of the container's root
3856 directory, and immediately removed when the container
3857 terminates again. Thus, a container can be started whose
3858 changes never alter the container's root directory, and are
3859 lost on container termination. This switch can also be used
3860 for starting a container off the root file system of the
3861 host without affecting the host OS. This switch is only
3862 available on btrfs file systems.
3864 * systemd-nspawn gained a new --template= switch. It takes the
3865 path to a container tree to use as template for the tree
3866 specified via --directory=, should that directory be
3867 missing. This allows instantiating containers dynamically,
3868 on first run. This switch is only available on btrfs file
3871 * When a .mount unit refers to a mount point on which multiple
3872 mounts are stacked, and the .mount unit is stopped all of
3873 the stacked mount points will now be unmounted until no
3874 mount point remains.
3876 * systemd now has an explicit notion of supported and
3877 unsupported unit types. Jobs enqueued for unsupported unit
3878 types will now fail with an "unsupported" error code. More
3879 specifically .swap, .automount and .device units are not
3880 supported in containers, .busname units are not supported on
3881 non-kdbus systems. .swap and .automount are also not
3882 supported if their respective kernel compile time options
3885 * machinectl gained support for two new "copy-from" and
3886 "copy-to" commands for copying files from a running
3887 container to the host or vice versa.
3889 * machinectl gained support for a new "bind" command to bind
3890 mount host directories into local containers. This is
3891 currently only supported for nspawn containers.
3893 * networkd gained support for configuring bridge forwarding
3894 database entries (fdb) from .network files.
3896 * A new tiny daemon "systemd-importd" has been added that can
3897 download container images in tar, raw, qcow2 or dkr formats,
3898 and make them available locally in /var/lib/machines, so
3899 that they can run as nspawn containers. The daemon can GPG
3900 verify the downloads (not supported for dkr, since it has no
3901 provisions for verifying downloads). It will transparently
3902 decompress bz2, xz, gzip compressed downloads if necessary,
3903 and restore sparse files on disk. The daemon uses privilege
3904 separation to ensure the actual download logic runs with
3905 fewer privileges than the daemon itself. machinectl has
3906 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
3907 make the functionality of importd available to the
3908 user. With this in place the Fedora and Ubuntu "Cloud"
3909 images can be downloaded and booted as containers unmodified
3910 (the Fedora images lack the appropriate GPG signature files
3911 currently, so they cannot be verified, but this will change
3912 soon, hopefully). Note that downloading images is currently
3913 only fully supported on btrfs.
3915 * machinectl is now able to list container images found in
3916 /var/lib/machines, along with some metadata about sizes of
3917 disk and similar. If the directory is located on btrfs and
3918 quota is enabled, this includes quota display. A new command
3919 "image-status" has been added that shows additional
3920 information about images.
3922 * machinectl is now able to clone container images
3923 efficiently, if the underlying file system (btrfs) supports
3924 it, with the new "machinectl clone" command. It also
3925 gained commands for renaming and removing images, as well as
3926 marking them read-only or read-write (supported also on
3927 legacy file systems).
3929 * networkd gained support for collecting LLDP network
3930 announcements, from hardware that supports this. This is
3931 shown in networkctl output.
3933 * systemd-run gained support for a new -t (--pty) switch for
3934 invoking a binary on a pty whose input and output is
3935 connected to the invoking terminal. This allows executing
3936 processes as system services while interactively
3937 communicating with them via the terminal. Most interestingly
3938 this is supported across container boundaries. Invoking
3939 "systemd-run -t /bin/bash" is an alternative to running a
3940 full login session, the difference being that the former
3941 will not register a session, nor go through the PAM session
3944 * tmpfiles gained support for a new "v" line type for creating
3945 btrfs subvolumes. If the underlying file system is a legacy
3946 file system, this automatically degrades to creating a
3947 normal directory. Among others /var/lib/machines is now
3948 created like this at boot, should it be missing.
3950 * The directory /var/lib/containers/ has been deprecated and
3951 been replaced by /var/lib/machines. The term "machines" has
3952 been used in the systemd context as generic term for both
3953 VMs and containers, and hence appears more appropriate for
3954 this, as the directory can also contain raw images bootable
3957 * systemd-nspawn when invoked with -M but without --directory=
3958 or --image= is now capable of searching for the container
3959 root directory, subvolume or disk image automatically, in
3960 /var/lib/machines. systemd-nspawn@.service has been updated
3961 to make use of this, thus allowing it to be used for raw
3964 * A new machines.target unit has been introduced that is
3965 supposed to group all containers/VMs invoked as services on
3966 the system. systemd-nspawn@.service has been updated to
3967 integrate with that.
3969 * machinectl gained a new "start" command, for invoking a
3970 container as a service. "machinectl start foo" is mostly
3971 equivalent to "systemctl start systemd-nspawn@foo.service",
3972 but handles escaping in a nicer way.
3974 * systemd-nspawn will now mount most of the cgroupfs tree
3975 read-only into each container, with the exception of the
3976 container's own subtree in the name=systemd hierarchy.
3978 * journald now sets the special FS_NOCOW file flag for its
3979 journal files. This should improve performance on btrfs, by
3980 avoiding heavy fragmentation when journald's write-pattern
3981 is used on COW file systems. It degrades btrfs' data
3982 integrity guarantees for the files to the same levels as for
3983 ext3/ext4 however. This should be OK though as journald does
3984 its own data integrity checks and all its objects are
3985 checksummed on disk. Also, journald should handle btrfs disk
3986 full events a lot more gracefully now, by processing SIGBUS
3987 errors, and not relying on fallocate() anymore.
3989 * When journald detects that journal files it is writing to
3990 have been deleted it will immediately start new journal
3993 * systemd now provides a way to store file descriptors
3994 per-service in PID 1. This is useful for daemons to ensure
3995 that fds they require are not lost during a daemon
3996 restart. The fds are passed to the daemon on the next
3997 invocation in the same way socket activation fds are
3998 passed. This is now used by journald to ensure that the
3999 various sockets connected to all the system's stdout/stderr
4000 are not lost when journald is restarted. File descriptors
4001 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
4002 an extension to sd_notify(). Note that a limit is enforced
4003 on the number of fds a service can store in PID 1, and it
4004 defaults to 0, so that no fds may be stored, unless this is
4005 explicitly turned on.
4007 * The default TERM variable to use for units connected to a
4008 terminal, when no other value is explicitly is set is now
4009 vt220 rather than vt102. This should be fairly safe still,
4010 but allows PgUp/PgDn work.
4012 * The /etc/crypttab option header= as known from Debian is now
4015 * "loginctl user-status" and "loginctl session-status" will
4016 now show the last 10 lines of log messages of the
4017 user/session following the status output. Similar,
4018 "machinectl status" will show the last 10 log lines
4019 associated with a virtual machine or container
4020 service. (Note that this is usually not the log messages
4021 done in the VM/container itself, but simply what the
4022 container manager logs. For nspawn this includes all console
4025 * "loginctl session-status" without further argument will now
4026 show the status of the session of the caller. Similar,
4027 "lock-session", "unlock-session", "activate",
4028 "enable-linger", "disable-linger" may now be called without
4029 session/user parameter in which case they apply to the
4030 caller's session/user.
4032 * An X11 session scriptlet is now shipped that uploads
4033 $DISPLAY and $XAUTHORITY into the environment of the systemd
4034 --user daemon if a session begins. This should improve
4035 compatibility with X11 enabled applications run as systemd
4038 * Generators are now subject to masking via /etc and /run, the
4039 same way as unit files.
4041 * networkd .network files gained support for configuring
4042 per-link IPv4/IPv6 packet forwarding as well as IPv4
4043 masquerading. This is by default turned on for veth links to
4044 containers, as registered by systemd-nspawn. This means that
4045 nspawn containers run with --network-veth will now get
4046 automatic routed access to the host's networks without any
4047 further configuration or setup, as long as networkd runs on
4050 * systemd-nspawn gained the --port= (-p) switch to expose TCP
4051 or UDP posts of a container on the host. With this in place
4052 it is possible to run containers with private veth links
4053 (--network-veth), and have their functionality exposed on
4054 the host as if their services were running directly on the
4057 * systemd-nspawn's --network-veth switch now gained a short
4058 version "-n", since with the changes above it is now truly
4059 useful out-of-the-box. The systemd-nspawn@.service has been
4060 updated to make use of it too by default.
4062 * systemd-nspawn will now maintain a per-image R/W lock, to
4063 ensure that the same image is not started more than once
4064 writable. (It's OK to run an image multiple times
4065 simultaneously in read-only mode.)
4067 * systemd-nspawn's --image= option is now capable of
4068 dissecting and booting MBR and GPT disk images that contain
4069 only a single active Linux partition. Previously it
4070 supported only GPT disk images with proper GPT type
4071 IDs. This allows running cloud images from major
4072 distributions directly with systemd-nspawn, without
4075 * In addition to collecting mouse dpi data in the udev
4076 hardware database, there's now support for collecting angle
4077 information for mouse scroll wheels. The database is
4078 supposed to guarantee similar scrolling behavior on mice
4079 that it knows about. There's also support for collecting
4080 information about Touchpad types.
4082 * udev's input_id built-in will now also collect touch screen
4083 dimension data and attach it to probed devices.
4085 * /etc/os-release gained support for a Distribution Privacy
4088 * networkd gained support for creating "ipvlan", "gretap",
4089 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
4091 * systemd-tmpfiles gained support for "a" lines for setting
4094 * systemd-nspawn will now mount /tmp in the container to
4095 tmpfs, automatically.
4097 * systemd now exposes the memory.usage_in_bytes cgroup
4098 attribute and shows it for each service in the "systemctl
4099 status" output, if available.
4101 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
4102 immediate reboot is triggered. This useful if shutdown is
4103 hung and is unable to complete, to expedite the
4104 operation. Note that this kind of reboot will still unmount
4105 all file systems, and hence should not result in fsck being
4108 * A .device unit for an optical block device will now be
4109 considered active only when a medium is in the drive. Also,
4110 mount units are now bound to their backing devices thus
4111 triggering automatic unmounting when devices become
4112 unavailable. With this in place systemd will now
4113 automatically unmount left-over mounts when a CD-ROM is
4114 ejected or an USB stick is yanked from the system.
4116 * networkd-wait-online now has support for waiting for
4117 specific interfaces only (with globbing), and for giving up
4118 after a configurable timeout.
4120 * networkd now exits when idle. It will be automatically
4121 restarted as soon as interfaces show up, are removed or
4122 change state. networkd will stay around as long as there is
4123 at least one DHCP state machine or similar around, that keep
4126 * networkd may now configure IPv6 link-local addressing in
4127 addition to IPv4 link-local addressing.
4129 * The IPv6 "token" for use in SLAAC may now be configured for
4130 each .network interface in networkd.
4132 * Routes configured with networkd may now be assigned a scope
4135 * networkd's [Match] sections now support globbing and lists
4136 of multiple space-separated matches per item.
4138 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
4139 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
4140 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
4141 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
4142 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
4143 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
4144 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
4145 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
4146 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
4147 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
4148 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
4149 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
4150 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
4151 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
4152 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
4153 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
4154 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
4155 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
4156 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
4157 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
4158 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
4159 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
4160 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
4161 Hoffmann, Zbigniew Jędrzejewski-Szmek
4163 — Berlin, 2015-02-16
4167 * When querying unit file enablement status (for example via
4168 "systemctl is-enabled"), a new state "indirect" is now known
4169 which indicates that a unit might not be enabled itself, but
4170 another unit listed in its Also= setting might be.
4172 * Similar to the various existing ConditionXYZ= settings for
4173 units, there are now matching AssertXYZ= settings. While
4174 failing conditions cause a unit to be skipped, but its job
4175 to succeed, failing assertions declared like this will cause
4176 a unit start operation and its job to fail.
4178 * hostnamed now knows a new chassis type "embedded".
4180 * systemctl gained a new "edit" command. When used on a unit
4181 file, this allows extending unit files with .d/ drop-in
4182 configuration snippets or editing the full file (after
4183 copying it from /usr/lib to /etc). This will invoke the
4184 user's editor (as configured with $EDITOR), and reload the
4185 modified configuration after editing.
4187 * "systemctl status" now shows the suggested enablement state
4188 for a unit, as declared in the (usually vendor-supplied)
4189 system preset files.
4191 * nss-myhostname will now resolve the single-label host name
4192 "gateway" to the locally configured default IP routing
4193 gateways, ordered by their metrics. This assigns a stable
4194 name to the used gateways, regardless which ones are
4195 currently configured. Note that the name will only be
4196 resolved after all other name sources (if nss-myhostname is
4197 configured properly) and should hence not negatively impact
4198 systems that use the single-label host name "gateway" in
4201 * systemd-inhibit now allows filtering by mode when listing
4204 * Scope and service units gained a new "Delegate" boolean
4205 property, which, when set, allows processes running inside the
4206 unit to further partition resources. This is primarily
4207 useful for systemd user instances as well as container
4210 * journald will now pick up audit messages directly from
4211 the kernel, and log them like any other log message. The
4212 audit fields are split up and fully indexed. This means that
4213 journalctl in many ways is now a (nicer!) alternative to
4214 ausearch, the traditional audit client. Note that this
4215 implements only a minimal audit client. If you want the
4216 special audit modes like reboot-on-log-overflow, please use
4217 the traditional auditd instead, which can be used in
4218 parallel to journald.
4220 * The ConditionSecurity= unit file option now understands the
4221 special string "audit" to check whether auditing is
4224 * journalctl gained two new commands --vacuum-size= and
4225 --vacuum-time= to delete old journal files until the
4226 remaining ones take up no more than the specified size on disk,
4227 or are not older than the specified time.
4229 * A new, native PPPoE library has been added to sd-network,
4230 systemd's library of light-weight networking protocols. This
4231 library will be used in a future version of networkd to
4232 enable PPPoE communication without an external pppd daemon.
4234 * The busctl tool now understands a new "capture" verb that
4235 works similar to "monitor", but writes a packet capture
4236 trace to STDOUT that can be redirected to a file which is
4237 compatible with libcap's capture file format. This can then
4238 be loaded in Wireshark and similar tools to inspect bus
4241 * The busctl tool now understands a new "tree" verb that shows
4242 the object trees of a specific service on the bus, or of all
4245 * The busctl tool now understands a new "introspect" verb that
4246 shows all interfaces and members of objects on the bus,
4247 including their signature and values. This is particularly
4248 useful to get more information about bus objects shown by
4249 the new "busctl tree" command.
4251 * The busctl tool now understands new verbs "call",
4252 "set-property" and "get-property" for invoking bus method
4253 calls, setting and getting bus object properties in a
4256 * busctl gained a new --augment-creds= argument that controls
4257 whether the tool shall augment credential information it
4258 gets from the bus with data from /proc, in a possibly
4261 * nspawn's --link-journal= switch gained two new values
4262 "try-guest" and "try-host" that work like "guest" and
4263 "host", but do not fail if the host has no persistent
4264 journalling enabled. -j is now equivalent to
4265 --link-journal=try-guest.
4267 * macvlan network devices created by nspawn will now have
4268 stable MAC addresses.
4270 * A new SmackProcessLabel= unit setting has been added, which
4271 controls the SMACK security label processes forked off by
4272 the respective unit shall use.
4274 * If compiled with --enable-xkbcommon, systemd-localed will
4275 verify x11 keymap settings by compiling the given keymap. It
4276 will spew out warnings if the compilation fails. This
4277 requires libxkbcommon to be installed.
4279 * When a coredump is collected, a larger number of metadata
4280 fields is now collected and included in the journal records
4281 created for it. More specifically, control group membership,
4282 environment variables, memory maps, working directory,
4283 chroot directory, /proc/$PID/status, and a list of open file
4284 descriptors is now stored in the log entry.
4286 * The udev hwdb now contains DPI information for mice. For
4289 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
4291 * All systemd programs that read standalone configuration
4292 files in /etc now also support a corresponding series of
4293 .conf.d configuration directories in /etc/, /run/,
4294 /usr/local/lib/, /usr/lib/, and (if configured with
4295 --enable-split-usr) /lib/. In particular, the following
4296 configuration files now have corresponding configuration
4297 directories: system.conf user.conf, logind.conf,
4298 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
4299 resolved.conf, timesyncd.conf, journal-remote.conf, and
4300 journal-upload.conf. Note that distributions should use the
4301 configuration directories in /usr/lib/; the directories in
4302 /etc/ are reserved for the system administrator.
4304 * systemd-rfkill will no longer take the rfkill device name
4305 into account when storing rfkill state on disk, as the name
4306 might be dynamically assigned and not stable. Instead, the
4307 ID_PATH udev variable combined with the rfkill type (wlan,
4308 bluetooth, ...) is used.
4310 * A new service systemd-machine-id-commit.service has been
4311 added. When used on systems where /etc is read-only during
4312 boot, and /etc/machine-id is not initialized (but an empty
4313 file), this service will copy the temporary machine ID
4314 created as replacement into /etc after the system is fully
4315 booted up. This is useful for systems that are freshly
4316 installed with a non-initialized machine ID, but should get
4317 a fixed machine ID for subsequent boots.
4319 * networkd's .netdev files now provide a large set of
4320 configuration parameters for VXLAN devices. Similarly, the
4321 bridge port cost parameter is now configurable in .network
4322 files. There's also new support for configuring IP source
4323 routing. networkd .link files gained support for a new
4324 OriginalName= match that is useful to match against the
4325 original interface name the kernel assigned. .network files
4326 may include MTU= and MACAddress= fields for altering the MTU
4327 and MAC address while being connected to a specific network
4330 * The LUKS logic gained supported for configuring
4331 UUID-specific key files. There's also new support for naming
4332 LUKS device from the kernel command line, using the new
4333 luks.name= argument.
4335 * Timer units may now be transiently created via the bus API
4336 (this was previously already available for scope and service
4337 units). In addition it is now possible to create multiple
4338 transient units at the same time with a single bus call. The
4339 "systemd-run" tool has been updated to make use of this for
4340 running commands on a specified time, in at(1)-style.
4342 * tmpfiles gained support for "t" lines, for assigning
4343 extended attributes to files. Among other uses this may be
4344 used to assign SMACK labels to files.
4346 Contributions from: Alin Rauta, Alison Chaiken, Andrej
4347 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
4348 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
4349 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
4350 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
4351 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
4352 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
4353 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
4354 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
4355 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
4356 Michael Biebl, Michael Chapman, Michael Marineau, Michal
4357 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
4358 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
4359 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
4360 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
4361 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4362 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
4363 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
4365 — Berlin, 2014-12-10
4369 * journalctl gained the new options -t/--identifier= to match
4370 on the syslog identifier (aka "tag"), as well as --utc to
4371 show log timestamps in the UTC timezone. journalctl now also
4372 accepts -n/--lines=all to disable line capping in a pager.
4374 * journalctl gained a new switch, --flush, that synchronously
4375 flushes logs from /run/log/journal to /var/log/journal if
4376 persistent storage is enabled. systemd-journal-flush.service
4377 now waits until the operation is complete.
4379 * Services can notify the manager before they start a reload
4380 (by sending RELOADING=1) or shutdown (by sending
4381 STOPPING=1). This allows the manager to track and show the
4382 internal state of daemons and closes a race condition when
4383 the process is still running but has closed its D-Bus
4386 * Services with Type=oneshot do not have to have any ExecStart
4389 * User units are now loaded also from
4390 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
4391 /run/systemd/user directory that was already previously
4392 supported, but is under the control of the user.
4394 * Job timeouts (i.e. time-outs on the time a job that is
4395 queued stays in the run queue) can now optionally result in
4396 immediate reboot or power-off actions (JobTimeoutAction= and
4397 JobTimeoutRebootArgument=). This is useful on ".target"
4398 units, to limit the maximum time a target remains
4399 undispatched in the run queue, and to trigger an emergency
4400 operation in such a case. This is now used by default to
4401 turn off the system if boot-up (as defined by everything in
4402 basic.target) hangs and does not complete for at least
4403 15min. Also, if power-off or reboot hang for at least 30min
4404 an immediate power-off/reboot operation is triggered. This
4405 functionality is particularly useful to increase reliability
4406 on embedded devices, but also on laptops which might
4407 accidentally get powered on when carried in a backpack and
4408 whose boot stays stuck in a hard disk encryption passphrase
4411 * systemd-logind can be configured to also handle lid switch
4412 events even when the machine is docked or multiple displays
4413 are attached (HandleLidSwitchDocked= option).
4415 * A helper binary and a service have been added which can be
4416 used to resume from hibernation in the initramfs. A
4417 generator will parse the resume= option on the kernel
4418 command line to trigger resume.
4420 * A user console daemon systemd-consoled has been
4421 added. Currently, it is a preview, and will so far open a
4422 single terminal on each session of the user marked as
4423 Desktop=systemd-console.
4425 * Route metrics can be specified for DHCP routes added by
4428 * The SELinux context of socket-activated services can be set
4429 from the information provided by the networking stack
4430 (SELinuxContextFromNet= option).
4432 * Userspace firmware loading support has been removed and
4433 the minimum supported kernel version is thus bumped to 3.7.
4435 * Timeout for udev workers has been increased from 1 to 3
4436 minutes, but a warning will be printed after 1 minute to
4437 help diagnose kernel modules that take a long time to load.
4439 * Udev rules can now remove tags on devices with TAG-="foobar".
4441 * systemd's readahead implementation has been removed. In many
4442 circumstances it didn't give expected benefits even for
4443 rotational disk drives and was becoming less relevant in the
4444 age of SSDs. As none of the developers has been using
4445 rotating media anymore, and nobody stepped up to actively
4446 maintain this component of systemd it has now been removed.
4448 * Swap units can use Options= to specify discard options.
4449 Discard options specified for swaps in /etc/fstab are now
4452 * Docker containers are now detected as a separate type of
4455 * The Password Agent protocol gained support for queries where
4456 the user input is shown, useful e.g. for user names.
4457 systemd-ask-password gained a new --echo option to turn that
4460 * The default sysctl.d/ snippets will now set:
4462 net.core.default_qdisc = fq_codel
4464 This selects Fair Queuing Controlled Delay as the default
4465 queuing discipline for network interfaces. fq_codel helps
4466 fight the network bufferbloat problem. It is believed to be
4467 a good default with no tuning required for most workloads.
4468 Downstream distributions may override this choice. On 10Gbit
4469 servers that do not do forwarding, "fq" may perform better.
4470 Systems without a good clocksource should use "pfifo_fast".
4472 * If kdbus is enabled during build a new option BusPolicy= is
4473 available for service units, that allows locking all service
4474 processes into a stricter bus policy, in order to limit
4475 access to various bus services, or even hide most of them
4476 from the service's view entirely.
4478 * networkctl will now show the .network and .link file
4479 networkd has applied to a specific interface.
4481 * sd-login gained a new API call sd_session_get_desktop() to
4482 query which desktop environment has been selected for a
4485 * UNIX utmp support is now compile-time optional to support
4486 legacy-free systems.
4488 * systemctl gained two new commands "add-wants" and
4489 "add-requires" for pulling in units from specific targets
4492 * If the word "rescue" is specified on the kernel command line
4493 the system will now boot into rescue mode (aka
4494 rescue.target), which was previously available only by
4495 specifying "1" or "systemd.unit=rescue.target" on the kernel
4496 command line. This new kernel command line option nicely
4497 mirrors the already existing "emergency" kernel command line
4500 * New kernel command line options mount.usr=, mount.usrflags=,
4501 mount.usrfstype= have been added that match root=, rootflags=,
4502 rootfstype= but allow mounting a specific file system to
4505 * The $NOTIFY_SOCKET is now also passed to control processes of
4506 services, not only the main process.
4508 * This version reenables support for fsck's -l switch. This
4509 means at least version v2.25 of util-linux is required for
4510 operation, otherwise dead-locks on device nodes may
4511 occur. Again: you need to update util-linux to at least
4512 v2.25 when updating systemd to v217.
4514 * The "multi-seat-x" tool has been removed from systemd, as
4515 its functionality has been integrated into X servers 1.16,
4516 and the tool is hence redundant. It is recommended to update
4517 display managers invoking this tool to simply invoke X
4518 directly from now on, again.
4520 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
4521 message flag has been added for all of systemd's PolicyKit
4522 authenticated method calls has been added. In particular
4523 this now allows optional interactive authorization via
4524 PolicyKit for many of PID1's privileged operations such as
4525 unit file enabling and disabling.
4527 * "udevadm hwdb --update" learnt a new switch "--usr" for
4528 placing the rebuilt hardware database in /usr instead of
4529 /etc. When used only hardware database entries stored in
4530 /usr will be used, and any user database entries in /etc are
4531 ignored. This functionality is useful for vendors to ship a
4532 pre-built database on systems where local configuration is
4533 unnecessary or unlikely.
4535 * Calendar time specifications in .timer units now also
4536 understand the strings "semi-annually", "quarterly" and
4537 "minutely" as shortcuts (in addition to the preexisting
4538 "anually", "hourly", ...).
4540 * systemd-tmpfiles will now correctly create files in /dev
4541 at boot which are marked for creation only at boot. It is
4542 recommended to always create static device nodes with 'c!'
4543 and 'b!', so that they are created only at boot and not
4544 overwritten at runtime.
4546 * When the watchdog logic is used for a service (WatchdogSec=)
4547 and the watchdog timeout is hit the service will now be
4548 terminated with SIGABRT (instead of just SIGTERM), in order
4549 to make sure a proper coredump and backtrace is
4550 generated. This ensures that hanging services will result in
4551 similar coredump/backtrace behaviour as services that hit a
4554 Contributions from: Andreas Henriksson, Andrei Borzenkov,
4555 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
4556 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
4557 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
4558 Herrmann, David Sommerseth, David Strauss, Emil Renner
4559 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
4560 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
4561 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
4562 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
4563 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
4564 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
4565 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
4566 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
4567 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
4568 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
4569 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
4570 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
4571 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
4572 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
4573 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
4574 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
4577 — Berlin, 2014-10-28
4581 * timedated no longer reads NTP implementation unit names from
4582 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
4583 implementations should add a
4585 Conflicts=systemd-timesyncd.service
4587 to their unit files to take over and replace systemd's NTP
4588 default functionality.
4590 * systemd-sysusers gained a new line type "r" for configuring
4591 which UID/GID ranges to allocate system users/groups
4592 from. Lines of type "u" may now add an additional column
4593 that specifies the home directory for the system user to be
4594 created. Also, systemd-sysusers may now optionally read user
4595 information from STDIN instead of a file. This is useful for
4596 invoking it from RPM preinst scriptlets that need to create
4597 users before the first RPM file is installed since these
4598 files might need to be owned by them. A new
4599 %sysusers_create_inline RPM macro has been introduced to do
4600 just that. systemd-sysusers now updates the shadow files as
4601 well as the user/group databases, which should enhance
4602 compatibility with certain tools like grpck.
4604 * A number of bus APIs of PID 1 now optionally consult
4605 PolicyKit to permit access for otherwise unprivileged
4606 clients under certain conditions. Note that this currently
4607 doesn't support interactive authentication yet, but this is
4608 expected to be added eventually, too.
4610 * /etc/machine-info now has new fields for configuring the
4611 deployment environment of the machine, as well as the
4612 location of the machine. hostnamectl has been updated with
4613 new command to update these fields.
4615 * systemd-timesyncd has been updated to automatically acquire
4616 NTP server information from systemd-networkd, which might
4617 have been discovered via DHCP.
4619 * systemd-resolved now includes a caching DNS stub resolver
4620 and a complete LLMNR name resolution implementation. A new
4621 NSS module "nss-resolve" has been added which can be used
4622 instead of glibc's own "nss-dns" to resolve hostnames via
4623 systemd-resolved. Hostnames, addresses and arbitrary RRs may
4624 be resolved via systemd-resolved D-Bus APIs. In contrast to
4625 the glibc internal resolver systemd-resolved is aware of
4626 multi-homed system, and keeps DNS server and caches separate
4627 and per-interface. Queries are sent simultaneously on all
4628 interfaces that have DNS servers configured, in order to
4629 properly handle VPNs and local LANs which might resolve
4630 separate sets of domain names. systemd-resolved may acquire
4631 DNS server information from systemd-networkd automatically,
4632 which in turn might have discovered them via DHCP. A tool
4633 "systemd-resolve-host" has been added that may be used to
4634 query the DNS logic in resolved. systemd-resolved implements
4635 IDNA and automatically uses IDNA or UTF-8 encoding depending
4636 on whether classic DNS or LLMNR is used as transport. In the
4637 next releases we intend to add a DNSSEC and mDNS/DNS-SD
4638 implementation to systemd-resolved.
4640 * A new NSS module nss-mymachines has been added, that
4641 automatically resolves the names of all local registered
4642 containers to their respective IP addresses.
4644 * A new client tool "networkctl" for systemd-networkd has been
4645 added. It currently is entirely passive and will query
4646 networking configuration from udev, rtnetlink and networkd,
4647 and present it to the user in a very friendly
4648 way. Eventually, we hope to extend it to become a full
4649 control utility for networkd.
4651 * .socket units gained a new DeferAcceptSec= setting that
4652 controls the kernels' TCP_DEFER_ACCEPT sockopt for
4653 TCP. Similarly, support for controlling TCP keep-alive
4654 settings has been added (KeepAliveTimeSec=,
4655 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
4656 turning off Nagle's algorithm on TCP has been added
4659 * logind learned a new session type "web", for use in projects
4660 like Cockpit which register web clients as PAM sessions.
4662 * timer units with at least one OnCalendar= setting will now
4663 be started only after time-sync.target has been
4664 reached. This way they will not elapse before the system
4665 clock has been corrected by a local NTP client or
4666 similar. This is particular useful on RTC-less embedded
4667 machines, that come up with an invalid system clock.
4669 * systemd-nspawn's --network-veth= switch should now result in
4670 stable MAC addresses for both the outer and the inner side
4673 * systemd-nspawn gained a new --volatile= switch for running
4674 container instances with /etc or /var unpopulated.
4676 * The kdbus client code has been updated to use the new Linux
4677 3.17 memfd subsystem instead of the old kdbus-specific one.
4679 * systemd-networkd's DHCP client and server now support
4680 FORCERENEW. There are also new configuration options to
4681 configure the vendor client identifier and broadcast mode
4684 * systemd will no longer inform the kernel about the current
4685 timezone, as this is necessarily incorrect and racy as the
4686 kernel has no understanding of DST and similar
4687 concepts. This hence means FAT timestamps will be always
4688 considered UTC, similar to what Android is already
4689 doing. Also, when the RTC is configured to the local time
4690 (rather than UTC) systemd will never synchronize back to it,
4691 as this might confuse Windows at a later boot.
4693 * systemd-analyze gained a new command "verify" for offline
4694 validation of unit files.
4696 * systemd-networkd gained support for a couple of additional
4697 settings for bonding networking setups. Also, the metric for
4698 statically configured routes may now be configured. For
4699 network interfaces where this is appropriate the peer IP
4700 address may now be configured.
4702 * systemd-networkd's DHCP client will no longer request
4703 broadcasting by default, as this tripped up some networks.
4704 For hardware where broadcast is required the feature should
4705 be switched back on using RequestBroadcast=yes.
4707 * systemd-networkd will now set up IPv4LL addresses (when
4708 enabled) even if DHCP is configured successfully.
4710 * udev will now default to respect network device names given
4711 by the kernel when the kernel indicates that these are
4712 predictable. This behavior can be tweaked by changing
4713 NamePolicy= in the relevant .link file.
4715 * A new library systemd-terminal has been added that
4716 implements full TTY stream parsing and rendering. This
4717 library is supposed to be used later on for implementing a
4718 full userspace VT subsystem, replacing the current kernel
4721 * A new tool systemd-journal-upload has been added to push
4722 journal data to a remote system running
4723 systemd-journal-remote.
4725 * journald will no longer forward all local data to another
4726 running syslog daemon. This change has been made because
4727 rsyslog (which appears to be the most commonly used syslog
4728 implementation these days) no longer makes use of this, and
4729 instead pulls the data out of the journal on its own. Since
4730 forwarding the messages to a non-existent syslog server is
4731 more expensive than we assumed we have now turned this
4732 off. If you run a syslog server that is not a recent rsyslog
4733 version, you have to turn this option on again
4734 (ForwardToSyslog= in journald.conf).
4736 * journald now optionally supports the LZ4 compressor for
4737 larger journal fields. This compressor should perform much
4738 better than XZ which was the previous default.
4740 * machinectl now shows the IP addresses of local containers,
4741 if it knows them, plus the interface name of the container.
4743 * A new tool "systemd-escape" has been added that makes it
4744 easy to escape strings to build unit names and similar.
4746 * sd_notify() messages may now include a new ERRNO= field
4747 which is parsed and collected by systemd and shown among the
4748 "systemctl status" output for a service.
4750 * A new component "systemd-firstboot" has been added that
4751 queries the most basic systemd information (timezone,
4752 hostname, root password) interactively on first
4753 boot. Alternatively it may also be used to provision these
4754 things offline on OS images installed into directories.
4756 * The default sysctl.d/ snippets will now set
4758 net.ipv4.conf.default.promote_secondaries=1
4760 This has the benefit of no flushing secondary IP addresses
4761 when primary addresses are removed.
4763 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
4764 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
4765 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
4766 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
4767 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
4768 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
4769 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
4770 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
4771 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
4772 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
4773 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
4774 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
4775 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
4776 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
4777 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
4779 — Berlin, 2014-08-19
4783 * A new tool systemd-sysusers has been added. This tool
4784 creates system users and groups in /etc/passwd and
4785 /etc/group, based on static declarative system user/group
4786 definitions in /usr/lib/sysusers.d/. This is useful to
4787 enable factory resets and volatile systems that boot up with
4788 an empty /etc directory, and thus need system users and
4789 groups created during early boot. systemd now also ships
4790 with two default sysusers.d/ files for the most basic
4791 users and groups systemd and the core operating system
4794 * A new tmpfiles snippet has been added that rebuilds the
4795 essential files in /etc on boot, should they be missing.
4797 * A directive for ensuring automatic clean-up of
4798 /var/cache/man/ has been removed from the default
4799 configuration. This line should now be shipped by the man
4800 implementation. The necessary change has been made to the
4801 man-db implementation. Note that you need to update your man
4802 implementation to one that ships this line, otherwise no
4803 automatic clean-up of /var/cache/man will take place.
4805 * A new condition ConditionNeedsUpdate= has been added that
4806 may conditionalize services to only run when /etc or /var
4807 are "older" than the vendor operating system resources in
4808 /usr. This is useful for reconstructing or updating /etc
4809 after an offline update of /usr or a factory reset, on the
4810 next reboot. Services that want to run once after such an
4811 update or reset should use this condition and order
4812 themselves before the new systemd-update-done.service, which
4813 will mark the two directories as fully updated. A number of
4814 service files have been added making use of this, to rebuild
4815 the udev hardware database, the journald message catalog and
4816 dynamic loader cache (ldconfig). The systemd-sysusers tool
4817 described above also makes use of this now. With this in
4818 place it is now possible to start up a minimal operating
4819 system with /etc empty cleanly. For more information on the
4820 concepts involved see this recent blog story:
4822 http://0pointer.de/blog/projects/stateless.html
4824 * A new system group "input" has been introduced, and all
4825 input device nodes get this group assigned. This is useful
4826 for system-level software to get access to input devices. It
4827 complements what is already done for "audio" and "video".
4829 * systemd-networkd learnt minimal DHCPv4 server support in
4830 addition to the existing DHCPv4 client support. It also
4831 learnt DHCPv6 client and IPv6 Router Solicitation client
4832 support. The DHCPv4 client gained support for static routes
4833 passed in from the server. Note that the [DHCPv4] section
4834 known in older systemd-networkd versions has been renamed to
4835 [DHCP] and is now also used by the DHCPv6 client. Existing
4836 .network files using settings of this section should be
4837 updated, though compatibility is maintained. Optionally, the
4838 client hostname may now be sent to the DHCP server.
4840 * networkd gained support for vxlan virtual networks as well
4841 as tun/tap and dummy devices.
4843 * networkd gained support for automatic allocation of address
4844 ranges for interfaces from a system-wide pool of
4845 addresses. This is useful for dynamically managing a large
4846 number of interfaces with a single network configuration
4847 file. In particular this is useful to easily assign
4848 appropriate IP addresses to the veth links of a large number
4849 of nspawn instances.
4851 * RPM macros for processing sysusers, sysctl and binfmt
4852 drop-in snippets at package installation time have been
4855 * The /etc/os-release file should now be placed in
4856 /usr/lib/os-release. The old location is automatically
4857 created as symlink. /usr/lib is the more appropriate
4858 location of this file, since it shall actually describe the
4859 vendor operating system shipped in /usr, and not the
4860 configuration stored in /etc.
4862 * .mount units gained a new boolean SloppyOptions= setting
4863 that maps to mount(8)'s -s option which enables permissive
4864 parsing of unknown mount options.
4866 * tmpfiles learnt a new "L+" directive which creates a symlink
4867 but (unlike "L") deletes a pre-existing file first, should
4868 it already exist and not already be the correct
4869 symlink. Similarly, "b+", "c+" and "p+" directives have been
4870 added as well, which create block and character devices, as
4871 well as fifos in the filesystem, possibly removing any
4872 pre-existing files of different types.
4874 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
4875 'argument' field (which so far specified the source to
4876 symlink/copy the files from) is now optional. If omitted the
4877 same file os copied from /usr/share/factory/ suffixed by the
4878 full destination path. This is useful for populating /etc
4879 with essential files, by copying them from vendor defaults
4880 shipped in /usr/share/factory/etc.
4882 * A new command "systemctl preset-all" has been added that
4883 applies the service preset settings to all installed unit
4884 files. A new switch --preset-mode= has been added that
4885 controls whether only enable or only disable operations
4888 * A new command "systemctl is-system-running" has been added
4889 that allows checking the overall state of the system, for
4890 example whether it is fully up and running.
4892 * When the system boots up with an empty /etc, the equivalent
4893 to "systemctl preset-all" is executed during early boot, to
4894 make sure all default services are enabled after a factory
4897 * systemd now contains a minimal preset file that enables the
4898 most basic services systemd ships by default.
4900 * Unit files' [Install] section gained a new DefaultInstance=
4901 field for defining the default instance to create if a
4902 template unit is enabled with no instance specified.
4904 * A new passive target cryptsetup-pre.target has been added
4905 that may be used by services that need to make they run and
4906 finish before the first LUKS cryptographic device is set up.
4908 * The /dev/loop-control and /dev/btrfs-control device nodes
4909 are now owned by the "disk" group by default, opening up
4910 access to this group.
4912 * systemd-coredump will now automatically generate a
4913 stack trace of all core dumps taking place on the system,
4914 based on elfutils' libdw library. This stack trace is logged
4917 * systemd-coredump may now optionally store coredumps directly
4918 on disk (in /var/lib/systemd/coredump, possibly compressed),
4919 instead of storing them unconditionally in the journal. This
4920 mode is the new default. A new configuration file
4921 /etc/systemd/coredump.conf has been added to configure this
4922 and other parameters of systemd-coredump.
4924 * coredumpctl gained a new "info" verb to show details about a
4925 specific coredump. A new switch "-1" has also been added
4926 that makes sure to only show information about the most
4927 recent entry instead of all entries. Also, as the tool is
4928 generally useful now the "systemd-" prefix of the binary
4929 name has been removed. Distributions that want to maintain
4930 compatibility with the old name should add a symlink from
4931 the old name to the new name.
4933 * journald's SplitMode= now defaults to "uid". This makes sure
4934 that unprivileged users can access their own coredumps with
4935 coredumpctl without restrictions.
4937 * New kernel command line options "systemd.wants=" (for
4938 pulling an additional unit during boot), "systemd.mask="
4939 (for masking a specific unit for the boot), and
4940 "systemd.debug-shell" (for enabling the debug shell on tty9)
4941 have been added. This is implemented in the new generator
4942 "systemd-debug-generator".
4944 * systemd-nspawn will now by default filter a couple of
4945 syscalls for containers, among them those required for
4946 kernel module loading, direct x86 IO port access, swap
4947 management, and kexec. Most importantly though
4948 open_by_handle_at() is now prohibited for containers,
4949 closing a hole similar to a recently discussed vulnerability
4950 in docker regarding access to files on file hierarchies the
4951 container should normally not have access to. Note that, for
4952 nspawn, we generally make no security claims anyway (and
4953 this is explicitly documented in the man page), so this is
4954 just a fix for one of the most obvious problems.
4956 * A new man page file-hierarchy(7) has been added that
4957 contains a minimized, modernized version of the file system
4958 layout systemd expects, similar in style to the FHS
4959 specification or hier(5). A new tool systemd-path(1) has
4960 been added to query many of these paths for the local
4963 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
4964 longer done. Since the directory now has a per-user size
4965 limit, and is cleaned on logout this appears unnecessary,
4966 in particular since this now brings the lifecycle of this
4967 directory closer in line with how IPC objects are handled.
4969 * systemd.pc now exports a number of additional directories,
4970 including $libdir (which is useful to identify the library
4971 path for the primary architecture of the system), and a
4972 couple of drop-in directories.
4974 * udev's predictable network interface names now use the dev_port
4975 sysfs attribute, introduced in linux 3.15 instead of dev_id to
4976 distinguish between ports of the same PCI function. dev_id should
4977 only be used for ports using the same HW address, hence the need
4980 * machined has been updated to export the OS version of a
4981 container (read from /etc/os-release and
4982 /usr/lib/os-release) on the bus. This is now shown in
4983 "machinectl status" for a machine.
4985 * A new service setting RestartForceExitStatus= has been
4986 added. If configured to a set of exit signals or process
4987 return values, the service will be restarted when the main
4988 daemon process exits with any of them, regardless of the
4991 * systemctl's -H switch for connecting to remote systemd
4992 machines has been extended so that it may be used to
4993 directly connect to a specific container on the
4994 host. "systemctl -H root@foobar:waldi" will now connect as
4995 user "root" to host "foobar", and then proceed directly to
4996 the container named "waldi". Note that currently you have to
4997 authenticate as user "root" for this to work, as entering
4998 containers is a privileged operation.
5000 Contributions from: Andreas Henriksson, Benjamin Steinwender,
5001 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
5002 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
5003 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
5004 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
5005 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
5006 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
5007 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
5008 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
5009 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
5010 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
5011 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
5013 — Berlin, 2014-07-03
5017 * As an experimental feature, udev now tries to lock the
5018 disk device node (flock(LOCK_SH|LOCK_NB)) while it
5019 executes events for the disk or any of its partitions.
5020 Applications like partitioning programs can lock the
5021 disk device node (flock(LOCK_EX)) and claim temporary
5022 device ownership that way; udev will entirely skip all event
5023 handling for this disk and its partitions. If the disk
5024 was opened for writing, the close will trigger a partition
5025 table rescan in udev's "watch" facility, and if needed
5026 synthesize "change" events for the disk and all its partitions.
5027 This is now unconditionally enabled, and if it turns out to
5028 cause major problems, we might turn it on only for specific
5029 devices, or might need to disable it entirely. Device Mapper
5030 devices are excluded from this logic.
5032 * We temporarily dropped the "-l" switch for fsck invocations,
5033 since they collide with the flock() logic above. util-linux
5034 upstream has been changed already to avoid this conflict,
5035 and we will readd "-l" as soon as util-linux with this
5036 change has been released.
5038 * The dependency on libattr has been removed. Since a long
5039 time, the extended attribute calls have moved to glibc, and
5040 libattr is thus unnecessary.
5042 * Virtualization detection works without privileges now. This
5043 means the systemd-detect-virt binary no longer requires
5044 CAP_SYS_PTRACE file capabilities, and our daemons can run
5045 with fewer privileges.
5047 * systemd-networkd now runs under its own "systemd-network"
5048 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
5049 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
5050 loses the ability to write to files owned by root this way.
5052 * Similarly, systemd-resolved now runs under its own
5053 "systemd-resolve" user with no capabilities remaining.
5055 * Similarly, systemd-bus-proxyd now runs under its own
5056 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
5058 * systemd-networkd gained support for setting up "veth"
5059 virtual Ethernet devices for container connectivity, as well
5060 as GRE and VTI tunnels.
5062 * systemd-networkd will no longer automatically attempt to
5063 manually load kernel modules necessary for certain tunnel
5064 transports. Instead, it is assumed the kernel loads them
5065 automatically when required. This only works correctly on
5066 very new kernels. On older kernels, please consider adding
5067 the kernel modules to /etc/modules-load.d/ as a work-around.
5069 * The resolv.conf file systemd-resolved generates has been
5070 moved to /run/systemd/resolve/. If you have a symlink from
5071 /etc/resolv.conf, it might be necessary to correct it.
5073 * Two new service settings, ProtectHome= and ProtectSystem=,
5074 have been added. When enabled, they will make the user data
5075 (such as /home) inaccessible or read-only and the system
5076 (such as /usr) read-only, for specific services. This allows
5077 very light-weight per-service sandboxing to avoid
5078 modifications of user data or system files from
5079 services. These two new switches have been enabled for all
5080 of systemd's long-running services, where appropriate.
5082 * Socket units gained new SocketUser= and SocketGroup=
5083 settings to set the owner user and group of AF_UNIX sockets
5084 and FIFOs in the file system.
5086 * Socket units gained a new RemoveOnStop= setting. If enabled,
5087 all FIFOS and sockets in the file system will be removed
5088 when the specific socket unit is stopped.
5090 * Socket units gained a new Symlinks= setting. It takes a list
5091 of symlinks to create to file system sockets or FIFOs
5092 created by the specific Unix sockets. This is useful to
5093 manage symlinks to socket nodes with the same life-cycle as
5096 * The /dev/log socket and /dev/initctl FIFO have been moved to
5097 /run, and have been replaced by symlinks. This allows
5098 connecting to these facilities even if PrivateDevices=yes is
5099 used for a service (which makes /dev/log itself unavailable,
5100 but /run is left). This also has the benefit of ensuring
5101 that /dev only contains device nodes, directories and
5102 symlinks, and nothing else.
5104 * sd-daemon gained two new calls sd_pid_notify() and
5105 sd_pid_notifyf(). They are similar to sd_notify() and
5106 sd_notifyf(), but allow overriding of the source PID of
5107 notification messages if permissions permit this. This is
5108 useful to send notify messages on behalf of a different
5109 process (for example, the parent process). The
5110 systemd-notify tool has been updated to make use of this
5111 when sending messages (so that notification messages now
5112 originate from the shell script invoking systemd-notify and
5113 not the systemd-notify process itself. This should minimize
5114 a race where systemd fails to associate notification
5115 messages to services when the originating process already
5118 * A new "on-abnormal" setting for Restart= has been added. If
5119 set, it will result in automatic restarts on all "abnormal"
5120 reasons for a process to exit, which includes unclean
5121 signals, core dumps, timeouts and watchdog timeouts, but
5122 does not include clean and unclean exit codes or clean
5123 signals. Restart=on-abnormal is an alternative for
5124 Restart=on-failure for services that shall be able to
5125 terminate and avoid restarts on certain errors, by
5126 indicating so with an unclean exit code. Restart=on-failure
5127 or Restart=on-abnormal is now the recommended setting for
5128 all long-running services.
5130 * If the InaccessibleDirectories= service setting points to a
5131 mount point (or if there are any submounts contained within
5132 it), it is now attempted to completely unmount it, to make
5133 the file systems truly unavailable for the respective
5136 * The ReadOnlyDirectories= service setting and
5137 systemd-nspawn's --read-only parameter are now recursively
5138 applied to all submounts, too.
5140 * Mount units may now be created transiently via the bus APIs.
5142 * The support for SysV and LSB init scripts has been removed
5143 from the systemd daemon itself. Instead, it is now
5144 implemented as a generator that creates native systemd units
5145 from these scripts when needed. This enables us to remove a
5146 substantial amount of legacy code from PID 1, following the
5147 fact that many distributions only ship a very small number
5148 of LSB/SysV init scripts nowadays.
5150 * Privileged Xen (dom0) domains are not considered
5151 virtualization anymore by the virtualization detection
5152 logic. After all, they generally have unrestricted access to
5153 the hardware and usually are used to manage the unprivileged
5156 * systemd-tmpfiles gained a new "C" line type, for copying
5157 files or entire directories.
5159 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
5160 lines. So far, they have been non-globbing versions of the
5161 latter, and have thus been redundant. In future, it is
5162 recommended to only use "z". "m" has hence been removed
5163 from the documentation, even though it stays supported.
5165 * A tmpfiles snippet to recreate the most basic structure in
5166 /var has been added. This is enough to create the /var/run →
5167 /run symlink and create a couple of structural
5168 directories. This allows systems to boot up with an empty or
5169 volatile /var. Of course, while with this change, the core OS
5170 now is capable with dealing with a volatile /var, not all
5171 user services are ready for it. However, we hope that sooner
5172 or later, many service daemons will be changed upstream so
5173 that they are able to automatically create their necessary
5174 directories in /var at boot, should they be missing. This is
5175 the first step to allow state-less systems that only require
5176 the vendor image for /usr to boot.
5178 * systemd-nspawn has gained a new --tmpfs= switch to mount an
5179 empty tmpfs instance to a specific directory. This is
5180 particularly useful for making use of the automatic
5181 reconstruction of /var (see above), by passing --tmpfs=/var.
5183 * Access modes specified in tmpfiles snippets may now be
5184 prefixed with "~", which indicates that they shall be masked
5185 by whether the existing file or directory is currently
5186 writable, readable or executable at all. Also, if specified,
5187 the sgid/suid/sticky bits will be masked for all
5190 * A new passive target unit "network-pre.target" has been
5191 added which is useful for services that shall run before any
5192 network is configured, for example firewall scripts.
5194 * The "floppy" group that previously owned the /dev/fd*
5195 devices is no longer used. The "disk" group is now used
5196 instead. Distributions should probably deprecate usage of
5199 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
5200 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
5201 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
5202 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
5203 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
5204 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
5207 — Berlin, 2014-06-11
5211 * A new "systemd-timesyncd" daemon has been added for
5212 synchronizing the system clock across the network. It
5213 implements an SNTP client. In contrast to NTP
5214 implementations such as chrony or the NTP reference server,
5215 this only implements a client side, and does not bother with
5216 the full NTP complexity, focusing only on querying time from
5217 one remote server and synchronizing the local clock to
5218 it. Unless you intend to serve NTP to networked clients or
5219 want to connect to local hardware clocks, this simple NTP
5220 client should be more than appropriate for most
5221 installations. The daemon runs with minimal privileges, and
5222 has been hooked up with networkd to only operate when
5223 network connectivity is available. The daemon saves the
5224 current clock to disk every time a new NTP sync has been
5225 acquired, and uses this to possibly correct the system clock
5226 early at bootup, in order to accommodate for systems that
5227 lack an RTC such as the Raspberry Pi and embedded devices,
5228 and to make sure that time monotonically progresses on these
5229 systems, even if it is not always correct. To make use of
5230 this daemon, a new system user and group "systemd-timesync"
5231 needs to be created on installation of systemd.
5233 * The queue "seqnum" interface of libudev has been disabled, as
5234 it was generally incompatible with device namespacing as
5235 sequence numbers of devices go "missing" if the devices are
5236 part of a different namespace.
5238 * "systemctl list-timers" and "systemctl list-sockets" gained
5239 a --recursive switch for showing units of these types also
5240 for all local containers, similar in style to the already
5241 supported --recursive switch for "systemctl list-units".
5243 * A new RebootArgument= setting has been added for service
5244 units, which may be used to specify a kernel reboot argument
5245 to use when triggering reboots with StartLimitAction=.
5247 * A new FailureAction= setting has been added for service
5248 units which may be used to specify an operation to trigger
5249 when a service fails. This works similarly to
5250 StartLimitAction=, but unlike it, controls what is done
5251 immediately rather than only after several attempts to
5252 restart the service in question.
5254 * hostnamed got updated to also expose the kernel name,
5255 release, and version on the bus. This is useful for
5256 executing commands like hostnamectl with the -H switch.
5257 systemd-analyze makes use of this to properly display
5258 details when running non-locally.
5260 * The bootchart tool can now show cgroup information in the
5261 graphs it generates.
5263 * The CFS CPU quota cgroup attribute is now exposed for
5264 services. The new CPUQuota= switch has been added for this
5265 which takes a percentage value. Setting this will have the
5266 result that a service may never get more CPU time than the
5267 specified percentage, even if the machine is otherwise idle.
5269 * systemd-networkd learned IPIP and SIT tunnel support.
5271 * LSB init scripts exposing a dependency on $network will now
5272 get a dependency on network-online.target rather than simply
5273 network.target. This should bring LSB handling closer to
5274 what it was on SysV systems.
5276 * A new fsck.repair= kernel option has been added to control
5277 how fsck shall deal with unclean file systems at boot.
5279 * The (.ini) configuration file parser will now silently
5280 ignore sections whose name begins with "X-". This may be
5281 used to maintain application-specific extension sections in unit
5284 * machined gained a new API to query the IP addresses of
5285 registered containers. "machinectl status" has been updated
5286 to show these addresses in its output.
5288 * A new call sd_uid_get_display() has been added to the
5289 sd-login APIs for querying the "primary" session of a
5290 user. The "primary" session of the user is elected from the
5291 user's sessions and generally a graphical session is
5292 preferred over a text one.
5294 * A minimal systemd-resolved daemon has been added. It
5295 currently simply acts as a companion to systemd-networkd and
5296 manages resolv.conf based on per-interface DNS
5297 configuration, possibly supplied via DHCP. In the long run
5298 we hope to extend this into a local DNSSEC enabled DNS and
5301 * The systemd-networkd-wait-online tool is now enabled by
5302 default. It will delay network-online.target until a network
5303 connection has been configured. The tool primarily integrates
5304 with networkd, but will also make a best effort to make sense
5305 of network configuration performed in some other way.
5307 * Two new service options StartupCPUShares= and
5308 StartupBlockIOWeight= have been added that work similarly to
5309 CPUShares= and BlockIOWeight= however only apply during
5310 system startup. This is useful to prioritize certain services
5311 differently during bootup than during normal runtime.
5313 * hostnamed has been changed to prefer the statically
5314 configured hostname in /etc/hostname (unless set to
5315 'localhost' or empty) over any dynamic one supplied by
5316 dhcp. With this change, the rules for picking the hostname
5317 match more closely the rules of other configuration settings
5318 where the local administrator's configuration in /etc always
5319 overrides any other settings.
5321 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
5322 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
5323 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
5324 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
5325 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
5326 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
5327 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
5328 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
5329 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
5330 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
5331 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
5332 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
5333 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
5334 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
5335 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
5336 Lindskog, WaLyong Cho, Will Woods, Zbigniew
5339 — Beijing, 2014-05-28
5343 * When restoring the screen brightness at boot, stay away from
5344 the darkest setting or from the lowest 5% of the available
5345 range, depending on which is the larger value of both. This
5346 should effectively protect the user from rebooting into a
5347 black screen, should the brightness have been set to minimum
5350 * sd-login gained a new sd_machine_get_class() call to
5351 determine the class ("vm" or "container") of a machine
5352 registered with machined.
5354 * sd-login gained new calls
5355 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
5356 to query the identity of the peer of a local AF_UNIX
5357 connection. They operate similarly to their sd_pid_get_xyz()
5360 * PID 1 will now maintain a system-wide system state engine
5361 with the states "starting", "running", "degraded",
5362 "maintenance", "stopping". These states are bound to system
5363 startup, normal runtime, runtime with at least one failed
5364 service, rescue/emergency mode and system shutdown. This
5365 state is shown in the "systemctl status" output when no unit
5366 name is passed. It is useful to determine system state, in
5367 particularly when doing so for many systems or containers at
5370 * A new command "list-machines" has been added to "systemctl"
5371 that lists all local OS containers and shows their system
5372 state (see above), if systemd runs inside of them.
5374 * systemctl gained a new "-r" switch to recursively enumerate
5375 units on all local containers, when used with the
5376 "list-unit" command (which is the default one that is
5377 executed when no parameters are specified).
5379 * The GPT automatic partition discovery logic will now honour
5380 two GPT partition flags: one may be set on a partition to
5381 cause it to be mounted read-only, and the other may be set
5382 on a partition to ignore it during automatic discovery.
5384 * Two new GPT type UUIDs have been added for automatic root
5385 partition discovery, for 32-bit and 64-bit ARM. This is not
5386 particularly useful for discovering the root directory on
5387 these architectures during bare-metal boots (since UEFI is
5388 not common there), but still very useful to allow booting of
5389 ARM disk images in nspawn with the -i option.
5391 * MAC addresses of interfaces created with nspawn's
5392 --network-interface= switch will now be generated from the
5393 machine name, and thus be stable between multiple invocations
5396 * logind will now automatically remove all IPC objects owned
5397 by a user if she or he fully logs out. This makes sure that
5398 users who are logged out cannot continue to consume IPC
5399 resources. This covers SysV memory, semaphores and message
5400 queues as well as POSIX shared memory and message
5401 queues. Traditionally, SysV and POSIX IPC had no life-cycle
5402 limits. With this functionality, that is corrected. This may
5403 be turned off by using the RemoveIPC= switch of logind.conf.
5405 * The systemd-machine-id-setup and tmpfiles tools gained a
5406 --root= switch to operate on a specific root directory,
5409 * journald can now forward logged messages to the TTYs of all
5410 logged in users ("wall"). This is the default for all
5411 emergency messages now.
5413 * A new tool systemd-journal-remote has been added to stream
5414 journal log messages across the network.
5416 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
5417 controller trees are mounted into it. Note that the
5418 directories mounted beneath it are not read-only. This is a
5419 security measure and is particularly useful because glibc
5420 actually includes a search logic to pick any tmpfs it can
5421 find to implement shm_open() if /dev/shm is not available
5422 (which it might very well be in namespaced setups).
5424 * machinectl gained a new "poweroff" command to cleanly power
5425 down a local OS container.
5427 * The PrivateDevices= unit file setting will now also drop the
5428 CAP_MKNOD capability from the capability bound set, and
5429 imply DevicePolicy=closed.
5431 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
5432 comprehensively on all long-running systemd services where
5433 this is appropriate.
5435 * systemd-udevd will now run in a disassociated mount
5436 namespace. To mount directories from udev rules, make sure to
5437 pull in mount units via SYSTEMD_WANTS properties.
5439 * The kdbus support gained support for uploading policy into
5440 the kernel. sd-bus gained support for creating "monitoring"
5441 connections that can eavesdrop into all bus communication
5442 for debugging purposes.
5444 * Timestamps may now be specified in seconds since the UNIX
5445 epoch Jan 1st, 1970 by specifying "@" followed by the value
5448 * Native tcpwrap support in systemd has been removed. tcpwrap
5449 is old code, not really maintained anymore and has serious
5450 shortcomings, and better options such as firewalls
5451 exist. For setups that require tcpwrap usage, please
5452 consider invoking your socket-activated service via tcpd,
5453 like on traditional inetd.
5455 * A new system.conf configuration option
5456 DefaultTimerAccuracySec= has been added that controls the
5457 default AccuracySec= setting of .timer units.
5459 * Timer units gained a new WakeSystem= switch. If enabled,
5460 timers configured this way will cause the system to resume
5461 from system suspend (if the system supports that, which most
5464 * Timer units gained a new Persistent= switch. If enabled,
5465 timers configured this way will save to disk when they have
5466 been last triggered. This information is then used on next
5467 reboot to possible execute overdue timer events, that
5468 could not take place because the system was powered off.
5469 This enables simple anacron-like behaviour for timer units.
5471 * systemctl's "list-timers" will now also list the time a
5472 timer unit was last triggered in addition to the next time
5473 it will be triggered.
5475 * systemd-networkd will now assign predictable IPv4LL
5476 addresses to its local interfaces.
5478 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
5479 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
5480 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
5481 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
5482 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
5483 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
5484 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
5485 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
5488 — Berlin, 2014-03-25
5492 * A new unit file setting RestrictAddressFamilies= has been
5493 added to restrict which socket address families unit
5494 processes gain access to. This takes address family names
5495 like "AF_INET" or "AF_UNIX", and is useful to minimize the
5496 attack surface of services via exotic protocol stacks. This
5497 is built on seccomp system call filters.
5499 * Two new unit file settings RuntimeDirectory= and
5500 RuntimeDirectoryMode= have been added that may be used to
5501 manage a per-daemon runtime directories below /run. This is
5502 an alternative for setting up directory permissions with
5503 tmpfiles snippets, and has the advantage that the runtime
5504 directory's lifetime is bound to the daemon runtime and that
5505 the daemon starts up with an empty directory each time. This
5506 is particularly useful when writing services that drop
5507 privileges using the User= or Group= setting.
5509 * The DeviceAllow= unit setting now supports globbing for
5510 matching against device group names.
5512 * The systemd configuration file system.conf gained new
5513 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
5514 DefaultMemoryAccounting= to globally turn on/off accounting
5515 for specific resources (cgroups) for all units. These
5516 settings may still be overridden individually in each unit
5519 * systemd-gpt-auto-generator is now able to discover /srv and
5520 root partitions in addition to /home and swap partitions. It
5521 also supports LUKS-encrypted partitions now. With this in
5522 place, automatic discovery of partitions to mount following
5523 the Discoverable Partitions Specification
5524 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
5525 is now a lot more complete. This allows booting without
5526 /etc/fstab and without root= on the kernel command line on
5527 systems prepared appropriately.
5529 * systemd-nspawn gained a new --image= switch which allows
5530 booting up disk images and Linux installations on any block
5531 device that follow the Discoverable Partitions Specification
5532 (see above). This means that installations made with
5533 appropriately updated installers may now be started and
5534 deployed using container managers, completely
5535 unmodified. (We hope that libvirt-lxc will add support for
5536 this feature soon, too.)
5538 * systemd-nspawn gained a new --network-macvlan= setting to
5539 set up a private macvlan interface for the
5540 container. Similarly, systemd-networkd gained a new
5541 Kind=macvlan setting in .netdev files.
5543 * systemd-networkd now supports configuring local addresses
5546 * A new tool systemd-network-wait-online has been added to
5547 synchronously wait for network connectivity using
5550 * The sd-bus.h bus API gained a new sd_bus_track object for
5551 tracking the life-cycle of bus peers. Note that sd-bus.h is
5552 still not a public API though (unless you specify
5553 --enable-kdbus on the configure command line, which however
5554 voids your warranty and you get no API stability guarantee).
5556 * The $XDG_RUNTIME_DIR runtime directories for each user are
5557 now individual tmpfs instances, which has the benefit of
5558 introducing separate pools for each user, with individual
5559 size limits, and thus making sure that unprivileged clients
5560 can no longer negatively impact the system or other users by
5561 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
5562 RuntimeDirectorySize= has been introduced that allows
5563 controlling the default size limit for all users. It
5564 defaults to 10% of the available physical memory. This is no
5565 replacement for quotas on tmpfs though (which the kernel
5566 still does not support), as /dev/shm and /tmp are still
5567 shared resources used by both the system and unprivileged
5570 * logind will now automatically turn off automatic suspending
5571 on laptop lid close when more than one display is
5572 connected. This was previously expected to be implemented
5573 individually in desktop environments (such as GNOME),
5574 however has been added to logind now, in order to fix a
5575 boot-time race where a desktop environment might not have
5576 been started yet and thus not been able to take an inhibitor
5577 lock at the time where logind already suspends the system
5578 due to a closed lid.
5580 * logind will now wait at least 30s after each system
5581 suspend/resume cycle, and 3min after system boot before
5582 suspending the system due to a closed laptop lid. This
5583 should give USB docking stations and similar enough time to
5584 be probed and configured after system resume and boot in
5585 order to then act as suspend blocker.
5587 * systemd-run gained a new --property= setting which allows
5588 initialization of resource control properties (and others)
5589 for the created scope or service unit. Example: "systemd-run
5590 --property=BlockIOWeight=10 updatedb" may be used to run
5591 updatedb at a low block IO scheduling weight.
5593 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
5594 now also work in --scope mode.
5596 * When systemd is compiled with kdbus support, basic support
5597 for enforced policies is now in place. (Note that enabling
5598 kdbus still voids your warranty and no API compatibility
5601 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
5602 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5603 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
5604 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
5605 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
5606 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
5607 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
5608 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
5609 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
5610 Zbigniew Jędrzejewski-Szmek
5612 — Berlin, 2014-03-12
5616 * systemd will now relabel /dev after loading the SMACK policy
5617 according to SMACK rules.
5619 * A new unit file option AppArmorProfile= has been added to
5620 set the AppArmor profile for the processes of a unit.
5622 * A new condition check ConditionArchitecture= has been added
5623 to conditionalize units based on the system architecture, as
5624 reported by uname()'s "machine" field.
5626 * systemd-networkd now supports matching on the system
5627 virtualization, architecture, kernel command line, host name
5630 * logind is now a lot more aggressive when suspending the
5631 machine due to a closed laptop lid. Instead of acting only
5632 on the lid close action, it will continuously watch the lid
5633 status and act on it. This is useful for laptops where the
5634 power button is on the outside of the chassis so that it can
5635 be reached without opening the lid (such as the Lenovo
5636 Yoga). On those machines, logind will now immediately
5637 re-suspend the machine if the power button has been
5638 accidentally pressed while the laptop was suspended and in a
5639 backpack or similar.
5641 * logind will now watch SW_DOCK switches and inhibit reaction
5642 to the lid switch if it is pressed. This means that logind
5643 will not suspend the machine anymore if the lid is closed
5644 and the system is docked, if the laptop supports SW_DOCK
5645 notifications via the input layer. Note that ACPI docking
5646 stations do not generate this currently. Also note that this
5647 logic is usually not fully sufficient and Desktop
5648 Environments should take a lid switch inhibitor lock when an
5649 external display is connected, as systemd will not watch
5652 * nspawn will now make use of the devices cgroup controller by
5653 default, and only permit creation of and access to the usual
5654 API device nodes like /dev/null or /dev/random, as well as
5655 access to (but not creation of) the pty devices.
5657 * We will now ship a default .network file for
5658 systemd-networkd that automatically configures DHCP for
5659 network interfaces created by nspawn's --network-veth or
5660 --network-bridge= switches.
5662 * systemd will now understand the usual M, K, G, T suffixes
5663 according to SI conventions (i.e. to the base 1000) when
5664 referring to throughput and hardware metrics. It will stay
5665 with IEC conventions (i.e. to the base 1024) for software
5666 metrics, according to what is customary according to
5667 Wikipedia. We explicitly document which base applies for
5668 each configuration option.
5670 * The DeviceAllow= setting in unit files now supports a syntax
5671 to whitelist an entire group of devices node majors at once,
5672 based on the /proc/devices listing. For example, with the
5673 string "char-pts", it is now possible to whitelist all
5674 current and future pseudo-TTYs at once.
5676 * sd-event learned a new "post" event source. Event sources of
5677 this type are triggered by the dispatching of any event
5678 source of a type that is not "post". This is useful for
5679 implementing clean-up and check event sources that are
5680 triggered by other work being done in the program.
5682 * systemd-networkd is no longer statically enabled, but uses
5683 the usual [Install] sections so that it can be
5684 enabled/disabled using systemctl. It still is enabled by
5687 * When creating a veth interface pair with systemd-nspawn, the
5688 host side will now be prefixed with "vb-" if
5689 --network-bridge= is used, and with "ve-" if --network-veth
5690 is used. This way, it is easy to distinguish these cases on
5691 the host, for example to apply different configuration to
5692 them with systemd-networkd.
5694 * The compatibility libraries for libsystemd-journal.so,
5695 libsystem-id128.so, libsystemd-login.so and
5696 libsystemd-daemon.so do not make use of IFUNC
5697 anymore. Instead, we now build libsystemd.so multiple times
5698 under these alternative names. This means that the footprint
5699 is drastically increased, but given that these are
5700 transitional compatibility libraries, this should not matter
5701 much. This change has been made necessary to support the ARM
5702 platform for these compatibility libraries, as the ARM
5703 toolchain is not really at the same level as the toolchain
5704 for other architectures like x86 and does not support
5705 IFUNC. Please make sure to use --enable-compat-libs only
5706 during a transitional period!
5708 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
5709 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
5710 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
5711 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
5712 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
5713 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
5714 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
5715 Zbigniew Jędrzejewski-Szmek
5717 — Berlin, 2014-02-24
5721 * A new component "systemd-networkd" has been added that can
5722 be used to configure local network interfaces statically or
5723 via DHCP. It is capable of bringing up bridges, VLANs, and
5724 bonding. Currently, no hook-ups for interactive network
5725 configuration are provided. Use this for your initrd,
5726 container, embedded, or server setup if you need a simple,
5727 yet powerful, network configuration solution. This
5728 configuration subsystem is quite nifty, as it allows wildcard
5729 hotplug matching in interfaces. For example, with a single
5730 configuration snippet, you can configure that all Ethernet
5731 interfaces showing up are automatically added to a bridge,
5732 or similar. It supports link-sensing and more.
5734 * A new tool "systemd-socket-proxyd" has been added which can
5735 act as a bidirectional proxy for TCP sockets. This is
5736 useful for adding socket activation support to services that
5737 do not actually support socket activation, including virtual
5738 machines and the like.
5740 * Add a new tool to save/restore rfkill state on
5743 * Save/restore state of keyboard backlights in addition to
5744 display backlights on shutdown/boot.
5746 * udev learned a new SECLABEL{} construct to label device
5747 nodes with a specific security label when they appear. For
5748 now, only SECLABEL{selinux} is supported, but the syntax is
5749 prepared for additional security frameworks.
5751 * udev gained a new scheme to configure link-level attributes
5752 from files in /etc/systemd/network/*.link. These files can
5753 match against MAC address, device path, driver name and type,
5754 and will apply attributes like the naming policy, link speed,
5755 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
5756 address assignment policy (randomized, ...).
5758 * The configuration of network interface naming rules for
5759 "permanent interface names" has changed: a new NamePolicy=
5760 setting in the [Link] section of .link files determines the
5761 priority of possible naming schemes (onboard, slot, MAC,
5762 path). The default value of this setting is determined by
5763 /usr/lib/net/links/99-default.link. Old
5764 80-net-name-slot.rules udev configuration file has been
5765 removed, so local configuration overriding this file should
5766 be adapted to override 99-default.link instead.
5768 * When the User= switch is used in a unit file, also
5769 initialize $SHELL= based on the user database entry.
5771 * systemd no longer depends on libdbus. All communication is
5772 now done with sd-bus, systemd's low-level bus library
5775 * kdbus support has been added to PID 1 itself. When kdbus is
5776 enabled, this causes PID 1 to set up the system bus and
5777 enable support for a new ".busname" unit type that
5778 encapsulates bus name activation on kdbus. It works a little
5779 bit like ".socket" units, except for bus names. A new
5780 generator has been added that converts classic dbus1 service
5781 activation files automatically into native systemd .busname
5784 * sd-bus: add a light-weight vtable implementation that allows
5785 defining objects on the bus with a simple static const
5786 vtable array of its methods, signals and properties.
5788 * systemd will not generate or install static dbus
5789 introspection data anymore to /usr/share/dbus-1/interfaces,
5790 as the precise format of these files is unclear, and
5791 nothing makes use of it.
5793 * A proxy daemon is now provided to proxy clients connecting
5794 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
5795 compatibility with classic D-Bus.
5797 * A bus driver implementation has been added that supports the
5798 classic D-Bus bus driver calls on kdbus, also for
5799 compatibility purposes.
5801 * A new API "sd-event.h" has been added that implements a
5802 minimal event loop API built around epoll. It provides a
5803 couple of features that direct epoll usage is lacking:
5804 prioritization of events, scales to large numbers of timer
5805 events, per-event timer slack (accuracy), system-wide
5806 coalescing of timer events, exit handlers, watchdog
5807 supervision support using systemd's sd_notify() API, child
5810 * A new API "sd-rntl.h" has been added that provides an API
5811 around the route netlink interface of the kernel, similar in
5812 style to "sd-bus.h".
5814 * A new API "sd-dhcp-client.h" has been added that provides a
5815 small DHCPv4 client-side implementation. This is used by
5818 * There is a new kernel command line option
5819 "systemd.restore_state=0|1". When set to "0", none of the
5820 systemd tools will restore saved runtime state to hardware
5821 devices. More specifically, the rfkill and backlight states
5824 * The FsckPassNo= compatibility option in mount/service units
5825 has been removed. The fstab generator will now add the
5826 necessary dependencies automatically, and does not require
5827 PID1's support for that anymore.
5829 * journalctl gained a new switch, --list-boots, that lists
5830 recent boots with their times and boot IDs.
5832 * The various tools like systemctl, loginctl, timedatectl,
5833 busctl, systemd-run, ... have gained a new switch "-M" to
5834 connect to a specific, local OS container (as direct
5835 connection, without requiring SSH). This works on any
5836 container that is registered with machined, such as those
5837 created by libvirt-lxc or nspawn.
5839 * systemd-run and systemd-analyze also gained support for "-H"
5840 to connect to remote hosts via SSH. This is particularly
5841 useful for systemd-run because it enables queuing of jobs
5842 onto remote systems.
5844 * machinectl gained a new command "login" to open a getty
5845 login in any local container. This works with any container
5846 that is registered with machined (such as those created by
5847 libvirt-lxc or nspawn), and which runs systemd inside.
5849 * machinectl gained a new "reboot" command that may be used to
5850 trigger a reboot on a specific container that is registered
5851 with machined. This works on any container that runs an init
5852 system of some kind.
5854 * systemctl gained a new "list-timers" command to print a nice
5855 listing of installed timer units with the times they elapse
5858 * Alternative reboot() parameters may now be specified on the
5859 "systemctl reboot" command line and are passed to the
5860 reboot() system call.
5862 * systemctl gained a new --job-mode= switch to configure the
5863 mode to queue a job with. This is a more generic version of
5864 --fail, --irreversible, and --ignore-dependencies, which are
5865 still available but not advertised anymore.
5867 * /etc/systemd/system.conf gained new settings to configure
5868 various default timeouts of units, as well as the default
5869 start limit interval and burst. These may still be overridden
5872 * PID1 will now export on the bus profile data of the security
5873 policy upload process (such as the SELinux policy upload to
5876 * journald: when forwarding logs to the console, include
5877 timestamps (following the setting in
5878 /sys/module/printk/parameters/time).
5880 * OnCalendar= in timer units now understands the special
5881 strings "yearly" and "annually". (Both are equivalent)
5883 * The accuracy of timer units is now configurable with the new
5884 AccuracySec= setting. It defaults to 1min.
5886 * A new dependency type JoinsNamespaceOf= has been added that
5887 allows running two services within the same /tmp and network
5888 namespace, if PrivateNetwork= or PrivateTmp= are used.
5890 * A new command "cat" has been added to systemctl. It outputs
5891 the original unit file of a unit, and concatenates the
5892 contents of additional "drop-in" unit file snippets, so that
5893 the full configuration is shown.
5895 * systemctl now supports globbing on the various "list-xyz"
5896 commands, like "list-units" or "list-sockets", as well as on
5897 those commands which take multiple unit names.
5899 * journalctl's --unit= switch gained support for globbing.
5901 * All systemd daemons now make use of the watchdog logic so
5902 that systemd automatically notices when they hang.
5904 * If the $container_ttys environment variable is set,
5905 getty-generator will automatically spawn a getty for each
5906 listed tty. This is useful for container managers to request
5907 login gettys to be spawned on as many ttys as needed.
5909 * %h, %s, %U specifier support is not available anymore when
5910 used in unit files for PID 1. This is because NSS calls are
5911 not safe from PID 1. They stay available for --user
5912 instances of systemd, and as special case for the root user.
5914 * loginctl gained a new "--no-legend" switch to turn off output
5917 * The "sd-login.h" API gained three new calls:
5918 sd_session_is_remote(), sd_session_get_remote_user(),
5919 sd_session_get_remote_host() to query information about
5922 * The udev hardware database now also carries vendor/product
5923 information of SDIO devices.
5925 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
5926 determine whether watchdog notifications are requested by
5929 * Socket-activated per-connection services now include a
5930 short description of the connection parameters in the
5933 * tmpfiles gained a new "--boot" option. When this is not used,
5934 only lines where the command character is not suffixed with
5935 "!" are executed. When this option is specified, those
5936 options are executed too. This partitions tmpfiles
5937 directives into those that can be safely executed at any
5938 time, and those which should be run only at boot (for
5939 example, a line that creates /run/nologin).
5941 * A new API "sd-resolve.h" has been added which provides a simple
5942 asynchronous wrapper around glibc NSS host name resolution
5943 calls, such as getaddrinfo(). In contrast to glibc's
5944 getaddrinfo_a(), it does not use signals. In contrast to most
5945 other asynchronous name resolution libraries, this one does
5946 not reimplement DNS, but reuses NSS, so that alternate
5947 host name resolution systems continue to work, such as mDNS,
5948 LDAP, etc. This API is based on libasyncns, but it has been
5949 cleaned up for inclusion in systemd.
5951 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
5952 "sd-daemon.h" are no longer found in individual libraries
5953 libsystemd-journal.so, libsystemd-login.so,
5954 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
5955 merged them into a single library, libsystemd.so, which
5956 provides all symbols. The reason for this is cyclic
5957 dependencies, as these libraries tend to use each other's
5958 symbols. So far, we have managed to workaround that by linking
5959 a copy of a good part of our code into each of these
5960 libraries again and again, which, however, makes certain
5961 things hard to do, like sharing static variables. Also, it
5962 substantially increases footprint. With this change, there
5963 is only one library for the basic APIs systemd
5964 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
5965 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
5966 library as well, however are subject to the --enable-kdbus
5967 switch (see below). Note that "sd-dhcp-client.h" is not part
5968 of this library (this is because it only consumes, never
5969 provides, services of/to other APIs). To make the transition
5970 easy from the separate libraries to the unified one, we
5971 provide the --enable-compat-libs compile-time switch which
5972 will generate stub libraries that are compatible with the
5973 old ones but redirect all calls to the new one.
5975 * All of the kdbus logic and the new APIs "sd-bus.h",
5976 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
5977 and "sd-utf8.h" are compile-time optional via the
5978 "--enable-kdbus" switch, and they are not compiled in by
5979 default. To make use of kdbus, you have to explicitly enable
5980 the switch. Note however, that neither the kernel nor the
5981 userspace API for all of this is considered stable yet. We
5982 want to maintain the freedom to still change the APIs for
5983 now. By specifying this build-time switch, you acknowledge
5984 that you are aware of the instability of the current
5987 * Also, note that while kdbus is pretty much complete,
5988 it lacks one thing: proper policy support. This means you
5989 can build a fully working system with all features; however,
5990 it will be highly insecure. Policy support will be added in
5991 one of the next releases, at the same time that we will
5992 declare the APIs stable.
5994 * When the kernel command line argument "kdbus" is specified,
5995 systemd will automatically load the kdbus.ko kernel module. At
5996 this stage of development, it is only useful for testing kdbus
5997 and should not be used in production. Note: if "--enable-kdbus"
5998 is specified, and the kdbus.ko kernel module is available, and
5999 "kdbus" is added to the kernel command line, the entire system
6000 runs with kdbus instead of dbus-daemon, with the above mentioned
6001 problem of missing the system policy enforcement. Also a future
6002 version of kdbus.ko or a newer systemd will not be compatible with
6003 each other, and will unlikely be able to boot the machine if only
6004 one of them is updated.
6006 * systemctl gained a new "import-environment" command which
6007 uploads the caller's environment (or parts thereof) into the
6008 service manager so that it is inherited by services started
6009 by the manager. This is useful to upload variables like
6010 $DISPLAY into the user service manager.
6012 * A new PrivateDevices= switch has been added to service units
6013 which allows running a service with a namespaced /dev
6014 directory that does not contain any device nodes for
6015 physical devices. More specifically, it only includes devices
6016 such as /dev/null, /dev/urandom, and /dev/zero which are API
6019 * logind has been extended to support behaviour like VT
6020 switching on seats that do not support a VT. This makes
6021 multi-session available on seats that are not the first seat
6022 (seat0), and on systems where kernel support for VTs has
6023 been disabled at compile-time.
6025 * If a process holds a delay lock for system sleep or shutdown
6026 and fails to release it in time, we will now log its
6027 identity. This makes it easier to identify processes that
6028 cause slow suspends or power-offs.
6030 * When parsing /etc/crypttab, support for a new key-slot=
6031 option as supported by Debian is added. It allows indicating
6032 which LUKS slot to use on disk, speeding up key loading.
6034 * The sd_journal_sendv() API call has been checked and
6035 officially declared to be async-signal-safe so that it may
6036 be invoked from signal handlers for logging purposes.
6038 * Boot-time status output is now enabled automatically after a
6039 short timeout if boot does not progress, in order to give
6040 the user an indication what she or he is waiting for.
6042 * The boot-time output has been improved to show how much time
6043 remains until jobs expire.
6045 * The KillMode= switch in service units gained a new possible
6046 value "mixed". If set, and the unit is shut down, then the
6047 initial SIGTERM signal is sent only to the main daemon
6048 process, while the following SIGKILL signal is sent to
6049 all remaining processes of the service.
6051 * When a scope unit is registered, a new property "Controller"
6052 may be set. If set to a valid bus name, systemd will send a
6053 RequestStop() signal to this name when it would like to shut
6054 down the scope. This may be used to hook manager logic into
6055 the shutdown logic of scope units. Also, scope units may now
6056 be put in a special "abandoned" state, in which case the
6057 manager process which created them takes no further
6058 responsibilities for it.
6060 * When reading unit files, systemd will now verify
6061 the access mode of these files, and warn about certain
6062 suspicious combinations. This has been added to make it
6063 easier to track down packaging bugs where unit files are
6064 marked executable or world-writable.
6066 * systemd-nspawn gained a new "--setenv=" switch to set
6067 container-wide environment variables. The similar option in
6068 systemd-activate was renamed from "--environment=" to
6069 "--setenv=" for consistency.
6071 * systemd-nspawn has been updated to create a new kdbus domain
6072 for each container that is invoked, thus allowing each
6073 container to have its own set of system and user buses,
6074 independent of the host.
6076 * systemd-nspawn gained a new --drop-capability= switch to run
6077 the container with less capabilities than the default. Both
6078 --drop-capability= and --capability= now take the special
6079 string "all" for dropping or keeping all capabilities.
6081 * systemd-nspawn gained new switches for executing containers
6082 with specific SELinux labels set.
6084 * systemd-nspawn gained a new --quiet switch to not generate
6085 any additional output but the container's own console
6088 * systemd-nspawn gained a new --share-system switch to run a
6089 container without PID namespacing enabled.
6091 * systemd-nspawn gained a new --register= switch to control
6092 whether the container is registered with systemd-machined or
6093 not. This is useful for containers that do not run full
6094 OS images, but only specific apps.
6096 * systemd-nspawn gained a new --keep-unit which may be used
6097 when invoked as the only program from a service unit, and
6098 results in registration of the unit service itself in
6099 systemd-machined, instead of a newly opened scope unit.
6101 * systemd-nspawn gained a new --network-interface= switch for
6102 moving arbitrary interfaces to the container. The new
6103 --network-veth switch creates a virtual Ethernet connection
6104 between host and container. The new --network-bridge=
6105 switch then allows assigning the host side of this virtual
6106 Ethernet connection to a bridge device.
6108 * systemd-nspawn gained a new --personality= switch for
6109 setting the kernel personality for the container. This is
6110 useful when running a 32-bit container on a 64-bit host. A
6111 similar option Personality= is now also available for service
6114 * logind will now also track a "Desktop" identifier for each
6115 session which encodes the desktop environment of it. This is
6116 useful for desktop environments that want to identify
6117 multiple running sessions of itself easily.
6119 * A new SELinuxContext= setting for service units has been
6120 added that allows setting a specific SELinux execution
6121 context for a service.
6123 * Most systemd client tools will now honour $SYSTEMD_LESS for
6124 settings of the "less" pager. By default, these tools will
6125 override $LESS to allow certain operations to work, such as
6126 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
6127 influence this logic.
6129 * systemd's "seccomp" hook-up has been changed to make use of
6130 the libseccomp library instead of using its own
6131 implementation. This has benefits for portability among
6134 * For usage together with SystemCallFilter=, a new
6135 SystemCallErrorNumber= setting has been introduced that
6136 allows configuration of a system error number to be returned
6137 on filtered system calls, instead of immediately killing the
6138 process. Also, SystemCallArchitectures= has been added to
6139 limit access to system calls of a particular architecture
6140 (in order to turn off support for unused secondary
6141 architectures). There is also a global
6142 SystemCallArchitectures= setting in system.conf now to turn
6143 off support for non-native system calls system-wide.
6145 * systemd requires a kernel with a working name_to_handle_at(),
6146 please see the kernel config requirements in the README file.
6148 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
6149 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
6150 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
6151 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
6152 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
6153 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
6154 Elia Pinto, Florian Weimer, George McCollister, Goffredo
6155 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
6156 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
6157 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
6158 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
6159 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
6160 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
6161 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
6162 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
6163 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
6164 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
6165 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
6166 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
6167 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
6168 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
6169 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
6170 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
6171 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
6173 — Berlin, 2014-02-20
6177 * logind has gained support for facilitating privileged input
6178 and drm device access for unprivileged clients. This work is
6179 useful to allow Wayland display servers (and similar
6180 programs, such as kmscon) to run under the user's ID and
6181 access input and drm devices which are normally
6182 protected. When this is used (and the kernel is new enough)
6183 logind will "mute" IO on the file descriptors passed to
6184 Wayland as long as it is in the background and "unmute" it
6185 if it returns into the foreground. This allows secure
6186 session switching without allowing background sessions to
6187 eavesdrop on input and display data. This also introduces
6188 session switching support if VT support is turned off in the
6189 kernel, and on seats that are not seat0.
6191 * A new kernel command line option luks.options= is understood
6192 now which allows specifying LUKS options for usage for LUKS
6193 encrypted partitions specified with luks.uuid=.
6195 * tmpfiles.d(5) snippets may now use specifier expansion in
6196 path names. More specifically %m, %b, %H, %v, are now
6197 replaced by the local machine id, boot id, hostname, and
6198 kernel version number.
6200 * A new tmpfiles.d(5) command "m" has been introduced which
6201 may be used to change the owner/group/access mode of a file
6202 or directory if it exists, but do nothing if it does not.
6204 * This release removes high-level support for the
6205 MemorySoftLimit= cgroup setting. The underlying kernel
6206 cgroup attribute memory.soft_limit= is currently badly
6207 designed and likely to be removed from the kernel API in its
6208 current form, hence we should not expose it for now.
6210 * The memory.use_hierarchy cgroup attribute is now enabled for
6211 all cgroups systemd creates in the memory cgroup
6212 hierarchy. This option is likely to be come the built-in
6213 default in the kernel anyway, and the non-hierarchical mode
6214 never made much sense in the intrinsically hierarchical
6217 * A new field _SYSTEMD_SLICE= is logged along with all journal
6218 messages containing the slice a message was generated
6219 from. This is useful to allow easy per-customer filtering of
6220 logs among other things.
6222 * systemd-journald will no longer adjust the group of journal
6223 files it creates to the "systemd-journal" group. Instead we
6224 rely on the journal directory to be owned by the
6225 "systemd-journal" group, and its setgid bit set, so that the
6226 kernel file system layer will automatically enforce that
6227 journal files inherit this group assignment. The reason for
6228 this change is that we cannot allow NSS look-ups from
6229 journald which would be necessary to resolve
6230 "systemd-journal" to a numeric GID, because this might
6231 create deadlocks if NSS involves synchronous queries to
6232 other daemons (such as nscd, or sssd) which in turn are
6233 logging clients of journald and might block on it, which
6234 would then dead lock. A tmpfiles.d(5) snippet included in
6235 systemd will make sure the setgid bit and group are
6236 properly set on the journal directory if it exists on every
6237 boot. However, we recommend adjusting it manually after
6238 upgrades too (or from RPM scriptlets), so that the change is
6239 not delayed until next reboot.
6241 * Backlight and random seed files in /var/lib/ have moved into
6242 the /var/lib/systemd/ directory, in order to centralize all
6243 systemd generated files in one directory.
6245 * Boot time performance measurements (as displayed by
6246 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
6247 performance information if that's available to determine how
6248 much time BIOS and boot loader initialization required. With
6249 a sufficiently new BIOS you hence no longer need to boot
6250 with Gummiboot to get access to such information.
6252 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
6253 Cristian Rodríguez, Dave Reisner, David Herrmann, David
6254 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
6255 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
6256 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
6257 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
6258 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6260 — Berlin, 2013-10-02
6264 * The Restart= option for services now understands a new
6265 on-watchdog setting, which will restart the service
6266 automatically if the service stops sending out watchdog keep
6267 alive messages (as configured with WatchdogSec=).
6269 * The getty generator (which is responsible for bringing up a
6270 getty on configured serial consoles) will no longer only
6271 start a getty on the primary kernel console but on all
6272 others, too. This makes the order in which console= is
6273 specified on the kernel command line less important.
6275 * libsystemd-logind gained a new sd_session_get_vt() call to
6276 retrieve the VT number of a session.
6278 * If the option "tries=0" is set for an entry of /etc/crypttab
6279 its passphrase is queried indefinitely instead of any
6280 maximum number of tries.
6282 * If a service with a configure PID file terminates its PID
6283 file will now be removed automatically if it still exists
6284 afterwards. This should put an end to stale PID files.
6286 * systemd-run will now also take relative binary path names
6287 for execution and no longer insists on absolute paths.
6289 * InaccessibleDirectories= and ReadOnlyDirectories= now take
6290 paths that are optionally prefixed with "-" to indicate that
6291 it should not be considered a failure if they do not exist.
6293 * journalctl -o (and similar commands) now understands a new
6294 output mode "short-precise", it is similar to "short" but
6295 shows timestamps with usec accuracy.
6297 * The option "discard" (as known from Debian) is now
6298 synonymous to "allow-discards" in /etc/crypttab. In fact,
6299 "discard" is preferred now (since it is easier to remember
6302 * Some licensing clean-ups were made, so that more code is now
6303 LGPL-2.1 licensed than before.
6305 * A minimal tool to save/restore the display backlight
6306 brightness across reboots has been added. It will store the
6307 backlight setting as late as possible at shutdown, and
6308 restore it as early as possible during reboot.
6310 * A logic to automatically discover and enable home and swap
6311 partitions on GPT disks has been added. With this in place
6312 /etc/fstab becomes optional for many setups as systemd can
6313 discover certain partitions located on the root disk
6314 automatically. Home partitions are recognized under their
6315 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
6316 partitions are recognized under their GPT type ID
6317 0657fd6da4ab43c484e50933c84b4f4f.
6319 * systemd will no longer pass any environment from the kernel
6320 or initrd to system services. If you want to set an
6321 environment for all services, do so via the kernel command
6322 line systemd.setenv= assignment.
6324 * The systemd-sysctl tool no longer natively reads the file
6325 /etc/sysctl.conf. If desired, the file should be symlinked
6326 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
6327 legacy support by a symlink rather than built-in code, it
6328 also makes the otherwise hidden order of application of the
6329 different files visible. (Note that this partly reverts to a
6330 pre-198 application order of sysctl knobs!)
6332 * The "systemctl set-log-level" and "systemctl dump" commands
6333 have been moved to systemd-analyze.
6335 * systemd-run learned the new --remain-after-exit switch,
6336 which causes the scope unit not to be cleaned up
6337 automatically after the process terminated.
6339 * tmpfiles learned a new --exclude-prefix= switch to exclude
6340 certain paths from operation.
6342 * journald will now automatically flush all messages to disk
6343 as soon as a message at the log level CRIT, ALERT or EMERG
6346 Contributions from: Andrew Cook, Brandon Philips, Christian
6347 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
6348 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
6349 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
6350 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
6351 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
6352 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
6353 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
6354 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
6355 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
6356 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
6357 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
6358 William Giokas, Zbigniew Jędrzejewski-Szmek
6360 — Berlin, 2013-09-13
6364 * The documentation has been updated to cover the various new
6365 concepts introduced with 205.
6367 * Unit files now understand the new %v specifier which
6368 resolves to the kernel version string as returned by "uname
6371 * systemctl now supports filtering the unit list output by
6372 load state, active state and sub state, using the new
6375 * "systemctl status" will now show the results of the
6376 condition checks (like ConditionPathExists= and similar) of
6377 the last start attempts of the unit. They are also logged to
6380 * "journalctl -b" may now be used to look for boot output of a
6381 specific boot. Try "journalctl -b -1" for the previous boot,
6382 but the syntax is substantially more powerful.
6384 * "journalctl --show-cursor" has been added which prints the
6385 cursor string the last shown log line. This may then be used
6386 with the new "journalctl --after-cursor=" switch to continue
6387 browsing logs from that point on.
6389 * "journalctl --force" may now be used to force regeneration
6392 * Creation of "dead" device nodes has been moved from udev
6393 into kmod and tmpfiles. Previously, udev would read the kmod
6394 databases to pre-generate dead device nodes based on meta
6395 information contained in kernel modules, so that these would
6396 be auto-loaded on access rather then at boot. As this
6397 does not really have much to do with the exposing actual
6398 kernel devices to userspace this has always been slightly
6399 alien in the udev codebase. Following the new scheme kmod
6400 will now generate a runtime snippet for tmpfiles from the
6401 module meta information and it now is tmpfiles' job to the
6402 create the nodes. This also allows overriding access and
6403 other parameters for the nodes using the usual tmpfiles
6404 facilities. As side effect this allows us to remove the
6405 CAP_SYS_MKNOD capability bit from udevd entirely.
6407 * logind's device ACLs may now be applied to these "dead"
6408 devices nodes too, thus finally allowing managed access to
6409 devices such as /dev/snd/sequencer without loading the
6410 backing module right-away.
6412 * A new RPM macro has been added that may be used to apply
6413 tmpfiles configuration during package installation.
6415 * systemd-detect-virt and ConditionVirtualization= now can
6416 detect User-Mode-Linux machines (UML).
6418 * journald will now implicitly log the effective capabilities
6419 set of processes in the message metadata.
6421 * systemd-cryptsetup has gained support for TrueCrypt volumes.
6423 * The initrd interface has been simplified (more specifically,
6424 support for passing performance data via environment
6425 variables and fsck results via files in /run has been
6426 removed). These features were non-essential, and are
6427 nowadays available in a much nicer way by having systemd in
6428 the initrd serialize its state and have the hosts systemd
6429 deserialize it again.
6431 * The udev "keymap" data files and tools to apply keyboard
6432 specific mappings of scan to key codes, and force-release
6433 scan code lists have been entirely replaced by a udev
6434 "keyboard" builtin and a hwdb data file.
6436 * systemd will now honour the kernel's "quiet" command line
6437 argument also during late shutdown, resulting in a
6438 completely silent shutdown when used.
6440 * There's now an option to control the SO_REUSEPORT socket
6441 option in .socket units.
6443 * Instance units will now automatically get a per-template
6444 subslice of system.slice unless something else is explicitly
6445 configured. For example, instances of sshd@.service will now
6446 implicitly be placed in system-sshd.slice rather than
6447 system.slice as before.
6449 * Test coverage support may now be enabled at build time.
6451 Contributions from: Dave Reisner, Frederic Crozat, Harald
6452 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
6453 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
6454 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
6455 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
6456 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
6457 Giokas, Zbigniew Jędrzejewski-Szmek
6459 — Berlin, 2013-07-23
6463 * Two new unit types have been introduced:
6465 Scope units are very similar to service units, however, are
6466 created out of pre-existing processes — instead of PID 1
6467 forking off the processes. By using scope units it is
6468 possible for system services and applications to group their
6469 own child processes (worker processes) in a powerful way
6470 which then maybe used to organize them, or kill them
6471 together, or apply resource limits on them.
6473 Slice units may be used to partition system resources in an
6474 hierarchical fashion and then assign other units to them. By
6475 default there are now three slices: system.slice (for all
6476 system services), user.slice (for all user sessions),
6477 machine.slice (for VMs and containers).
6479 Slices and scopes have been introduced primarily in
6480 context of the work to move cgroup handling to a
6481 single-writer scheme, where only PID 1
6482 creates/removes/manages cgroups.
6484 * There's a new concept of "transient" units. In contrast to
6485 normal units these units are created via an API at runtime,
6486 not from configuration from disk. More specifically this
6487 means it is now possible to run arbitrary programs as
6488 independent services, with all execution parameters passed
6489 in via bus APIs rather than read from disk. Transient units
6490 make systemd substantially more dynamic then it ever was,
6491 and useful as a general batch manager.
6493 * logind has been updated to make use of scope and slice units
6494 for managing user sessions. As a user logs in he will get
6495 his own private slice unit, to which all sessions are added
6496 as scope units. We also added support for automatically
6497 adding an instance of user@.service for the user into the
6498 slice. Effectively logind will no longer create cgroup
6499 hierarchies on its own now, it will defer entirely to PID 1
6500 for this by means of scope, service and slice units. Since
6501 user sessions this way become entities managed by PID 1
6502 the output of "systemctl" is now a lot more comprehensive.
6504 * A new mini-daemon "systemd-machined" has been added which
6505 may be used by virtualization managers to register local
6506 VMs/containers. nspawn has been updated accordingly, and
6507 libvirt will be updated shortly. machined will collect a bit
6508 of meta information about the VMs/containers, and assign
6509 them their own scope unit (see above). The collected
6510 meta-data is then made available via the "machinectl" tool,
6511 and exposed in "ps" and similar tools. machined/machinectl
6512 is compile-time optional.
6514 * As discussed earlier, the low-level cgroup configuration
6515 options ControlGroup=, ControlGroupModify=,
6516 ControlGroupPersistent=, ControlGroupAttribute= have been
6517 removed. Please use high-level attribute settings instead as
6518 well as slice units.
6520 * A new bus call SetUnitProperties() has been added to alter
6521 various runtime parameters of a unit. This is primarily
6522 useful to alter cgroup parameters dynamically in a nice way,
6523 but will be extended later on to make more properties
6524 modifiable at runtime. systemctl gained a new set-properties
6525 command that wraps this call.
6527 * A new tool "systemd-run" has been added which can be used to
6528 run arbitrary command lines as transient services or scopes,
6529 while configuring a number of settings via the command
6530 line. This tool is currently very basic, however already
6531 very useful. We plan to extend this tool to even allow
6532 queuing of execution jobs with time triggers from the
6533 command line, similar in fashion to "at".
6535 * nspawn will now inform the user explicitly that kernels with
6536 audit enabled break containers, and suggest the user to turn
6539 * Support for detecting the IMA and AppArmor security
6540 frameworks with ConditionSecurity= has been added.
6542 * journalctl gained a new "-k" switch for showing only kernel
6543 messages, mimicking dmesg output; in addition to "--user"
6544 and "--system" switches for showing only user's own logs
6547 * systemd-delta can now show information about drop-in
6548 snippets extending unit files.
6550 * libsystemd-bus has been substantially updated but is still
6551 not available as public API.
6553 * systemd will now look for the "debug" argument on the kernel
6554 command line and enable debug logging, similar to what
6555 "systemd.log_level=debug" already did before.
6557 * "systemctl set-default", "systemctl get-default" has been
6558 added to configure the default.target symlink, which
6559 controls what to boot into by default.
6561 * "systemctl set-log-level" has been added as a convenient
6562 way to raise and lower systemd logging threshold.
6564 * "systemd-analyze plot" will now show the time the various
6565 generators needed for execution, as well as information
6566 about the unit file loading.
6568 * libsystemd-journal gained a new sd_journal_open_files() call
6569 for opening specific journal files. journactl also gained a
6570 new switch to expose this new functionality. Previously we
6571 only supported opening all files from a directory, or all
6572 files from the system, as opening individual files only is
6573 racy due to journal file rotation.
6575 * systemd gained the new DefaultEnvironment= setting in
6576 /etc/systemd/system.conf to set environment variables for
6579 * If a privileged process logs a journal message with the
6580 OBJECT_PID= field set, then journald will automatically
6581 augment this with additional OBJECT_UID=, OBJECT_GID=,
6582 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
6583 system services want to log events about specific client
6584 processes. journactl/systemctl has been updated to make use
6585 of this information if all log messages regarding a specific
6588 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
6589 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
6590 Reisner, David Coppa, David King, David Strauss, Eelco
6591 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
6592 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
6593 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
6594 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
6595 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
6596 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
6597 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
6598 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
6599 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
6600 Łukasz Stelmach, 장동준
6604 * The Python bindings gained some minimal support for the APIs
6605 exposed by libsystemd-logind.
6607 * ConditionSecurity= gained support for detecting SMACK. Since
6608 this condition already supports SELinux and AppArmor we only
6609 miss IMA for this. Patches welcome!
6611 Contributions from: Karol Lewandowski, Lennart Poettering,
6612 Zbigniew Jędrzejewski-Szmek
6616 * systemd-nspawn will now create /etc/resolv.conf if
6617 necessary, before bind-mounting the host's file onto it.
6619 * systemd-nspawn will now store meta information about a
6620 container on the container's cgroup as extended attribute
6621 fields, including the root directory.
6623 * The cgroup hierarchy has been reworked in many ways. All
6624 objects any of the components systemd creates in the cgroup
6625 tree are now suffixed. More specifically, user sessions are
6626 now placed in cgroups suffixed with ".session", users in
6627 cgroups suffixed with ".user", and nspawn containers in
6628 cgroups suffixed with ".nspawn". Furthermore, all cgroup
6629 names are now escaped in a simple scheme to avoid collision
6630 of userspace object names with kernel filenames. This work
6631 is preparation for making these objects relocatable in the
6632 cgroup tree, in order to allow easy resource partitioning of
6633 these objects without causing naming conflicts.
6635 * systemctl list-dependencies gained the new switches
6636 --plain, --reverse, --after and --before.
6638 * systemd-inhibit now shows the process name of processes that
6639 have taken an inhibitor lock.
6641 * nss-myhostname will now also resolve "localhost"
6642 implicitly. This makes /etc/hosts an optional file and
6643 nicely handles that on IPv6 ::1 maps to both "localhost" and
6646 * libsystemd-logind.so gained a new call
6647 sd_get_machine_names() to enumerate running containers and
6648 VMs (currently only supported by very new libvirt and
6649 nspawn). sd_login_monitor can now be used to watch
6650 VMs/containers coming and going.
6652 * .include is not allowed recursively anymore, and only in
6653 unit files. Usually it is better to use drop-in snippets in
6654 .d/*.conf anyway, as introduced with systemd 198.
6656 * systemd-analyze gained a new "critical-chain" command that
6657 determines the slowest chain of units run during system
6658 boot-up. It is very useful for tracking down where
6659 optimizing boot time is the most beneficial.
6661 * systemd will no longer allow manipulating service paths in
6662 the name=systemd:/system cgroup tree using ControlGroup= in
6663 units. (But is still fine with it in all other dirs.)
6665 * There's a new systemd-nspawn@.service service file that may
6666 be used to easily run nspawn containers as system
6667 services. With the container's root directory in
6668 /var/lib/container/foobar it is now sufficient to run
6669 "systemctl start systemd-nspawn@foobar.service" to boot it.
6671 * systemd-cgls gained a new parameter "--machine" to list only
6672 the processes within a certain container.
6674 * ConditionSecurity= now can check for "apparmor". We still
6675 are lacking checks for SMACK and IMA for this condition
6676 check though. Patches welcome!
6678 * A new configuration file /etc/systemd/sleep.conf has been
6679 added that may be used to configure which kernel operation
6680 systemd is supposed to execute when "suspend", "hibernate"
6681 or "hybrid-sleep" is requested. This makes the new kernel
6682 "freeze" state accessible to the user.
6684 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
6685 the passed argument if applicable.
6687 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
6688 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
6689 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
6690 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
6691 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
6692 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
6693 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
6698 * The output of 'systemctl list-jobs' got some polishing. The
6699 '--type=' argument may now be passed more than once. A new
6700 command 'systemctl list-sockets' has been added which shows
6701 a list of kernel sockets systemd is listening on with the
6702 socket units they belong to, plus the units these socket
6705 * The experimental libsystemd-bus library got substantial
6706 updates to work in conjunction with the (also experimental)
6707 kdbus kernel project. It works well enough to exchange
6708 messages with some sophistication. Note that kdbus is not
6709 ready yet, and the library is mostly an elaborate test case
6710 for now, and not installable.
6712 * systemd gained a new unit 'systemd-static-nodes.service'
6713 that generates static device nodes earlier during boot, and
6714 can run in conjunction with udev.
6716 * libsystemd-login gained a new call sd_pid_get_user_unit()
6717 to retrieve the user systemd unit a process is running
6718 in. This is useful for systems where systemd is used as
6721 * systemd-nspawn now places all containers in the new /machine
6722 top-level cgroup directory in the name=systemd
6723 hierarchy. libvirt will soon do the same, so that we get a
6724 uniform separation of /system, /user and /machine for system
6725 services, user processes and containers/virtual
6726 machines. This new cgroup hierarchy is also useful to stick
6727 stable names to specific container instances, which can be
6728 recognized later this way (this name may be controlled
6729 via systemd-nspawn's new -M switch). libsystemd-login also
6730 gained a new call sd_pid_get_machine_name() to retrieve the
6731 name of the container/VM a specific process belongs to.
6733 * bootchart can now store its data in the journal.
6735 * libsystemd-journal gained a new call
6736 sd_journal_add_conjunction() for AND expressions to the
6737 matching logic. This can be used to express more complex
6738 logical expressions.
6740 * journactl can now take multiple --unit= and --user-unit=
6743 * The cryptsetup logic now understands the "luks.key=" kernel
6744 command line switch for specifying a file to read the
6745 decryption key from. Also, if a configured key file is not
6746 found the tool will now automatically fall back to prompting
6749 * Python systemd.journal module was updated to wrap recently
6750 added functions from libsystemd-journal. The interface was
6751 changed to bring the low level interface in s.j._Reader
6752 closer to the C API, and the high level interface in
6753 s.j.Reader was updated to wrap and convert all data about
6756 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
6757 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
6758 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
6759 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
6760 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
6761 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6765 * journalctl --update-catalog now understands a new --root=
6766 option to operate on catalogs found in a different root
6769 * During shutdown after systemd has terminated all running
6770 services a final killing loop kills all remaining left-over
6771 processes. We will now print the name of these processes
6772 when we send SIGKILL to them, since this usually indicates a
6775 * If /etc/crypttab refers to password files stored on
6776 configured mount points automatic dependencies will now be
6777 generated to ensure the specific mount is established first
6778 before the key file is attempted to be read.
6780 * 'systemctl status' will now show information about the
6781 network sockets a socket unit is listening on.
6783 * 'systemctl status' will also shown information about any
6784 drop-in configuration file for units. (Drop-In configuration
6785 files in this context are files such as
6786 /etc/systemd/systemd/foobar.service.d/*.conf)
6788 * systemd-cgtop now optionally shows summed up CPU times of
6789 cgroups. Press '%' while running cgtop to switch between
6790 percentage and absolute mode. This is useful to determine
6791 which cgroups use up the most CPU time over the entire
6792 runtime of the system. systemd-cgtop has also been updated
6793 to be 'pipeable' for processing with further shell tools.
6795 * 'hostnamectl set-hostname' will now allow setting of FQDN
6798 * The formatting and parsing of time span values has been
6799 changed. The parser now understands fractional expressions
6800 such as "5.5h". The formatter will now output fractional
6801 expressions for all time spans under 1min, i.e. "5.123456s"
6802 rather than "5s 123ms 456us". For time spans under 1s
6803 millisecond values are shown, for those under 1ms
6804 microsecond values are shown. This should greatly improve
6805 all time-related output of systemd.
6807 * libsystemd-login and libsystemd-journal gained new
6808 functions for querying the poll() events mask and poll()
6809 timeout value for integration into arbitrary event
6812 * localectl gained the ability to list available X11 keymaps
6813 (models, layouts, variants, options).
6815 * 'systemd-analyze dot' gained the ability to filter for
6816 specific units via shell-style globs, to create smaller,
6817 more useful graphs. I.e. it is now possible to create simple
6818 graphs of all the dependencies between only target units, or
6819 of all units that Avahi has dependencies with.
6821 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
6822 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
6823 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
6824 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
6825 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
6826 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
6827 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
6831 * The boot-time readahead implementation for rotating media
6832 will now read the read-ahead data in multiple passes which
6833 consist of all read requests made in equidistant time
6834 intervals. This means instead of strictly reading read-ahead
6835 data in its physical order on disk we now try to find a
6836 middle ground between physical and access time order.
6838 * /etc/os-release files gained a new BUILD_ID= field for usage
6839 on operating systems that provide continuous builds of OS
6842 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
6843 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
6844 William Douglas, Zbigniew Jędrzejewski-Szmek
6848 * systemd-python gained an API exposing libsystemd-daemon.
6850 * The SMACK setup logic gained support for uploading CIPSO
6853 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
6854 ReadOnlyDirectories= and InaccessibleDirectories= has
6855 changed. The private /tmp and /var/tmp directories are now
6856 shared by all processes of a service (which means
6857 ExecStartPre= may now leave data in /tmp that ExecStart= of
6858 the same service can still access). When a service is
6859 stopped its temporary directories are immediately deleted
6860 (normal clean-up with tmpfiles is still done in addition to
6863 * By default, systemd will now set a couple of sysctl
6864 variables in the kernel: the safe sysrq options are turned
6865 on, IP route verification is turned on, and source routing
6866 disabled. The recently added hardlink and softlink
6867 protection of the kernel is turned on. These settings should
6868 be reasonably safe, and good defaults for all new systems.
6870 * The predictable network naming logic may now be turned off
6871 with a new kernel command line switch: net.ifnames=0.
6873 * A new libsystemd-bus module has been added that implements a
6874 pretty complete D-Bus client library. For details see:
6876 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
6878 * journald will now explicitly flush the journal files to disk
6879 at the latest 5min after each write. The file will then also
6880 be marked offline until the next write. This should increase
6881 reliability in case of a crash. The synchronization delay
6882 can be configured via SyncIntervalSec= in journald.conf.
6884 * There's a new remote-fs-setup.target unit that can be used
6885 to pull in specific services when at least one remote file
6886 system is to be mounted.
6888 * There are new targets timers.target and paths.target as
6889 canonical targets to pull user timer and path units in
6890 from. This complements sockets.target with a similar
6891 purpose for socket units.
6893 * libudev gained a new call udev_device_set_attribute_value()
6894 to set sysfs attributes of a device.
6896 * The udev daemon now sets the default number of worker
6897 processes executed in parallel based on the number of available
6898 CPUs instead of the amount of available RAM. This is supposed
6899 to provide a more reliable default and limit a too aggressive
6900 parallelism for setups with 1000s of devices connected.
6902 Contributions from: Auke Kok, Colin Walters, Cristian
6903 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
6904 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
6905 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
6906 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
6907 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
6908 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
6909 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
6910 Zbigniew Jędrzejewski-Szmek
6914 * Configuration of unit files may now be extended via drop-in
6915 files without having to edit/override the unit files
6916 themselves. More specifically, if the administrator wants to
6917 change one value for a service file foobar.service he can
6918 now do so by dropping in a configuration snippet into
6919 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
6920 will load all these snippets and apply them on top of the
6921 main unit configuration file, possibly extending or
6922 overriding its settings. Using these drop-in snippets is
6923 generally nicer than the two earlier options for changing
6924 unit files locally: copying the files from
6925 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
6926 them there; or creating a new file in /etc/systemd/system/
6927 that incorporates the original one via ".include". Drop-in
6928 snippets into these .d/ directories can be placed in any
6929 directory systemd looks for units in, and the usual
6930 overriding semantics between /usr/lib, /etc and /run apply
6933 * Most unit file settings which take lists of items can now be
6934 reset by assigning the empty string to them. For example,
6935 normally, settings such as Environment=FOO=BAR append a new
6936 environment variable assignment to the environment block,
6937 each time they are used. By assigning Environment= the empty
6938 string the environment block can be reset to empty. This is
6939 particularly useful with the .d/*.conf drop-in snippets
6940 mentioned above, since this adds the ability to reset list
6941 settings from vendor unit files via these drop-ins.
6943 * systemctl gained a new "list-dependencies" command for
6944 listing the dependencies of a unit recursively.
6946 * Inhibitors are now honored and listed by "systemctl
6947 suspend", "systemctl poweroff" (and similar) too, not only
6948 GNOME. These commands will also list active sessions by
6951 * Resource limits (as exposed by the various control group
6952 controllers) can now be controlled dynamically at runtime
6953 for all units. More specifically, you can now use a command
6954 like "systemctl set-cgroup-attr foobar.service cpu.shares
6955 2000" to alter the CPU shares a specific service gets. These
6956 settings are stored persistently on disk, and thus allow the
6957 administrator to easily adjust the resource usage of
6958 services with a few simple commands. This dynamic resource
6959 management logic is also available to other programs via the
6960 bus. Almost any kernel cgroup attribute and controller is
6963 * systemd-vconsole-setup will now copy all font settings to
6964 all allocated VTs, where it previously applied them only to
6967 * libsystemd-login gained the new sd_session_get_tty() API
6970 * This release drops support for a few legacy or
6971 distribution-specific LSB facility names when parsing init
6972 scripts: $x-display-manager, $mail-transfer-agent,
6973 $mail-transport-agent, $mail-transfer-agent, $smtp,
6974 $null. Also, the mail-transfer-agent.target unit backing
6975 this has been removed. Distributions which want to retain
6976 compatibility with this should carry the burden for
6977 supporting this themselves and patch support for these back
6978 in, if they really need to. Also, the facilities $syslog and
6979 $local_fs are now ignored, since systemd does not support
6980 early-boot LSB init scripts anymore, and these facilities
6981 are implied anyway for normal services. syslog.target has
6984 * There are new bus calls on PID1's Manager object for
6985 cancelling jobs, and removing snapshot units. Previously,
6986 both calls were only available on the Job and Snapshot
6989 * systemd-journal-gatewayd gained SSL support.
6991 * The various "environment" files, such as /etc/locale.conf
6992 now support continuation lines with a backslash ("\") as
6993 last character in the line, similarly in style (but different)
6994 to how this is supported in shells.
6996 * For normal user processes the _SYSTEMD_USER_UNIT= field is
6997 now implicitly appended to every log entry logged. systemctl
6998 has been updated to filter by this field when operating on a
6999 user systemd instance.
7001 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
7002 CAP_AUDIT_CONTROL capabilities to the capabilities set for
7003 the container. This makes it easier to boot unmodified
7004 Fedora systems in a container, which however still requires
7005 audit=0 to be passed on the kernel command line. Auditing in
7006 kernel and userspace is unfortunately still too broken in
7007 context of containers, hence we recommend compiling it out
7008 of the kernel or using audit=0. Hopefully this will be fixed
7009 one day for good in the kernel.
7011 * nspawn gained the new --bind= and --bind-ro= parameters to
7012 bind mount specific directories from the host into the
7015 * nspawn will now mount its own devpts file system instance
7016 into the container, in order not to leak pty devices from
7017 the host into the container.
7019 * systemd will now read the firmware boot time performance
7020 information from the EFI variables, if the used boot loader
7021 supports this, and takes it into account for boot performance
7022 analysis via "systemd-analyze". This is currently supported
7023 only in conjunction with Gummiboot, but could be supported
7024 by other boot loaders too. For details see:
7026 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
7028 * A new generator has been added that automatically mounts the
7029 EFI System Partition (ESP) to /boot, if that directory
7030 exists, is empty, and no other file system has been
7031 configured to be mounted there.
7033 * logind will now send out PrepareForSleep(false) out
7034 unconditionally, after coming back from suspend. This may be
7035 used by applications as asynchronous notification for
7036 system resume events.
7038 * "systemctl unlock-sessions" has been added, that allows
7039 unlocking the screens of all user sessions at once, similar
7040 to how "systemctl lock-sessions" already locked all users
7041 sessions. This is backed by a new D-Bus call UnlockSessions().
7043 * "loginctl seat-status" will now show the master device of a
7044 seat. (i.e. the device of a seat that needs to be around for
7045 the seat to be considered available, usually the graphics
7048 * tmpfiles gained a new "X" line type, that allows
7049 configuration of files and directories (with wildcards) that
7050 shall be excluded from automatic cleanup ("aging").
7052 * udev default rules set the device node permissions now only
7053 at "add" events, and do not change them any longer with a
7054 later "change" event.
7056 * The log messages for lid events and power/sleep keypresses
7057 now carry a message ID.
7059 * We now have a substantially larger unit test suite, but this
7060 continues to be work in progress.
7062 * udevadm hwdb gained a new --root= parameter to change the
7063 root directory to operate relative to.
7065 * logind will now issue a background sync() request to the kernel
7066 early at shutdown, so that dirty buffers are flushed to disk early
7067 instead of at the last moment, in order to optimize shutdown
7070 * A new bootctl tool has been added that is an interface for
7071 certain boot loader operations. This is currently a preview
7072 and is likely to be extended into a small mechanism daemon
7073 like timedated, localed, hostnamed, and can be used by
7074 graphical UIs to enumerate available boot options, and
7075 request boot into firmware operations.
7077 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
7078 the rest of the package. It also has been updated to work
7079 correctly in initrds.
7081 * Policykit previously has been runtime optional, and is now
7082 also compile time optional via a configure switch.
7084 * systemd-analyze has been reimplemented in C. Also "systemctl
7085 dot" has moved into systemd-analyze.
7087 * "systemctl status" with no further parameters will now print
7088 the status of all active or failed units.
7090 * Operations such as "systemctl start" can now be executed
7091 with a new mode "--irreversible" which may be used to queue
7092 operations that cannot accidentally be reversed by a later
7093 job queuing. This is by default used to make shutdown
7094 requests more robust.
7096 * The Python API of systemd now gained a new module for
7097 reading journal files.
7099 * A new tool kernel-install has been added that can install
7100 kernel images according to the Boot Loader Specification:
7102 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
7104 * Boot time console output has been improved to provide
7105 animated boot time output for hanging jobs.
7107 * A new tool systemd-activate has been added which can be used
7108 to test socket activation with, directly from the command
7109 line. This should make it much easier to test and debug
7110 socket activation in daemons.
7112 * journalctl gained a new "--reverse" (or -r) option to show
7113 journal output in reverse order (i.e. newest line first).
7115 * journalctl gained a new "--pager-end" (or -e) option to jump
7116 to immediately jump to the end of the journal in the
7117 pager. This is only supported in conjunction with "less".
7119 * journalctl gained a new "--user-unit=" option, that works
7120 similarly to "--unit=" but filters for user units rather than
7123 * A number of unit files to ease adoption of systemd in
7124 initrds has been added. This moves some minimal logic from
7125 the various initrd implementations into systemd proper.
7127 * The journal files are now owned by a new group
7128 "systemd-journal", which exists specifically to allow access
7129 to the journal, and nothing else. Previously, we used the
7130 "adm" group for that, which however possibly covers more
7131 than just journal/log file access. This new group is now
7132 already used by systemd-journal-gatewayd to ensure this
7133 daemon gets access to the journal files and as little else
7134 as possible. Note that "make install" will also set FS ACLs
7135 up for /var/log/journal to give "adm" and "wheel" read
7136 access to it, in addition to "systemd-journal" which owns
7137 the journal files. We recommend that packaging scripts also
7138 add read access to "adm" + "wheel" to /var/log/journal, and
7139 all existing/future journal files. To normal users and
7140 administrators little changes, however packagers need to
7141 ensure to create the "systemd-journal" system group at
7142 package installation time.
7144 * The systemd-journal-gatewayd now runs as unprivileged user
7145 systemd-journal-gateway:systemd-journal-gateway. Packaging
7146 scripts need to create these system user/group at
7149 * timedated now exposes a new boolean property CanNTP that
7150 indicates whether a local NTP service is available or not.
7152 * systemd-detect-virt will now also detect xen PVs
7154 * The pstore file system is now mounted by default, if it is
7157 * In addition to the SELinux and IMA policies we will now also
7158 load SMACK policies at early boot.
7160 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
7161 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
7162 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
7163 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
7164 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
7165 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
7166 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
7167 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
7168 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
7169 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
7170 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
7171 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
7172 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
7173 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
7177 * Timer units now support calendar time events in addition to
7178 monotonic time events. That means you can now trigger a unit
7179 based on a calendar time specification such as "Thu,Fri
7180 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
7181 or fifth day of any month of the year 2013, given that it is
7182 a thursday or friday. This brings timer event support
7183 considerably closer to cron's capabilities. For details on
7184 the supported calendar time specification language see
7187 * udev now supports a number of different naming policies for
7188 network interfaces for predictable names, and a combination
7189 of these policies is now the default. Please see this wiki
7190 document for details:
7192 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
7194 * Auke Kok's bootchart implementation has been added to the
7195 systemd tree. It is an optional component that can graph the
7196 boot in quite some detail. It is one of the best bootchart
7197 implementations around and minimal in its code and
7200 * nss-myhostname has been integrated into the systemd source
7201 tree. nss-myhostname guarantees that the local hostname
7202 always stays resolvable via NSS. It has been a weak
7203 requirement of systemd-hostnamed since a long time, and
7204 since its code is actually trivial we decided to just
7205 include it in systemd's source tree. It can be turned off
7206 with a configure switch.
7208 * The read-ahead logic is now capable of properly detecting
7209 whether a btrfs file system is on SSD or rotating media, in
7210 order to optimize the read-ahead scheme. Previously, it was
7211 only capable of detecting this on traditional file systems
7214 * In udev, additional device properties are now read from the
7215 IAB in addition to the OUI database. Also, Bluetooth company
7216 identities are attached to the devices as well.
7218 * In service files %U may be used as specifier that is
7219 replaced by the configured user name of the service.
7221 * nspawn may now be invoked without a controlling TTY. This
7222 makes it suitable for invocation as its own service. This
7223 may be used to set up a simple containerized server system
7224 using only core OS tools.
7226 * systemd and nspawn can now accept socket file descriptors
7227 when they are started for socket activation. This enables
7228 implementation of socket activated nspawn
7229 containers. i.e. think about autospawning an entire OS image
7230 when the first SSH or HTTP connection is received. We expect
7231 that similar functionality will also be added to libvirt-lxc
7234 * journalctl will now suppress ANSI color codes when
7235 presenting log data.
7237 * systemctl will no longer show control group information for
7238 a unit if the control group is empty anyway.
7240 * logind can now automatically suspend/hibernate/shutdown the
7243 * /etc/machine-info and hostnamed now also expose the chassis
7244 type of the system. This can be used to determine whether
7245 the local system is a laptop, desktop, handset or
7246 tablet. This information may either be configured by the
7247 user/vendor or is automatically determined from ACPI and DMI
7248 information if possible.
7250 * A number of PolicyKit actions are now bound together with
7251 "imply" rules. This should simplify creating UIs because
7252 many actions will now authenticate similar ones as well.
7254 * Unit files learnt a new condition ConditionACPower= which
7255 may be used to conditionalize a unit depending on whether an
7256 AC power source is connected or not, of whether the system
7257 is running on battery power.
7259 * systemctl gained a new "is-failed" verb that may be used in
7260 shell scripts and suchlike to check whether a specific unit
7261 is in the "failed" state.
7263 * The EnvironmentFile= setting in unit files now supports file
7264 globbing, and can hence be used to easily read a number of
7265 environment files at once.
7267 * systemd will no longer detect and recognize specific
7268 distributions. All distribution-specific #ifdeffery has been
7269 removed, systemd is now fully generic and
7270 distribution-agnostic. Effectively, not too much is lost as
7271 a lot of the code is still accessible via explicit configure
7272 switches. However, support for some distribution specific
7273 legacy configuration file formats has been dropped. We
7274 recommend distributions to simply adopt the configuration
7275 files everybody else uses now and convert the old
7276 configuration from packaging scripts. Most distributions
7277 already did that. If that's not possible or desirable,
7278 distributions are welcome to forward port the specific
7279 pieces of code locally from the git history.
7281 * When logging a message about a unit systemd will now always
7282 log the unit name in the message meta data.
7284 * localectl will now also discover system locale data that is
7285 not stored in locale archives, but directly unpacked.
7287 * logind will no longer unconditionally use framebuffer
7288 devices as seat masters, i.e. as devices that are required
7289 to be existing before a seat is considered preset. Instead,
7290 it will now look for all devices that are tagged as
7291 "seat-master" in udev. By default, framebuffer devices will
7292 be marked as such, but depending on local systems, other
7293 devices might be marked as well. This may be used to
7294 integrate graphics cards using closed source drivers (such
7295 as NVidia ones) more nicely into logind. Note however, that
7296 we recommend using the open source NVidia drivers instead,
7297 and no udev rules for the closed-source drivers will be
7298 shipped from us upstream.
7300 Contributions from: Adam Williamson, Alessandro Crismani, Auke
7301 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
7302 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
7303 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
7304 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
7305 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
7306 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
7307 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
7308 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
7309 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
7310 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
7311 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
7316 * udev gained support for loading additional device properties
7317 from an indexed database that is keyed by vendor/product IDs
7318 and similar device identifiers. For the beginning this
7319 "hwdb" is populated with data from the well-known PCI and
7320 USB database, but also includes PNP, ACPI and OID data. In
7321 the longer run this indexed database shall grow into
7322 becoming the one central database for non-essential
7323 userspace device metadata. Previously, data from the PCI/USB
7324 database was only attached to select devices, since the
7325 lookup was a relatively expensive operation due to O(n) time
7326 complexity (with n being the number of entries in the
7327 database). Since this is now O(1), we decided to add in this
7328 data for all devices where this is available, by
7329 default. Note that the indexed database needs to be rebuilt
7330 when new data files are installed. To achieve this you need
7331 to update your packaging scripts to invoke "udevadm hwdb
7332 --update" after installation of hwdb data files. For
7333 RPM-based distributions we introduced the new
7334 %udev_hwdb_update macro for this purpose.
7336 * The Journal gained support for the "Message Catalog", an
7337 indexed database to link up additional information with
7338 journal entries. For further details please check:
7340 https://www.freedesktop.org/wiki/Software/systemd/catalog
7342 The indexed message catalog database also needs to be
7343 rebuilt after installation of message catalog files. Use
7344 "journalctl --update-catalog" for this. For RPM-based
7345 distributions we introduced the %journal_catalog_update
7346 macro for this purpose.
7348 * The Python Journal bindings gained support for the standard
7349 Python logging framework.
7351 * The Journal API gained new functions for checking whether
7352 the underlying file system of a journal file is capable of
7353 properly reporting file change notifications, or whether
7354 applications that want to reflect journal changes "live"
7355 need to recheck journal files continuously in appropriate
7358 * It is now possible to set the "age" field for tmpfiles
7359 entries to 0, indicating that files matching this entry
7360 shall always be removed when the directories are cleaned up.
7362 * coredumpctl gained a new "gdb" verb which invokes gdb
7363 right-away on the selected coredump.
7365 * There's now support for "hybrid sleep" on kernels that
7366 support this, in addition to "suspend" and "hibernate". Use
7367 "systemctl hybrid-sleep" to make use of this.
7369 * logind's HandleSuspendKey= setting (and related settings)
7370 now gained support for a new "lock" setting to simply
7371 request the screen lock on all local sessions, instead of
7372 actually executing a suspend or hibernation.
7374 * systemd will now mount the EFI variables file system by
7377 * Socket units now gained support for configuration of the
7378 SMACK security label.
7380 * timedatectl will now output the time of the last and next
7381 daylight saving change.
7383 * We dropped support for various legacy and distro-specific
7384 concepts, such as insserv, early-boot SysV services
7385 (i.e. those for non-standard runlevels such as 'b' or 'S')
7386 or ArchLinux /etc/rc.conf support. We recommend the
7387 distributions who still need support this to either continue
7388 to maintain the necessary patches downstream, or find a
7389 different solution. (Talk to us if you have questions!)
7391 * Various systemd components will now bypass PolicyKit checks
7392 for root and otherwise handle properly if PolicyKit is not
7393 found to be around. This should fix most issues for
7394 PolicyKit-less systems. Quite frankly this should have been
7395 this way since day one. It is absolutely our intention to
7396 make systemd work fine on PolicyKit-less systems, and we
7397 consider it a bug if something does not work as it should if
7398 PolicyKit is not around.
7400 * For embedded systems it is now possible to build udev and
7401 systemd without blkid and/or kmod support.
7403 * "systemctl switch-root" is now capable of switching root
7404 more than once. I.e. in addition to transitions from the
7405 initrd to the host OS it is now possible to transition to
7406 further OS images from the host. This is useful to implement
7407 offline updating tools.
7409 * Various other additions have been made to the RPM macros
7410 shipped with systemd. Use %udev_rules_update() after
7411 installing new udev rules files. %_udevhwdbdir,
7412 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
7413 %_sysctldir are now available which resolve to the right
7414 directories for packages to place various data files in.
7416 * journalctl gained the new --full switch (in addition to
7417 --all, to disable ellipsation for long messages.
7419 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
7420 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
7421 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
7422 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
7423 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
7424 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
7425 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
7426 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
7427 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
7431 * journalctl gained new --since= and --until= switches to
7432 filter by time. It also now supports nice filtering for
7433 units via --unit=/-u.
7435 * Type=oneshot services may use ExecReload= and do the
7438 * The journal daemon now supports time-based rotation and
7439 vacuuming, in addition to the usual disk-space based
7442 * The journal will now index the available field values for
7443 each field name. This enables clients to show pretty drop
7444 downs of available match values when filtering. The bash
7445 completion of journalctl has been updated
7446 accordingly. journalctl gained a new switch -F to list all
7447 values a certain field takes in the journal database.
7449 * More service events are now written as structured messages
7450 to the journal, and made recognizable via message IDs.
7452 * The timedated, localed and hostnamed mini-services which
7453 previously only provided support for changing time, locale
7454 and hostname settings from graphical DEs such as GNOME now
7455 also have a minimal (but very useful) text-based client
7456 utility each. This is probably the nicest way to changing
7457 these settings from the command line now, especially since
7458 it lists available options and is fully integrated with bash
7461 * There's now a new tool "systemd-coredumpctl" to list and
7462 extract coredumps from the journal.
7464 * We now install a README each in /var/log/ and
7465 /etc/rc.d/init.d explaining where the system logs and init
7466 scripts went. This hopefully should help folks who go to
7467 that dirs and look into the otherwise now empty void and
7468 scratch their heads.
7470 * When user-services are invoked (by systemd --user) the
7471 $MANAGERPID env var is set to the PID of systemd.
7473 * SIGRTMIN+24 when sent to a --user instance will now result
7474 in immediate termination of systemd.
7476 * gatewayd received numerous feature additions such as a
7477 "follow" mode, for live syncing and filtering.
7479 * browse.html now allows filtering and showing detailed
7480 information on specific entries. Keyboard navigation and
7481 mouse screen support has been added.
7483 * gatewayd/journalctl now supports HTML5/JSON
7484 Server-Sent-Events as output.
7486 * The SysV init script compatibility logic will now
7487 heuristically determine whether a script supports the
7488 "reload" verb, and only then make this available as
7491 * "systemctl status --follow" has been removed, use "journalctl
7494 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
7495 have been removed since they are hardly useful to be
7498 * And I'd like to take the opportunity to specifically mention
7499 Zbigniew for his great contributions. Zbigniew, you rock!
7501 Contributions from: Andrew Eikum, Christian Hesse, Colin
7502 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
7503 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
7504 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
7505 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
7506 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
7507 Jędrzejewski-Szmek, Сковорода Никита Андреевич
7511 * If /etc/vconsole.conf is non-existent or empty we will no
7512 longer load any console font or key map at boot by
7513 default. Instead the kernel defaults will be left
7514 intact. This is definitely the right thing to do, as no
7515 configuration should mean no configuration, and hard-coding
7516 font names that are different on all archs is probably a bad
7517 idea. Also, the kernel default key map and font should be
7518 good enough for most cases anyway, and mostly identical to
7519 the userspace fonts/key maps we previously overloaded them
7520 with. If distributions want to continue to default to a
7521 non-kernel font or key map they should ship a default
7522 /etc/vconsole.conf with the appropriate contents.
7524 Contributions from: Colin Walters, Daniel J Walsh, Dave
7525 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
7526 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
7530 * journalctl gained a new --cursor= switch to show entries
7531 starting from the specified location in the journal.
7533 * We now enforce a size limit on journal entry fields exported
7534 with "-o json" in journalctl. Fields larger than 4K will be
7535 assigned null. This can be turned off with --all.
7537 * An (optional) journal gateway daemon is now available as
7538 "systemd-journal-gatewayd.service". This service provides
7539 access to the journal via HTTP and JSON. This functionality
7540 will be used to implement live log synchronization in both
7541 pull and push modes, but has various other users too, such
7542 as easy log access for debugging of embedded devices. Right
7543 now it is already useful to retrieve the journal via HTTP:
7545 # systemctl start systemd-journal-gatewayd.service
7546 # wget http://localhost:19531/entries
7548 This will download the journal contents in a
7549 /var/log/messages compatible format. The same as JSON:
7551 # curl -H"Accept: application/json" http://localhost:19531/entries
7553 This service is also accessible via a web browser where a
7554 single static HTML5 app is served that uses the JSON logic
7555 to enable the user to do some basic browsing of the
7556 journal. This will be extended later on. Here's an example
7557 screenshot of this app in its current state:
7559 http://0pointer.de/public/journal-gatewayd
7561 Contributions from: Kay Sievers, Lennart Poettering, Robert
7562 Milasan, Tom Gundersen
7566 * The bash completion logic is now available for journalctl
7569 * We do not mount the "cpuset" controller anymore together with
7570 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
7571 started if no parameters are assigned to it. "cpuset" hence
7572 broke code that assumed it could create "cpu" groups and
7575 * journalctl -f will now subscribe to terminal size changes,
7576 and line break accordingly.
7578 Contributions from: Dave Reisner, Kay Sievers, Lennart
7579 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
7583 * nspawn will now create a symlink /etc/localtime in the
7584 container environment, copying the host's timezone
7585 setting. Previously this has been done via a bind mount, but
7586 since symlinks cannot be bind mounted this has now been
7587 changed to create/update the appropriate symlink.
7589 * journalctl -n's line number argument is now optional, and
7590 will default to 10 if omitted.
7592 * journald will now log the maximum size the journal files may
7593 take up on disk. This is particularly useful if the default
7594 built-in logic of determining this parameter from the file
7595 system size is used. Use "systemctl status
7596 systemd-journald.service" to see this information.
7598 * The multi-seat X wrapper tool has been stripped down. As X
7599 is now capable of enumerating graphics devices via udev in a
7600 seat-aware way the wrapper is not strictly necessary
7601 anymore. A stripped down temporary stop-gap is still shipped
7602 until the upstream display managers have been updated to
7603 fully support the new X logic. Expect this wrapper to be
7604 removed entirely in one of the next releases.
7606 * HandleSleepKey= in logind.conf has been split up into
7607 HandleSuspendKey= and HandleHibernateKey=. The old setting
7608 is not available anymore. X11 and the kernel are
7609 distinguishing between these keys and we should too. This
7610 also means the inhibition lock for these keys has been split
7613 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
7614 Poettering, Lukas Nykryn, Václav Pavlín
7618 * Whenever a unit changes state we will now log this to the
7619 journal and show along the unit's own log output in
7622 * ConditionPathIsMountPoint= can now properly detect bind
7623 mount points too. (Previously, a bind mount of one file
7624 system to another place in the same file system could not be
7625 detected as mount, since they shared struct stat's st_dev
7628 * We will now mount the cgroup controllers cpu, cpuacct,
7629 cpuset and the controllers net_cls, net_prio together by
7632 * nspawn containers will now have a virtualized boot
7633 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
7634 over with a randomized ID at container initialization). This
7635 has the effect of making "journalctl -b" do the right thing
7638 * The JSON output journal serialization has been updated not
7639 to generate "endless" list objects anymore, but rather one
7640 JSON object per line. This is more in line how most JSON
7641 parsers expect JSON objects. The new output mode
7642 "json-pretty" has been added to provide similar output, but
7643 neatly aligned for readability by humans.
7645 * We dropped all explicit sync() invocations in the shutdown
7646 code. The kernel does this implicitly anyway in the kernel
7647 reboot() syscall. halt(8)'s -n option is now a compatibility
7650 * We now support virtualized reboot() in containers, as
7651 supported by newer kernels. We will fall back to exit() if
7652 CAP_SYS_REBOOT is not available to the container. Also,
7653 nspawn makes use of this now and will actually reboot the
7654 container if the containerized OS asks for that.
7656 * journalctl will only show local log output by default
7657 now. Use --merge (-m) to show remote log output, too.
7659 * libsystemd-journal gained the new sd_journal_get_usage()
7660 call to determine the current disk usage of all journal
7661 files. This is exposed in the new "journalctl --disk-usage"
7664 * journald gained a new configuration setting SplitMode= in
7665 journald.conf which may be used to control how user journals
7666 are split off. See journald.conf(5) for details.
7668 * A new condition type ConditionFileNotEmpty= has been added.
7670 * tmpfiles' "w" lines now support file globbing, to write
7671 multiple files at once.
7673 * We added Python bindings for the journal submission
7674 APIs. More Python APIs for a number of selected APIs will
7675 likely follow. Note that we intend to add native bindings
7676 only for the Python language, as we consider it common
7677 enough to deserve bindings shipped within systemd. There are
7678 various projects outside of systemd that provide bindings
7679 for languages such as PHP or Lua.
7681 * Many conditions will now resolve specifiers such as %i. In
7682 addition, PathChanged= and related directives of .path units
7683 now support specifiers as well.
7685 * There's now a new RPM macro definition for the system preset
7688 * journald will now warn if it ca not forward a message to the
7689 syslog daemon because its socket is full.
7691 * timedated will no longer write or process /etc/timezone,
7692 except on Debian. As we do not support late mounted /usr
7693 anymore /etc/localtime always being a symlink is now safe,
7694 and hence the information in /etc/timezone is not necessary
7697 * logind will now always reserve one VT for a text getty (VT6
7698 by default). Previously if more than 6 X sessions where
7699 started they took up all the VTs with auto-spawned gettys,
7700 so that no text gettys were available anymore.
7702 * udev will now automatically inform the btrfs kernel logic
7703 about btrfs RAID components showing up. This should make
7704 simple hotplug based btrfs RAID assembly work.
7706 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
7707 (but not for its children which will stay at the kernel
7708 default). This should allow setups with a lot more listening
7711 * systemd will now always pass the configured timezone to the
7712 kernel at boot. timedated will do the same when the timezone
7715 * logind's inhibition logic has been updated. By default,
7716 logind will now handle the lid switch, the power and sleep
7717 keys all the time, even in graphical sessions. If DEs want
7718 to handle these events on their own they should take the new
7719 handle-power-key, handle-sleep-key and handle-lid-switch
7720 inhibitors during their runtime. A simple way to achieve
7721 that is to invoke the DE wrapped in an invocation of:
7723 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
7725 * Access to unit operations is now checked via SELinux taking
7726 the unit file label and client process label into account.
7728 * systemd will now notify the administrator in the journal
7729 when he over-mounts a non-empty directory.
7731 * There are new specifiers that are resolved in unit files,
7732 for the host name (%H), the machine ID (%m) and the boot ID
7735 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
7736 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
7737 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
7738 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
7739 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
7740 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
7741 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
7745 * Support for reading structured kernel messages from
7746 /dev/kmsg has now been added and is enabled by default.
7748 * Support for reading kernel messages from /proc/kmsg has now
7749 been removed. If you want kernel messages in the journal
7750 make sure to run a recent kernel (>= 3.5) that supports
7751 reading structured messages from /dev/kmsg (see
7752 above). /proc/kmsg is now exclusive property of classic
7753 syslog daemons again.
7755 * The libudev API gained the new
7756 udev_device_new_from_device_id() call.
7758 * The logic for file system namespace (ReadOnlyDirectory=,
7759 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
7760 require pivot_root() anymore. This means fewer temporary
7761 directories are created below /tmp for this feature.
7763 * nspawn containers will now see and receive all submounts
7764 made on the host OS below the root file system of the
7767 * Forward Secure Sealing is now supported for Journal files,
7768 which provide cryptographical sealing of journal files so
7769 that attackers cannot alter log history anymore without this
7770 being detectable. Lennart will soon post a blog story about
7771 this explaining it in more detail.
7773 * There are two new service settings RestartPreventExitStatus=
7774 and SuccessExitStatus= which allow configuration of exit
7775 status (exit code or signal) which will be excepted from the
7776 restart logic, resp. consider successful.
7778 * journalctl gained the new --verify switch that can be used
7779 to check the integrity of the structure of journal files and
7780 (if Forward Secure Sealing is enabled) the contents of
7783 * nspawn containers will now be run with /dev/stdin, /dev/fd/
7784 and similar symlinks pre-created. This makes running shells
7785 as container init process a lot more fun.
7787 * The fstab support can now handle PARTUUID= and PARTLABEL=
7790 * A new ConditionHost= condition has been added to match
7791 against the hostname (with globs) and machine ID. This is
7792 useful for clusters where a single OS image is used to
7793 provision a large number of hosts which shall run slightly
7794 different sets of services.
7796 * Services which hit the restart limit will now be placed in a
7799 Contributions from: Bertram Poettering, Dave Reisner, Huang
7800 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
7801 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
7805 * When running in --user mode systemd will now become a
7806 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
7807 tree a lot more organized.
7809 * A new PartOf= unit dependency type has been introduced that
7810 may be used to group services in a natural way.
7812 * "systemctl enable" may now be used to enable instances of
7815 * journalctl now prints error log levels in red, and
7816 warning/notice log levels in bright white. It also supports
7817 filtering by log level now.
7819 * cgtop gained a new -n switch (similar to top), to configure
7820 the maximum number of iterations to run for. It also gained
7821 -b, to run in batch mode (accepting no input).
7823 * The suffix ".service" may now be omitted on most systemctl
7824 command lines involving service unit names.
7826 * There's a new bus call in logind to lock all sessions, as
7827 well as a loginctl verb for it "lock-sessions".
7829 * libsystemd-logind.so gained a new call sd_journal_perror()
7830 that works similar to libc perror() but logs to the journal
7831 and encodes structured information about the error number.
7833 * /etc/crypttab entries now understand the new keyfile-size=
7836 * shutdown(8) now can send a (configurable) wall message when
7837 a shutdown is cancelled.
7839 * The mount propagation mode for the root file system will now
7840 default to "shared", which is useful to make containers work
7841 nicely out-of-the-box so that they receive new mounts from
7842 the host. This can be undone locally by running "mount
7843 --make-rprivate /" if needed.
7845 * The prefdm.service file has been removed. Distributions
7846 should maintain this unit downstream if they intend to keep
7847 it around. However, we recommend writing normal unit files
7848 for display managers instead.
7850 * Since systemd is a crucial part of the OS we will now
7851 default to a number of compiler switches that improve
7852 security (hardening) such as read-only relocations, stack
7853 protection, and suchlike.
7855 * The TimeoutSec= setting for services is now split into
7856 TimeoutStartSec= and TimeoutStopSec= to allow configuration
7857 of individual time outs for the start and the stop phase of
7860 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
7861 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
7862 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
7863 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
7864 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
7865 Gundersen, Zbigniew Jędrzejewski-Szmek
7869 * The journal and id128 C APIs are now fully documented as man
7872 * Extra safety checks have been added when transitioning from
7873 the initial RAM disk to the main system to avoid accidental
7876 * /etc/crypttab entries now understand the new keyfile-offset=
7879 * systemctl -t can now be used to filter by unit load state.
7881 * The journal C API gained the new sd_journal_wait() call to
7882 make writing synchronous journal clients easier.
7884 * journalctl gained the new -D switch to show journals from a
7887 * journalctl now displays a special marker between log
7888 messages of two different boots.
7890 * The journal is now explicitly flushed to /var via a service
7891 systemd-journal-flush.service, rather than implicitly simply
7892 by seeing /var/log/journal to be writable.
7894 * journalctl (and the journal C APIs) can now match for much
7895 more complex expressions, with alternatives and
7898 * When transitioning from the initial RAM disk to the main
7899 system we will now kill all processes in a killing spree to
7900 ensure no processes stay around by accident.
7902 * Three new specifiers may be used in unit files: %u, %h, %s
7903 resolve to the user name, user home directory resp. user
7904 shell. This is useful for running systemd user instances.
7906 * We now automatically rotate journal files if their data
7907 object hash table gets a fill level > 75%. We also size the
7908 hash table based on the configured maximum file size. This
7909 together should lower hash collisions drastically and thus
7910 speed things up a bit.
7912 * journalctl gained the new "--header" switch to introspect
7913 header data of journal files.
7915 * A new setting SystemCallFilters= has been added to services
7916 which may be used to apply blacklists or whitelists to
7917 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
7919 * nspawn gained a new --link-journal= switch (and quicker: -j)
7920 to link the container journal with the host. This makes it
7921 very easy to centralize log viewing on the host for all
7922 guests while still keeping the journal files separated.
7924 * Many bugfixes and optimizations
7926 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
7927 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
7928 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
7933 * Several tools now understand kernel command line arguments,
7934 which are only read when run in an initial RAM disk. They
7935 usually follow closely their normal counterparts, but are
7938 * There's a new tool to analyze the readahead files that are
7939 automatically generated at boot. Use:
7941 /usr/lib/systemd/systemd-readahead analyze /.readahead
7943 * We now provide an early debug shell on tty9 if this enabled. Use:
7945 systemctl enable debug-shell.service
7947 * All plymouth related units have been moved into the Plymouth
7948 package. Please make sure to upgrade your Plymouth version
7951 * systemd-tmpfiles now supports getting passed the basename of
7952 a configuration file only, in which case it will look for it
7953 in all appropriate directories automatically.
7955 * udevadm info now takes a /dev or /sys path as argument, and
7956 does the right thing. Example:
7958 udevadm info /dev/sda
7959 udevadm info /sys/class/block/sda
7961 * systemctl now prints a warning if a unit is stopped but a
7962 unit that might trigger it continues to run. Example: a
7963 service is stopped but the socket that activates it is left
7966 * "systemctl status" will now mention if the log output was
7967 shortened due to rotation since a service has been started.
7969 * The journal API now exposes functions to determine the
7970 "cutoff" times due to rotation.
7972 * journald now understands SIGUSR1 and SIGUSR2 for triggering
7973 immediately flushing of runtime logs to /var if possible,
7974 resp. for triggering immediate rotation of the journal
7977 * It is now considered an error if a service is attempted to
7978 be stopped that is not loaded.
7980 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
7982 * systemd-analyze now supports Python 3
7984 * tmpfiles now supports cleaning up directories via aging
7985 where the first level dirs are always kept around but
7986 directories beneath it automatically aged. This is enabled
7987 by prefixing the age field with '~'.
7989 * Seat objects now expose CanGraphical, CanTTY properties
7990 which is required to deal with very fast bootups where the
7991 display manager might be running before the graphics drivers
7992 completed initialization.
7994 * Seat objects now expose a State property.
7996 * We now include RPM macros for service enabling/disabling
7997 based on the preset logic. We recommend RPM based
7998 distributions to make use of these macros if possible. This
7999 makes it simpler to reuse RPM spec files across
8002 * We now make sure that the collected systemd unit name is
8003 always valid when services log to the journal via
8006 * There's a new man page kernel-command-line(7) detailing all
8007 command line options we understand.
8009 * The fstab generator may now be disabled at boot by passing
8010 fstab=0 on the kernel command line.
8012 * A new kernel command line option modules-load= is now understood
8013 to load a specific kernel module statically, early at boot.
8015 * Unit names specified on the systemctl command line are now
8016 automatically escaped as needed. Also, if file system or
8017 device paths are specified they are automatically turned
8018 into the appropriate mount or device unit names. Example:
8020 systemctl status /home
8021 systemctl status /dev/sda
8023 * The SysVConsole= configuration option has been removed from
8024 system.conf parsing.
8026 * The SysV search path is no longer exported on the D-Bus
8029 * The Names= option has been removed from unit file parsing.
8031 * There's a new man page bootup(7) detailing the boot process.
8033 * Every unit and every generator we ship with systemd now
8034 comes with full documentation. The self-explanatory boot is
8037 * A couple of services gained "systemd-" prefixes in their
8038 name if they wrap systemd code, rather than only external
8039 code. Among them fsck@.service which is now
8040 systemd-fsck@.service.
8042 * The HaveWatchdog property has been removed from the D-Bus
8045 * systemd.confirm_spawn= on the kernel command line should now
8048 * There's a new man page crypttab(5) which details all options
8049 we actually understand.
8051 * systemd-nspawn gained a new --capability= switch to pass
8052 additional capabilities to the container.
8054 * timedated will now read known NTP implementation unit names
8055 from /usr/lib/systemd/ntp-units.d/*.list,
8056 systemd-timedated-ntp.target has been removed.
8058 * journalctl gained a new switch "-b" that lists log data of
8059 the current boot only.
8061 * The notify socket is in the abstract namespace again, in
8062 order to support daemons which chroot() at start-up.
8064 * There is a new Storage= configuration option for journald
8065 which allows configuration of where log data should go. This
8066 also provides a way to disable journal logging entirely, so
8067 that data collected is only forwarded to the console, the
8068 kernel log buffer or another syslog implementation.
8070 * Many bugfixes and optimizations
8072 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
8073 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
8074 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
8075 Shawn Landden, Tom Gundersen
8079 * "systemctl help <unit>" now shows the man page if one is
8082 * Several new man pages have been added.
8084 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
8085 MaxLevelConsole= can now be specified in
8086 journald.conf. These options allow reducing the amount of
8087 data stored on disk or forwarded by the log level.
8089 * TimerSlackNSec= can now be specified in system.conf for
8090 PID1. This allows system-wide power savings.
8092 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
8093 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
8098 * logind is now capable of (optionally) handling power and
8099 sleep keys as well as the lid switch.
8101 * journalctl now understands the syntax "journalctl
8102 /usr/bin/avahi-daemon" to get all log output of a specific
8105 * CapabilityBoundingSet= in system.conf now also influences
8106 the capability bound set of usermode helpers of the kernel.
8108 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
8109 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
8110 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
8111 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
8115 * Note that we skipped 139 releases here in order to set the
8116 new version to something that is greater than both udev's
8117 and systemd's most recent version number.
8119 * udev: all udev sources are merged into the systemd source tree now.
8120 All future udev development will happen in the systemd tree. It
8121 is still fully supported to use the udev daemon and tools without
8122 systemd running, like in initramfs or other init systems. Building
8123 udev though, will require the *build* of the systemd tree, but
8124 udev can be properly *run* without systemd.
8126 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
8127 should be used to create dead device nodes as workarounds for broken
8130 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
8131 no longer supported. udev_monitor_new_from_netlink() needs to be
8132 used to subscribe to events.
8134 * udev: when udevd is started by systemd, processes which are left
8135 behind by forking them off of udev rules, are unconditionally cleaned
8136 up and killed now after the event handling has finished. Services or
8137 daemons must be started as systemd services. Services can be
8138 pulled-in by udev to get started, but they can no longer be directly
8139 forked by udev rules.
8141 * udev: the daemon binary is called systemd-udevd now and installed
8142 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
8143 to adapt to that, create symlink, or rename the binary after building
8146 * libudev no longer provides these symbols:
8147 udev_monitor_from_socket()
8148 udev_queue_get_failed_list_entry()
8149 udev_get_{dev,sys,run}_path()
8150 The versions number was bumped and symbol versioning introduced.
8152 * systemd-loginctl and systemd-journalctl have been renamed
8153 to loginctl and journalctl to match systemctl.
8155 * The config files: /etc/systemd/systemd-logind.conf and
8156 /etc/systemd/systemd-journald.conf have been renamed to
8157 logind.conf and journald.conf. Package updates should rename
8158 the files to the new names on upgrade.
8160 * For almost all files the license is now LGPL2.1+, changed
8161 from the previous GPL2.0+. Exceptions are some minor stuff
8162 of udev (which will be changed to LGPL2.1 eventually, too),
8163 and the MIT licensed sd-daemon.[ch] library that is suitable
8164 to be used as drop-in files.
8166 * systemd and logind now handle system sleep states, in
8167 particular suspending and hibernating.
8169 * logind now implements a sleep/shutdown/idle inhibiting logic
8170 suitable for a variety of uses. Soonishly Lennart will blog
8171 about this in more detail.
8173 * var-run.mount and var-lock.mount are no longer provided
8174 (which previously bind mounted these directories to their new
8175 places). Distributions which have not converted these
8176 directories to symlinks should consider stealing these files
8177 from git history and add them downstream.
8179 * We introduced the Documentation= field for units and added
8180 this to all our shipped units. This is useful to make it
8181 easier to explore the boot and the purpose of the various
8184 * All smaller setup units (such as
8185 systemd-vconsole-setup.service) now detect properly if they
8186 are run in a container and are skipped when
8187 appropriate. This guarantees an entirely noise-free boot in
8188 Linux container environments such as systemd-nspawn.
8190 * A framework for implementing offline system updates is now
8191 integrated, for details see:
8192 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
8194 * A new service type Type=idle is available now which helps us
8195 avoiding ugly interleaving of getty output and boot status
8198 * There's now a system-wide CapabilityBoundingSet= option to
8199 globally reduce the set of capabilities for the
8200 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
8201 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
8202 even CAP_NET_ADMIN system-wide for secure systems.
8204 * There are now system-wide DefaultLimitXXX= options to
8205 globally change the defaults of the various resource limits
8206 for all units started by PID 1.
8208 * Harald Hoyer's systemd test suite has been integrated into
8209 systemd which allows easy testing of systemd builds in qemu
8210 and nspawn. (This is really awesome! Ask us for details!)
8212 * The fstab parser is now implemented as generator, not inside
8215 * systemctl will now warn you if .mount units generated from
8216 /etc/fstab are out of date due to changes in fstab that
8217 have not been read by systemd yet.
8219 * systemd is now suitable for usage in initrds. Dracut has
8220 already been updated to make use of this. With this in place
8221 initrds get a slight bit faster but primarily are much
8222 easier to introspect and debug since "systemctl status" in
8223 the host system can be used to introspect initrd services,
8224 and the journal from the initrd is kept around too.
8226 * systemd-delta has been added, a tool to explore differences
8227 between user/admin configuration and vendor defaults.
8229 * PrivateTmp= now affects both /tmp and /var/tmp.
8231 * Boot time status messages are now much prettier and feature
8232 proper english language. Booting up systemd has never been
8235 * Read-ahead pack files now include the inode number of all
8236 files to pre-cache. When the inode changes the pre-caching
8237 is not attempted. This should be nicer to deal with updated
8238 packages which might result in changes of read-ahead
8241 * We now temporaritly lower the kernel's read_ahead_kb variable
8242 when collecting read-ahead data to ensure the kernel's
8243 built-in read-ahead does not add noise to our measurements
8244 of necessary blocks to pre-cache.
8246 * There's now RequiresMountsFor= to add automatic dependencies
8247 for all mounts necessary for a specific file system path.
8249 * MountAuto= and SwapAuto= have been removed from
8250 system.conf. Mounting file systems at boot has to take place
8253 * nspawn now learned a new switch --uuid= to set the machine
8254 ID on the command line.
8256 * nspawn now learned the -b switch to automatically search
8259 * vt102 is now the default TERM for serial TTYs, upgraded from
8262 * systemd-logind now works on VT-less systems.
8264 * The build tree has been reorganized. The individual
8265 components now have directories of their own.
8267 * A new condition type ConditionPathIsReadWrite= is now available.
8269 * nspawn learned the new -C switch to create cgroups for the
8270 container in other hierarchies.
8272 * We now have support for hardware watchdogs, configurable in
8275 * The scheduled shutdown logic now has a public API.
8277 * We now mount /tmp as tmpfs by default, but this can be
8278 masked and /etc/fstab can override it.
8280 * Since udisks does not make use of /media anymore we are not
8281 mounting a tmpfs on it anymore.
8283 * journalctl gained a new --local switch to only interleave
8284 locally generated journal files.
8286 * We can now load the IMA policy at boot automatically.
8288 * The GTK tools have been split off into a systemd-ui.
8290 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
8291 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
8292 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
8293 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
8294 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
8295 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
8296 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
8297 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
8298 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
8303 * This is mostly a bugfix release
8305 * Support optional initialization of the machine ID from the
8306 KVM or container configured UUID.
8308 * Support immediate reboots with "systemctl reboot -ff"
8310 * Show /etc/os-release data in systemd-analyze output
8312 * Many bugfixes for the journal, including endianness fixes and
8313 ensuring that disk space enforcement works
8315 * sd-login.h is C++ compatible again
8317 * Extend the /etc/os-release format on request of the Debian
8320 * We now refuse non-UTF8 strings used in various configuration
8321 and unit files. This is done to ensure we do not pass invalid
8322 data over D-Bus or expose it elsewhere.
8324 * Register Mimo USB Screens as suitable for automatic seat
8327 * Read SELinux client context from journal clients in a race
8330 * Reorder configuration file lookup order. /etc now always
8331 overrides /run in order to allow the administrator to always
8332 and unconditionally override vendor-supplied or
8333 automatically generated data.
8335 * The various user visible bits of the journal now have man
8336 pages. We still lack man pages for the journal API calls
8339 * We now ship all man pages in HTML format again in the
8342 Contributions from: Dave Reisner, Dirk Eibach, Frederic
8343 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
8344 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
8349 * This is mostly a bugfix release
8351 * systems lacking /etc/os-release are no longer supported.
8353 * Various functionality updates to libsystemd-login.so
8355 * Track class of PAM logins to distinguish greeters from
8358 Contributions from: Kay Sievers, Lennart Poettering, Michael
8363 * This is an important bugfix release for v41.
8365 * Building man pages is now optional which should be useful
8366 for those building systemd from git but unwilling to install
8369 * Watchdog support for supervising services is now usable. In
8370 a future release support for hardware watchdogs
8371 (i.e. /dev/watchdog) will be added building on this.
8373 * Service start rate limiting is now configurable and can be
8374 turned off per service. When a start rate limit is hit a
8375 reboot can automatically be triggered.
8377 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
8379 Contributions from: Benjamin Franzke, Bill Nottingham,
8380 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
8381 Schmidt, Michał Górny, Piotr Drąg
8385 * The systemd binary is installed /usr/lib/systemd/systemd now;
8386 An existing /sbin/init symlink needs to be adapted with the
8389 * The code that loads kernel modules has been ported to invoke
8390 libkmod directly, instead of modprobe. This means we do not
8391 support systems with module-init-tools anymore.
8393 * Watchdog support is now already useful, but still not
8396 * A new kernel command line option systemd.setenv= is
8397 understood to set system wide environment variables
8398 dynamically at boot.
8400 * We now limit the set of capabilities of systemd-journald.
8402 * We now set SIGPIPE to ignore by default, since it only is
8403 useful in shell pipelines, and has little use in general
8404 code. This can be disabled with IgnoreSIPIPE=no in unit
8407 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
8408 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
8413 * This is mostly a bugfix release
8415 * We now expose the reason why a service failed in the
8416 "Result" D-Bus property.
8418 * Rudimentary service watchdog support (will be completed over
8419 the next few releases.)
8421 * When systemd forks off in order execute some service we will
8422 now immediately changes its argv[0] to reflect which process
8423 it will execute. This is useful to minimize the time window
8424 with a generic argv[0], which makes bootcharts more useful
8426 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
8427 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
8428 Mike Kazantsev, Ray Strode
8432 * This is mostly a test release, but incorporates many
8435 * New systemd-cgtop tool to show control groups by their
8438 * Linking against libacl for ACLs is optional again. If
8439 disabled, support tracking device access for active logins
8440 goes becomes unavailable, and so does access to the user
8441 journals by the respective users.
8443 * If a group "adm" exists, journal files are automatically
8444 owned by them, thus allow members of this group full access
8445 to the system journal as well as all user journals.
8447 * The journal now stores the SELinux context of the logging
8448 client for all entries.
8450 * Add C++ inclusion guards to all public headers
8452 * New output mode "cat" in the journal to print only text
8453 messages, without any meta data like date or time.
8455 * Include tiny X server wrapper as a temporary stop-gap to
8456 teach XOrg udev display enumeration. This is used by display
8457 managers such as gdm, and will go away as soon as XOrg
8458 learned native udev hotplugging for display devices.
8460 * Add new systemd-cat tool for executing arbitrary programs
8461 with STDERR/STDOUT connected to the journal. Can also act as
8462 BSD logger replacement, and does so by default.
8464 * Optionally store all locally generated coredumps in the
8465 journal along with meta data.
8467 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
8468 writing short strings to files (for usage for /sys), and for
8469 creating symlinks, character and block device nodes.
8471 * New unit file option ControlGroupPersistent= to make cgroups
8472 persistent, following the mechanisms outlined in
8473 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
8475 * Support multiple local RTCs in a sane way
8477 * No longer monopolize IO when replaying readahead data on
8478 rotating disks, since we might starve non-file-system IO to
8479 death, since fanotify() will not see accesses done by blkid,
8482 * Do not show kernel threads in systemd-cgls anymore, unless
8483 requested with new -k switch.
8485 Contributions from: Dan Horák, Kay Sievers, Lennart
8486 Poettering, Michal Schmidt
8490 * This is mostly a test release, but incorporates many
8493 * The git repository moved to:
8494 git://anongit.freedesktop.org/systemd/systemd
8495 ssh://git.freedesktop.org/git/systemd/systemd
8497 * First release with the journal
8498 http://0pointer.de/blog/projects/the-journal.html
8500 * The journal replaces both systemd-kmsg-syslogd and
8501 systemd-stdout-bridge.
8503 * New sd_pid_get_unit() API call in libsystemd-logind
8505 * Many systemadm clean-ups
8507 * Introduce remote-fs-pre.target which is ordered before all
8508 remote mounts and may be used to start services before all
8511 * Added Mageia support
8513 * Add bash completion for systemd-loginctl
8515 * Actively monitor PID file creation for daemons which exit in
8516 the parent process before having finished writing the PID
8517 file in the daemon process. Daemons which do this need to be
8518 fixed (i.e. PID file creation must have finished before the
8519 parent exits), but we now react a bit more gracefully to them.
8521 * Add colourful boot output, mimicking the well-known output
8522 of existing distributions.
8524 * New option PassCredentials= for socket units, for
8525 compatibility with a recent kernel ABI breakage.
8527 * /etc/rc.local is now hooked in via a generator binary, and
8528 thus will no longer act as synchronization point during
8531 * systemctl list-unit-files now supports --root=.
8533 * systemd-tmpfiles now understands two new commands: z, Z for
8534 relabelling files according to the SELinux database. This is
8535 useful to apply SELinux labels to specific files in /sys,
8538 * Output of SysV services is now forwarded to both the console
8539 and the journal by default, not only just the console.
8541 * New man pages for all APIs from libsystemd-login.
8543 * The build tree got reorganized and the build system is a
8544 lot more modular allowing embedded setups to specifically
8545 select the components of systemd they are interested in.
8547 * Support for Linux systems lacking the kernel VT subsystem is
8550 * configure's --with-rootdir= got renamed to
8551 --with-rootprefix= to follow the naming used by udev and
8554 * Unless specified otherwise we will now install to /usr instead
8555 of /usr/local by default.
8557 * Processes with '@' in argv[0][0] are now excluded from the
8558 final shut-down killing spree, following the logic explained
8560 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
8562 * All processes remaining in a service cgroup when we enter
8563 the START or START_PRE states are now killed with
8564 SIGKILL. That means it is no longer possible to spawn
8565 background processes from ExecStart= lines (which was never
8566 supported anyway, and bad style).
8568 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
8569 reloading of units together.
8571 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
8572 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
8573 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
8574 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
8575 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek