1 systemd System and Service Manager
5 * Two new unit types have been introduced:
7 Scope units are very similar to service units, however, are
8 created out of pre-existing processes -- instead of PID 1
9 forking off the processes. By using scope units it is
10 possible for system services and applications to group their
11 own child processes (worker processes) in a powerful way
12 which then maybe used to organize them, or kill them
13 together, or apply resource limits on them.
15 Slice units may be used to partition system resources in an
16 hierarchial fashion and then assign other units to them. By
17 default there are now three slices: system.slice (for all
18 system services), user.slice (for all user sessions),
19 machine.slice (for VMs and containers).
21 Slices and scopes have been introduced primarily in
22 context of the work to move cgroup handling to a
23 single-writer scheme, where only PID 1
24 creates/removes/manages cgroups.
26 * There's a new concept of "transient" units. In contrast to
27 normal units these units are created via an API at runtime,
28 not from configuration from disk. More specifically this
29 means it is now possible to run arbitrary programs as
30 independent services, with all execution parameters passed
31 in via bus APIs rather than read from disk. Transient units
32 make systemd substantially more dynamic then it ever was,
33 and useful as a general batch manager.
35 * logind has been updated to make use of scope and slice units
36 for managing user sessions. As a user logs in he will get
37 his own private slice unit, to which all sessions are added
38 as scope units. We also added support for automatically
39 adding an instance of user@.service for the user into the
40 slice. Effectively logind will no longer create cgroup
41 hierarchies on its own now, it will defer entirely to PID 1
42 for this by means of scope, service and slice units. Since
43 user sessions this way become entities managed by PID 1
44 the output of "systemctl" is now a lot more comprehensive.
46 * A new mini-daemon "systemd-machined" has been added which
47 may be used by virtualization managers to register local
48 VMs/containers. nspawn has been updated accordingly, and
49 libvirt will be updated shortly. machined will collect a bit
50 of meta information about the VMs/containers, and assign
51 them their own scope unit (see above). The collected
52 meta-data is then made available via the "machinectl" tool,
53 and exposed in "ps" and similar tools. machined/machinectl
54 is compile-time optional.
56 * As discussed earlier, the low-level cgroup configuration
57 options ControlGroup=, ControlGroupModify=,
58 ControlGroupPersistent=, ControlGroupAttribute= have been
59 removed. Please use high-level attribute settings instead as
62 * A new bus call SetUnitProperties() has been added to alter
63 various runtime parameters of a unit. This is primarily
64 useful to alter cgroup parameters dynamically in a nice way,
65 but will be extended later on to make more properties
66 modifiable at runtime. systemctl gained a new set-properties
67 command that wraps this call.
69 * A new tool "systemd-run" has been added which can be used to
70 run arbitrary command lines as transient services or scopes,
71 while configuring a number of settings via the command
72 line. This tool is currently very basic, however already
73 very useful. We plan to extend this tool to even allow
74 queuing of execution jobs with time triggers from the
75 command line, similar in fashion to "at".
77 * nspawn will now inform the user explicitly that kernels with
78 audit enabled break containers, and suggest the user to turn
81 * Support for detecting the IMA and AppArmor security
82 frameworks with ConditionSecurity= has been added.
84 * journalctl gained a new "-k" switch for showing only kernel
85 messages, mimicking dmesg output; in addition to "--user"
86 and "--system" switches for showing only user's own logs
89 * systemd-delta can now show information about drop-in
90 snippets extending unit files.
92 * libsystemd-bus has been substantially updated but is still
93 not available as public API.
95 * systemd will now look for the "debug" argument on the kernel
96 command line and enable debug logging, similar to
97 "systemd.log_level=debug" already did before.
99 * "systemctl set-default", "systemctl get-default" has been
100 added to configure the default.target symlink, which
101 controls what to boot into by default.
103 * "systemctl set-log-level" has been added as a convenient
104 way to raise and lower systemd logging threshold.
106 * "systemd-analyze plot" will now show the time the various
107 generators needed for execution, as well as information
108 about the unit file loading.
110 * libsystemd-journal gained a new sd_journal_open_files() call
111 for opening specific journal files. journactl also gained a
112 new switch to expose this new functionality. Previously we
113 only supported opening all files from a directory, or all
114 files from the system, as opening individual files only is
115 racy due to journal file rotation.
117 * systemd gained the new DefaultEnvironment= setting in
118 /etc/systemd/system.conf to set environment variables for
121 * If a privileged process logs a journal message with the
122 OBJECT_PID= field set, then journald will automatically
123 augment this with additional OBJECT_UID=, OBJECT_GID=,
124 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
125 system services want to log events about specific client
126 processes. journactl/systemctl has been updated to make use
127 of this information if all log messages regarding a specific
130 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
131 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
132 Reisner, David Coppa, David King, David Strauss, Eelco
133 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
134 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
135 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
136 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
137 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
138 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
139 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
140 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
141 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
146 * The Python bindings gained some minimal support for the APIs
147 exposed by libsystemd-logind.
149 * ConditionSecurity= gained support for detecting SMACK. Since
150 this condition already supports SELinux and AppArmor we only
151 miss IMA for this. Patches welcome!
153 Contributions from: Karol Lewandowski, Lennart Poettering,
154 Zbigniew Jędrzejewski-Szmek
158 * systemd-nspawn will now create /etc/resolv.conf if
159 necessary, before bind-mounting the host's file onto it.
161 * systemd-nspawn will now store meta information about a
162 container on the container's cgroup as extended attribute
163 fields, including the root directory.
165 * The cgroup hierarchy has been reworked in many ways. All
166 objects any of the components systemd creates in the cgroup
167 tree are now suffixed. More specifically, user sessions are
168 now placed in cgroups suffixed with ".session", users in
169 cgroups suffixed with ".user", and nspawn containers in
170 cgroups suffixed with ".nspawn". Furthermore, all cgroup
171 names are now escaped in a simple scheme to avoid collision
172 of userspace object names with kernel filenames. This work
173 is preparation for making these objects relocatable in the
174 cgroup tree, in order to allow easy resource partitioning of
175 these objects without causing naming conflicts.
177 * systemctl list-dependencies gained the new switches
178 --plain, --reverse, --after and --before.
180 * systemd-inhibit now shows the process name of processes that
181 have taken an inhibitor lock.
183 * nss-myhostname will now also resolve "localhost"
184 implicitly. This makes /etc/hosts an optional file and
185 nicely handles that on IPv6 ::1 maps to both "localhost" and
188 * libsystemd-logind.so gained a new call
189 sd_get_machine_names() to enumerate running containers and
190 VMs (currently only supported by very new libvirt and
191 nspawn). sd_login_monitor can now be used to watch
192 VMs/containers coming and going.
194 * .include is not allowed recursively anymore, and only in
195 unit files. Usually it is better to use drop-in snippets in
196 .d/*.conf anyway, as introduced with systemd 198.
198 * systemd-analyze gained a new "critical-chain" command that
199 determines the slowest chain of units run during system
200 boot-up. It is very useful for tracking down where
201 optimizing boot time is the most beneficial.
203 * systemd will no longer allow manipulating service paths in
204 the name=systemd:/system cgroup tree using ControlGroup= in
205 units. (But is still fine with it in all other dirs.)
207 * There's a new systemd-nspawn@.service service file that may
208 be used to easily run nspawn containers as system
209 services. With the container's root directory in
210 /var/lib/container/foobar it is now sufficient to run
211 "systemctl start systemd-nspawn@foobar.service" to boot it.
213 * systemd-cgls gained a new parameter "--machine" to list only
214 the processes within a certain container.
216 * ConditionSecurity= now can check for "apparmor". We still
217 are lacking checks for SMACK and IMA for this condition
218 check though. Patches welcome!
220 * A new configuration file /etc/systemd/sleep.conf has been
221 added that may be used to configure which kernel operation
222 systemd is supposed to execute when "suspend", "hibernate"
223 or "hybrid-sleep" is requested. This makes the new kernel
224 "freeze" state accessible to the user.
226 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
227 the passed argument if applicable.
229 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
230 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
231 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
232 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
233 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
234 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
235 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
240 * The output of 'systemctl list-jobs' got some polishing. The
241 '--type=' argument may now be passed more than once. A new
242 command 'systemctl list-sockets' has been added which shows
243 a list of kernel sockets systemd is listening on with the
244 socket units they belong to, plus the units these socket
247 * The experimental libsystemd-bus library got substantial
248 updates to work in conjunction with the (also experimental)
249 kdbus kernel project. It works well enough to exchange
250 messages with some sophistication. Note that kdbus is not
251 ready yet, and the library is mostly an elaborate test case
252 for now, and not installable.
254 * systemd gained a new unit 'systemd-static-nodes.service'
255 that generates static device nodes earlier during boot, and
256 can run in conjunction with udev.
258 * libsystemd-login gained a new call sd_pid_get_user_unit()
259 to retrieve the user systemd unit a process is running
260 in. This is useful for systems where systemd is used as
263 * systemd-nspawn now places all containers in the new /machine
264 top-level cgroup directory in the name=systemd
265 hierarchy. libvirt will soon do the same, so that we get a
266 uniform separation of /system, /user and /machine for system
267 services, user processes and containers/virtual
268 machines. This new cgroup hierarchy is also useful to stick
269 stable names to specific container instances, which can be
270 recognized later this way (this name may be controlled
271 via systemd-nspawn's new -M switch). libsystemd-login also
272 gained a new call sd_pid_get_machine_name() to retrieve the
273 name of the container/VM a specific process belongs to.
275 * bootchart can now store its data in the journal.
277 * libsystemd-journal gained a new call
278 sd_journal_add_conjunction() for AND expressions to the
279 matching logic. This can be used to express more complex
282 * journactl can now take multiple --unit= and --user-unit=
285 * The cryptsetup logic now understands the "luks.key=" kernel
286 command line switch for specifying a file to read the
287 decryption key from. Also, if a configured key file is not
288 found the tool will now automatically fall back to prompting
291 * Python systemd.journal module was updated to wrap recently
292 added functions from libsystemd-journal. The interface was
293 changed to bring the low level interface in s.j._Reader
294 closer to the C API, and the high level interface in
295 s.j.Reader was updated to wrap and convert all data about
298 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
299 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
300 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
301 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
302 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
303 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
307 * journalctl --update-catalog now understands a new --root=
308 option to operate on catalogs found in a different root
311 * During shutdown after systemd has terminated all running
312 services a final killing loop kills all remaining left-over
313 processes. We will now print the name of these processes
314 when we send SIGKILL to them, since this usually indicates a
317 * If /etc/crypttab refers to password files stored on
318 configured mount points automatic dependencies will now be
319 generated to ensure the specific mount is established first
320 before the key file is attempted to be read.
322 * 'systemctl status' will now show information about the
323 network sockets a socket unit is listening on.
325 * 'systemctl status' will also shown information about any
326 drop-in configuration file for units. (Drop-In configuration
327 files in this context are files such as
328 /etc/systemd/systemd/foobar.service.d/*.conf)
330 * systemd-cgtop now optionally shows summed up CPU times of
331 cgroups. Press '%' while running cgtop to switch between
332 percentage and absolute mode. This is useful to determine
333 which cgroups use up the most CPU time over the entire
334 runtime of the system. systemd-cgtop has also been updated
335 to be 'pipeable' for processing with further shell tools.
337 * 'hostnamectl set-hostname' will now allow setting of FQDN
340 * The formatting and parsing of time span values has been
341 changed. The parser now understands fractional expressions
342 such as "5.5h". The formatter will now output fractional
343 expressions for all time spans under 1min, i.e. "5.123456s"
344 rather than "5s 123ms 456us". For time spans under 1s
345 millisecond values are shown, for those under 1ms
346 microsecond values are shown. This should greatly improve
347 all time-related output of systemd.
349 * libsystemd-login and libsystemd-journal gained new
350 functions for querying the poll() events mask and poll()
351 timeout value for integration into arbitrary event
354 * localectl gained the ability to list available X11 keymaps
355 (models, layouts, variants, options).
357 * 'systemd-analyze dot' gained the ability to filter for
358 specific units via shell-style globs, to create smaller,
359 more useful graphs. I.e. it's now possible to create simple
360 graphs of all the dependencies between only target units, or
361 of all units that Avahi has dependencies with.
363 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
364 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
365 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
366 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
367 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
368 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
369 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
373 * The boot-time readahead implementation for rotating media
374 will now read the read-ahead data in multiple passes which
375 consist of all read requests made in equidistant time
376 intervals. This means instead of strictly reading read-ahead
377 data in its physical order on disk we now try to find a
378 middle ground between physical and access time order.
380 * /etc/os-release files gained a new BUILD_ID= field for usage
381 on operating systems that provide continuous builds of OS
384 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
385 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
386 William Douglas, Zbigniew Jędrzejewski-Szmek
390 * systemd-python gained an API exposing libsystemd-daemon.
392 * The SMACK setup logic gained support for uploading CIPSO
395 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
396 ReadOnlyDirectories= and InaccessibleDirectories= has
397 changed. The private /tmp and /var/tmp directories are now
398 shared by all processes of a service (which means
399 ExecStartPre= may now leave data in /tmp that ExecStart= of
400 the same service can still access). When a service is
401 stopped its temporary directories are immediately deleted
402 (normal clean-up with tmpfiles is still done in addition to
405 * By default, systemd will now set a couple of sysctl
406 variables in the kernel: the safe sysrq options are turned
407 on, IP route verification is turned on, and source routing
408 disabled. The recently added hardlink and softlink
409 protection of the kernel is turned on. These settings should
410 be reasonably safe, and good defaults for all new systems.
412 * The predictable network naming logic may now be turned off
413 with a new kernel command line switch: net.ifnames=0.
415 * A new libsystemd-bus module has been added that implements a
416 pretty complete D-Bus client library. For details see:
418 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
420 * journald will now explicitly flush the journal files to disk
421 at the latest 5min after each write. The file will then also
422 be marked offline until the next write. This should increase
423 reliability in case of a crash. The synchronization delay
424 can be configured via SyncIntervalSec= in journald.conf.
426 * There's a new remote-fs-setup.target unit that can be used
427 to pull in specific services when at least one remote file
428 system is to be mounted.
430 * There are new targets timers.target and paths.target as
431 canonical targets to pull user timer and path units in
432 from. This complements sockets.target with a similar
433 purpose for socket units.
435 * libudev gained a new call udev_device_set_attribute_value()
436 to set sysfs attributes of a device.
438 * The udev daemon now sets the default number of worker
439 processes executed in parallel based on the number of available
440 CPUs instead of the amount of available RAM. This is supposed
441 to provide a more reliable default and limit a too aggressive
442 paralellism for setups with 1000s of devices connected.
444 Contributions from: Auke Kok, Colin Walters, Cristian
445 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
446 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
447 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
448 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
449 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
450 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
451 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
452 Zbigniew Jędrzejewski-Szmek
456 * Configuration of unit files may now be extended via drop-in
457 files without having to edit/override the unit files
458 themselves. More specifically, if the administrator wants to
459 change one value for a service file foobar.service he can
460 now do so by dropping in a configuration snippet into
461 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
462 will load all these snippets and apply them on top of the
463 main unit configuration file, possibly extending or
464 overriding its settings. Using these drop-in snippets is
465 generally nicer than the two earlier options for changing
466 unit files locally: copying the files from
467 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
468 them there; or creating a new file in /etc/systemd/system/
469 that incorporates the original one via ".include". Drop-in
470 snippets into these .d/ directories can be placed in any
471 directory systemd looks for units in, and the usual
472 overriding semantics between /usr/lib, /etc and /run apply
475 * Most unit file settings which take lists of items can now be
476 reset by assigning the empty string to them. For example,
477 normally, settings such as Environment=FOO=BAR append a new
478 environment variable assignment to the environment block,
479 each time they are used. By assigning Environment= the empty
480 string the environment block can be reset to empty. This is
481 particularly useful with the .d/*.conf drop-in snippets
482 mentioned above, since this adds the ability to reset list
483 settings from vendor unit files via these drop-ins.
485 * systemctl gained a new "list-dependencies" command for
486 listing the dependencies of a unit recursively.
488 * Inhibitors are now honored and listed by "systemctl
489 suspend", "systemctl poweroff" (and similar) too, not only
490 GNOME. These commands will also list active sessions by
493 * Resource limits (as exposed by the various control group
494 controllers) can now be controlled dynamically at runtime
495 for all units. More specifically, you can now use a command
496 like "systemctl set-cgroup-attr foobar.service cpu.shares
497 2000" to alter the CPU shares a specific service gets. These
498 settings are stored persistently on disk, and thus allow the
499 administrator to easily adjust the resource usage of
500 services with a few simple commands. This dynamic resource
501 management logic is also available to other programs via the
502 bus. Almost any kernel cgroup attribute and controller is
505 * systemd-vconsole-setup will now copy all font settings to
506 all allocated VTs, where it previously applied them only to
509 * libsystemd-login gained the new sd_session_get_tty() API
512 * This release drops support for a few legacy or
513 distribution-specific LSB facility names when parsing init
514 scripts: $x-display-manager, $mail-transfer-agent,
515 $mail-transport-agent, $mail-transfer-agent, $smtp,
516 $null. Also, the mail-transfer-agent.target unit backing
517 this has been removed. Distributions which want to retain
518 compatibility with this should carry the burden for
519 supporting this themselves and patch support for these back
520 in, if they really need to. Also, the facilities $syslog and
521 $local_fs are now ignored, since systemd does not support
522 early-boot LSB init scripts anymore, and these facilities
523 are implied anyway for normal services. syslog.target has
526 * There are new bus calls on PID1's Manager object for
527 cancelling jobs, and removing snapshot units. Previously,
528 both calls were only available on the Job and Snapshot
531 * systemd-journal-gatewayd gained SSL support.
533 * The various "environment" files, such as /etc/locale.conf
534 now support continuation lines with a backslash ("\") as
535 last character in the line, similar in style (but different)
536 to how this is supported in shells.
538 * For normal user processes the _SYSTEMD_USER_UNIT= field is
539 now implicitly appended to every log entry logged. systemctl
540 has been updated to filter by this field when operating on a
541 user systemd instance.
543 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
544 CAP_AUDIT_CONTROL capabilities to the capabilities set for
545 the container. This makes it easier to boot unmodified
546 Fedora systems in a container, which however still requires
547 audit=0 to be passed on the kernel command line. Auditing in
548 kernel and userspace is unfortunately still too broken in
549 context of containers, hence we recommend compiling it out
550 of the kernel or using audit=0. Hopefully this will be fixed
551 one day for good in the kernel.
553 * nspawn gained the new --bind= and --bind-ro= parameters to
554 bind mount specific directories from the host into the
557 * nspawn will now mount its own devpts file system instance
558 into the container, in order not to leak pty devices from
559 the host into the container.
561 * systemd will now read the firmware boot time performance
562 information from the EFI variables, if the used boot loader
563 supports this, and takes it into account for boot performance
564 analysis via "systemd-analyze". This is currently supported
565 only in conjunction with Gummiboot, but could be supported
566 by other boot loaders too. For details see:
568 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
570 * A new generator has been added that automatically mounts the
571 EFI System Partition (ESP) to /boot, if that directory
572 exists, is empty, and no other file system has been
573 configured to be mounted there.
575 * logind will now send out PrepareForSleep(false) out
576 unconditionally, after coming back from suspend. This may be
577 used by applications as asynchronous notification for
578 system resume events.
580 * "systemctl unlock-sessions" has been added, that allows
581 unlocking the screens of all user sessions at once, similar
582 how "systemctl lock-sessions" already locked all users
583 sessions. This is backed by a new D-Bus call UnlockSessions().
585 * "loginctl seat-status" will now show the master device of a
586 seat. (i.e. the device of a seat that needs to be around for
587 the seat to be considered available, usually the graphics
590 * tmpfiles gained a new "X" line type, that allows
591 configuration of files and directories (with wildcards) that
592 shall be excluded from automatic cleanup ("aging").
594 * udev default rules set the device node permissions now only
595 at "add" events, and do not change them any longer with a
596 later "change" event.
598 * The log messages for lid events and power/sleep keypresses
599 now carry a message ID.
601 * We now have a substantially larger unit test suite, but this
602 continues to be work in progress.
604 * udevadm hwdb gained a new --root= parameter to change the
605 root directory to operate relative to.
607 * logind will now issue a background sync() request to the kernel
608 early at shutdown, so that dirty buffers are flushed to disk early
609 instead of at the last moment, in order to optimize shutdown
612 * A new bootctl tool has been added that is an interface for
613 certain boot loader operations. This is currently a preview
614 and is likely to be extended into a small mechanism daemon
615 like timedated, localed, hostnamed, and can be used by
616 graphical UIs to enumerate available boot options, and
617 request boot into firmware operations.
619 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
620 the rest of the package. It also has been updated to work
621 correctly in initrds.
623 * Policykit previously has been runtime optional, and is now
624 also compile time optional via a configure switch.
626 * systemd-analyze has been reimplemented in C. Also "systemctl
627 dot" has moved into systemd-analyze.
629 * "systemctl status" with no further parameters will now print
630 the status of all active or failed units.
632 * Operations such as "systemctl start" can now be executed
633 with a new mode "--irreversible" which may be used to queue
634 operations that cannot accidentally be reversed by a later
635 job queuing. This is by default used to make shutdown
636 requests more robust.
638 * The Python API of systemd now gained a new module for
639 reading journal files.
641 * A new tool kernel-install has been added that can install
642 kernel images according to the Boot Loader Specification:
644 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
646 * Boot time console output has been improved to provide
647 animated boot time output for hanging jobs.
649 * A new tool systemd-activate has been added which can be used
650 to test socket activation with, directly from the command
651 line. This should make it much easier to test and debug
652 socket activation in daemons.
654 * journalctl gained a new "--reverse" (or -r) option to show
655 journal output in reverse order (i.e. newest line first).
657 * journalctl gained a new "--pager-end" (or -e) option to jump
658 to immediately jump to the end of the journal in the
659 pager. This is only supported in conjunction with "less".
661 * journalctl gained a new "--user-unit=" option, that works
662 similar to "--unit=" but filters for user units rather than
665 * A number of unit files to ease adoption of systemd in
666 initrds has been added. This moves some minimal logic from
667 the various initrd implementations into systemd proper.
669 * The journal files are now owned by a new group
670 "systemd-journal", which exists specifically to allow access
671 to the journal, and nothing else. Previously, we used the
672 "adm" group for that, which however possibly covers more
673 than just journal/log file access. This new group is now
674 already used by systemd-journal-gatewayd to ensure this
675 daemon gets access to the journal files and as little else
676 as possible. Note that "make install" will also set FS ACLs
677 up for /var/log/journal to give "adm" and "wheel" read
678 access to it, in addition to "systemd-journal" which owns
679 the journal files. We recommend that packaging scripts also
680 add read access to "adm" + "wheel" to /var/log/journal, and
681 all existing/future journal files. To normal users and
682 administrators little changes, however packagers need to
683 ensure to create the "systemd-journal" system group at
684 package installation time.
686 * The systemd-journal-gatewayd now runs as unprivileged user
687 systemd-journal-gateway:systemd-journal-gateway. Packaging
688 scripts need to create these system user/group at
691 * timedated now exposes a new boolean property CanNTP that
692 indicates whether a local NTP service is available or not.
694 * systemd-detect-virt will now also detect xen PVs
696 * The pstore file system is now mounted by default, if it is
699 * In addition to the SELinux and IMA policies we will now also
700 load SMACK policies at early boot.
702 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
703 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
704 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
705 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
706 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
707 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
708 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
709 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
710 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
711 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
712 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
713 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
714 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
715 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
719 * Timer units now support calendar time events in addition to
720 monotonic time events. That means you can now trigger a unit
721 based on a calendar time specification such as "Thu,Fri
722 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
723 or fifth day of any month of the year 2013, given that it is
724 a thursday or friday. This brings timer event support
725 considerably closer to cron's capabilities. For details on
726 the supported calendar time specification language see
729 * udev now supports a number of different naming policies for
730 network interfaces for predictable names, and a combination
731 of these policies is now the default. Please see this wiki
732 document for details:
734 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
736 * Auke Kok's bootchart implementation has been added to the
737 systemd tree. It's an optional component that can graph the
738 boot in quite some detail. It's one of the best bootchart
739 implementations around and minimal in its code and
742 * nss-myhostname has been integrated into the systemd source
743 tree. nss-myhostname guarantees that the local hostname
744 always stays resolvable via NSS. It has been a weak
745 requirement of systemd-hostnamed since a long time, and
746 since its code is actually trivial we decided to just
747 include it in systemd's source tree. It can be turned off
748 with a configure switch.
750 * The read-ahead logic is now capable of properly detecting
751 whether a btrfs file system is on SSD or rotating media, in
752 order to optimize the read-ahead scheme. Previously, it was
753 only capable of detecting this on traditional file systems
756 * In udev, additional device properties are now read from the
757 IAB in addition to the OUI database. Also, Bluetooth company
758 identities are attached to the devices as well.
760 * In service files %U may be used as specifier that is
761 replaced by the configured user name of the service.
763 * nspawn may now be invoked without a controlling TTY. This
764 makes it suitable for invocation as its own service. This
765 may be used to set up a simple containerized server system
766 using only core OS tools.
768 * systemd and nspawn can now accept socket file descriptors
769 when they are started for socket activation. This enables
770 implementation of socket activated nspawn
771 containers. i.e. think about autospawning an entire OS image
772 when the first SSH or HTTP connection is received. We expect
773 that similar functionality will also be added to libvirt-lxc
776 * journalctl will now suppress ANSI color codes when
779 * systemctl will no longer show control group information for
780 a unit if a the control group is empty anyway.
782 * logind can now automatically suspend/hibernate/shutdown the
785 * /etc/machine-info and hostnamed now also expose the chassis
786 type of the system. This can be used to determine whether
787 the local system is a laptop, desktop, handset or
788 tablet. This information may either be configured by the
789 user/vendor or is automatically determined from ACPI and DMI
790 information if possible.
792 * A number of PolicyKit actions are now bound together with
793 "imply" rules. This should simplify creating UIs because
794 many actions will now authenticate similar ones as well.
796 * Unit files learnt a new condition ConditionACPower= which
797 may be used to conditionalize a unit depending on whether an
798 AC power source is connected or not, of whether the system
799 is running on battery power.
801 * systemctl gained a new "is-failed" verb that may be used in
802 shell scripts and suchlike to check whether a specific unit
803 is in the "failed" state.
805 * The EnvironmentFile= setting in unit files now supports file
806 globbing, and can hence be used to easily read a number of
807 environment files at once.
809 * systemd will no longer detect and recognize specific
810 distributions. All distribution-specific #ifdeffery has been
811 removed, systemd is now fully generic and
812 distribution-agnostic. Effectively, not too much is lost as
813 a lot of the code is still accessible via explicit configure
814 switches. However, support for some distribution specific
815 legacy configuration file formats has been dropped. We
816 recommend distributions to simply adopt the configuration
817 files everybody else uses now and convert the old
818 configuration from packaging scripts. Most distributions
819 already did that. If that's not possible or desirable,
820 distributions are welcome to forward port the specific
821 pieces of code locally from the git history.
823 * When logging a message about a unit systemd will now always
824 log the unit name in the message meta data.
826 * localectl will now also discover system locale data that is
827 not stored in locale archives, but directly unpacked.
829 * logind will no longer unconditionally use framebuffer
830 devices as seat masters, i.e. as devices that are required
831 to be existing before a seat is considered preset. Instead,
832 it will now look for all devices that are tagged as
833 "seat-master" in udev. By default framebuffer devices will
834 be marked as such, but depending on local systems other
835 devices might be marked as well. This may be used to
836 integrate graphics cards using closed source drivers (such
837 as NVidia ones) more nicely into logind. Note however, that
838 we recommend using the open source NVidia drivers instead,
839 and no udev rules for the closed-source drivers will be
840 shipped from us upstream.
842 Contributions from: Adam Williamson, Alessandro Crismani, Auke
843 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
844 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
845 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
846 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
847 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
848 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
849 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
850 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
851 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
852 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
853 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
858 * udev gained support for loading additional device properties
859 from an indexed database that is keyed by vendor/product IDs
860 and similar device identifiers. For the beginning this
861 "hwdb" is populated with data from the well-known PCI and
862 USB database, but also includes PNP, ACPI and OID data. In
863 the longer run this indexed database shall grow into
864 becoming the one central database for non-essential
865 userspace device metadata. Previously, data from the PCI/USB
866 database was only attached to select devices, since the
867 lookup was a relatively expensive operation due to O(n) time
868 complexity (with n being the number of entries in the
869 database). Since this is now O(1), we decided to add in this
870 data for all devices where this is available, by
871 default. Note that the indexed database needs to be rebuilt
872 when new data files are installed. To achieve this you need
873 to update your packaging scripts to invoke "udevadm hwdb
874 --update" after installation of hwdb data files. For
875 RPM-based distributions we introduced the new
876 %udev_hwdb_update macro for this purpose.
878 * The Journal gained support for the "Message Catalog", an
879 indexed database to link up additional information with
880 journal entries. For further details please check:
882 http://www.freedesktop.org/wiki/Software/systemd/catalog
884 The indexed message catalog database also needs to be
885 rebuilt after installation of message catalog files. Use
886 "journalctl --update-catalog" for this. For RPM-based
887 distributions we introduced the %journal_catalog_update
888 macro for this purpose.
890 * The Python Journal bindings gained support for the standard
891 Python logging framework.
893 * The Journal API gained new functions for checking whether
894 the underlying file system of a journal file is capable of
895 properly reporting file change notifications, or whether
896 applications that want to reflect journal changes "live"
897 need to recheck journal files continuously in appropriate
900 * It is now possible to set the "age" field for tmpfiles
901 entries to 0, indicating that files matching this entry
902 shall always be removed when the directories are cleaned up.
904 * coredumpctl gained a new "gdb" verb which invokes gdb
905 right-away on the selected coredump.
907 * There's now support for "hybrid sleep" on kernels that
908 support this, in addition to "suspend" and "hibernate". Use
909 "systemctl hybrid-sleep" to make use of this.
911 * logind's HandleSuspendKey= setting (and related settings)
912 now gained support for a new "lock" setting to simply
913 request the screen lock on all local sessions, instead of
914 actually executing a suspend or hibernation.
916 * systemd will now mount the EFI variables file system by
919 * Socket units now gained support for configuration of the
920 SMACK security label.
922 * timedatectl will now output the time of the last and next
923 daylight saving change.
925 * We dropped support for various legacy and distro-specific
926 concepts, such as insserv, early-boot SysV services
927 (i.e. those for non-standard runlevels such as 'b' or 'S')
928 or ArchLinux /etc/rc.conf support. We recommend the
929 distributions who still need support this to either continue
930 to maintain the necessary patches downstream, or find a
931 different solution. (Talk to us if you have questions!)
933 * Various systemd components will now bypass PolicyKit checks
934 for root and otherwise handle properly if PolicyKit is not
935 found to be around. This should fix most issues for
936 PolicyKit-less systems. Quite frankly this should have been
937 this way since day one. It is absolutely our intention to
938 make systemd work fine on PolicyKit-less systems, and we
939 consider it a bug if something doesn't work as it should if
940 PolicyKit is not around.
942 * For embedded systems it is now possible to build udev and
943 systemd without blkid and/or kmod support.
945 * "systemctl switch-root" is now capable of switching root
946 more than once. I.e. in addition to transitions from the
947 initrd to the host OS it is now possible to transition to
948 further OS images from the host. This is useful to implement
949 offline updating tools.
951 * Various other additions have been made to the RPM macros
952 shipped with systemd. Use %udev_rules_update() after
953 installing new udev rules files. %_udevhwdbdir,
954 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
955 %_sysctldir are now available which resolve to the right
956 directories for packages to place various data files in.
958 * journalctl gained the new --full switch (in addition to
959 --all, to disable ellipsation for long messages.
961 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
962 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
963 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
964 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
965 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
966 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
967 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
968 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
969 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
973 * journalctl gained new --since= and --until= switches to
974 filter by time. It also now supports nice filtering for
975 units via --unit=/-u.
977 * Type=oneshot services may use ExecReload= and do the
980 * The journal daemon now supports time-based rotation and
981 vacuuming, in addition to the usual disk-space based
984 * The journal will now index the available field values for
985 each field name. This enables clients to show pretty drop
986 downs of available match values when filtering. The bash
987 completion of journalctl has been updated
988 accordingly. journalctl gained a new switch -F to list all
989 values a certain field takes in the journal database.
991 * More service events are now written as structured messages
992 to the journal, and made recognizable via message IDs.
994 * The timedated, localed and hostnamed mini-services which
995 previously only provided support for changing time, locale
996 and hostname settings from graphical DEs such as GNOME now
997 also have a minimal (but very useful) text-based client
998 utility each. This is probably the nicest way to changing
999 these settings from the command line now, especially since
1000 it lists available options and is fully integrated with bash
1003 * There's now a new tool "systemd-coredumpctl" to list and
1004 extract coredumps from the journal.
1006 * We now install a README each in /var/log/ and
1007 /etc/rc.d/init.d explaining where the system logs and init
1008 scripts went. This hopefully should help folks who go to
1009 that dirs and look into the otherwise now empty void and
1010 scratch their heads.
1012 * When user-services are invoked (by systemd --user) the
1013 $MANAGERPID env var is set to the PID of systemd.
1015 * SIGRTMIN+24 when sent to a --user instance will now result
1016 in immediate termination of systemd.
1018 * gatewayd received numerous feature additions such as a
1019 "follow" mode, for live syncing and filtering.
1021 * browse.html now allows filtering and showing detailed
1022 information on specific entries. Keyboard navigation and
1023 mouse screen support has been added.
1025 * gatewayd/journalctl now supports HTML5/JSON
1026 Server-Sent-Events as output.
1028 * The SysV init script compatibility logic will now
1029 heuristically determine whether a script supports the
1030 "reload" verb, and only then make this available as
1033 * "systemctl status --follow" has been removed, use "journalctl
1036 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
1037 have been removed since they are hardly useful to be
1040 * And I'd like to take the opportunity to specifically mention
1041 Zbigniew for his great contributions. Zbigniew, you rock!
1043 Contributions from: Andrew Eikum, Christian Hesse, Colin
1044 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
1045 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
1046 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
1047 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
1048 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
1049 Jędrzejewski-Szmek, Сковорода Никита Андреевич
1053 * If /etc/vconsole.conf is non-existent or empty we will no
1054 longer load any console font or key map at boot by
1055 default. Instead the kernel defaults will be left
1056 intact. This is definitely the right thing to do, as no
1057 configuration should mean no configuration, and hard-coding
1058 font names that are different on all archs is probably a bad
1059 idea. Also, the kernel default key map and font should be
1060 good enough for most cases anyway, and mostly identical to
1061 the userspace fonts/key maps we previously overloaded them
1062 with. If distributions want to continue to default to a
1063 non-kernel font or key map they should ship a default
1064 /etc/vconsole.conf with the appropriate contents.
1066 Contributions from: Colin Walters, Daniel J Walsh, Dave
1067 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
1068 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1072 * journalctl gained a new --cursor= switch to show entries
1073 starting from the specified location in the journal.
1075 * We now enforce a size limit on journal entry fields exported
1076 with "-o json" in journalctl. Fields larger than 4K will be
1077 assigned null. This can be turned off with --all.
1079 * An (optional) journal gateway daemon is now available as
1080 "systemd-journal-gatewayd.service". This service provides
1081 access to the journal via HTTP and JSON. This functionality
1082 will be used to implement live log synchronization in both
1083 pull and push modes, but has various other users too, such
1084 as easy log access for debugging of embedded devices. Right
1085 now it is already useful to retrieve the journal via HTTP:
1087 # systemctl start systemd-journal-gatewayd.service
1088 # wget http://localhost:19531/entries
1090 This will download the journal contents in a
1091 /var/log/messages compatible format. The same as JSON:
1093 # curl -H"Accept: application/json" http://localhost:19531/entries
1095 This service is also accessible via a web browser where a
1096 single static HTML5 app is served that uses the JSON logic
1097 to enable the user to do some basic browsing of the
1098 journal. This will be extended later on. Here's an example
1099 screenshot of this app in its current state:
1101 http://0pointer.de/public/journal-gatewayd
1103 Contributions from: Kay Sievers, Lennart Poettering, Robert
1104 Milasan, Tom Gundersen
1108 * The bash completion logic is now available for journalctl
1111 * We don't mount the "cpuset" controller anymore together with
1112 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
1113 started if no parameters are assigned to it. "cpuset" hence
1114 broke code that assumed it it could create "cpu" groups and
1117 * journalctl -f will now subscribe to terminal size changes,
1118 and line break accordingly.
1120 Contributions from: Dave Reisner, Kay Sievers, Lennart
1121 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
1125 * nspawn will now create a symlink /etc/localtime in the
1126 container environment, copying the host's timezone
1127 setting. Previously this has been done via a bind mount, but
1128 since symlinks cannot be bind mounted this has now been
1129 changed to create/update the appropriate symlink.
1131 * journalctl -n's line number argument is now optional, and
1132 will default to 10 if omitted.
1134 * journald will now log the maximum size the journal files may
1135 take up on disk. This is particularly useful if the default
1136 built-in logic of determining this parameter from the file
1137 system size is used. Use "systemctl status
1138 systemd-journald.service" to see this information.
1140 * The multi-seat X wrapper tool has been stripped down. As X
1141 is now capable of enumerating graphics devices via udev in a
1142 seat-aware way the wrapper is not strictly necessary
1143 anymore. A stripped down temporary stop-gap is still shipped
1144 until the upstream display managers have been updated to
1145 fully support the new X logic. Expect this wrapper to be
1146 removed entirely in one of the next releases.
1148 * HandleSleepKey= in logind.conf has been split up into
1149 HandleSuspendKey= and HandleHibernateKey=. The old setting
1150 is not available anymore. X11 and the kernel are
1151 distuingishing between these keys and we should too. This
1152 also means the inhibition lock for these keys has been split
1155 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
1156 Poettering, Lukas Nykryn, Václav Pavlín
1160 * Whenever a unit changes state we'll now log this to the
1161 journal and show along the unit's own log output in
1164 * ConditionPathIsMountPoint= can now properly detect bind
1165 mount points too. (Previously, a bind mount of one file
1166 system to another place in the same file system could not be
1167 detected as mount, since they shared struct stat's st_dev
1170 * We will now mount the cgroup controllers cpu, cpuacct,
1171 cpuset and the controllers net_cls, net_prio together by
1174 * nspawn containers will now have a virtualized boot
1175 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
1176 over with a randomized ID at container initialization). This
1177 has the effect of making "journalctl -b" do the right thing
1180 * The JSON output journal serialization has been updated not
1181 to generate "endless" list objects anymore, but rather one
1182 JSON object per line. This is more in line how most JSON
1183 parsers expect JSON objects. The new output mode
1184 "json-pretty" has been added to provide similar output, but
1185 neatly aligned for readability by humans.
1187 * We dropped all explicit sync() invocations in the shutdown
1188 code. The kernel does this implicitly anyway in the kernel
1189 reboot() syscall. halt(8)'s -n option is now a compatibility
1192 * We now support virtualized reboot() in containers, as
1193 supported by newer kernels. We will fall back to exit() if
1194 CAP_SYS_REBOOT is not available to the container. Also,
1195 nspawn makes use of this now and will actually reboot the
1196 container if the containerized OS asks for that.
1198 * journalctl will only show local log output by default
1199 now. Use --merge (-m) to show remote log output, too.
1201 * libsystemd-journal gained the new sd_journal_get_usage()
1202 call to determine the current disk usage of all journal
1203 files. This is exposed in the new "journalctl --disk-usage"
1206 * journald gained a new configuration setting SplitMode= in
1207 journald.conf which may be used to control how user journals
1208 are split off. See journald.conf(5) for details.
1210 * A new condition type ConditionFileNotEmpty= has been added.
1212 * tmpfiles' "w" lines now support file globbing, to write
1213 multiple files at once.
1215 * We added Python bindings for the journal submission
1216 APIs. More Python APIs for a number of selected APIs will
1217 likely follow. Note that we intend to add native bindings
1218 only for the Python language, as we consider it common
1219 enough to deserve bindings shipped within systemd. There are
1220 various projects outside of systemd that provide bindings
1221 for languages such as PHP or Lua.
1223 * Many conditions will now resolve specifiers such as %i. In
1224 addition, PathChanged= and related directives of .path units
1225 now support specifiers as well.
1227 * There's now a new RPM macro definition for the system preset
1230 * journald will now warn if it can't forward a message to the
1231 syslog daemon because it's socket is full.
1233 * timedated will no longer write or process /etc/timezone,
1234 except on Debian. As we do not support late mounted /usr
1235 anymore /etc/localtime always being a symlink is now safe,
1236 and hence the information in /etc/timezone is not necessary
1239 * logind will now always reserve one VT for a text getty (VT6
1240 by default). Previously if more than 6 X sessions where
1241 started they took up all the VTs with auto-spawned gettys,
1242 so that no text gettys were available anymore.
1244 * udev will now automatically inform the btrfs kernel logic
1245 about btrfs RAID components showing up. This should make
1246 simple hotplug based btrfs RAID assembly work.
1248 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
1249 (but not for its children which will stay at the kernel
1250 default). This should allow setups with a lot more listening
1253 * systemd will now always pass the configured timezone to the
1254 kernel at boot. timedated will do the same when the timezone
1257 * logind's inhibition logic has been updated. By default,
1258 logind will now handle the lid switch, the power and sleep
1259 keys all the time, even in graphical sessions. If DEs want
1260 to handle these events on their own they should take the new
1261 handle-power-key, handle-sleep-key and handle-lid-switch
1262 inhibitors during their runtime. A simple way to achiveve
1263 that is to invoke the DE wrapped in an invocation of:
1265 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
1267 * Access to unit operations is now checked via SELinux taking
1268 the unit file label and client process label into account.
1270 * systemd will now notify the administrator in the journal
1271 when he over-mounts a non-empty directory.
1273 * There are new specifiers that are resolved in unit files,
1274 for the host name (%H), the machine ID (%m) and the boot ID
1277 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
1278 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
1279 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
1280 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
1281 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
1282 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
1283 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
1287 * Support for reading structured kernel messages from
1288 /dev/kmsg has now been added and is enabled by default.
1290 * Support for reading kernel messages from /proc/kmsg has now
1291 been removed. If you want kernel messages in the journal
1292 make sure to run a recent kernel (>= 3.5) that supports
1293 reading structured messages from /dev/kmsg (see
1294 above). /proc/kmsg is now exclusive property of classic
1295 syslog daemons again.
1297 * The libudev API gained the new
1298 udev_device_new_from_device_id() call.
1300 * The logic for file system namespace (ReadOnlyDirectory=,
1301 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
1302 require pivot_root() anymore. This means fewer temporary
1303 directories are created below /tmp for this feature.
1305 * nspawn containers will now see and receive all submounts
1306 made on the host OS below the root file system of the
1309 * Forward Secure Sealing is now supported for Journal files,
1310 which provide cryptographical sealing of journal files so
1311 that attackers cannot alter log history anymore without this
1312 being detectable. Lennart will soon post a blog story about
1313 this explaining it in more detail.
1315 * There are two new service settings RestartPreventExitStatus=
1316 and SuccessExitStatus= which allow configuration of exit
1317 status (exit code or signal) which will be excepted from the
1318 restart logic, resp. consider successful.
1320 * journalctl gained the new --verify switch that can be used
1321 to check the integrity of the structure of journal files and
1322 (if Forward Secure Sealing is enabled) the contents of
1325 * nspawn containers will now be run with /dev/stdin, /dev/fd/
1326 and similar symlinks pre-created. This makes running shells
1327 as container init process a lot more fun.
1329 * The fstab support can now handle PARTUUID= and PARTLABEL=
1332 * A new ConditionHost= condition has been added to match
1333 against the hostname (with globs) and machine ID. This is
1334 useful for clusters where a single OS image is used to
1335 provision a large number of hosts which shall run slightly
1336 different sets of services.
1338 * Services which hit the restart limit will now be placed in a
1341 Contributions from: Bertram Poettering, Dave Reisner, Huang
1342 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
1343 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
1347 * When running in --user mode systemd will now become a
1348 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
1349 tree a lot more organized.
1351 * A new PartOf= unit dependency type has been introduced that
1352 may be used to group services in a natural way.
1354 * "systemctl enable" may now be used to enable instances of
1357 * journalctl now prints error log levels in red, and
1358 warning/notice log levels in bright white. It also supports
1359 filtering by log level now.
1361 * cgtop gained a new -n switch (similar to top), to configure
1362 the maximum number of iterations to run for. It also gained
1363 -b, to run in batch mode (accepting no input).
1365 * The suffix ".service" may now be omitted on most systemctl
1366 command lines involving service unit names.
1368 * There's a new bus call in logind to lock all sessions, as
1369 well as a loginctl verb for it "lock-sessions".
1371 * libsystemd-logind.so gained a new call sd_journal_perror()
1372 that works similar to libc perror() but logs to the journal
1373 and encodes structured information about the error number.
1375 * /etc/crypttab entries now understand the new keyfile-size=
1378 * shutdown(8) now can send a (configurable) wall message when
1379 a shutdown is cancelled.
1381 * The mount propagation mode for the root file system will now
1382 default to "shared", which is useful to make containers work
1383 nicely out-of-the-box so that they receive new mounts from
1384 the host. This can be undone locally by running "mount
1385 --make-rprivate /" if needed.
1387 * The prefdm.service file has been removed. Distributions
1388 should maintain this unit downstream if they intend to keep
1389 it around. However, we recommend writing normal unit files
1390 for display managers instead.
1392 * Since systemd is a crucial part of the OS we will now
1393 default to a number of compiler switches that improve
1394 security (hardening) such as read-only relocations, stack
1395 protection, and suchlike.
1397 * The TimeoutSec= setting for services is now split into
1398 TimeoutStartSec= and TimeoutStopSec= to allow configuration
1399 of individual time outs for the start and the stop phase of
1402 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
1403 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
1404 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
1405 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
1406 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
1407 Gundersen, Zbigniew Jędrzejewski-Szmek
1411 * The journal and id128 C APIs are now fully documented as man
1414 * Extra safety checks have been added when transitioning from
1415 the initial RAM disk to the main system to avoid accidental
1418 * /etc/crypttab entries now understand the new keyfile-offset=
1421 * systemctl -t can now be used to filter by unit load state.
1423 * The journal C API gained the new sd_journal_wait() call to
1424 make writing synchronous journal clients easier.
1426 * journalctl gained the new -D switch to show journals from a
1429 * journalctl now displays a special marker between log
1430 messages of two different boots.
1432 * The journal is now explicitly flushed to /var via a service
1433 systemd-journal-flush.service, rather than implicitly simply
1434 by seeing /var/log/journal to be writable.
1436 * journalctl (and the journal C APIs) can now match for much
1437 more complex expressions, with alternatives and
1440 * When transitioning from the initial RAM disk to the main
1441 system we will now kill all processes in a killing spree to
1442 ensure no processes stay around by accident.
1444 * Three new specifiers may be used in unit files: %u, %h, %s
1445 resolve to the user name, user home directory resp. user
1446 shell. This is useful for running systemd user instances.
1448 * We now automatically rotate journal files if their data
1449 object hash table gets a fill level > 75%. We also size the
1450 hash table based on the configured maximum file size. This
1451 together should lower hash collisions drastically and thus
1452 speed things up a bit.
1454 * journalctl gained the new "--header" switch to introspect
1455 header data of journal files.
1457 * A new setting SystemCallFilters= has been added to services
1458 which may be used to apply blacklists or whitelists to
1459 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
1461 * nspawn gained a new --link-journal= switch (and quicker: -j)
1462 to link the container journal with the host. This makes it
1463 very easy to centralize log viewing on the host for all
1464 guests while still keeping the journal files separated.
1466 * Many bugfixes and optimizations
1468 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
1469 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
1470 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
1475 * Several tools now understand kernel command line arguments,
1476 which are only read when run in an initial RAM disk. They
1477 usually follow closely their normal counterparts, but are
1480 * There's a new tool to analyze the readahead files that are
1481 automatically generated at boot. Use:
1483 /usr/lib/systemd/systemd-readahead analyze /.readahead
1485 * We now provide an early debug shell on tty9 if this enabled. Use:
1487 systemctl enable debug-shell.service
1489 * All plymouth related units have been moved into the Plymouth
1490 package. Please make sure to upgrade your Plymouth version
1493 * systemd-tmpfiles now supports getting passed the basename of
1494 a configuration file only, in which case it will look for it
1495 in all appropriate directories automatically.
1497 * udevadm info now takes a /dev or /sys path as argument, and
1498 does the right thing. Example:
1500 udevadm info /dev/sda
1501 udevadm info /sys/class/block/sda
1503 * systemctl now prints a warning if a unit is stopped but a
1504 unit that might trigger it continues to run. Example: a
1505 service is stopped but the socket that activates it is left
1508 * "systemctl status" will now mention if the log output was
1509 shortened due to rotation since a service has been started.
1511 * The journal API now exposes functions to determine the
1512 "cutoff" times due to rotation.
1514 * journald now understands SIGUSR1 and SIGUSR2 for triggering
1515 immediately flushing of runtime logs to /var if possible,
1516 resp. for triggering immediate rotation of the journal
1519 * It is now considered an error if a service is attempted to
1520 be stopped that is not loaded.
1522 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
1524 * systemd-analyze now supports Python 3
1526 * tmpfiles now supports cleaning up directories via aging
1527 where the first level dirs are always kept around but
1528 directories beneath it automatically aged. This is enabled
1529 by prefixing the age field with '~'.
1531 * Seat objects now expose CanGraphical, CanTTY properties
1532 which is required to deal with very fast bootups where the
1533 display manager might be running before the graphics drivers
1534 completed initialization.
1536 * Seat objects now expose a State property.
1538 * We now include RPM macros for service enabling/disabling
1539 based on the preset logic. We recommend RPM based
1540 distributions to make use of these macros if possible. This
1541 makes it simpler to reuse RPM spec files across
1544 * We now make sure that the collected systemd unit name is
1545 always valid when services log to the journal via
1548 * There's a new man page kernel-command-line(7) detailing all
1549 command line options we understand.
1551 * The fstab generator may now be disabled at boot by passing
1552 fstab=0 on the kernel command line.
1554 * A new kernel command line option modules-load= is now understood
1555 to load a specific kernel module statically, early at boot.
1557 * Unit names specified on the systemctl command line are now
1558 automatically escaped as needed. Also, if file system or
1559 device paths are specified they are automatically turned
1560 into the appropriate mount or device unit names. Example:
1562 systemctl status /home
1563 systemctl status /dev/sda
1565 * The SysVConsole= configuration option has been removed from
1566 system.conf parsing.
1568 * The SysV search path is no longer exported on the D-Bus
1571 * The Names= option is been removed from unit file parsing.
1573 * There's a new man page bootup(7) detailing the boot process.
1575 * Every unit and every generator we ship with systemd now
1576 comes with full documentation. The self-explanatory boot is
1579 * A couple of services gained "systemd-" prefixes in their
1580 name if they wrap systemd code, rather than only external
1581 code. Among them fsck@.service which is now
1582 systemd-fsck@.service.
1584 * The HaveWatchdog property has been removed from the D-Bus
1587 * systemd.confirm_spawn= on the kernel command line should now
1590 * There's a new man page crypttab(5) which details all options
1591 we actually understand.
1593 * systemd-nspawn gained a new --capability= switch to pass
1594 additional capabilities to the container.
1596 * timedated will now read known NTP implementation unit names
1597 from /usr/lib/systemd/ntp-units.d/*.list,
1598 systemd-timedated-ntp.target has been removed.
1600 * journalctl gained a new switch "-b" that lists log data of
1601 the current boot only.
1603 * The notify socket is in the abstract namespace again, in
1604 order to support daemons which chroot() at start-up.
1606 * There is a new Storage= configuration option for journald
1607 which allows configuration of where log data should go. This
1608 also provides a way to disable journal logging entirely, so
1609 that data collected is only forwarded to the console, the
1610 kernel log buffer or another syslog implementation.
1612 * Many bugfixes and optimizations
1614 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
1615 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
1616 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
1617 Shawn Landden, Tom Gundersen
1621 * "systemctl help <unit>" now shows the man page if one is
1624 * Several new man pages have been added.
1626 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
1627 MaxLevelConsole= can now be specified in
1628 journald.conf. These options allow reducing the amount of
1629 data stored on disk or forwarded by the log level.
1631 * TimerSlackNSec= can now be specified in system.conf for
1632 PID1. This allows system-wide power savings.
1634 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
1635 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
1640 * logind is now capable of (optionally) handling power and
1641 sleep keys as well as the lid switch.
1643 * journalctl now understands the syntax "journalctl
1644 /usr/bin/avahi-daemon" to get all log output of a specific
1647 * CapabilityBoundingSet= in system.conf now also influences
1648 the capability bound set of usermode helpers of the kernel.
1650 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
1651 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
1652 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
1653 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
1657 * Note that we skipped 139 releases here in order to set the
1658 new version to something that is greater than both udev's
1659 and systemd's most recent version number.
1661 * udev: all udev sources are merged into the systemd source tree now.
1662 All future udev development will happen in the systemd tree. It
1663 is still fully supported to use the udev daemon and tools without
1664 systemd running, like in initramfs or other init systems. Building
1665 udev though, will require the *build* of the systemd tree, but
1666 udev can be properly *run* without systemd.
1668 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
1669 should be used to create dead device nodes as workarounds for broken
1672 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
1673 no longer supported. udev_monitor_new_from_netlink() needs to be
1674 used to subscribe to events.
1676 * udev: when udevd is started by systemd, processes which are left
1677 behind by forking them off of udev rules, are unconditionally cleaned
1678 up and killed now after the event handling has finished. Services or
1679 daemons must be started as systemd services. Services can be
1680 pulled-in by udev to get started, but they can no longer be directly
1681 forked by udev rules.
1683 * udev: the daemon binary is called systemd-udevd now and installed
1684 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
1685 to adapt to that, create symlink, or rename the binary after building
1688 * libudev no longer provides these symbols:
1689 udev_monitor_from_socket()
1690 udev_queue_get_failed_list_entry()
1691 udev_get_{dev,sys,run}_path()
1692 The versions number was bumped and symbol versioning introduced.
1694 * systemd-loginctl and systemd-journalctl have been renamed
1695 to loginctl and journalctl to match systemctl.
1697 * The config files: /etc/systemd/systemd-logind.conf and
1698 /etc/systemd/systemd-journald.conf have been renamed to
1699 logind.conf and journald.conf. Package updates should rename
1700 the files to the new names on upgrade.
1702 * For almost all files the license is now LGPL2.1+, changed
1703 from the previous GPL2.0+. Exceptions are some minor stuff
1704 of udev (which will be changed to LGPL2.1 eventually, too),
1705 and the MIT licensed sd-daemon.[ch] library that is suitable
1706 to be used as drop-in files.
1708 * systemd and logind now handle system sleep states, in
1709 particular suspending and hibernating.
1711 * logind now implements a sleep/shutdown/idle inhibiting logic
1712 suitable for a variety of uses. Soonishly Lennart will blog
1713 about this in more detail.
1715 * var-run.mount and var-lock.mount are no longer provided
1716 (which prevously bind mounted these directories to their new
1717 places). Distributions which have not converted these
1718 directories to symlinks should consider stealing these files
1719 from git history and add them downstream.
1721 * We introduced the Documentation= field for units and added
1722 this to all our shipped units. This is useful to make it
1723 easier to explore the boot and the purpose of the various
1726 * All smaller setup units (such as
1727 systemd-vconsole-setup.service) now detect properly if they
1728 are run in a container and are skipped when
1729 appropriate. This guarantees an entirely noise-free boot in
1730 Linux container environments such as systemd-nspawn.
1732 * A framework for implementing offline system updates is now
1733 integrated, for details see:
1734 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
1736 * A new service type Type=idle is available now which helps us
1737 avoiding ugly interleaving of getty output and boot status
1740 * There's now a system-wide CapabilityBoundingSet= option to
1741 globally reduce the set of capabilities for the
1742 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
1743 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
1744 even CAP_NET_ADMIN system-wide for secure systems.
1746 * There are now system-wide DefaultLimitXXX= options to
1747 globally change the defaults of the various resource limits
1748 for all units started by PID 1.
1750 * Harald Hoyer's systemd test suite has been integrated into
1751 systemd which allows easy testing of systemd builds in qemu
1752 and nspawn. (This is really awesome! Ask us for details!)
1754 * The fstab parser is now implemented as generator, not inside
1757 * systemctl will now warn you if .mount units generated from
1758 /etc/fstab are out of date due to changes in fstab that
1759 haven't been read by systemd yet.
1761 * systemd is now suitable for usage in initrds. Dracut has
1762 already been updated to make use of this. With this in place
1763 initrds get a slight bit faster but primarily are much
1764 easier to introspect and debug since "systemctl status" in
1765 the host system can be used to introspect initrd services,
1766 and the journal from the initrd is kept around too.
1768 * systemd-delta has been added, a tool to explore differences
1769 between user/admin configuration and vendor defaults.
1771 * PrivateTmp= now affects both /tmp and /var/tmp.
1773 * Boot time status messages are now much prettier and feature
1774 proper english language. Booting up systemd has never been
1777 * Read-ahead pack files now include the inode number of all
1778 files to pre-cache. When the inode changes the pre-caching
1779 is not attempted. This should be nicer to deal with updated
1780 packages which might result in changes of read-ahead
1783 * We now temporaritly lower the kernel's read_ahead_kb variable
1784 when collecting read-ahead data to ensure the kernel's
1785 built-in read-ahead does not add noise to our measurements
1786 of necessary blocks to pre-cache.
1788 * There's now RequiresMountsFor= to add automatic dependencies
1789 for all mounts necessary for a specific file system path.
1791 * MountAuto= and SwapAuto= have been removed from
1792 system.conf. Mounting file systems at boot has to take place
1795 * nspawn now learned a new switch --uuid= to set the machine
1796 ID on the command line.
1798 * nspawn now learned the -b switch to automatically search
1801 * vt102 is now the default TERM for serial TTYs, upgraded from
1804 * systemd-logind now works on VT-less systems.
1806 * The build tree has been reorganized. The individual
1807 components now have directories of their own.
1809 * A new condition type ConditionPathIsReadWrite= is now available.
1811 * nspawn learned the new -C switch to create cgroups for the
1812 container in other hierarchies.
1814 * We now have support for hardware watchdogs, configurable in
1817 * The scheduled shutdown logic now has a public API.
1819 * We now mount /tmp as tmpfs by default, but this can be
1820 masked and /etc/fstab can override it.
1822 * Since udisks doesn't make use of /media anymore we are not
1823 mounting a tmpfs on it anymore.
1825 * journalctl gained a new --local switch to only interleave
1826 locally generated journal files.
1828 * We can now load the IMA policy at boot automatically.
1830 * The GTK tools have been split off into a systemd-ui.
1832 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
1833 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
1834 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
1835 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
1836 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
1837 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
1838 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
1839 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
1840 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
1845 * This is mostly a bugfix release
1847 * Support optional initialization of the machine ID from the
1848 KVM or container configured UUID.
1850 * Support immediate reboots with "systemctl reboot -ff"
1852 * Show /etc/os-release data in systemd-analyze output
1854 * Many bugfixes for the journal, including endianness fixes and
1855 ensuring that disk space enforcement works
1857 * sd-login.h is C++ comptaible again
1859 * Extend the /etc/os-release format on request of the Debian
1862 * We now refuse non-UTF8 strings used in various configuration
1863 and unit files. This is done to ensure we don't pass invalid
1864 data over D-Bus or expose it elsewhere.
1866 * Register Mimo USB Screens as suitable for automatic seat
1869 * Read SELinux client context from journal clients in a race
1872 * Reorder configuration file lookup order. /etc now always
1873 overrides /run in order to allow the administrator to always
1874 and unconditionally override vendor supplied or
1875 automatically generated data.
1877 * The various user visible bits of the journal now have man
1878 pages. We still lack man pages for the journal API calls
1881 * We now ship all man pages in HTML format again in the
1884 Contributions from: Dave Reisner, Dirk Eibach, Frederic
1885 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
1886 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
1891 * This is mostly a bugfix release
1893 * systems lacking /etc/os-release are no longer supported.
1895 * Various functionality updates to libsystemd-login.so
1897 * Track class of PAM logins to distuingish greeters from
1900 Contributions from: Kay Sievers, Lennart Poettering, Michael
1905 * This is an important bugfix release for v41.
1907 * Building man pages is now optional which should be useful
1908 for those building systemd from git but unwilling to install
1911 * Watchdog support for supervising services is now usable. In
1912 a future release support for hardware watchdogs
1913 (i.e. /dev/watchdog) will be added building on this.
1915 * Service start rate limiting is now configurable and can be
1916 turned off per service. When a start rate limit is hit a
1917 reboot can automatically be triggered.
1919 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
1921 Contributions from: Benjamin Franzke, Bill Nottingham,
1922 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
1923 Schmidt, Michał Górny, Piotr Drąg
1927 * The systemd binary is installed /usr/lib/systemd/systemd now;
1928 An existing /sbin/init symlink needs to be adapted with the
1931 * The code that loads kernel modules has been ported to invoke
1932 libkmod directly, instead of modprobe. This means we do not
1933 support systems with module-init-tools anymore.
1935 * Watchdog support is now already useful, but still not
1938 * A new kernel command line option systemd.setenv= is
1939 understood to set system wide environment variables
1940 dynamically at boot.
1942 * We now limit the set of capabilities of systemd-journald.
1944 * We now set SIGPIPE to ignore by default, since it only is
1945 useful in shell pipelines, and has little use in general
1946 code. This can be disabled with IgnoreSIPIPE=no in unit
1949 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
1950 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
1955 * This is mostly a bugfix release
1957 * We now expose the reason why a service failed in the
1958 "Result" D-Bus property.
1960 * Rudimentary service watchdog support (will be completed over
1961 the next few releases.)
1963 * When systemd forks off in order execute some service we will
1964 now immediately changes its argv[0] to reflect which process
1965 it will execute. This is useful to minimize the time window
1966 with a generic argv[0], which makes bootcharts more useful
1968 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
1969 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
1970 Mike Kazantsev, Ray Strode
1974 * This is mostly a test release, but incorporates many
1977 * New systemd-cgtop tool to show control groups by their
1980 * Linking against libacl for ACLs is optional again. If
1981 disabled, support tracking device access for active logins
1982 goes becomes unavailable, and so does access to the user
1983 journals by the respective users.
1985 * If a group "adm" exists, journal files are automatically
1986 owned by them, thus allow members of this group full access
1987 to the system journal as well as all user journals.
1989 * The journal now stores the SELinux context of the logging
1990 client for all entries.
1992 * Add C++ inclusion guards to all public headers
1994 * New output mode "cat" in the journal to print only text
1995 messages, without any meta data like date or time.
1997 * Include tiny X server wrapper as a temporary stop-gap to
1998 teach XOrg udev display enumeration. This is used by display
1999 managers such as gdm, and will go away as soon as XOrg
2000 learned native udev hotplugging for display devices.
2002 * Add new systemd-cat tool for executing arbitrary programs
2003 with STDERR/STDOUT connected to the journal. Can also act as
2004 BSD logger replacement, and does so by default.
2006 * Optionally store all locally generated coredumps in the
2007 journal along with meta data.
2009 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
2010 writing short strings to files (for usage for /sys), and for
2011 creating symlinks, character and block device nodes.
2013 * New unit file option ControlGroupPersistent= to make cgroups
2014 persistent, following the mechanisms outlined in
2015 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
2017 * Support multiple local RTCs in a sane way
2019 * No longer monopolize IO when replaying readahead data on
2020 rotating disks, since we might starve non-file-system IO to
2021 death, since fanotify() will not see accesses done by blkid,
2024 * Don't show kernel threads in systemd-cgls anymore, unless
2025 requested with new -k switch.
2027 Contributions from: Dan Horák, Kay Sievers, Lennart
2028 Poettering, Michal Schmidt
2032 * This is mostly a test release, but incorporates many
2035 * The git repository moved to:
2036 git://anongit.freedesktop.org/systemd/systemd
2037 ssh://git.freedesktop.org/git/systemd/systemd
2039 * First release with the journal
2040 http://0pointer.de/blog/projects/the-journal.html
2042 * The journal replaces both systemd-kmsg-syslogd and
2043 systemd-stdout-bridge.
2045 * New sd_pid_get_unit() API call in libsystemd-logind
2047 * Many systemadm clean-ups
2049 * Introduce remote-fs-pre.target which is ordered before all
2050 remote mounts and may be used to start services before all
2053 * Added Mageia support
2055 * Add bash completion for systemd-loginctl
2057 * Actively monitor PID file creation for daemons which exit in
2058 the parent process before having finished writing the PID
2059 file in the daemon process. Daemons which do this need to be
2060 fixed (i.e. PID file creation must have finished before the
2061 parent exits), but we now react a bit more gracefully to them.
2063 * Add colourful boot output, mimicking the well-known output
2064 of existing distributions.
2066 * New option PassCredentials= for socket units, for
2067 compatibility with a recent kernel ABI breakage.
2069 * /etc/rc.local is now hooked in via a generator binary, and
2070 thus will no longer act as synchronization point during
2073 * systemctl list-unit-files now supports --root=.
2075 * systemd-tmpfiles now understands two new commands: z, Z for
2076 relabelling files according to the SELinux database. This is
2077 useful to apply SELinux labels to specific files in /sys,
2080 * Output of SysV services is now forwarded to both the console
2081 and the journal by default, not only just the console.
2083 * New man pages for all APIs from libsystemd-login.
2085 * The build tree got reorganized and a the build system is a
2086 lot more modular allowing embedded setups to specifically
2087 select the components of systemd they are interested in.
2089 * Support for Linux systems lacking the kernel VT subsystem is
2092 * configure's --with-rootdir= got renamed to
2093 --with-rootprefix= to follow the naming used by udev and
2096 * Unless specified otherwise we'll now install to /usr instead
2097 of /usr/local by default.
2099 * Processes with '@' in argv[0][0] are now excluded from the
2100 final shut-down killing spree, following the logic explained
2102 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
2104 * All processes remaining in a service cgroup when we enter
2105 the START or START_PRE states are now killed with
2106 SIGKILL. That means it is no longer possible to spawn
2107 background processes from ExecStart= lines (which was never
2108 supported anyway, and bad style).
2110 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
2111 reloading of units together.
2113 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
2114 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
2115 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
2116 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
2117 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek