1 systemd System and Service Manager
5 * The documentation has been updated to cover the various new
6 concepts introduced with 205.
8 * Unit files now understand the new %v specifier which
9 resolves to the kernel version string as returned by "uname
12 * systemctl now supports filtering the unit list output by
13 load state, active state and sub state, using the new
16 * "systemctl status" will now show the results of the
17 condition checks (like ConditionPathExists= and similar) of
18 the last start attempts of the unit. They are also logged to
21 * "journalctl -b" may now be used to look for boot output of a
22 specific boot. Try "journalctl -b -1" for the previous boot,
23 but the syntax is substantially more powerful.
25 * "journalctl --show-cursor" has been added which prints the
26 cursor string the last shown log line. This may then be used
27 with the new "journalctl --after-cursor=" switch to continue
28 browsing logs from that point on.
30 * "journalctl --force" may now be used to force regeneration
33 * Device ACLs may now be applied to "dead" devices nodes too,
34 i.e. device nodes that are created based on modalias
35 information of unloaded kernel modules.
37 * A new RPM macro has been added that may be used to apply
38 tmpfiles configuration during package installation.
40 * systemd-detect-virt and ConditionVirtualization= now can
41 detect User-Mode-Linux machines (UML).
43 * journald will now log the effective capabilities set of
44 processes in the message metadata.
46 * systemd-cryptsetup has gained support for TrueCrypt volumes.
48 * The initrd interface has been simplified (more specifically,
49 support for passing performance data via environment
50 variables and fsck results via files in /run has been
51 removed). These features were non-essential, and are
52 nowadays available in a much nicer way by having systemd in
53 the initrd serialize its state and have the hosts systemd
60 * Two new unit types have been introduced:
62 Scope units are very similar to service units, however, are
63 created out of pre-existing processes -- instead of PID 1
64 forking off the processes. By using scope units it is
65 possible for system services and applications to group their
66 own child processes (worker processes) in a powerful way
67 which then maybe used to organize them, or kill them
68 together, or apply resource limits on them.
70 Slice units may be used to partition system resources in an
71 hierarchial fashion and then assign other units to them. By
72 default there are now three slices: system.slice (for all
73 system services), user.slice (for all user sessions),
74 machine.slice (for VMs and containers).
76 Slices and scopes have been introduced primarily in
77 context of the work to move cgroup handling to a
78 single-writer scheme, where only PID 1
79 creates/removes/manages cgroups.
81 * There's a new concept of "transient" units. In contrast to
82 normal units these units are created via an API at runtime,
83 not from configuration from disk. More specifically this
84 means it is now possible to run arbitrary programs as
85 independent services, with all execution parameters passed
86 in via bus APIs rather than read from disk. Transient units
87 make systemd substantially more dynamic then it ever was,
88 and useful as a general batch manager.
90 * logind has been updated to make use of scope and slice units
91 for managing user sessions. As a user logs in he will get
92 his own private slice unit, to which all sessions are added
93 as scope units. We also added support for automatically
94 adding an instance of user@.service for the user into the
95 slice. Effectively logind will no longer create cgroup
96 hierarchies on its own now, it will defer entirely to PID 1
97 for this by means of scope, service and slice units. Since
98 user sessions this way become entities managed by PID 1
99 the output of "systemctl" is now a lot more comprehensive.
101 * A new mini-daemon "systemd-machined" has been added which
102 may be used by virtualization managers to register local
103 VMs/containers. nspawn has been updated accordingly, and
104 libvirt will be updated shortly. machined will collect a bit
105 of meta information about the VMs/containers, and assign
106 them their own scope unit (see above). The collected
107 meta-data is then made available via the "machinectl" tool,
108 and exposed in "ps" and similar tools. machined/machinectl
109 is compile-time optional.
111 * As discussed earlier, the low-level cgroup configuration
112 options ControlGroup=, ControlGroupModify=,
113 ControlGroupPersistent=, ControlGroupAttribute= have been
114 removed. Please use high-level attribute settings instead as
117 * A new bus call SetUnitProperties() has been added to alter
118 various runtime parameters of a unit. This is primarily
119 useful to alter cgroup parameters dynamically in a nice way,
120 but will be extended later on to make more properties
121 modifiable at runtime. systemctl gained a new set-properties
122 command that wraps this call.
124 * A new tool "systemd-run" has been added which can be used to
125 run arbitrary command lines as transient services or scopes,
126 while configuring a number of settings via the command
127 line. This tool is currently very basic, however already
128 very useful. We plan to extend this tool to even allow
129 queuing of execution jobs with time triggers from the
130 command line, similar in fashion to "at".
132 * nspawn will now inform the user explicitly that kernels with
133 audit enabled break containers, and suggest the user to turn
136 * Support for detecting the IMA and AppArmor security
137 frameworks with ConditionSecurity= has been added.
139 * journalctl gained a new "-k" switch for showing only kernel
140 messages, mimicking dmesg output; in addition to "--user"
141 and "--system" switches for showing only user's own logs
144 * systemd-delta can now show information about drop-in
145 snippets extending unit files.
147 * libsystemd-bus has been substantially updated but is still
148 not available as public API.
150 * systemd will now look for the "debug" argument on the kernel
151 command line and enable debug logging, similar to
152 "systemd.log_level=debug" already did before.
154 * "systemctl set-default", "systemctl get-default" has been
155 added to configure the default.target symlink, which
156 controls what to boot into by default.
158 * "systemctl set-log-level" has been added as a convenient
159 way to raise and lower systemd logging threshold.
161 * "systemd-analyze plot" will now show the time the various
162 generators needed for execution, as well as information
163 about the unit file loading.
165 * libsystemd-journal gained a new sd_journal_open_files() call
166 for opening specific journal files. journactl also gained a
167 new switch to expose this new functionality. Previously we
168 only supported opening all files from a directory, or all
169 files from the system, as opening individual files only is
170 racy due to journal file rotation.
172 * systemd gained the new DefaultEnvironment= setting in
173 /etc/systemd/system.conf to set environment variables for
176 * If a privileged process logs a journal message with the
177 OBJECT_PID= field set, then journald will automatically
178 augment this with additional OBJECT_UID=, OBJECT_GID=,
179 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
180 system services want to log events about specific client
181 processes. journactl/systemctl has been updated to make use
182 of this information if all log messages regarding a specific
185 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
186 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
187 Reisner, David Coppa, David King, David Strauss, Eelco
188 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
189 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
190 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
191 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
192 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
193 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
194 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
195 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
196 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
201 * The Python bindings gained some minimal support for the APIs
202 exposed by libsystemd-logind.
204 * ConditionSecurity= gained support for detecting SMACK. Since
205 this condition already supports SELinux and AppArmor we only
206 miss IMA for this. Patches welcome!
208 Contributions from: Karol Lewandowski, Lennart Poettering,
209 Zbigniew Jędrzejewski-Szmek
213 * systemd-nspawn will now create /etc/resolv.conf if
214 necessary, before bind-mounting the host's file onto it.
216 * systemd-nspawn will now store meta information about a
217 container on the container's cgroup as extended attribute
218 fields, including the root directory.
220 * The cgroup hierarchy has been reworked in many ways. All
221 objects any of the components systemd creates in the cgroup
222 tree are now suffixed. More specifically, user sessions are
223 now placed in cgroups suffixed with ".session", users in
224 cgroups suffixed with ".user", and nspawn containers in
225 cgroups suffixed with ".nspawn". Furthermore, all cgroup
226 names are now escaped in a simple scheme to avoid collision
227 of userspace object names with kernel filenames. This work
228 is preparation for making these objects relocatable in the
229 cgroup tree, in order to allow easy resource partitioning of
230 these objects without causing naming conflicts.
232 * systemctl list-dependencies gained the new switches
233 --plain, --reverse, --after and --before.
235 * systemd-inhibit now shows the process name of processes that
236 have taken an inhibitor lock.
238 * nss-myhostname will now also resolve "localhost"
239 implicitly. This makes /etc/hosts an optional file and
240 nicely handles that on IPv6 ::1 maps to both "localhost" and
243 * libsystemd-logind.so gained a new call
244 sd_get_machine_names() to enumerate running containers and
245 VMs (currently only supported by very new libvirt and
246 nspawn). sd_login_monitor can now be used to watch
247 VMs/containers coming and going.
249 * .include is not allowed recursively anymore, and only in
250 unit files. Usually it is better to use drop-in snippets in
251 .d/*.conf anyway, as introduced with systemd 198.
253 * systemd-analyze gained a new "critical-chain" command that
254 determines the slowest chain of units run during system
255 boot-up. It is very useful for tracking down where
256 optimizing boot time is the most beneficial.
258 * systemd will no longer allow manipulating service paths in
259 the name=systemd:/system cgroup tree using ControlGroup= in
260 units. (But is still fine with it in all other dirs.)
262 * There's a new systemd-nspawn@.service service file that may
263 be used to easily run nspawn containers as system
264 services. With the container's root directory in
265 /var/lib/container/foobar it is now sufficient to run
266 "systemctl start systemd-nspawn@foobar.service" to boot it.
268 * systemd-cgls gained a new parameter "--machine" to list only
269 the processes within a certain container.
271 * ConditionSecurity= now can check for "apparmor". We still
272 are lacking checks for SMACK and IMA for this condition
273 check though. Patches welcome!
275 * A new configuration file /etc/systemd/sleep.conf has been
276 added that may be used to configure which kernel operation
277 systemd is supposed to execute when "suspend", "hibernate"
278 or "hybrid-sleep" is requested. This makes the new kernel
279 "freeze" state accessible to the user.
281 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
282 the passed argument if applicable.
284 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
285 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
286 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
287 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
288 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
289 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
290 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
295 * The output of 'systemctl list-jobs' got some polishing. The
296 '--type=' argument may now be passed more than once. A new
297 command 'systemctl list-sockets' has been added which shows
298 a list of kernel sockets systemd is listening on with the
299 socket units they belong to, plus the units these socket
302 * The experimental libsystemd-bus library got substantial
303 updates to work in conjunction with the (also experimental)
304 kdbus kernel project. It works well enough to exchange
305 messages with some sophistication. Note that kdbus is not
306 ready yet, and the library is mostly an elaborate test case
307 for now, and not installable.
309 * systemd gained a new unit 'systemd-static-nodes.service'
310 that generates static device nodes earlier during boot, and
311 can run in conjunction with udev.
313 * libsystemd-login gained a new call sd_pid_get_user_unit()
314 to retrieve the user systemd unit a process is running
315 in. This is useful for systems where systemd is used as
318 * systemd-nspawn now places all containers in the new /machine
319 top-level cgroup directory in the name=systemd
320 hierarchy. libvirt will soon do the same, so that we get a
321 uniform separation of /system, /user and /machine for system
322 services, user processes and containers/virtual
323 machines. This new cgroup hierarchy is also useful to stick
324 stable names to specific container instances, which can be
325 recognized later this way (this name may be controlled
326 via systemd-nspawn's new -M switch). libsystemd-login also
327 gained a new call sd_pid_get_machine_name() to retrieve the
328 name of the container/VM a specific process belongs to.
330 * bootchart can now store its data in the journal.
332 * libsystemd-journal gained a new call
333 sd_journal_add_conjunction() for AND expressions to the
334 matching logic. This can be used to express more complex
337 * journactl can now take multiple --unit= and --user-unit=
340 * The cryptsetup logic now understands the "luks.key=" kernel
341 command line switch for specifying a file to read the
342 decryption key from. Also, if a configured key file is not
343 found the tool will now automatically fall back to prompting
346 * Python systemd.journal module was updated to wrap recently
347 added functions from libsystemd-journal. The interface was
348 changed to bring the low level interface in s.j._Reader
349 closer to the C API, and the high level interface in
350 s.j.Reader was updated to wrap and convert all data about
353 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
354 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
355 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
356 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
357 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
358 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
362 * journalctl --update-catalog now understands a new --root=
363 option to operate on catalogs found in a different root
366 * During shutdown after systemd has terminated all running
367 services a final killing loop kills all remaining left-over
368 processes. We will now print the name of these processes
369 when we send SIGKILL to them, since this usually indicates a
372 * If /etc/crypttab refers to password files stored on
373 configured mount points automatic dependencies will now be
374 generated to ensure the specific mount is established first
375 before the key file is attempted to be read.
377 * 'systemctl status' will now show information about the
378 network sockets a socket unit is listening on.
380 * 'systemctl status' will also shown information about any
381 drop-in configuration file for units. (Drop-In configuration
382 files in this context are files such as
383 /etc/systemd/systemd/foobar.service.d/*.conf)
385 * systemd-cgtop now optionally shows summed up CPU times of
386 cgroups. Press '%' while running cgtop to switch between
387 percentage and absolute mode. This is useful to determine
388 which cgroups use up the most CPU time over the entire
389 runtime of the system. systemd-cgtop has also been updated
390 to be 'pipeable' for processing with further shell tools.
392 * 'hostnamectl set-hostname' will now allow setting of FQDN
395 * The formatting and parsing of time span values has been
396 changed. The parser now understands fractional expressions
397 such as "5.5h". The formatter will now output fractional
398 expressions for all time spans under 1min, i.e. "5.123456s"
399 rather than "5s 123ms 456us". For time spans under 1s
400 millisecond values are shown, for those under 1ms
401 microsecond values are shown. This should greatly improve
402 all time-related output of systemd.
404 * libsystemd-login and libsystemd-journal gained new
405 functions for querying the poll() events mask and poll()
406 timeout value for integration into arbitrary event
409 * localectl gained the ability to list available X11 keymaps
410 (models, layouts, variants, options).
412 * 'systemd-analyze dot' gained the ability to filter for
413 specific units via shell-style globs, to create smaller,
414 more useful graphs. I.e. it's now possible to create simple
415 graphs of all the dependencies between only target units, or
416 of all units that Avahi has dependencies with.
418 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
419 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
420 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
421 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
422 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
423 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
424 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
428 * The boot-time readahead implementation for rotating media
429 will now read the read-ahead data in multiple passes which
430 consist of all read requests made in equidistant time
431 intervals. This means instead of strictly reading read-ahead
432 data in its physical order on disk we now try to find a
433 middle ground between physical and access time order.
435 * /etc/os-release files gained a new BUILD_ID= field for usage
436 on operating systems that provide continuous builds of OS
439 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
440 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
441 William Douglas, Zbigniew Jędrzejewski-Szmek
445 * systemd-python gained an API exposing libsystemd-daemon.
447 * The SMACK setup logic gained support for uploading CIPSO
450 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
451 ReadOnlyDirectories= and InaccessibleDirectories= has
452 changed. The private /tmp and /var/tmp directories are now
453 shared by all processes of a service (which means
454 ExecStartPre= may now leave data in /tmp that ExecStart= of
455 the same service can still access). When a service is
456 stopped its temporary directories are immediately deleted
457 (normal clean-up with tmpfiles is still done in addition to
460 * By default, systemd will now set a couple of sysctl
461 variables in the kernel: the safe sysrq options are turned
462 on, IP route verification is turned on, and source routing
463 disabled. The recently added hardlink and softlink
464 protection of the kernel is turned on. These settings should
465 be reasonably safe, and good defaults for all new systems.
467 * The predictable network naming logic may now be turned off
468 with a new kernel command line switch: net.ifnames=0.
470 * A new libsystemd-bus module has been added that implements a
471 pretty complete D-Bus client library. For details see:
473 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
475 * journald will now explicitly flush the journal files to disk
476 at the latest 5min after each write. The file will then also
477 be marked offline until the next write. This should increase
478 reliability in case of a crash. The synchronization delay
479 can be configured via SyncIntervalSec= in journald.conf.
481 * There's a new remote-fs-setup.target unit that can be used
482 to pull in specific services when at least one remote file
483 system is to be mounted.
485 * There are new targets timers.target and paths.target as
486 canonical targets to pull user timer and path units in
487 from. This complements sockets.target with a similar
488 purpose for socket units.
490 * libudev gained a new call udev_device_set_attribute_value()
491 to set sysfs attributes of a device.
493 * The udev daemon now sets the default number of worker
494 processes executed in parallel based on the number of available
495 CPUs instead of the amount of available RAM. This is supposed
496 to provide a more reliable default and limit a too aggressive
497 paralellism for setups with 1000s of devices connected.
499 Contributions from: Auke Kok, Colin Walters, Cristian
500 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
501 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
502 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
503 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
504 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
505 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
506 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
507 Zbigniew Jędrzejewski-Szmek
511 * Configuration of unit files may now be extended via drop-in
512 files without having to edit/override the unit files
513 themselves. More specifically, if the administrator wants to
514 change one value for a service file foobar.service he can
515 now do so by dropping in a configuration snippet into
516 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
517 will load all these snippets and apply them on top of the
518 main unit configuration file, possibly extending or
519 overriding its settings. Using these drop-in snippets is
520 generally nicer than the two earlier options for changing
521 unit files locally: copying the files from
522 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
523 them there; or creating a new file in /etc/systemd/system/
524 that incorporates the original one via ".include". Drop-in
525 snippets into these .d/ directories can be placed in any
526 directory systemd looks for units in, and the usual
527 overriding semantics between /usr/lib, /etc and /run apply
530 * Most unit file settings which take lists of items can now be
531 reset by assigning the empty string to them. For example,
532 normally, settings such as Environment=FOO=BAR append a new
533 environment variable assignment to the environment block,
534 each time they are used. By assigning Environment= the empty
535 string the environment block can be reset to empty. This is
536 particularly useful with the .d/*.conf drop-in snippets
537 mentioned above, since this adds the ability to reset list
538 settings from vendor unit files via these drop-ins.
540 * systemctl gained a new "list-dependencies" command for
541 listing the dependencies of a unit recursively.
543 * Inhibitors are now honored and listed by "systemctl
544 suspend", "systemctl poweroff" (and similar) too, not only
545 GNOME. These commands will also list active sessions by
548 * Resource limits (as exposed by the various control group
549 controllers) can now be controlled dynamically at runtime
550 for all units. More specifically, you can now use a command
551 like "systemctl set-cgroup-attr foobar.service cpu.shares
552 2000" to alter the CPU shares a specific service gets. These
553 settings are stored persistently on disk, and thus allow the
554 administrator to easily adjust the resource usage of
555 services with a few simple commands. This dynamic resource
556 management logic is also available to other programs via the
557 bus. Almost any kernel cgroup attribute and controller is
560 * systemd-vconsole-setup will now copy all font settings to
561 all allocated VTs, where it previously applied them only to
564 * libsystemd-login gained the new sd_session_get_tty() API
567 * This release drops support for a few legacy or
568 distribution-specific LSB facility names when parsing init
569 scripts: $x-display-manager, $mail-transfer-agent,
570 $mail-transport-agent, $mail-transfer-agent, $smtp,
571 $null. Also, the mail-transfer-agent.target unit backing
572 this has been removed. Distributions which want to retain
573 compatibility with this should carry the burden for
574 supporting this themselves and patch support for these back
575 in, if they really need to. Also, the facilities $syslog and
576 $local_fs are now ignored, since systemd does not support
577 early-boot LSB init scripts anymore, and these facilities
578 are implied anyway for normal services. syslog.target has
581 * There are new bus calls on PID1's Manager object for
582 cancelling jobs, and removing snapshot units. Previously,
583 both calls were only available on the Job and Snapshot
586 * systemd-journal-gatewayd gained SSL support.
588 * The various "environment" files, such as /etc/locale.conf
589 now support continuation lines with a backslash ("\") as
590 last character in the line, similar in style (but different)
591 to how this is supported in shells.
593 * For normal user processes the _SYSTEMD_USER_UNIT= field is
594 now implicitly appended to every log entry logged. systemctl
595 has been updated to filter by this field when operating on a
596 user systemd instance.
598 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
599 CAP_AUDIT_CONTROL capabilities to the capabilities set for
600 the container. This makes it easier to boot unmodified
601 Fedora systems in a container, which however still requires
602 audit=0 to be passed on the kernel command line. Auditing in
603 kernel and userspace is unfortunately still too broken in
604 context of containers, hence we recommend compiling it out
605 of the kernel or using audit=0. Hopefully this will be fixed
606 one day for good in the kernel.
608 * nspawn gained the new --bind= and --bind-ro= parameters to
609 bind mount specific directories from the host into the
612 * nspawn will now mount its own devpts file system instance
613 into the container, in order not to leak pty devices from
614 the host into the container.
616 * systemd will now read the firmware boot time performance
617 information from the EFI variables, if the used boot loader
618 supports this, and takes it into account for boot performance
619 analysis via "systemd-analyze". This is currently supported
620 only in conjunction with Gummiboot, but could be supported
621 by other boot loaders too. For details see:
623 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
625 * A new generator has been added that automatically mounts the
626 EFI System Partition (ESP) to /boot, if that directory
627 exists, is empty, and no other file system has been
628 configured to be mounted there.
630 * logind will now send out PrepareForSleep(false) out
631 unconditionally, after coming back from suspend. This may be
632 used by applications as asynchronous notification for
633 system resume events.
635 * "systemctl unlock-sessions" has been added, that allows
636 unlocking the screens of all user sessions at once, similar
637 how "systemctl lock-sessions" already locked all users
638 sessions. This is backed by a new D-Bus call UnlockSessions().
640 * "loginctl seat-status" will now show the master device of a
641 seat. (i.e. the device of a seat that needs to be around for
642 the seat to be considered available, usually the graphics
645 * tmpfiles gained a new "X" line type, that allows
646 configuration of files and directories (with wildcards) that
647 shall be excluded from automatic cleanup ("aging").
649 * udev default rules set the device node permissions now only
650 at "add" events, and do not change them any longer with a
651 later "change" event.
653 * The log messages for lid events and power/sleep keypresses
654 now carry a message ID.
656 * We now have a substantially larger unit test suite, but this
657 continues to be work in progress.
659 * udevadm hwdb gained a new --root= parameter to change the
660 root directory to operate relative to.
662 * logind will now issue a background sync() request to the kernel
663 early at shutdown, so that dirty buffers are flushed to disk early
664 instead of at the last moment, in order to optimize shutdown
667 * A new bootctl tool has been added that is an interface for
668 certain boot loader operations. This is currently a preview
669 and is likely to be extended into a small mechanism daemon
670 like timedated, localed, hostnamed, and can be used by
671 graphical UIs to enumerate available boot options, and
672 request boot into firmware operations.
674 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
675 the rest of the package. It also has been updated to work
676 correctly in initrds.
678 * Policykit previously has been runtime optional, and is now
679 also compile time optional via a configure switch.
681 * systemd-analyze has been reimplemented in C. Also "systemctl
682 dot" has moved into systemd-analyze.
684 * "systemctl status" with no further parameters will now print
685 the status of all active or failed units.
687 * Operations such as "systemctl start" can now be executed
688 with a new mode "--irreversible" which may be used to queue
689 operations that cannot accidentally be reversed by a later
690 job queuing. This is by default used to make shutdown
691 requests more robust.
693 * The Python API of systemd now gained a new module for
694 reading journal files.
696 * A new tool kernel-install has been added that can install
697 kernel images according to the Boot Loader Specification:
699 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
701 * Boot time console output has been improved to provide
702 animated boot time output for hanging jobs.
704 * A new tool systemd-activate has been added which can be used
705 to test socket activation with, directly from the command
706 line. This should make it much easier to test and debug
707 socket activation in daemons.
709 * journalctl gained a new "--reverse" (or -r) option to show
710 journal output in reverse order (i.e. newest line first).
712 * journalctl gained a new "--pager-end" (or -e) option to jump
713 to immediately jump to the end of the journal in the
714 pager. This is only supported in conjunction with "less".
716 * journalctl gained a new "--user-unit=" option, that works
717 similar to "--unit=" but filters for user units rather than
720 * A number of unit files to ease adoption of systemd in
721 initrds has been added. This moves some minimal logic from
722 the various initrd implementations into systemd proper.
724 * The journal files are now owned by a new group
725 "systemd-journal", which exists specifically to allow access
726 to the journal, and nothing else. Previously, we used the
727 "adm" group for that, which however possibly covers more
728 than just journal/log file access. This new group is now
729 already used by systemd-journal-gatewayd to ensure this
730 daemon gets access to the journal files and as little else
731 as possible. Note that "make install" will also set FS ACLs
732 up for /var/log/journal to give "adm" and "wheel" read
733 access to it, in addition to "systemd-journal" which owns
734 the journal files. We recommend that packaging scripts also
735 add read access to "adm" + "wheel" to /var/log/journal, and
736 all existing/future journal files. To normal users and
737 administrators little changes, however packagers need to
738 ensure to create the "systemd-journal" system group at
739 package installation time.
741 * The systemd-journal-gatewayd now runs as unprivileged user
742 systemd-journal-gateway:systemd-journal-gateway. Packaging
743 scripts need to create these system user/group at
746 * timedated now exposes a new boolean property CanNTP that
747 indicates whether a local NTP service is available or not.
749 * systemd-detect-virt will now also detect xen PVs
751 * The pstore file system is now mounted by default, if it is
754 * In addition to the SELinux and IMA policies we will now also
755 load SMACK policies at early boot.
757 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
758 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
759 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
760 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
761 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
762 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
763 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
764 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
765 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
766 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
767 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
768 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
769 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
770 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
774 * Timer units now support calendar time events in addition to
775 monotonic time events. That means you can now trigger a unit
776 based on a calendar time specification such as "Thu,Fri
777 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
778 or fifth day of any month of the year 2013, given that it is
779 a thursday or friday. This brings timer event support
780 considerably closer to cron's capabilities. For details on
781 the supported calendar time specification language see
784 * udev now supports a number of different naming policies for
785 network interfaces for predictable names, and a combination
786 of these policies is now the default. Please see this wiki
787 document for details:
789 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
791 * Auke Kok's bootchart implementation has been added to the
792 systemd tree. It's an optional component that can graph the
793 boot in quite some detail. It's one of the best bootchart
794 implementations around and minimal in its code and
797 * nss-myhostname has been integrated into the systemd source
798 tree. nss-myhostname guarantees that the local hostname
799 always stays resolvable via NSS. It has been a weak
800 requirement of systemd-hostnamed since a long time, and
801 since its code is actually trivial we decided to just
802 include it in systemd's source tree. It can be turned off
803 with a configure switch.
805 * The read-ahead logic is now capable of properly detecting
806 whether a btrfs file system is on SSD or rotating media, in
807 order to optimize the read-ahead scheme. Previously, it was
808 only capable of detecting this on traditional file systems
811 * In udev, additional device properties are now read from the
812 IAB in addition to the OUI database. Also, Bluetooth company
813 identities are attached to the devices as well.
815 * In service files %U may be used as specifier that is
816 replaced by the configured user name of the service.
818 * nspawn may now be invoked without a controlling TTY. This
819 makes it suitable for invocation as its own service. This
820 may be used to set up a simple containerized server system
821 using only core OS tools.
823 * systemd and nspawn can now accept socket file descriptors
824 when they are started for socket activation. This enables
825 implementation of socket activated nspawn
826 containers. i.e. think about autospawning an entire OS image
827 when the first SSH or HTTP connection is received. We expect
828 that similar functionality will also be added to libvirt-lxc
831 * journalctl will now suppress ANSI color codes when
834 * systemctl will no longer show control group information for
835 a unit if a the control group is empty anyway.
837 * logind can now automatically suspend/hibernate/shutdown the
840 * /etc/machine-info and hostnamed now also expose the chassis
841 type of the system. This can be used to determine whether
842 the local system is a laptop, desktop, handset or
843 tablet. This information may either be configured by the
844 user/vendor or is automatically determined from ACPI and DMI
845 information if possible.
847 * A number of PolicyKit actions are now bound together with
848 "imply" rules. This should simplify creating UIs because
849 many actions will now authenticate similar ones as well.
851 * Unit files learnt a new condition ConditionACPower= which
852 may be used to conditionalize a unit depending on whether an
853 AC power source is connected or not, of whether the system
854 is running on battery power.
856 * systemctl gained a new "is-failed" verb that may be used in
857 shell scripts and suchlike to check whether a specific unit
858 is in the "failed" state.
860 * The EnvironmentFile= setting in unit files now supports file
861 globbing, and can hence be used to easily read a number of
862 environment files at once.
864 * systemd will no longer detect and recognize specific
865 distributions. All distribution-specific #ifdeffery has been
866 removed, systemd is now fully generic and
867 distribution-agnostic. Effectively, not too much is lost as
868 a lot of the code is still accessible via explicit configure
869 switches. However, support for some distribution specific
870 legacy configuration file formats has been dropped. We
871 recommend distributions to simply adopt the configuration
872 files everybody else uses now and convert the old
873 configuration from packaging scripts. Most distributions
874 already did that. If that's not possible or desirable,
875 distributions are welcome to forward port the specific
876 pieces of code locally from the git history.
878 * When logging a message about a unit systemd will now always
879 log the unit name in the message meta data.
881 * localectl will now also discover system locale data that is
882 not stored in locale archives, but directly unpacked.
884 * logind will no longer unconditionally use framebuffer
885 devices as seat masters, i.e. as devices that are required
886 to be existing before a seat is considered preset. Instead,
887 it will now look for all devices that are tagged as
888 "seat-master" in udev. By default framebuffer devices will
889 be marked as such, but depending on local systems other
890 devices might be marked as well. This may be used to
891 integrate graphics cards using closed source drivers (such
892 as NVidia ones) more nicely into logind. Note however, that
893 we recommend using the open source NVidia drivers instead,
894 and no udev rules for the closed-source drivers will be
895 shipped from us upstream.
897 Contributions from: Adam Williamson, Alessandro Crismani, Auke
898 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
899 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
900 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
901 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
902 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
903 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
904 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
905 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
906 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
907 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
908 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
913 * udev gained support for loading additional device properties
914 from an indexed database that is keyed by vendor/product IDs
915 and similar device identifiers. For the beginning this
916 "hwdb" is populated with data from the well-known PCI and
917 USB database, but also includes PNP, ACPI and OID data. In
918 the longer run this indexed database shall grow into
919 becoming the one central database for non-essential
920 userspace device metadata. Previously, data from the PCI/USB
921 database was only attached to select devices, since the
922 lookup was a relatively expensive operation due to O(n) time
923 complexity (with n being the number of entries in the
924 database). Since this is now O(1), we decided to add in this
925 data for all devices where this is available, by
926 default. Note that the indexed database needs to be rebuilt
927 when new data files are installed. To achieve this you need
928 to update your packaging scripts to invoke "udevadm hwdb
929 --update" after installation of hwdb data files. For
930 RPM-based distributions we introduced the new
931 %udev_hwdb_update macro for this purpose.
933 * The Journal gained support for the "Message Catalog", an
934 indexed database to link up additional information with
935 journal entries. For further details please check:
937 http://www.freedesktop.org/wiki/Software/systemd/catalog
939 The indexed message catalog database also needs to be
940 rebuilt after installation of message catalog files. Use
941 "journalctl --update-catalog" for this. For RPM-based
942 distributions we introduced the %journal_catalog_update
943 macro for this purpose.
945 * The Python Journal bindings gained support for the standard
946 Python logging framework.
948 * The Journal API gained new functions for checking whether
949 the underlying file system of a journal file is capable of
950 properly reporting file change notifications, or whether
951 applications that want to reflect journal changes "live"
952 need to recheck journal files continuously in appropriate
955 * It is now possible to set the "age" field for tmpfiles
956 entries to 0, indicating that files matching this entry
957 shall always be removed when the directories are cleaned up.
959 * coredumpctl gained a new "gdb" verb which invokes gdb
960 right-away on the selected coredump.
962 * There's now support for "hybrid sleep" on kernels that
963 support this, in addition to "suspend" and "hibernate". Use
964 "systemctl hybrid-sleep" to make use of this.
966 * logind's HandleSuspendKey= setting (and related settings)
967 now gained support for a new "lock" setting to simply
968 request the screen lock on all local sessions, instead of
969 actually executing a suspend or hibernation.
971 * systemd will now mount the EFI variables file system by
974 * Socket units now gained support for configuration of the
975 SMACK security label.
977 * timedatectl will now output the time of the last and next
978 daylight saving change.
980 * We dropped support for various legacy and distro-specific
981 concepts, such as insserv, early-boot SysV services
982 (i.e. those for non-standard runlevels such as 'b' or 'S')
983 or ArchLinux /etc/rc.conf support. We recommend the
984 distributions who still need support this to either continue
985 to maintain the necessary patches downstream, or find a
986 different solution. (Talk to us if you have questions!)
988 * Various systemd components will now bypass PolicyKit checks
989 for root and otherwise handle properly if PolicyKit is not
990 found to be around. This should fix most issues for
991 PolicyKit-less systems. Quite frankly this should have been
992 this way since day one. It is absolutely our intention to
993 make systemd work fine on PolicyKit-less systems, and we
994 consider it a bug if something doesn't work as it should if
995 PolicyKit is not around.
997 * For embedded systems it is now possible to build udev and
998 systemd without blkid and/or kmod support.
1000 * "systemctl switch-root" is now capable of switching root
1001 more than once. I.e. in addition to transitions from the
1002 initrd to the host OS it is now possible to transition to
1003 further OS images from the host. This is useful to implement
1004 offline updating tools.
1006 * Various other additions have been made to the RPM macros
1007 shipped with systemd. Use %udev_rules_update() after
1008 installing new udev rules files. %_udevhwdbdir,
1009 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
1010 %_sysctldir are now available which resolve to the right
1011 directories for packages to place various data files in.
1013 * journalctl gained the new --full switch (in addition to
1014 --all, to disable ellipsation for long messages.
1016 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
1017 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
1018 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
1019 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
1020 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
1021 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
1022 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
1023 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
1024 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
1028 * journalctl gained new --since= and --until= switches to
1029 filter by time. It also now supports nice filtering for
1030 units via --unit=/-u.
1032 * Type=oneshot services may use ExecReload= and do the
1035 * The journal daemon now supports time-based rotation and
1036 vacuuming, in addition to the usual disk-space based
1039 * The journal will now index the available field values for
1040 each field name. This enables clients to show pretty drop
1041 downs of available match values when filtering. The bash
1042 completion of journalctl has been updated
1043 accordingly. journalctl gained a new switch -F to list all
1044 values a certain field takes in the journal database.
1046 * More service events are now written as structured messages
1047 to the journal, and made recognizable via message IDs.
1049 * The timedated, localed and hostnamed mini-services which
1050 previously only provided support for changing time, locale
1051 and hostname settings from graphical DEs such as GNOME now
1052 also have a minimal (but very useful) text-based client
1053 utility each. This is probably the nicest way to changing
1054 these settings from the command line now, especially since
1055 it lists available options and is fully integrated with bash
1058 * There's now a new tool "systemd-coredumpctl" to list and
1059 extract coredumps from the journal.
1061 * We now install a README each in /var/log/ and
1062 /etc/rc.d/init.d explaining where the system logs and init
1063 scripts went. This hopefully should help folks who go to
1064 that dirs and look into the otherwise now empty void and
1065 scratch their heads.
1067 * When user-services are invoked (by systemd --user) the
1068 $MANAGERPID env var is set to the PID of systemd.
1070 * SIGRTMIN+24 when sent to a --user instance will now result
1071 in immediate termination of systemd.
1073 * gatewayd received numerous feature additions such as a
1074 "follow" mode, for live syncing and filtering.
1076 * browse.html now allows filtering and showing detailed
1077 information on specific entries. Keyboard navigation and
1078 mouse screen support has been added.
1080 * gatewayd/journalctl now supports HTML5/JSON
1081 Server-Sent-Events as output.
1083 * The SysV init script compatibility logic will now
1084 heuristically determine whether a script supports the
1085 "reload" verb, and only then make this available as
1088 * "systemctl status --follow" has been removed, use "journalctl
1091 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
1092 have been removed since they are hardly useful to be
1095 * And I'd like to take the opportunity to specifically mention
1096 Zbigniew for his great contributions. Zbigniew, you rock!
1098 Contributions from: Andrew Eikum, Christian Hesse, Colin
1099 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
1100 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
1101 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
1102 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
1103 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
1104 Jędrzejewski-Szmek, Сковорода Никита Андреевич
1108 * If /etc/vconsole.conf is non-existent or empty we will no
1109 longer load any console font or key map at boot by
1110 default. Instead the kernel defaults will be left
1111 intact. This is definitely the right thing to do, as no
1112 configuration should mean no configuration, and hard-coding
1113 font names that are different on all archs is probably a bad
1114 idea. Also, the kernel default key map and font should be
1115 good enough for most cases anyway, and mostly identical to
1116 the userspace fonts/key maps we previously overloaded them
1117 with. If distributions want to continue to default to a
1118 non-kernel font or key map they should ship a default
1119 /etc/vconsole.conf with the appropriate contents.
1121 Contributions from: Colin Walters, Daniel J Walsh, Dave
1122 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
1123 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1127 * journalctl gained a new --cursor= switch to show entries
1128 starting from the specified location in the journal.
1130 * We now enforce a size limit on journal entry fields exported
1131 with "-o json" in journalctl. Fields larger than 4K will be
1132 assigned null. This can be turned off with --all.
1134 * An (optional) journal gateway daemon is now available as
1135 "systemd-journal-gatewayd.service". This service provides
1136 access to the journal via HTTP and JSON. This functionality
1137 will be used to implement live log synchronization in both
1138 pull and push modes, but has various other users too, such
1139 as easy log access for debugging of embedded devices. Right
1140 now it is already useful to retrieve the journal via HTTP:
1142 # systemctl start systemd-journal-gatewayd.service
1143 # wget http://localhost:19531/entries
1145 This will download the journal contents in a
1146 /var/log/messages compatible format. The same as JSON:
1148 # curl -H"Accept: application/json" http://localhost:19531/entries
1150 This service is also accessible via a web browser where a
1151 single static HTML5 app is served that uses the JSON logic
1152 to enable the user to do some basic browsing of the
1153 journal. This will be extended later on. Here's an example
1154 screenshot of this app in its current state:
1156 http://0pointer.de/public/journal-gatewayd
1158 Contributions from: Kay Sievers, Lennart Poettering, Robert
1159 Milasan, Tom Gundersen
1163 * The bash completion logic is now available for journalctl
1166 * We don't mount the "cpuset" controller anymore together with
1167 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
1168 started if no parameters are assigned to it. "cpuset" hence
1169 broke code that assumed it it could create "cpu" groups and
1172 * journalctl -f will now subscribe to terminal size changes,
1173 and line break accordingly.
1175 Contributions from: Dave Reisner, Kay Sievers, Lennart
1176 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
1180 * nspawn will now create a symlink /etc/localtime in the
1181 container environment, copying the host's timezone
1182 setting. Previously this has been done via a bind mount, but
1183 since symlinks cannot be bind mounted this has now been
1184 changed to create/update the appropriate symlink.
1186 * journalctl -n's line number argument is now optional, and
1187 will default to 10 if omitted.
1189 * journald will now log the maximum size the journal files may
1190 take up on disk. This is particularly useful if the default
1191 built-in logic of determining this parameter from the file
1192 system size is used. Use "systemctl status
1193 systemd-journald.service" to see this information.
1195 * The multi-seat X wrapper tool has been stripped down. As X
1196 is now capable of enumerating graphics devices via udev in a
1197 seat-aware way the wrapper is not strictly necessary
1198 anymore. A stripped down temporary stop-gap is still shipped
1199 until the upstream display managers have been updated to
1200 fully support the new X logic. Expect this wrapper to be
1201 removed entirely in one of the next releases.
1203 * HandleSleepKey= in logind.conf has been split up into
1204 HandleSuspendKey= and HandleHibernateKey=. The old setting
1205 is not available anymore. X11 and the kernel are
1206 distuingishing between these keys and we should too. This
1207 also means the inhibition lock for these keys has been split
1210 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
1211 Poettering, Lukas Nykryn, Václav Pavlín
1215 * Whenever a unit changes state we'll now log this to the
1216 journal and show along the unit's own log output in
1219 * ConditionPathIsMountPoint= can now properly detect bind
1220 mount points too. (Previously, a bind mount of one file
1221 system to another place in the same file system could not be
1222 detected as mount, since they shared struct stat's st_dev
1225 * We will now mount the cgroup controllers cpu, cpuacct,
1226 cpuset and the controllers net_cls, net_prio together by
1229 * nspawn containers will now have a virtualized boot
1230 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
1231 over with a randomized ID at container initialization). This
1232 has the effect of making "journalctl -b" do the right thing
1235 * The JSON output journal serialization has been updated not
1236 to generate "endless" list objects anymore, but rather one
1237 JSON object per line. This is more in line how most JSON
1238 parsers expect JSON objects. The new output mode
1239 "json-pretty" has been added to provide similar output, but
1240 neatly aligned for readability by humans.
1242 * We dropped all explicit sync() invocations in the shutdown
1243 code. The kernel does this implicitly anyway in the kernel
1244 reboot() syscall. halt(8)'s -n option is now a compatibility
1247 * We now support virtualized reboot() in containers, as
1248 supported by newer kernels. We will fall back to exit() if
1249 CAP_SYS_REBOOT is not available to the container. Also,
1250 nspawn makes use of this now and will actually reboot the
1251 container if the containerized OS asks for that.
1253 * journalctl will only show local log output by default
1254 now. Use --merge (-m) to show remote log output, too.
1256 * libsystemd-journal gained the new sd_journal_get_usage()
1257 call to determine the current disk usage of all journal
1258 files. This is exposed in the new "journalctl --disk-usage"
1261 * journald gained a new configuration setting SplitMode= in
1262 journald.conf which may be used to control how user journals
1263 are split off. See journald.conf(5) for details.
1265 * A new condition type ConditionFileNotEmpty= has been added.
1267 * tmpfiles' "w" lines now support file globbing, to write
1268 multiple files at once.
1270 * We added Python bindings for the journal submission
1271 APIs. More Python APIs for a number of selected APIs will
1272 likely follow. Note that we intend to add native bindings
1273 only for the Python language, as we consider it common
1274 enough to deserve bindings shipped within systemd. There are
1275 various projects outside of systemd that provide bindings
1276 for languages such as PHP or Lua.
1278 * Many conditions will now resolve specifiers such as %i. In
1279 addition, PathChanged= and related directives of .path units
1280 now support specifiers as well.
1282 * There's now a new RPM macro definition for the system preset
1285 * journald will now warn if it can't forward a message to the
1286 syslog daemon because it's socket is full.
1288 * timedated will no longer write or process /etc/timezone,
1289 except on Debian. As we do not support late mounted /usr
1290 anymore /etc/localtime always being a symlink is now safe,
1291 and hence the information in /etc/timezone is not necessary
1294 * logind will now always reserve one VT for a text getty (VT6
1295 by default). Previously if more than 6 X sessions where
1296 started they took up all the VTs with auto-spawned gettys,
1297 so that no text gettys were available anymore.
1299 * udev will now automatically inform the btrfs kernel logic
1300 about btrfs RAID components showing up. This should make
1301 simple hotplug based btrfs RAID assembly work.
1303 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
1304 (but not for its children which will stay at the kernel
1305 default). This should allow setups with a lot more listening
1308 * systemd will now always pass the configured timezone to the
1309 kernel at boot. timedated will do the same when the timezone
1312 * logind's inhibition logic has been updated. By default,
1313 logind will now handle the lid switch, the power and sleep
1314 keys all the time, even in graphical sessions. If DEs want
1315 to handle these events on their own they should take the new
1316 handle-power-key, handle-sleep-key and handle-lid-switch
1317 inhibitors during their runtime. A simple way to achiveve
1318 that is to invoke the DE wrapped in an invocation of:
1320 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
1322 * Access to unit operations is now checked via SELinux taking
1323 the unit file label and client process label into account.
1325 * systemd will now notify the administrator in the journal
1326 when he over-mounts a non-empty directory.
1328 * There are new specifiers that are resolved in unit files,
1329 for the host name (%H), the machine ID (%m) and the boot ID
1332 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
1333 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
1334 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
1335 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
1336 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
1337 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
1338 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
1342 * Support for reading structured kernel messages from
1343 /dev/kmsg has now been added and is enabled by default.
1345 * Support for reading kernel messages from /proc/kmsg has now
1346 been removed. If you want kernel messages in the journal
1347 make sure to run a recent kernel (>= 3.5) that supports
1348 reading structured messages from /dev/kmsg (see
1349 above). /proc/kmsg is now exclusive property of classic
1350 syslog daemons again.
1352 * The libudev API gained the new
1353 udev_device_new_from_device_id() call.
1355 * The logic for file system namespace (ReadOnlyDirectory=,
1356 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
1357 require pivot_root() anymore. This means fewer temporary
1358 directories are created below /tmp for this feature.
1360 * nspawn containers will now see and receive all submounts
1361 made on the host OS below the root file system of the
1364 * Forward Secure Sealing is now supported for Journal files,
1365 which provide cryptographical sealing of journal files so
1366 that attackers cannot alter log history anymore without this
1367 being detectable. Lennart will soon post a blog story about
1368 this explaining it in more detail.
1370 * There are two new service settings RestartPreventExitStatus=
1371 and SuccessExitStatus= which allow configuration of exit
1372 status (exit code or signal) which will be excepted from the
1373 restart logic, resp. consider successful.
1375 * journalctl gained the new --verify switch that can be used
1376 to check the integrity of the structure of journal files and
1377 (if Forward Secure Sealing is enabled) the contents of
1380 * nspawn containers will now be run with /dev/stdin, /dev/fd/
1381 and similar symlinks pre-created. This makes running shells
1382 as container init process a lot more fun.
1384 * The fstab support can now handle PARTUUID= and PARTLABEL=
1387 * A new ConditionHost= condition has been added to match
1388 against the hostname (with globs) and machine ID. This is
1389 useful for clusters where a single OS image is used to
1390 provision a large number of hosts which shall run slightly
1391 different sets of services.
1393 * Services which hit the restart limit will now be placed in a
1396 Contributions from: Bertram Poettering, Dave Reisner, Huang
1397 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
1398 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
1402 * When running in --user mode systemd will now become a
1403 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
1404 tree a lot more organized.
1406 * A new PartOf= unit dependency type has been introduced that
1407 may be used to group services in a natural way.
1409 * "systemctl enable" may now be used to enable instances of
1412 * journalctl now prints error log levels in red, and
1413 warning/notice log levels in bright white. It also supports
1414 filtering by log level now.
1416 * cgtop gained a new -n switch (similar to top), to configure
1417 the maximum number of iterations to run for. It also gained
1418 -b, to run in batch mode (accepting no input).
1420 * The suffix ".service" may now be omitted on most systemctl
1421 command lines involving service unit names.
1423 * There's a new bus call in logind to lock all sessions, as
1424 well as a loginctl verb for it "lock-sessions".
1426 * libsystemd-logind.so gained a new call sd_journal_perror()
1427 that works similar to libc perror() but logs to the journal
1428 and encodes structured information about the error number.
1430 * /etc/crypttab entries now understand the new keyfile-size=
1433 * shutdown(8) now can send a (configurable) wall message when
1434 a shutdown is cancelled.
1436 * The mount propagation mode for the root file system will now
1437 default to "shared", which is useful to make containers work
1438 nicely out-of-the-box so that they receive new mounts from
1439 the host. This can be undone locally by running "mount
1440 --make-rprivate /" if needed.
1442 * The prefdm.service file has been removed. Distributions
1443 should maintain this unit downstream if they intend to keep
1444 it around. However, we recommend writing normal unit files
1445 for display managers instead.
1447 * Since systemd is a crucial part of the OS we will now
1448 default to a number of compiler switches that improve
1449 security (hardening) such as read-only relocations, stack
1450 protection, and suchlike.
1452 * The TimeoutSec= setting for services is now split into
1453 TimeoutStartSec= and TimeoutStopSec= to allow configuration
1454 of individual time outs for the start and the stop phase of
1457 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
1458 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
1459 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
1460 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
1461 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
1462 Gundersen, Zbigniew Jędrzejewski-Szmek
1466 * The journal and id128 C APIs are now fully documented as man
1469 * Extra safety checks have been added when transitioning from
1470 the initial RAM disk to the main system to avoid accidental
1473 * /etc/crypttab entries now understand the new keyfile-offset=
1476 * systemctl -t can now be used to filter by unit load state.
1478 * The journal C API gained the new sd_journal_wait() call to
1479 make writing synchronous journal clients easier.
1481 * journalctl gained the new -D switch to show journals from a
1484 * journalctl now displays a special marker between log
1485 messages of two different boots.
1487 * The journal is now explicitly flushed to /var via a service
1488 systemd-journal-flush.service, rather than implicitly simply
1489 by seeing /var/log/journal to be writable.
1491 * journalctl (and the journal C APIs) can now match for much
1492 more complex expressions, with alternatives and
1495 * When transitioning from the initial RAM disk to the main
1496 system we will now kill all processes in a killing spree to
1497 ensure no processes stay around by accident.
1499 * Three new specifiers may be used in unit files: %u, %h, %s
1500 resolve to the user name, user home directory resp. user
1501 shell. This is useful for running systemd user instances.
1503 * We now automatically rotate journal files if their data
1504 object hash table gets a fill level > 75%. We also size the
1505 hash table based on the configured maximum file size. This
1506 together should lower hash collisions drastically and thus
1507 speed things up a bit.
1509 * journalctl gained the new "--header" switch to introspect
1510 header data of journal files.
1512 * A new setting SystemCallFilters= has been added to services
1513 which may be used to apply blacklists or whitelists to
1514 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
1516 * nspawn gained a new --link-journal= switch (and quicker: -j)
1517 to link the container journal with the host. This makes it
1518 very easy to centralize log viewing on the host for all
1519 guests while still keeping the journal files separated.
1521 * Many bugfixes and optimizations
1523 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
1524 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
1525 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
1530 * Several tools now understand kernel command line arguments,
1531 which are only read when run in an initial RAM disk. They
1532 usually follow closely their normal counterparts, but are
1535 * There's a new tool to analyze the readahead files that are
1536 automatically generated at boot. Use:
1538 /usr/lib/systemd/systemd-readahead analyze /.readahead
1540 * We now provide an early debug shell on tty9 if this enabled. Use:
1542 systemctl enable debug-shell.service
1544 * All plymouth related units have been moved into the Plymouth
1545 package. Please make sure to upgrade your Plymouth version
1548 * systemd-tmpfiles now supports getting passed the basename of
1549 a configuration file only, in which case it will look for it
1550 in all appropriate directories automatically.
1552 * udevadm info now takes a /dev or /sys path as argument, and
1553 does the right thing. Example:
1555 udevadm info /dev/sda
1556 udevadm info /sys/class/block/sda
1558 * systemctl now prints a warning if a unit is stopped but a
1559 unit that might trigger it continues to run. Example: a
1560 service is stopped but the socket that activates it is left
1563 * "systemctl status" will now mention if the log output was
1564 shortened due to rotation since a service has been started.
1566 * The journal API now exposes functions to determine the
1567 "cutoff" times due to rotation.
1569 * journald now understands SIGUSR1 and SIGUSR2 for triggering
1570 immediately flushing of runtime logs to /var if possible,
1571 resp. for triggering immediate rotation of the journal
1574 * It is now considered an error if a service is attempted to
1575 be stopped that is not loaded.
1577 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
1579 * systemd-analyze now supports Python 3
1581 * tmpfiles now supports cleaning up directories via aging
1582 where the first level dirs are always kept around but
1583 directories beneath it automatically aged. This is enabled
1584 by prefixing the age field with '~'.
1586 * Seat objects now expose CanGraphical, CanTTY properties
1587 which is required to deal with very fast bootups where the
1588 display manager might be running before the graphics drivers
1589 completed initialization.
1591 * Seat objects now expose a State property.
1593 * We now include RPM macros for service enabling/disabling
1594 based on the preset logic. We recommend RPM based
1595 distributions to make use of these macros if possible. This
1596 makes it simpler to reuse RPM spec files across
1599 * We now make sure that the collected systemd unit name is
1600 always valid when services log to the journal via
1603 * There's a new man page kernel-command-line(7) detailing all
1604 command line options we understand.
1606 * The fstab generator may now be disabled at boot by passing
1607 fstab=0 on the kernel command line.
1609 * A new kernel command line option modules-load= is now understood
1610 to load a specific kernel module statically, early at boot.
1612 * Unit names specified on the systemctl command line are now
1613 automatically escaped as needed. Also, if file system or
1614 device paths are specified they are automatically turned
1615 into the appropriate mount or device unit names. Example:
1617 systemctl status /home
1618 systemctl status /dev/sda
1620 * The SysVConsole= configuration option has been removed from
1621 system.conf parsing.
1623 * The SysV search path is no longer exported on the D-Bus
1626 * The Names= option is been removed from unit file parsing.
1628 * There's a new man page bootup(7) detailing the boot process.
1630 * Every unit and every generator we ship with systemd now
1631 comes with full documentation. The self-explanatory boot is
1634 * A couple of services gained "systemd-" prefixes in their
1635 name if they wrap systemd code, rather than only external
1636 code. Among them fsck@.service which is now
1637 systemd-fsck@.service.
1639 * The HaveWatchdog property has been removed from the D-Bus
1642 * systemd.confirm_spawn= on the kernel command line should now
1645 * There's a new man page crypttab(5) which details all options
1646 we actually understand.
1648 * systemd-nspawn gained a new --capability= switch to pass
1649 additional capabilities to the container.
1651 * timedated will now read known NTP implementation unit names
1652 from /usr/lib/systemd/ntp-units.d/*.list,
1653 systemd-timedated-ntp.target has been removed.
1655 * journalctl gained a new switch "-b" that lists log data of
1656 the current boot only.
1658 * The notify socket is in the abstract namespace again, in
1659 order to support daemons which chroot() at start-up.
1661 * There is a new Storage= configuration option for journald
1662 which allows configuration of where log data should go. This
1663 also provides a way to disable journal logging entirely, so
1664 that data collected is only forwarded to the console, the
1665 kernel log buffer or another syslog implementation.
1667 * Many bugfixes and optimizations
1669 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
1670 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
1671 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
1672 Shawn Landden, Tom Gundersen
1676 * "systemctl help <unit>" now shows the man page if one is
1679 * Several new man pages have been added.
1681 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
1682 MaxLevelConsole= can now be specified in
1683 journald.conf. These options allow reducing the amount of
1684 data stored on disk or forwarded by the log level.
1686 * TimerSlackNSec= can now be specified in system.conf for
1687 PID1. This allows system-wide power savings.
1689 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
1690 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
1695 * logind is now capable of (optionally) handling power and
1696 sleep keys as well as the lid switch.
1698 * journalctl now understands the syntax "journalctl
1699 /usr/bin/avahi-daemon" to get all log output of a specific
1702 * CapabilityBoundingSet= in system.conf now also influences
1703 the capability bound set of usermode helpers of the kernel.
1705 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
1706 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
1707 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
1708 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
1712 * Note that we skipped 139 releases here in order to set the
1713 new version to something that is greater than both udev's
1714 and systemd's most recent version number.
1716 * udev: all udev sources are merged into the systemd source tree now.
1717 All future udev development will happen in the systemd tree. It
1718 is still fully supported to use the udev daemon and tools without
1719 systemd running, like in initramfs or other init systems. Building
1720 udev though, will require the *build* of the systemd tree, but
1721 udev can be properly *run* without systemd.
1723 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
1724 should be used to create dead device nodes as workarounds for broken
1727 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
1728 no longer supported. udev_monitor_new_from_netlink() needs to be
1729 used to subscribe to events.
1731 * udev: when udevd is started by systemd, processes which are left
1732 behind by forking them off of udev rules, are unconditionally cleaned
1733 up and killed now after the event handling has finished. Services or
1734 daemons must be started as systemd services. Services can be
1735 pulled-in by udev to get started, but they can no longer be directly
1736 forked by udev rules.
1738 * udev: the daemon binary is called systemd-udevd now and installed
1739 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
1740 to adapt to that, create symlink, or rename the binary after building
1743 * libudev no longer provides these symbols:
1744 udev_monitor_from_socket()
1745 udev_queue_get_failed_list_entry()
1746 udev_get_{dev,sys,run}_path()
1747 The versions number was bumped and symbol versioning introduced.
1749 * systemd-loginctl and systemd-journalctl have been renamed
1750 to loginctl and journalctl to match systemctl.
1752 * The config files: /etc/systemd/systemd-logind.conf and
1753 /etc/systemd/systemd-journald.conf have been renamed to
1754 logind.conf and journald.conf. Package updates should rename
1755 the files to the new names on upgrade.
1757 * For almost all files the license is now LGPL2.1+, changed
1758 from the previous GPL2.0+. Exceptions are some minor stuff
1759 of udev (which will be changed to LGPL2.1 eventually, too),
1760 and the MIT licensed sd-daemon.[ch] library that is suitable
1761 to be used as drop-in files.
1763 * systemd and logind now handle system sleep states, in
1764 particular suspending and hibernating.
1766 * logind now implements a sleep/shutdown/idle inhibiting logic
1767 suitable for a variety of uses. Soonishly Lennart will blog
1768 about this in more detail.
1770 * var-run.mount and var-lock.mount are no longer provided
1771 (which prevously bind mounted these directories to their new
1772 places). Distributions which have not converted these
1773 directories to symlinks should consider stealing these files
1774 from git history and add them downstream.
1776 * We introduced the Documentation= field for units and added
1777 this to all our shipped units. This is useful to make it
1778 easier to explore the boot and the purpose of the various
1781 * All smaller setup units (such as
1782 systemd-vconsole-setup.service) now detect properly if they
1783 are run in a container and are skipped when
1784 appropriate. This guarantees an entirely noise-free boot in
1785 Linux container environments such as systemd-nspawn.
1787 * A framework for implementing offline system updates is now
1788 integrated, for details see:
1789 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
1791 * A new service type Type=idle is available now which helps us
1792 avoiding ugly interleaving of getty output and boot status
1795 * There's now a system-wide CapabilityBoundingSet= option to
1796 globally reduce the set of capabilities for the
1797 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
1798 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
1799 even CAP_NET_ADMIN system-wide for secure systems.
1801 * There are now system-wide DefaultLimitXXX= options to
1802 globally change the defaults of the various resource limits
1803 for all units started by PID 1.
1805 * Harald Hoyer's systemd test suite has been integrated into
1806 systemd which allows easy testing of systemd builds in qemu
1807 and nspawn. (This is really awesome! Ask us for details!)
1809 * The fstab parser is now implemented as generator, not inside
1812 * systemctl will now warn you if .mount units generated from
1813 /etc/fstab are out of date due to changes in fstab that
1814 haven't been read by systemd yet.
1816 * systemd is now suitable for usage in initrds. Dracut has
1817 already been updated to make use of this. With this in place
1818 initrds get a slight bit faster but primarily are much
1819 easier to introspect and debug since "systemctl status" in
1820 the host system can be used to introspect initrd services,
1821 and the journal from the initrd is kept around too.
1823 * systemd-delta has been added, a tool to explore differences
1824 between user/admin configuration and vendor defaults.
1826 * PrivateTmp= now affects both /tmp and /var/tmp.
1828 * Boot time status messages are now much prettier and feature
1829 proper english language. Booting up systemd has never been
1832 * Read-ahead pack files now include the inode number of all
1833 files to pre-cache. When the inode changes the pre-caching
1834 is not attempted. This should be nicer to deal with updated
1835 packages which might result in changes of read-ahead
1838 * We now temporaritly lower the kernel's read_ahead_kb variable
1839 when collecting read-ahead data to ensure the kernel's
1840 built-in read-ahead does not add noise to our measurements
1841 of necessary blocks to pre-cache.
1843 * There's now RequiresMountsFor= to add automatic dependencies
1844 for all mounts necessary for a specific file system path.
1846 * MountAuto= and SwapAuto= have been removed from
1847 system.conf. Mounting file systems at boot has to take place
1850 * nspawn now learned a new switch --uuid= to set the machine
1851 ID on the command line.
1853 * nspawn now learned the -b switch to automatically search
1856 * vt102 is now the default TERM for serial TTYs, upgraded from
1859 * systemd-logind now works on VT-less systems.
1861 * The build tree has been reorganized. The individual
1862 components now have directories of their own.
1864 * A new condition type ConditionPathIsReadWrite= is now available.
1866 * nspawn learned the new -C switch to create cgroups for the
1867 container in other hierarchies.
1869 * We now have support for hardware watchdogs, configurable in
1872 * The scheduled shutdown logic now has a public API.
1874 * We now mount /tmp as tmpfs by default, but this can be
1875 masked and /etc/fstab can override it.
1877 * Since udisks doesn't make use of /media anymore we are not
1878 mounting a tmpfs on it anymore.
1880 * journalctl gained a new --local switch to only interleave
1881 locally generated journal files.
1883 * We can now load the IMA policy at boot automatically.
1885 * The GTK tools have been split off into a systemd-ui.
1887 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
1888 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
1889 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
1890 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
1891 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
1892 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
1893 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
1894 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
1895 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
1900 * This is mostly a bugfix release
1902 * Support optional initialization of the machine ID from the
1903 KVM or container configured UUID.
1905 * Support immediate reboots with "systemctl reboot -ff"
1907 * Show /etc/os-release data in systemd-analyze output
1909 * Many bugfixes for the journal, including endianness fixes and
1910 ensuring that disk space enforcement works
1912 * sd-login.h is C++ comptaible again
1914 * Extend the /etc/os-release format on request of the Debian
1917 * We now refuse non-UTF8 strings used in various configuration
1918 and unit files. This is done to ensure we don't pass invalid
1919 data over D-Bus or expose it elsewhere.
1921 * Register Mimo USB Screens as suitable for automatic seat
1924 * Read SELinux client context from journal clients in a race
1927 * Reorder configuration file lookup order. /etc now always
1928 overrides /run in order to allow the administrator to always
1929 and unconditionally override vendor supplied or
1930 automatically generated data.
1932 * The various user visible bits of the journal now have man
1933 pages. We still lack man pages for the journal API calls
1936 * We now ship all man pages in HTML format again in the
1939 Contributions from: Dave Reisner, Dirk Eibach, Frederic
1940 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
1941 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
1946 * This is mostly a bugfix release
1948 * systems lacking /etc/os-release are no longer supported.
1950 * Various functionality updates to libsystemd-login.so
1952 * Track class of PAM logins to distuingish greeters from
1955 Contributions from: Kay Sievers, Lennart Poettering, Michael
1960 * This is an important bugfix release for v41.
1962 * Building man pages is now optional which should be useful
1963 for those building systemd from git but unwilling to install
1966 * Watchdog support for supervising services is now usable. In
1967 a future release support for hardware watchdogs
1968 (i.e. /dev/watchdog) will be added building on this.
1970 * Service start rate limiting is now configurable and can be
1971 turned off per service. When a start rate limit is hit a
1972 reboot can automatically be triggered.
1974 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
1976 Contributions from: Benjamin Franzke, Bill Nottingham,
1977 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
1978 Schmidt, Michał Górny, Piotr Drąg
1982 * The systemd binary is installed /usr/lib/systemd/systemd now;
1983 An existing /sbin/init symlink needs to be adapted with the
1986 * The code that loads kernel modules has been ported to invoke
1987 libkmod directly, instead of modprobe. This means we do not
1988 support systems with module-init-tools anymore.
1990 * Watchdog support is now already useful, but still not
1993 * A new kernel command line option systemd.setenv= is
1994 understood to set system wide environment variables
1995 dynamically at boot.
1997 * We now limit the set of capabilities of systemd-journald.
1999 * We now set SIGPIPE to ignore by default, since it only is
2000 useful in shell pipelines, and has little use in general
2001 code. This can be disabled with IgnoreSIPIPE=no in unit
2004 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
2005 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
2010 * This is mostly a bugfix release
2012 * We now expose the reason why a service failed in the
2013 "Result" D-Bus property.
2015 * Rudimentary service watchdog support (will be completed over
2016 the next few releases.)
2018 * When systemd forks off in order execute some service we will
2019 now immediately changes its argv[0] to reflect which process
2020 it will execute. This is useful to minimize the time window
2021 with a generic argv[0], which makes bootcharts more useful
2023 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
2024 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
2025 Mike Kazantsev, Ray Strode
2029 * This is mostly a test release, but incorporates many
2032 * New systemd-cgtop tool to show control groups by their
2035 * Linking against libacl for ACLs is optional again. If
2036 disabled, support tracking device access for active logins
2037 goes becomes unavailable, and so does access to the user
2038 journals by the respective users.
2040 * If a group "adm" exists, journal files are automatically
2041 owned by them, thus allow members of this group full access
2042 to the system journal as well as all user journals.
2044 * The journal now stores the SELinux context of the logging
2045 client for all entries.
2047 * Add C++ inclusion guards to all public headers
2049 * New output mode "cat" in the journal to print only text
2050 messages, without any meta data like date or time.
2052 * Include tiny X server wrapper as a temporary stop-gap to
2053 teach XOrg udev display enumeration. This is used by display
2054 managers such as gdm, and will go away as soon as XOrg
2055 learned native udev hotplugging for display devices.
2057 * Add new systemd-cat tool for executing arbitrary programs
2058 with STDERR/STDOUT connected to the journal. Can also act as
2059 BSD logger replacement, and does so by default.
2061 * Optionally store all locally generated coredumps in the
2062 journal along with meta data.
2064 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
2065 writing short strings to files (for usage for /sys), and for
2066 creating symlinks, character and block device nodes.
2068 * New unit file option ControlGroupPersistent= to make cgroups
2069 persistent, following the mechanisms outlined in
2070 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
2072 * Support multiple local RTCs in a sane way
2074 * No longer monopolize IO when replaying readahead data on
2075 rotating disks, since we might starve non-file-system IO to
2076 death, since fanotify() will not see accesses done by blkid,
2079 * Don't show kernel threads in systemd-cgls anymore, unless
2080 requested with new -k switch.
2082 Contributions from: Dan Horák, Kay Sievers, Lennart
2083 Poettering, Michal Schmidt
2087 * This is mostly a test release, but incorporates many
2090 * The git repository moved to:
2091 git://anongit.freedesktop.org/systemd/systemd
2092 ssh://git.freedesktop.org/git/systemd/systemd
2094 * First release with the journal
2095 http://0pointer.de/blog/projects/the-journal.html
2097 * The journal replaces both systemd-kmsg-syslogd and
2098 systemd-stdout-bridge.
2100 * New sd_pid_get_unit() API call in libsystemd-logind
2102 * Many systemadm clean-ups
2104 * Introduce remote-fs-pre.target which is ordered before all
2105 remote mounts and may be used to start services before all
2108 * Added Mageia support
2110 * Add bash completion for systemd-loginctl
2112 * Actively monitor PID file creation for daemons which exit in
2113 the parent process before having finished writing the PID
2114 file in the daemon process. Daemons which do this need to be
2115 fixed (i.e. PID file creation must have finished before the
2116 parent exits), but we now react a bit more gracefully to them.
2118 * Add colourful boot output, mimicking the well-known output
2119 of existing distributions.
2121 * New option PassCredentials= for socket units, for
2122 compatibility with a recent kernel ABI breakage.
2124 * /etc/rc.local is now hooked in via a generator binary, and
2125 thus will no longer act as synchronization point during
2128 * systemctl list-unit-files now supports --root=.
2130 * systemd-tmpfiles now understands two new commands: z, Z for
2131 relabelling files according to the SELinux database. This is
2132 useful to apply SELinux labels to specific files in /sys,
2135 * Output of SysV services is now forwarded to both the console
2136 and the journal by default, not only just the console.
2138 * New man pages for all APIs from libsystemd-login.
2140 * The build tree got reorganized and a the build system is a
2141 lot more modular allowing embedded setups to specifically
2142 select the components of systemd they are interested in.
2144 * Support for Linux systems lacking the kernel VT subsystem is
2147 * configure's --with-rootdir= got renamed to
2148 --with-rootprefix= to follow the naming used by udev and
2151 * Unless specified otherwise we'll now install to /usr instead
2152 of /usr/local by default.
2154 * Processes with '@' in argv[0][0] are now excluded from the
2155 final shut-down killing spree, following the logic explained
2157 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
2159 * All processes remaining in a service cgroup when we enter
2160 the START or START_PRE states are now killed with
2161 SIGKILL. That means it is no longer possible to spawn
2162 background processes from ExecStart= lines (which was never
2163 supported anyway, and bad style).
2165 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
2166 reloading of units together.
2168 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
2169 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
2170 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
2171 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
2172 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek