1 systemd System and Service Manager
5 * The systemd-sysctl tool does no longer natively read the
6 file /etc/sysctl.conf. If desired, the file should be
7 symlinked from /etc/sysctl.d/99-sysctl.conf. Apart from
8 providing legacy support by a symlink rather than built-in
9 code, it also makes the otherwise hidden order of application
10 of the different files visible.
14 * The documentation has been updated to cover the various new
15 concepts introduced with 205.
17 * Unit files now understand the new %v specifier which
18 resolves to the kernel version string as returned by "uname
21 * systemctl now supports filtering the unit list output by
22 load state, active state and sub state, using the new
25 * "systemctl status" will now show the results of the
26 condition checks (like ConditionPathExists= and similar) of
27 the last start attempts of the unit. They are also logged to
30 * "journalctl -b" may now be used to look for boot output of a
31 specific boot. Try "journalctl -b -1" for the previous boot,
32 but the syntax is substantially more powerful.
34 * "journalctl --show-cursor" has been added which prints the
35 cursor string the last shown log line. This may then be used
36 with the new "journalctl --after-cursor=" switch to continue
37 browsing logs from that point on.
39 * "journalctl --force" may now be used to force regeneration
42 * Creation of "dead" device nodes has been moved from udev
43 into kmod and tmpfiles. Previously, udev would read the kmod
44 databases to pre-generate dead device nodes based on meta
45 information contained in kernel modules, so that these would
46 be auto-loaded on access rather then at boot. As this
47 doesn't really have much to do with the exposing actual
48 kernel devices to userspace this has always been slightly
49 alien in the udev codebase. Following the new scheme kmod
50 will now generate a runtime snippet for tmpfiles from the
51 module meta information and it now is tmpfiles' job to the
52 create the nodes. This also allows overriding access and
53 other parameters for the nodes using the usual tmpfiles
54 facilities. As side effect this allows us to remove the
55 CAP_SYS_MKNOD capability bit from udevd entirely.
57 * logind's device ACLs may now be applied to these "dead"
58 devices nodes too, thus finally allowing managed access to
59 devices such as /dev/snd/sequencer whithout loading the
60 backing module right-away.
62 * A new RPM macro has been added that may be used to apply
63 tmpfiles configuration during package installation.
65 * systemd-detect-virt and ConditionVirtualization= now can
66 detect User-Mode-Linux machines (UML).
68 * journald will now implicitly log the effective capabilities
69 set of processes in the message metadata.
71 * systemd-cryptsetup has gained support for TrueCrypt volumes.
73 * The initrd interface has been simplified (more specifically,
74 support for passing performance data via environment
75 variables and fsck results via files in /run has been
76 removed). These features were non-essential, and are
77 nowadays available in a much nicer way by having systemd in
78 the initrd serialize its state and have the hosts systemd
81 * The udev "keymap" data files and tools to apply keyboard
82 specific mappings of scan to key codes, and force-release
83 scan code lists have been entirely replaced by a udev
84 "keyboard" builtin and a hwdb data file.
86 * systemd will now honour the kernel's "quiet" command line
87 argument also during late shutdown, resulting in a
88 completely silent shutdown when used.
90 * There's now an option to control the SO_REUSEPORT socket
91 option in .socket units.
93 * Instance units will now automatically get a per-template
94 subslice of system.slice unless something else is explicitly
95 configured. For example, instances of sshd@.service will now
96 implicitly be placed in system-sshd.slice rather than
97 system.slice as before.
99 * Test coverage support may now be enabled at build time.
101 Contributions from: Dave Reisner, Frederic Crozat, Harald
102 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
103 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
104 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
105 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
106 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
107 Giokas, Zbigniew Jędrzejewski-Szmek
111 * Two new unit types have been introduced:
113 Scope units are very similar to service units, however, are
114 created out of pre-existing processes -- instead of PID 1
115 forking off the processes. By using scope units it is
116 possible for system services and applications to group their
117 own child processes (worker processes) in a powerful way
118 which then maybe used to organize them, or kill them
119 together, or apply resource limits on them.
121 Slice units may be used to partition system resources in an
122 hierarchial fashion and then assign other units to them. By
123 default there are now three slices: system.slice (for all
124 system services), user.slice (for all user sessions),
125 machine.slice (for VMs and containers).
127 Slices and scopes have been introduced primarily in
128 context of the work to move cgroup handling to a
129 single-writer scheme, where only PID 1
130 creates/removes/manages cgroups.
132 * There's a new concept of "transient" units. In contrast to
133 normal units these units are created via an API at runtime,
134 not from configuration from disk. More specifically this
135 means it is now possible to run arbitrary programs as
136 independent services, with all execution parameters passed
137 in via bus APIs rather than read from disk. Transient units
138 make systemd substantially more dynamic then it ever was,
139 and useful as a general batch manager.
141 * logind has been updated to make use of scope and slice units
142 for managing user sessions. As a user logs in he will get
143 his own private slice unit, to which all sessions are added
144 as scope units. We also added support for automatically
145 adding an instance of user@.service for the user into the
146 slice. Effectively logind will no longer create cgroup
147 hierarchies on its own now, it will defer entirely to PID 1
148 for this by means of scope, service and slice units. Since
149 user sessions this way become entities managed by PID 1
150 the output of "systemctl" is now a lot more comprehensive.
152 * A new mini-daemon "systemd-machined" has been added which
153 may be used by virtualization managers to register local
154 VMs/containers. nspawn has been updated accordingly, and
155 libvirt will be updated shortly. machined will collect a bit
156 of meta information about the VMs/containers, and assign
157 them their own scope unit (see above). The collected
158 meta-data is then made available via the "machinectl" tool,
159 and exposed in "ps" and similar tools. machined/machinectl
160 is compile-time optional.
162 * As discussed earlier, the low-level cgroup configuration
163 options ControlGroup=, ControlGroupModify=,
164 ControlGroupPersistent=, ControlGroupAttribute= have been
165 removed. Please use high-level attribute settings instead as
168 * A new bus call SetUnitProperties() has been added to alter
169 various runtime parameters of a unit. This is primarily
170 useful to alter cgroup parameters dynamically in a nice way,
171 but will be extended later on to make more properties
172 modifiable at runtime. systemctl gained a new set-properties
173 command that wraps this call.
175 * A new tool "systemd-run" has been added which can be used to
176 run arbitrary command lines as transient services or scopes,
177 while configuring a number of settings via the command
178 line. This tool is currently very basic, however already
179 very useful. We plan to extend this tool to even allow
180 queuing of execution jobs with time triggers from the
181 command line, similar in fashion to "at".
183 * nspawn will now inform the user explicitly that kernels with
184 audit enabled break containers, and suggest the user to turn
187 * Support for detecting the IMA and AppArmor security
188 frameworks with ConditionSecurity= has been added.
190 * journalctl gained a new "-k" switch for showing only kernel
191 messages, mimicking dmesg output; in addition to "--user"
192 and "--system" switches for showing only user's own logs
195 * systemd-delta can now show information about drop-in
196 snippets extending unit files.
198 * libsystemd-bus has been substantially updated but is still
199 not available as public API.
201 * systemd will now look for the "debug" argument on the kernel
202 command line and enable debug logging, similar to
203 "systemd.log_level=debug" already did before.
205 * "systemctl set-default", "systemctl get-default" has been
206 added to configure the default.target symlink, which
207 controls what to boot into by default.
209 * "systemctl set-log-level" has been added as a convenient
210 way to raise and lower systemd logging threshold.
212 * "systemd-analyze plot" will now show the time the various
213 generators needed for execution, as well as information
214 about the unit file loading.
216 * libsystemd-journal gained a new sd_journal_open_files() call
217 for opening specific journal files. journactl also gained a
218 new switch to expose this new functionality. Previously we
219 only supported opening all files from a directory, or all
220 files from the system, as opening individual files only is
221 racy due to journal file rotation.
223 * systemd gained the new DefaultEnvironment= setting in
224 /etc/systemd/system.conf to set environment variables for
227 * If a privileged process logs a journal message with the
228 OBJECT_PID= field set, then journald will automatically
229 augment this with additional OBJECT_UID=, OBJECT_GID=,
230 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
231 system services want to log events about specific client
232 processes. journactl/systemctl has been updated to make use
233 of this information if all log messages regarding a specific
236 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
237 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
238 Reisner, David Coppa, David King, David Strauss, Eelco
239 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
240 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
241 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
242 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
243 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
244 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
245 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
246 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
247 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
252 * The Python bindings gained some minimal support for the APIs
253 exposed by libsystemd-logind.
255 * ConditionSecurity= gained support for detecting SMACK. Since
256 this condition already supports SELinux and AppArmor we only
257 miss IMA for this. Patches welcome!
259 Contributions from: Karol Lewandowski, Lennart Poettering,
260 Zbigniew Jędrzejewski-Szmek
264 * systemd-nspawn will now create /etc/resolv.conf if
265 necessary, before bind-mounting the host's file onto it.
267 * systemd-nspawn will now store meta information about a
268 container on the container's cgroup as extended attribute
269 fields, including the root directory.
271 * The cgroup hierarchy has been reworked in many ways. All
272 objects any of the components systemd creates in the cgroup
273 tree are now suffixed. More specifically, user sessions are
274 now placed in cgroups suffixed with ".session", users in
275 cgroups suffixed with ".user", and nspawn containers in
276 cgroups suffixed with ".nspawn". Furthermore, all cgroup
277 names are now escaped in a simple scheme to avoid collision
278 of userspace object names with kernel filenames. This work
279 is preparation for making these objects relocatable in the
280 cgroup tree, in order to allow easy resource partitioning of
281 these objects without causing naming conflicts.
283 * systemctl list-dependencies gained the new switches
284 --plain, --reverse, --after and --before.
286 * systemd-inhibit now shows the process name of processes that
287 have taken an inhibitor lock.
289 * nss-myhostname will now also resolve "localhost"
290 implicitly. This makes /etc/hosts an optional file and
291 nicely handles that on IPv6 ::1 maps to both "localhost" and
294 * libsystemd-logind.so gained a new call
295 sd_get_machine_names() to enumerate running containers and
296 VMs (currently only supported by very new libvirt and
297 nspawn). sd_login_monitor can now be used to watch
298 VMs/containers coming and going.
300 * .include is not allowed recursively anymore, and only in
301 unit files. Usually it is better to use drop-in snippets in
302 .d/*.conf anyway, as introduced with systemd 198.
304 * systemd-analyze gained a new "critical-chain" command that
305 determines the slowest chain of units run during system
306 boot-up. It is very useful for tracking down where
307 optimizing boot time is the most beneficial.
309 * systemd will no longer allow manipulating service paths in
310 the name=systemd:/system cgroup tree using ControlGroup= in
311 units. (But is still fine with it in all other dirs.)
313 * There's a new systemd-nspawn@.service service file that may
314 be used to easily run nspawn containers as system
315 services. With the container's root directory in
316 /var/lib/container/foobar it is now sufficient to run
317 "systemctl start systemd-nspawn@foobar.service" to boot it.
319 * systemd-cgls gained a new parameter "--machine" to list only
320 the processes within a certain container.
322 * ConditionSecurity= now can check for "apparmor". We still
323 are lacking checks for SMACK and IMA for this condition
324 check though. Patches welcome!
326 * A new configuration file /etc/systemd/sleep.conf has been
327 added that may be used to configure which kernel operation
328 systemd is supposed to execute when "suspend", "hibernate"
329 or "hybrid-sleep" is requested. This makes the new kernel
330 "freeze" state accessible to the user.
332 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
333 the passed argument if applicable.
335 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
336 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
337 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
338 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
339 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
340 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
341 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
346 * The output of 'systemctl list-jobs' got some polishing. The
347 '--type=' argument may now be passed more than once. A new
348 command 'systemctl list-sockets' has been added which shows
349 a list of kernel sockets systemd is listening on with the
350 socket units they belong to, plus the units these socket
353 * The experimental libsystemd-bus library got substantial
354 updates to work in conjunction with the (also experimental)
355 kdbus kernel project. It works well enough to exchange
356 messages with some sophistication. Note that kdbus is not
357 ready yet, and the library is mostly an elaborate test case
358 for now, and not installable.
360 * systemd gained a new unit 'systemd-static-nodes.service'
361 that generates static device nodes earlier during boot, and
362 can run in conjunction with udev.
364 * libsystemd-login gained a new call sd_pid_get_user_unit()
365 to retrieve the user systemd unit a process is running
366 in. This is useful for systems where systemd is used as
369 * systemd-nspawn now places all containers in the new /machine
370 top-level cgroup directory in the name=systemd
371 hierarchy. libvirt will soon do the same, so that we get a
372 uniform separation of /system, /user and /machine for system
373 services, user processes and containers/virtual
374 machines. This new cgroup hierarchy is also useful to stick
375 stable names to specific container instances, which can be
376 recognized later this way (this name may be controlled
377 via systemd-nspawn's new -M switch). libsystemd-login also
378 gained a new call sd_pid_get_machine_name() to retrieve the
379 name of the container/VM a specific process belongs to.
381 * bootchart can now store its data in the journal.
383 * libsystemd-journal gained a new call
384 sd_journal_add_conjunction() for AND expressions to the
385 matching logic. This can be used to express more complex
388 * journactl can now take multiple --unit= and --user-unit=
391 * The cryptsetup logic now understands the "luks.key=" kernel
392 command line switch for specifying a file to read the
393 decryption key from. Also, if a configured key file is not
394 found the tool will now automatically fall back to prompting
397 * Python systemd.journal module was updated to wrap recently
398 added functions from libsystemd-journal. The interface was
399 changed to bring the low level interface in s.j._Reader
400 closer to the C API, and the high level interface in
401 s.j.Reader was updated to wrap and convert all data about
404 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
405 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
406 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
407 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
408 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
409 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
413 * journalctl --update-catalog now understands a new --root=
414 option to operate on catalogs found in a different root
417 * During shutdown after systemd has terminated all running
418 services a final killing loop kills all remaining left-over
419 processes. We will now print the name of these processes
420 when we send SIGKILL to them, since this usually indicates a
423 * If /etc/crypttab refers to password files stored on
424 configured mount points automatic dependencies will now be
425 generated to ensure the specific mount is established first
426 before the key file is attempted to be read.
428 * 'systemctl status' will now show information about the
429 network sockets a socket unit is listening on.
431 * 'systemctl status' will also shown information about any
432 drop-in configuration file for units. (Drop-In configuration
433 files in this context are files such as
434 /etc/systemd/systemd/foobar.service.d/*.conf)
436 * systemd-cgtop now optionally shows summed up CPU times of
437 cgroups. Press '%' while running cgtop to switch between
438 percentage and absolute mode. This is useful to determine
439 which cgroups use up the most CPU time over the entire
440 runtime of the system. systemd-cgtop has also been updated
441 to be 'pipeable' for processing with further shell tools.
443 * 'hostnamectl set-hostname' will now allow setting of FQDN
446 * The formatting and parsing of time span values has been
447 changed. The parser now understands fractional expressions
448 such as "5.5h". The formatter will now output fractional
449 expressions for all time spans under 1min, i.e. "5.123456s"
450 rather than "5s 123ms 456us". For time spans under 1s
451 millisecond values are shown, for those under 1ms
452 microsecond values are shown. This should greatly improve
453 all time-related output of systemd.
455 * libsystemd-login and libsystemd-journal gained new
456 functions for querying the poll() events mask and poll()
457 timeout value for integration into arbitrary event
460 * localectl gained the ability to list available X11 keymaps
461 (models, layouts, variants, options).
463 * 'systemd-analyze dot' gained the ability to filter for
464 specific units via shell-style globs, to create smaller,
465 more useful graphs. I.e. it's now possible to create simple
466 graphs of all the dependencies between only target units, or
467 of all units that Avahi has dependencies with.
469 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
470 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
471 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
472 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
473 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
474 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
475 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
479 * The boot-time readahead implementation for rotating media
480 will now read the read-ahead data in multiple passes which
481 consist of all read requests made in equidistant time
482 intervals. This means instead of strictly reading read-ahead
483 data in its physical order on disk we now try to find a
484 middle ground between physical and access time order.
486 * /etc/os-release files gained a new BUILD_ID= field for usage
487 on operating systems that provide continuous builds of OS
490 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
491 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
492 William Douglas, Zbigniew Jędrzejewski-Szmek
496 * systemd-python gained an API exposing libsystemd-daemon.
498 * The SMACK setup logic gained support for uploading CIPSO
501 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
502 ReadOnlyDirectories= and InaccessibleDirectories= has
503 changed. The private /tmp and /var/tmp directories are now
504 shared by all processes of a service (which means
505 ExecStartPre= may now leave data in /tmp that ExecStart= of
506 the same service can still access). When a service is
507 stopped its temporary directories are immediately deleted
508 (normal clean-up with tmpfiles is still done in addition to
511 * By default, systemd will now set a couple of sysctl
512 variables in the kernel: the safe sysrq options are turned
513 on, IP route verification is turned on, and source routing
514 disabled. The recently added hardlink and softlink
515 protection of the kernel is turned on. These settings should
516 be reasonably safe, and good defaults for all new systems.
518 * The predictable network naming logic may now be turned off
519 with a new kernel command line switch: net.ifnames=0.
521 * A new libsystemd-bus module has been added that implements a
522 pretty complete D-Bus client library. For details see:
524 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
526 * journald will now explicitly flush the journal files to disk
527 at the latest 5min after each write. The file will then also
528 be marked offline until the next write. This should increase
529 reliability in case of a crash. The synchronization delay
530 can be configured via SyncIntervalSec= in journald.conf.
532 * There's a new remote-fs-setup.target unit that can be used
533 to pull in specific services when at least one remote file
534 system is to be mounted.
536 * There are new targets timers.target and paths.target as
537 canonical targets to pull user timer and path units in
538 from. This complements sockets.target with a similar
539 purpose for socket units.
541 * libudev gained a new call udev_device_set_attribute_value()
542 to set sysfs attributes of a device.
544 * The udev daemon now sets the default number of worker
545 processes executed in parallel based on the number of available
546 CPUs instead of the amount of available RAM. This is supposed
547 to provide a more reliable default and limit a too aggressive
548 paralellism for setups with 1000s of devices connected.
550 Contributions from: Auke Kok, Colin Walters, Cristian
551 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
552 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
553 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
554 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
555 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
556 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
557 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
558 Zbigniew Jędrzejewski-Szmek
562 * Configuration of unit files may now be extended via drop-in
563 files without having to edit/override the unit files
564 themselves. More specifically, if the administrator wants to
565 change one value for a service file foobar.service he can
566 now do so by dropping in a configuration snippet into
567 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
568 will load all these snippets and apply them on top of the
569 main unit configuration file, possibly extending or
570 overriding its settings. Using these drop-in snippets is
571 generally nicer than the two earlier options for changing
572 unit files locally: copying the files from
573 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
574 them there; or creating a new file in /etc/systemd/system/
575 that incorporates the original one via ".include". Drop-in
576 snippets into these .d/ directories can be placed in any
577 directory systemd looks for units in, and the usual
578 overriding semantics between /usr/lib, /etc and /run apply
581 * Most unit file settings which take lists of items can now be
582 reset by assigning the empty string to them. For example,
583 normally, settings such as Environment=FOO=BAR append a new
584 environment variable assignment to the environment block,
585 each time they are used. By assigning Environment= the empty
586 string the environment block can be reset to empty. This is
587 particularly useful with the .d/*.conf drop-in snippets
588 mentioned above, since this adds the ability to reset list
589 settings from vendor unit files via these drop-ins.
591 * systemctl gained a new "list-dependencies" command for
592 listing the dependencies of a unit recursively.
594 * Inhibitors are now honored and listed by "systemctl
595 suspend", "systemctl poweroff" (and similar) too, not only
596 GNOME. These commands will also list active sessions by
599 * Resource limits (as exposed by the various control group
600 controllers) can now be controlled dynamically at runtime
601 for all units. More specifically, you can now use a command
602 like "systemctl set-cgroup-attr foobar.service cpu.shares
603 2000" to alter the CPU shares a specific service gets. These
604 settings are stored persistently on disk, and thus allow the
605 administrator to easily adjust the resource usage of
606 services with a few simple commands. This dynamic resource
607 management logic is also available to other programs via the
608 bus. Almost any kernel cgroup attribute and controller is
611 * systemd-vconsole-setup will now copy all font settings to
612 all allocated VTs, where it previously applied them only to
615 * libsystemd-login gained the new sd_session_get_tty() API
618 * This release drops support for a few legacy or
619 distribution-specific LSB facility names when parsing init
620 scripts: $x-display-manager, $mail-transfer-agent,
621 $mail-transport-agent, $mail-transfer-agent, $smtp,
622 $null. Also, the mail-transfer-agent.target unit backing
623 this has been removed. Distributions which want to retain
624 compatibility with this should carry the burden for
625 supporting this themselves and patch support for these back
626 in, if they really need to. Also, the facilities $syslog and
627 $local_fs are now ignored, since systemd does not support
628 early-boot LSB init scripts anymore, and these facilities
629 are implied anyway for normal services. syslog.target has
632 * There are new bus calls on PID1's Manager object for
633 cancelling jobs, and removing snapshot units. Previously,
634 both calls were only available on the Job and Snapshot
637 * systemd-journal-gatewayd gained SSL support.
639 * The various "environment" files, such as /etc/locale.conf
640 now support continuation lines with a backslash ("\") as
641 last character in the line, similar in style (but different)
642 to how this is supported in shells.
644 * For normal user processes the _SYSTEMD_USER_UNIT= field is
645 now implicitly appended to every log entry logged. systemctl
646 has been updated to filter by this field when operating on a
647 user systemd instance.
649 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
650 CAP_AUDIT_CONTROL capabilities to the capabilities set for
651 the container. This makes it easier to boot unmodified
652 Fedora systems in a container, which however still requires
653 audit=0 to be passed on the kernel command line. Auditing in
654 kernel and userspace is unfortunately still too broken in
655 context of containers, hence we recommend compiling it out
656 of the kernel or using audit=0. Hopefully this will be fixed
657 one day for good in the kernel.
659 * nspawn gained the new --bind= and --bind-ro= parameters to
660 bind mount specific directories from the host into the
663 * nspawn will now mount its own devpts file system instance
664 into the container, in order not to leak pty devices from
665 the host into the container.
667 * systemd will now read the firmware boot time performance
668 information from the EFI variables, if the used boot loader
669 supports this, and takes it into account for boot performance
670 analysis via "systemd-analyze". This is currently supported
671 only in conjunction with Gummiboot, but could be supported
672 by other boot loaders too. For details see:
674 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
676 * A new generator has been added that automatically mounts the
677 EFI System Partition (ESP) to /boot, if that directory
678 exists, is empty, and no other file system has been
679 configured to be mounted there.
681 * logind will now send out PrepareForSleep(false) out
682 unconditionally, after coming back from suspend. This may be
683 used by applications as asynchronous notification for
684 system resume events.
686 * "systemctl unlock-sessions" has been added, that allows
687 unlocking the screens of all user sessions at once, similar
688 how "systemctl lock-sessions" already locked all users
689 sessions. This is backed by a new D-Bus call UnlockSessions().
691 * "loginctl seat-status" will now show the master device of a
692 seat. (i.e. the device of a seat that needs to be around for
693 the seat to be considered available, usually the graphics
696 * tmpfiles gained a new "X" line type, that allows
697 configuration of files and directories (with wildcards) that
698 shall be excluded from automatic cleanup ("aging").
700 * udev default rules set the device node permissions now only
701 at "add" events, and do not change them any longer with a
702 later "change" event.
704 * The log messages for lid events and power/sleep keypresses
705 now carry a message ID.
707 * We now have a substantially larger unit test suite, but this
708 continues to be work in progress.
710 * udevadm hwdb gained a new --root= parameter to change the
711 root directory to operate relative to.
713 * logind will now issue a background sync() request to the kernel
714 early at shutdown, so that dirty buffers are flushed to disk early
715 instead of at the last moment, in order to optimize shutdown
718 * A new bootctl tool has been added that is an interface for
719 certain boot loader operations. This is currently a preview
720 and is likely to be extended into a small mechanism daemon
721 like timedated, localed, hostnamed, and can be used by
722 graphical UIs to enumerate available boot options, and
723 request boot into firmware operations.
725 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
726 the rest of the package. It also has been updated to work
727 correctly in initrds.
729 * Policykit previously has been runtime optional, and is now
730 also compile time optional via a configure switch.
732 * systemd-analyze has been reimplemented in C. Also "systemctl
733 dot" has moved into systemd-analyze.
735 * "systemctl status" with no further parameters will now print
736 the status of all active or failed units.
738 * Operations such as "systemctl start" can now be executed
739 with a new mode "--irreversible" which may be used to queue
740 operations that cannot accidentally be reversed by a later
741 job queuing. This is by default used to make shutdown
742 requests more robust.
744 * The Python API of systemd now gained a new module for
745 reading journal files.
747 * A new tool kernel-install has been added that can install
748 kernel images according to the Boot Loader Specification:
750 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
752 * Boot time console output has been improved to provide
753 animated boot time output for hanging jobs.
755 * A new tool systemd-activate has been added which can be used
756 to test socket activation with, directly from the command
757 line. This should make it much easier to test and debug
758 socket activation in daemons.
760 * journalctl gained a new "--reverse" (or -r) option to show
761 journal output in reverse order (i.e. newest line first).
763 * journalctl gained a new "--pager-end" (or -e) option to jump
764 to immediately jump to the end of the journal in the
765 pager. This is only supported in conjunction with "less".
767 * journalctl gained a new "--user-unit=" option, that works
768 similar to "--unit=" but filters for user units rather than
771 * A number of unit files to ease adoption of systemd in
772 initrds has been added. This moves some minimal logic from
773 the various initrd implementations into systemd proper.
775 * The journal files are now owned by a new group
776 "systemd-journal", which exists specifically to allow access
777 to the journal, and nothing else. Previously, we used the
778 "adm" group for that, which however possibly covers more
779 than just journal/log file access. This new group is now
780 already used by systemd-journal-gatewayd to ensure this
781 daemon gets access to the journal files and as little else
782 as possible. Note that "make install" will also set FS ACLs
783 up for /var/log/journal to give "adm" and "wheel" read
784 access to it, in addition to "systemd-journal" which owns
785 the journal files. We recommend that packaging scripts also
786 add read access to "adm" + "wheel" to /var/log/journal, and
787 all existing/future journal files. To normal users and
788 administrators little changes, however packagers need to
789 ensure to create the "systemd-journal" system group at
790 package installation time.
792 * The systemd-journal-gatewayd now runs as unprivileged user
793 systemd-journal-gateway:systemd-journal-gateway. Packaging
794 scripts need to create these system user/group at
797 * timedated now exposes a new boolean property CanNTP that
798 indicates whether a local NTP service is available or not.
800 * systemd-detect-virt will now also detect xen PVs
802 * The pstore file system is now mounted by default, if it is
805 * In addition to the SELinux and IMA policies we will now also
806 load SMACK policies at early boot.
808 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
809 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
810 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
811 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
812 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
813 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
814 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
815 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
816 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
817 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
818 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
819 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
820 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
821 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
825 * Timer units now support calendar time events in addition to
826 monotonic time events. That means you can now trigger a unit
827 based on a calendar time specification such as "Thu,Fri
828 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
829 or fifth day of any month of the year 2013, given that it is
830 a thursday or friday. This brings timer event support
831 considerably closer to cron's capabilities. For details on
832 the supported calendar time specification language see
835 * udev now supports a number of different naming policies for
836 network interfaces for predictable names, and a combination
837 of these policies is now the default. Please see this wiki
838 document for details:
840 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
842 * Auke Kok's bootchart implementation has been added to the
843 systemd tree. It's an optional component that can graph the
844 boot in quite some detail. It's one of the best bootchart
845 implementations around and minimal in its code and
848 * nss-myhostname has been integrated into the systemd source
849 tree. nss-myhostname guarantees that the local hostname
850 always stays resolvable via NSS. It has been a weak
851 requirement of systemd-hostnamed since a long time, and
852 since its code is actually trivial we decided to just
853 include it in systemd's source tree. It can be turned off
854 with a configure switch.
856 * The read-ahead logic is now capable of properly detecting
857 whether a btrfs file system is on SSD or rotating media, in
858 order to optimize the read-ahead scheme. Previously, it was
859 only capable of detecting this on traditional file systems
862 * In udev, additional device properties are now read from the
863 IAB in addition to the OUI database. Also, Bluetooth company
864 identities are attached to the devices as well.
866 * In service files %U may be used as specifier that is
867 replaced by the configured user name of the service.
869 * nspawn may now be invoked without a controlling TTY. This
870 makes it suitable for invocation as its own service. This
871 may be used to set up a simple containerized server system
872 using only core OS tools.
874 * systemd and nspawn can now accept socket file descriptors
875 when they are started for socket activation. This enables
876 implementation of socket activated nspawn
877 containers. i.e. think about autospawning an entire OS image
878 when the first SSH or HTTP connection is received. We expect
879 that similar functionality will also be added to libvirt-lxc
882 * journalctl will now suppress ANSI color codes when
885 * systemctl will no longer show control group information for
886 a unit if a the control group is empty anyway.
888 * logind can now automatically suspend/hibernate/shutdown the
891 * /etc/machine-info and hostnamed now also expose the chassis
892 type of the system. This can be used to determine whether
893 the local system is a laptop, desktop, handset or
894 tablet. This information may either be configured by the
895 user/vendor or is automatically determined from ACPI and DMI
896 information if possible.
898 * A number of PolicyKit actions are now bound together with
899 "imply" rules. This should simplify creating UIs because
900 many actions will now authenticate similar ones as well.
902 * Unit files learnt a new condition ConditionACPower= which
903 may be used to conditionalize a unit depending on whether an
904 AC power source is connected or not, of whether the system
905 is running on battery power.
907 * systemctl gained a new "is-failed" verb that may be used in
908 shell scripts and suchlike to check whether a specific unit
909 is in the "failed" state.
911 * The EnvironmentFile= setting in unit files now supports file
912 globbing, and can hence be used to easily read a number of
913 environment files at once.
915 * systemd will no longer detect and recognize specific
916 distributions. All distribution-specific #ifdeffery has been
917 removed, systemd is now fully generic and
918 distribution-agnostic. Effectively, not too much is lost as
919 a lot of the code is still accessible via explicit configure
920 switches. However, support for some distribution specific
921 legacy configuration file formats has been dropped. We
922 recommend distributions to simply adopt the configuration
923 files everybody else uses now and convert the old
924 configuration from packaging scripts. Most distributions
925 already did that. If that's not possible or desirable,
926 distributions are welcome to forward port the specific
927 pieces of code locally from the git history.
929 * When logging a message about a unit systemd will now always
930 log the unit name in the message meta data.
932 * localectl will now also discover system locale data that is
933 not stored in locale archives, but directly unpacked.
935 * logind will no longer unconditionally use framebuffer
936 devices as seat masters, i.e. as devices that are required
937 to be existing before a seat is considered preset. Instead,
938 it will now look for all devices that are tagged as
939 "seat-master" in udev. By default framebuffer devices will
940 be marked as such, but depending on local systems other
941 devices might be marked as well. This may be used to
942 integrate graphics cards using closed source drivers (such
943 as NVidia ones) more nicely into logind. Note however, that
944 we recommend using the open source NVidia drivers instead,
945 and no udev rules for the closed-source drivers will be
946 shipped from us upstream.
948 Contributions from: Adam Williamson, Alessandro Crismani, Auke
949 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
950 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
951 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
952 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
953 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
954 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
955 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
956 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
957 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
958 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
959 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
964 * udev gained support for loading additional device properties
965 from an indexed database that is keyed by vendor/product IDs
966 and similar device identifiers. For the beginning this
967 "hwdb" is populated with data from the well-known PCI and
968 USB database, but also includes PNP, ACPI and OID data. In
969 the longer run this indexed database shall grow into
970 becoming the one central database for non-essential
971 userspace device metadata. Previously, data from the PCI/USB
972 database was only attached to select devices, since the
973 lookup was a relatively expensive operation due to O(n) time
974 complexity (with n being the number of entries in the
975 database). Since this is now O(1), we decided to add in this
976 data for all devices where this is available, by
977 default. Note that the indexed database needs to be rebuilt
978 when new data files are installed. To achieve this you need
979 to update your packaging scripts to invoke "udevadm hwdb
980 --update" after installation of hwdb data files. For
981 RPM-based distributions we introduced the new
982 %udev_hwdb_update macro for this purpose.
984 * The Journal gained support for the "Message Catalog", an
985 indexed database to link up additional information with
986 journal entries. For further details please check:
988 http://www.freedesktop.org/wiki/Software/systemd/catalog
990 The indexed message catalog database also needs to be
991 rebuilt after installation of message catalog files. Use
992 "journalctl --update-catalog" for this. For RPM-based
993 distributions we introduced the %journal_catalog_update
994 macro for this purpose.
996 * The Python Journal bindings gained support for the standard
997 Python logging framework.
999 * The Journal API gained new functions for checking whether
1000 the underlying file system of a journal file is capable of
1001 properly reporting file change notifications, or whether
1002 applications that want to reflect journal changes "live"
1003 need to recheck journal files continuously in appropriate
1006 * It is now possible to set the "age" field for tmpfiles
1007 entries to 0, indicating that files matching this entry
1008 shall always be removed when the directories are cleaned up.
1010 * coredumpctl gained a new "gdb" verb which invokes gdb
1011 right-away on the selected coredump.
1013 * There's now support for "hybrid sleep" on kernels that
1014 support this, in addition to "suspend" and "hibernate". Use
1015 "systemctl hybrid-sleep" to make use of this.
1017 * logind's HandleSuspendKey= setting (and related settings)
1018 now gained support for a new "lock" setting to simply
1019 request the screen lock on all local sessions, instead of
1020 actually executing a suspend or hibernation.
1022 * systemd will now mount the EFI variables file system by
1025 * Socket units now gained support for configuration of the
1026 SMACK security label.
1028 * timedatectl will now output the time of the last and next
1029 daylight saving change.
1031 * We dropped support for various legacy and distro-specific
1032 concepts, such as insserv, early-boot SysV services
1033 (i.e. those for non-standard runlevels such as 'b' or 'S')
1034 or ArchLinux /etc/rc.conf support. We recommend the
1035 distributions who still need support this to either continue
1036 to maintain the necessary patches downstream, or find a
1037 different solution. (Talk to us if you have questions!)
1039 * Various systemd components will now bypass PolicyKit checks
1040 for root and otherwise handle properly if PolicyKit is not
1041 found to be around. This should fix most issues for
1042 PolicyKit-less systems. Quite frankly this should have been
1043 this way since day one. It is absolutely our intention to
1044 make systemd work fine on PolicyKit-less systems, and we
1045 consider it a bug if something doesn't work as it should if
1046 PolicyKit is not around.
1048 * For embedded systems it is now possible to build udev and
1049 systemd without blkid and/or kmod support.
1051 * "systemctl switch-root" is now capable of switching root
1052 more than once. I.e. in addition to transitions from the
1053 initrd to the host OS it is now possible to transition to
1054 further OS images from the host. This is useful to implement
1055 offline updating tools.
1057 * Various other additions have been made to the RPM macros
1058 shipped with systemd. Use %udev_rules_update() after
1059 installing new udev rules files. %_udevhwdbdir,
1060 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
1061 %_sysctldir are now available which resolve to the right
1062 directories for packages to place various data files in.
1064 * journalctl gained the new --full switch (in addition to
1065 --all, to disable ellipsation for long messages.
1067 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
1068 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
1069 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
1070 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
1071 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
1072 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
1073 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
1074 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
1075 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
1079 * journalctl gained new --since= and --until= switches to
1080 filter by time. It also now supports nice filtering for
1081 units via --unit=/-u.
1083 * Type=oneshot services may use ExecReload= and do the
1086 * The journal daemon now supports time-based rotation and
1087 vacuuming, in addition to the usual disk-space based
1090 * The journal will now index the available field values for
1091 each field name. This enables clients to show pretty drop
1092 downs of available match values when filtering. The bash
1093 completion of journalctl has been updated
1094 accordingly. journalctl gained a new switch -F to list all
1095 values a certain field takes in the journal database.
1097 * More service events are now written as structured messages
1098 to the journal, and made recognizable via message IDs.
1100 * The timedated, localed and hostnamed mini-services which
1101 previously only provided support for changing time, locale
1102 and hostname settings from graphical DEs such as GNOME now
1103 also have a minimal (but very useful) text-based client
1104 utility each. This is probably the nicest way to changing
1105 these settings from the command line now, especially since
1106 it lists available options and is fully integrated with bash
1109 * There's now a new tool "systemd-coredumpctl" to list and
1110 extract coredumps from the journal.
1112 * We now install a README each in /var/log/ and
1113 /etc/rc.d/init.d explaining where the system logs and init
1114 scripts went. This hopefully should help folks who go to
1115 that dirs and look into the otherwise now empty void and
1116 scratch their heads.
1118 * When user-services are invoked (by systemd --user) the
1119 $MANAGERPID env var is set to the PID of systemd.
1121 * SIGRTMIN+24 when sent to a --user instance will now result
1122 in immediate termination of systemd.
1124 * gatewayd received numerous feature additions such as a
1125 "follow" mode, for live syncing and filtering.
1127 * browse.html now allows filtering and showing detailed
1128 information on specific entries. Keyboard navigation and
1129 mouse screen support has been added.
1131 * gatewayd/journalctl now supports HTML5/JSON
1132 Server-Sent-Events as output.
1134 * The SysV init script compatibility logic will now
1135 heuristically determine whether a script supports the
1136 "reload" verb, and only then make this available as
1139 * "systemctl status --follow" has been removed, use "journalctl
1142 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
1143 have been removed since they are hardly useful to be
1146 * And I'd like to take the opportunity to specifically mention
1147 Zbigniew for his great contributions. Zbigniew, you rock!
1149 Contributions from: Andrew Eikum, Christian Hesse, Colin
1150 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
1151 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
1152 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
1153 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
1154 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
1155 Jędrzejewski-Szmek, Сковорода Никита Андреевич
1159 * If /etc/vconsole.conf is non-existent or empty we will no
1160 longer load any console font or key map at boot by
1161 default. Instead the kernel defaults will be left
1162 intact. This is definitely the right thing to do, as no
1163 configuration should mean no configuration, and hard-coding
1164 font names that are different on all archs is probably a bad
1165 idea. Also, the kernel default key map and font should be
1166 good enough for most cases anyway, and mostly identical to
1167 the userspace fonts/key maps we previously overloaded them
1168 with. If distributions want to continue to default to a
1169 non-kernel font or key map they should ship a default
1170 /etc/vconsole.conf with the appropriate contents.
1172 Contributions from: Colin Walters, Daniel J Walsh, Dave
1173 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
1174 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1178 * journalctl gained a new --cursor= switch to show entries
1179 starting from the specified location in the journal.
1181 * We now enforce a size limit on journal entry fields exported
1182 with "-o json" in journalctl. Fields larger than 4K will be
1183 assigned null. This can be turned off with --all.
1185 * An (optional) journal gateway daemon is now available as
1186 "systemd-journal-gatewayd.service". This service provides
1187 access to the journal via HTTP and JSON. This functionality
1188 will be used to implement live log synchronization in both
1189 pull and push modes, but has various other users too, such
1190 as easy log access for debugging of embedded devices. Right
1191 now it is already useful to retrieve the journal via HTTP:
1193 # systemctl start systemd-journal-gatewayd.service
1194 # wget http://localhost:19531/entries
1196 This will download the journal contents in a
1197 /var/log/messages compatible format. The same as JSON:
1199 # curl -H"Accept: application/json" http://localhost:19531/entries
1201 This service is also accessible via a web browser where a
1202 single static HTML5 app is served that uses the JSON logic
1203 to enable the user to do some basic browsing of the
1204 journal. This will be extended later on. Here's an example
1205 screenshot of this app in its current state:
1207 http://0pointer.de/public/journal-gatewayd
1209 Contributions from: Kay Sievers, Lennart Poettering, Robert
1210 Milasan, Tom Gundersen
1214 * The bash completion logic is now available for journalctl
1217 * We don't mount the "cpuset" controller anymore together with
1218 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
1219 started if no parameters are assigned to it. "cpuset" hence
1220 broke code that assumed it it could create "cpu" groups and
1223 * journalctl -f will now subscribe to terminal size changes,
1224 and line break accordingly.
1226 Contributions from: Dave Reisner, Kay Sievers, Lennart
1227 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
1231 * nspawn will now create a symlink /etc/localtime in the
1232 container environment, copying the host's timezone
1233 setting. Previously this has been done via a bind mount, but
1234 since symlinks cannot be bind mounted this has now been
1235 changed to create/update the appropriate symlink.
1237 * journalctl -n's line number argument is now optional, and
1238 will default to 10 if omitted.
1240 * journald will now log the maximum size the journal files may
1241 take up on disk. This is particularly useful if the default
1242 built-in logic of determining this parameter from the file
1243 system size is used. Use "systemctl status
1244 systemd-journald.service" to see this information.
1246 * The multi-seat X wrapper tool has been stripped down. As X
1247 is now capable of enumerating graphics devices via udev in a
1248 seat-aware way the wrapper is not strictly necessary
1249 anymore. A stripped down temporary stop-gap is still shipped
1250 until the upstream display managers have been updated to
1251 fully support the new X logic. Expect this wrapper to be
1252 removed entirely in one of the next releases.
1254 * HandleSleepKey= in logind.conf has been split up into
1255 HandleSuspendKey= and HandleHibernateKey=. The old setting
1256 is not available anymore. X11 and the kernel are
1257 distuingishing between these keys and we should too. This
1258 also means the inhibition lock for these keys has been split
1261 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
1262 Poettering, Lukas Nykryn, Václav Pavlín
1266 * Whenever a unit changes state we'll now log this to the
1267 journal and show along the unit's own log output in
1270 * ConditionPathIsMountPoint= can now properly detect bind
1271 mount points too. (Previously, a bind mount of one file
1272 system to another place in the same file system could not be
1273 detected as mount, since they shared struct stat's st_dev
1276 * We will now mount the cgroup controllers cpu, cpuacct,
1277 cpuset and the controllers net_cls, net_prio together by
1280 * nspawn containers will now have a virtualized boot
1281 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
1282 over with a randomized ID at container initialization). This
1283 has the effect of making "journalctl -b" do the right thing
1286 * The JSON output journal serialization has been updated not
1287 to generate "endless" list objects anymore, but rather one
1288 JSON object per line. This is more in line how most JSON
1289 parsers expect JSON objects. The new output mode
1290 "json-pretty" has been added to provide similar output, but
1291 neatly aligned for readability by humans.
1293 * We dropped all explicit sync() invocations in the shutdown
1294 code. The kernel does this implicitly anyway in the kernel
1295 reboot() syscall. halt(8)'s -n option is now a compatibility
1298 * We now support virtualized reboot() in containers, as
1299 supported by newer kernels. We will fall back to exit() if
1300 CAP_SYS_REBOOT is not available to the container. Also,
1301 nspawn makes use of this now and will actually reboot the
1302 container if the containerized OS asks for that.
1304 * journalctl will only show local log output by default
1305 now. Use --merge (-m) to show remote log output, too.
1307 * libsystemd-journal gained the new sd_journal_get_usage()
1308 call to determine the current disk usage of all journal
1309 files. This is exposed in the new "journalctl --disk-usage"
1312 * journald gained a new configuration setting SplitMode= in
1313 journald.conf which may be used to control how user journals
1314 are split off. See journald.conf(5) for details.
1316 * A new condition type ConditionFileNotEmpty= has been added.
1318 * tmpfiles' "w" lines now support file globbing, to write
1319 multiple files at once.
1321 * We added Python bindings for the journal submission
1322 APIs. More Python APIs for a number of selected APIs will
1323 likely follow. Note that we intend to add native bindings
1324 only for the Python language, as we consider it common
1325 enough to deserve bindings shipped within systemd. There are
1326 various projects outside of systemd that provide bindings
1327 for languages such as PHP or Lua.
1329 * Many conditions will now resolve specifiers such as %i. In
1330 addition, PathChanged= and related directives of .path units
1331 now support specifiers as well.
1333 * There's now a new RPM macro definition for the system preset
1336 * journald will now warn if it can't forward a message to the
1337 syslog daemon because it's socket is full.
1339 * timedated will no longer write or process /etc/timezone,
1340 except on Debian. As we do not support late mounted /usr
1341 anymore /etc/localtime always being a symlink is now safe,
1342 and hence the information in /etc/timezone is not necessary
1345 * logind will now always reserve one VT for a text getty (VT6
1346 by default). Previously if more than 6 X sessions where
1347 started they took up all the VTs with auto-spawned gettys,
1348 so that no text gettys were available anymore.
1350 * udev will now automatically inform the btrfs kernel logic
1351 about btrfs RAID components showing up. This should make
1352 simple hotplug based btrfs RAID assembly work.
1354 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
1355 (but not for its children which will stay at the kernel
1356 default). This should allow setups with a lot more listening
1359 * systemd will now always pass the configured timezone to the
1360 kernel at boot. timedated will do the same when the timezone
1363 * logind's inhibition logic has been updated. By default,
1364 logind will now handle the lid switch, the power and sleep
1365 keys all the time, even in graphical sessions. If DEs want
1366 to handle these events on their own they should take the new
1367 handle-power-key, handle-sleep-key and handle-lid-switch
1368 inhibitors during their runtime. A simple way to achiveve
1369 that is to invoke the DE wrapped in an invocation of:
1371 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
1373 * Access to unit operations is now checked via SELinux taking
1374 the unit file label and client process label into account.
1376 * systemd will now notify the administrator in the journal
1377 when he over-mounts a non-empty directory.
1379 * There are new specifiers that are resolved in unit files,
1380 for the host name (%H), the machine ID (%m) and the boot ID
1383 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
1384 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
1385 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
1386 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
1387 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
1388 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
1389 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
1393 * Support for reading structured kernel messages from
1394 /dev/kmsg has now been added and is enabled by default.
1396 * Support for reading kernel messages from /proc/kmsg has now
1397 been removed. If you want kernel messages in the journal
1398 make sure to run a recent kernel (>= 3.5) that supports
1399 reading structured messages from /dev/kmsg (see
1400 above). /proc/kmsg is now exclusive property of classic
1401 syslog daemons again.
1403 * The libudev API gained the new
1404 udev_device_new_from_device_id() call.
1406 * The logic for file system namespace (ReadOnlyDirectory=,
1407 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
1408 require pivot_root() anymore. This means fewer temporary
1409 directories are created below /tmp for this feature.
1411 * nspawn containers will now see and receive all submounts
1412 made on the host OS below the root file system of the
1415 * Forward Secure Sealing is now supported for Journal files,
1416 which provide cryptographical sealing of journal files so
1417 that attackers cannot alter log history anymore without this
1418 being detectable. Lennart will soon post a blog story about
1419 this explaining it in more detail.
1421 * There are two new service settings RestartPreventExitStatus=
1422 and SuccessExitStatus= which allow configuration of exit
1423 status (exit code or signal) which will be excepted from the
1424 restart logic, resp. consider successful.
1426 * journalctl gained the new --verify switch that can be used
1427 to check the integrity of the structure of journal files and
1428 (if Forward Secure Sealing is enabled) the contents of
1431 * nspawn containers will now be run with /dev/stdin, /dev/fd/
1432 and similar symlinks pre-created. This makes running shells
1433 as container init process a lot more fun.
1435 * The fstab support can now handle PARTUUID= and PARTLABEL=
1438 * A new ConditionHost= condition has been added to match
1439 against the hostname (with globs) and machine ID. This is
1440 useful for clusters where a single OS image is used to
1441 provision a large number of hosts which shall run slightly
1442 different sets of services.
1444 * Services which hit the restart limit will now be placed in a
1447 Contributions from: Bertram Poettering, Dave Reisner, Huang
1448 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
1449 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
1453 * When running in --user mode systemd will now become a
1454 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
1455 tree a lot more organized.
1457 * A new PartOf= unit dependency type has been introduced that
1458 may be used to group services in a natural way.
1460 * "systemctl enable" may now be used to enable instances of
1463 * journalctl now prints error log levels in red, and
1464 warning/notice log levels in bright white. It also supports
1465 filtering by log level now.
1467 * cgtop gained a new -n switch (similar to top), to configure
1468 the maximum number of iterations to run for. It also gained
1469 -b, to run in batch mode (accepting no input).
1471 * The suffix ".service" may now be omitted on most systemctl
1472 command lines involving service unit names.
1474 * There's a new bus call in logind to lock all sessions, as
1475 well as a loginctl verb for it "lock-sessions".
1477 * libsystemd-logind.so gained a new call sd_journal_perror()
1478 that works similar to libc perror() but logs to the journal
1479 and encodes structured information about the error number.
1481 * /etc/crypttab entries now understand the new keyfile-size=
1484 * shutdown(8) now can send a (configurable) wall message when
1485 a shutdown is cancelled.
1487 * The mount propagation mode for the root file system will now
1488 default to "shared", which is useful to make containers work
1489 nicely out-of-the-box so that they receive new mounts from
1490 the host. This can be undone locally by running "mount
1491 --make-rprivate /" if needed.
1493 * The prefdm.service file has been removed. Distributions
1494 should maintain this unit downstream if they intend to keep
1495 it around. However, we recommend writing normal unit files
1496 for display managers instead.
1498 * Since systemd is a crucial part of the OS we will now
1499 default to a number of compiler switches that improve
1500 security (hardening) such as read-only relocations, stack
1501 protection, and suchlike.
1503 * The TimeoutSec= setting for services is now split into
1504 TimeoutStartSec= and TimeoutStopSec= to allow configuration
1505 of individual time outs for the start and the stop phase of
1508 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
1509 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
1510 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
1511 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
1512 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
1513 Gundersen, Zbigniew Jędrzejewski-Szmek
1517 * The journal and id128 C APIs are now fully documented as man
1520 * Extra safety checks have been added when transitioning from
1521 the initial RAM disk to the main system to avoid accidental
1524 * /etc/crypttab entries now understand the new keyfile-offset=
1527 * systemctl -t can now be used to filter by unit load state.
1529 * The journal C API gained the new sd_journal_wait() call to
1530 make writing synchronous journal clients easier.
1532 * journalctl gained the new -D switch to show journals from a
1535 * journalctl now displays a special marker between log
1536 messages of two different boots.
1538 * The journal is now explicitly flushed to /var via a service
1539 systemd-journal-flush.service, rather than implicitly simply
1540 by seeing /var/log/journal to be writable.
1542 * journalctl (and the journal C APIs) can now match for much
1543 more complex expressions, with alternatives and
1546 * When transitioning from the initial RAM disk to the main
1547 system we will now kill all processes in a killing spree to
1548 ensure no processes stay around by accident.
1550 * Three new specifiers may be used in unit files: %u, %h, %s
1551 resolve to the user name, user home directory resp. user
1552 shell. This is useful for running systemd user instances.
1554 * We now automatically rotate journal files if their data
1555 object hash table gets a fill level > 75%. We also size the
1556 hash table based on the configured maximum file size. This
1557 together should lower hash collisions drastically and thus
1558 speed things up a bit.
1560 * journalctl gained the new "--header" switch to introspect
1561 header data of journal files.
1563 * A new setting SystemCallFilters= has been added to services
1564 which may be used to apply blacklists or whitelists to
1565 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
1567 * nspawn gained a new --link-journal= switch (and quicker: -j)
1568 to link the container journal with the host. This makes it
1569 very easy to centralize log viewing on the host for all
1570 guests while still keeping the journal files separated.
1572 * Many bugfixes and optimizations
1574 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
1575 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
1576 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
1581 * Several tools now understand kernel command line arguments,
1582 which are only read when run in an initial RAM disk. They
1583 usually follow closely their normal counterparts, but are
1586 * There's a new tool to analyze the readahead files that are
1587 automatically generated at boot. Use:
1589 /usr/lib/systemd/systemd-readahead analyze /.readahead
1591 * We now provide an early debug shell on tty9 if this enabled. Use:
1593 systemctl enable debug-shell.service
1595 * All plymouth related units have been moved into the Plymouth
1596 package. Please make sure to upgrade your Plymouth version
1599 * systemd-tmpfiles now supports getting passed the basename of
1600 a configuration file only, in which case it will look for it
1601 in all appropriate directories automatically.
1603 * udevadm info now takes a /dev or /sys path as argument, and
1604 does the right thing. Example:
1606 udevadm info /dev/sda
1607 udevadm info /sys/class/block/sda
1609 * systemctl now prints a warning if a unit is stopped but a
1610 unit that might trigger it continues to run. Example: a
1611 service is stopped but the socket that activates it is left
1614 * "systemctl status" will now mention if the log output was
1615 shortened due to rotation since a service has been started.
1617 * The journal API now exposes functions to determine the
1618 "cutoff" times due to rotation.
1620 * journald now understands SIGUSR1 and SIGUSR2 for triggering
1621 immediately flushing of runtime logs to /var if possible,
1622 resp. for triggering immediate rotation of the journal
1625 * It is now considered an error if a service is attempted to
1626 be stopped that is not loaded.
1628 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
1630 * systemd-analyze now supports Python 3
1632 * tmpfiles now supports cleaning up directories via aging
1633 where the first level dirs are always kept around but
1634 directories beneath it automatically aged. This is enabled
1635 by prefixing the age field with '~'.
1637 * Seat objects now expose CanGraphical, CanTTY properties
1638 which is required to deal with very fast bootups where the
1639 display manager might be running before the graphics drivers
1640 completed initialization.
1642 * Seat objects now expose a State property.
1644 * We now include RPM macros for service enabling/disabling
1645 based on the preset logic. We recommend RPM based
1646 distributions to make use of these macros if possible. This
1647 makes it simpler to reuse RPM spec files across
1650 * We now make sure that the collected systemd unit name is
1651 always valid when services log to the journal via
1654 * There's a new man page kernel-command-line(7) detailing all
1655 command line options we understand.
1657 * The fstab generator may now be disabled at boot by passing
1658 fstab=0 on the kernel command line.
1660 * A new kernel command line option modules-load= is now understood
1661 to load a specific kernel module statically, early at boot.
1663 * Unit names specified on the systemctl command line are now
1664 automatically escaped as needed. Also, if file system or
1665 device paths are specified they are automatically turned
1666 into the appropriate mount or device unit names. Example:
1668 systemctl status /home
1669 systemctl status /dev/sda
1671 * The SysVConsole= configuration option has been removed from
1672 system.conf parsing.
1674 * The SysV search path is no longer exported on the D-Bus
1677 * The Names= option is been removed from unit file parsing.
1679 * There's a new man page bootup(7) detailing the boot process.
1681 * Every unit and every generator we ship with systemd now
1682 comes with full documentation. The self-explanatory boot is
1685 * A couple of services gained "systemd-" prefixes in their
1686 name if they wrap systemd code, rather than only external
1687 code. Among them fsck@.service which is now
1688 systemd-fsck@.service.
1690 * The HaveWatchdog property has been removed from the D-Bus
1693 * systemd.confirm_spawn= on the kernel command line should now
1696 * There's a new man page crypttab(5) which details all options
1697 we actually understand.
1699 * systemd-nspawn gained a new --capability= switch to pass
1700 additional capabilities to the container.
1702 * timedated will now read known NTP implementation unit names
1703 from /usr/lib/systemd/ntp-units.d/*.list,
1704 systemd-timedated-ntp.target has been removed.
1706 * journalctl gained a new switch "-b" that lists log data of
1707 the current boot only.
1709 * The notify socket is in the abstract namespace again, in
1710 order to support daemons which chroot() at start-up.
1712 * There is a new Storage= configuration option for journald
1713 which allows configuration of where log data should go. This
1714 also provides a way to disable journal logging entirely, so
1715 that data collected is only forwarded to the console, the
1716 kernel log buffer or another syslog implementation.
1718 * Many bugfixes and optimizations
1720 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
1721 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
1722 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
1723 Shawn Landden, Tom Gundersen
1727 * "systemctl help <unit>" now shows the man page if one is
1730 * Several new man pages have been added.
1732 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
1733 MaxLevelConsole= can now be specified in
1734 journald.conf. These options allow reducing the amount of
1735 data stored on disk or forwarded by the log level.
1737 * TimerSlackNSec= can now be specified in system.conf for
1738 PID1. This allows system-wide power savings.
1740 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
1741 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
1746 * logind is now capable of (optionally) handling power and
1747 sleep keys as well as the lid switch.
1749 * journalctl now understands the syntax "journalctl
1750 /usr/bin/avahi-daemon" to get all log output of a specific
1753 * CapabilityBoundingSet= in system.conf now also influences
1754 the capability bound set of usermode helpers of the kernel.
1756 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
1757 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
1758 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
1759 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
1763 * Note that we skipped 139 releases here in order to set the
1764 new version to something that is greater than both udev's
1765 and systemd's most recent version number.
1767 * udev: all udev sources are merged into the systemd source tree now.
1768 All future udev development will happen in the systemd tree. It
1769 is still fully supported to use the udev daemon and tools without
1770 systemd running, like in initramfs or other init systems. Building
1771 udev though, will require the *build* of the systemd tree, but
1772 udev can be properly *run* without systemd.
1774 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
1775 should be used to create dead device nodes as workarounds for broken
1778 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
1779 no longer supported. udev_monitor_new_from_netlink() needs to be
1780 used to subscribe to events.
1782 * udev: when udevd is started by systemd, processes which are left
1783 behind by forking them off of udev rules, are unconditionally cleaned
1784 up and killed now after the event handling has finished. Services or
1785 daemons must be started as systemd services. Services can be
1786 pulled-in by udev to get started, but they can no longer be directly
1787 forked by udev rules.
1789 * udev: the daemon binary is called systemd-udevd now and installed
1790 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
1791 to adapt to that, create symlink, or rename the binary after building
1794 * libudev no longer provides these symbols:
1795 udev_monitor_from_socket()
1796 udev_queue_get_failed_list_entry()
1797 udev_get_{dev,sys,run}_path()
1798 The versions number was bumped and symbol versioning introduced.
1800 * systemd-loginctl and systemd-journalctl have been renamed
1801 to loginctl and journalctl to match systemctl.
1803 * The config files: /etc/systemd/systemd-logind.conf and
1804 /etc/systemd/systemd-journald.conf have been renamed to
1805 logind.conf and journald.conf. Package updates should rename
1806 the files to the new names on upgrade.
1808 * For almost all files the license is now LGPL2.1+, changed
1809 from the previous GPL2.0+. Exceptions are some minor stuff
1810 of udev (which will be changed to LGPL2.1 eventually, too),
1811 and the MIT licensed sd-daemon.[ch] library that is suitable
1812 to be used as drop-in files.
1814 * systemd and logind now handle system sleep states, in
1815 particular suspending and hibernating.
1817 * logind now implements a sleep/shutdown/idle inhibiting logic
1818 suitable for a variety of uses. Soonishly Lennart will blog
1819 about this in more detail.
1821 * var-run.mount and var-lock.mount are no longer provided
1822 (which prevously bind mounted these directories to their new
1823 places). Distributions which have not converted these
1824 directories to symlinks should consider stealing these files
1825 from git history and add them downstream.
1827 * We introduced the Documentation= field for units and added
1828 this to all our shipped units. This is useful to make it
1829 easier to explore the boot and the purpose of the various
1832 * All smaller setup units (such as
1833 systemd-vconsole-setup.service) now detect properly if they
1834 are run in a container and are skipped when
1835 appropriate. This guarantees an entirely noise-free boot in
1836 Linux container environments such as systemd-nspawn.
1838 * A framework for implementing offline system updates is now
1839 integrated, for details see:
1840 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
1842 * A new service type Type=idle is available now which helps us
1843 avoiding ugly interleaving of getty output and boot status
1846 * There's now a system-wide CapabilityBoundingSet= option to
1847 globally reduce the set of capabilities for the
1848 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
1849 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
1850 even CAP_NET_ADMIN system-wide for secure systems.
1852 * There are now system-wide DefaultLimitXXX= options to
1853 globally change the defaults of the various resource limits
1854 for all units started by PID 1.
1856 * Harald Hoyer's systemd test suite has been integrated into
1857 systemd which allows easy testing of systemd builds in qemu
1858 and nspawn. (This is really awesome! Ask us for details!)
1860 * The fstab parser is now implemented as generator, not inside
1863 * systemctl will now warn you if .mount units generated from
1864 /etc/fstab are out of date due to changes in fstab that
1865 haven't been read by systemd yet.
1867 * systemd is now suitable for usage in initrds. Dracut has
1868 already been updated to make use of this. With this in place
1869 initrds get a slight bit faster but primarily are much
1870 easier to introspect and debug since "systemctl status" in
1871 the host system can be used to introspect initrd services,
1872 and the journal from the initrd is kept around too.
1874 * systemd-delta has been added, a tool to explore differences
1875 between user/admin configuration and vendor defaults.
1877 * PrivateTmp= now affects both /tmp and /var/tmp.
1879 * Boot time status messages are now much prettier and feature
1880 proper english language. Booting up systemd has never been
1883 * Read-ahead pack files now include the inode number of all
1884 files to pre-cache. When the inode changes the pre-caching
1885 is not attempted. This should be nicer to deal with updated
1886 packages which might result in changes of read-ahead
1889 * We now temporaritly lower the kernel's read_ahead_kb variable
1890 when collecting read-ahead data to ensure the kernel's
1891 built-in read-ahead does not add noise to our measurements
1892 of necessary blocks to pre-cache.
1894 * There's now RequiresMountsFor= to add automatic dependencies
1895 for all mounts necessary for a specific file system path.
1897 * MountAuto= and SwapAuto= have been removed from
1898 system.conf. Mounting file systems at boot has to take place
1901 * nspawn now learned a new switch --uuid= to set the machine
1902 ID on the command line.
1904 * nspawn now learned the -b switch to automatically search
1907 * vt102 is now the default TERM for serial TTYs, upgraded from
1910 * systemd-logind now works on VT-less systems.
1912 * The build tree has been reorganized. The individual
1913 components now have directories of their own.
1915 * A new condition type ConditionPathIsReadWrite= is now available.
1917 * nspawn learned the new -C switch to create cgroups for the
1918 container in other hierarchies.
1920 * We now have support for hardware watchdogs, configurable in
1923 * The scheduled shutdown logic now has a public API.
1925 * We now mount /tmp as tmpfs by default, but this can be
1926 masked and /etc/fstab can override it.
1928 * Since udisks doesn't make use of /media anymore we are not
1929 mounting a tmpfs on it anymore.
1931 * journalctl gained a new --local switch to only interleave
1932 locally generated journal files.
1934 * We can now load the IMA policy at boot automatically.
1936 * The GTK tools have been split off into a systemd-ui.
1938 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
1939 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
1940 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
1941 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
1942 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
1943 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
1944 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
1945 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
1946 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
1951 * This is mostly a bugfix release
1953 * Support optional initialization of the machine ID from the
1954 KVM or container configured UUID.
1956 * Support immediate reboots with "systemctl reboot -ff"
1958 * Show /etc/os-release data in systemd-analyze output
1960 * Many bugfixes for the journal, including endianness fixes and
1961 ensuring that disk space enforcement works
1963 * sd-login.h is C++ comptaible again
1965 * Extend the /etc/os-release format on request of the Debian
1968 * We now refuse non-UTF8 strings used in various configuration
1969 and unit files. This is done to ensure we don't pass invalid
1970 data over D-Bus or expose it elsewhere.
1972 * Register Mimo USB Screens as suitable for automatic seat
1975 * Read SELinux client context from journal clients in a race
1978 * Reorder configuration file lookup order. /etc now always
1979 overrides /run in order to allow the administrator to always
1980 and unconditionally override vendor supplied or
1981 automatically generated data.
1983 * The various user visible bits of the journal now have man
1984 pages. We still lack man pages for the journal API calls
1987 * We now ship all man pages in HTML format again in the
1990 Contributions from: Dave Reisner, Dirk Eibach, Frederic
1991 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
1992 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
1997 * This is mostly a bugfix release
1999 * systems lacking /etc/os-release are no longer supported.
2001 * Various functionality updates to libsystemd-login.so
2003 * Track class of PAM logins to distuingish greeters from
2006 Contributions from: Kay Sievers, Lennart Poettering, Michael
2011 * This is an important bugfix release for v41.
2013 * Building man pages is now optional which should be useful
2014 for those building systemd from git but unwilling to install
2017 * Watchdog support for supervising services is now usable. In
2018 a future release support for hardware watchdogs
2019 (i.e. /dev/watchdog) will be added building on this.
2021 * Service start rate limiting is now configurable and can be
2022 turned off per service. When a start rate limit is hit a
2023 reboot can automatically be triggered.
2025 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
2027 Contributions from: Benjamin Franzke, Bill Nottingham,
2028 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
2029 Schmidt, Michał Górny, Piotr Drąg
2033 * The systemd binary is installed /usr/lib/systemd/systemd now;
2034 An existing /sbin/init symlink needs to be adapted with the
2037 * The code that loads kernel modules has been ported to invoke
2038 libkmod directly, instead of modprobe. This means we do not
2039 support systems with module-init-tools anymore.
2041 * Watchdog support is now already useful, but still not
2044 * A new kernel command line option systemd.setenv= is
2045 understood to set system wide environment variables
2046 dynamically at boot.
2048 * We now limit the set of capabilities of systemd-journald.
2050 * We now set SIGPIPE to ignore by default, since it only is
2051 useful in shell pipelines, and has little use in general
2052 code. This can be disabled with IgnoreSIPIPE=no in unit
2055 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
2056 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
2061 * This is mostly a bugfix release
2063 * We now expose the reason why a service failed in the
2064 "Result" D-Bus property.
2066 * Rudimentary service watchdog support (will be completed over
2067 the next few releases.)
2069 * When systemd forks off in order execute some service we will
2070 now immediately changes its argv[0] to reflect which process
2071 it will execute. This is useful to minimize the time window
2072 with a generic argv[0], which makes bootcharts more useful
2074 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
2075 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
2076 Mike Kazantsev, Ray Strode
2080 * This is mostly a test release, but incorporates many
2083 * New systemd-cgtop tool to show control groups by their
2086 * Linking against libacl for ACLs is optional again. If
2087 disabled, support tracking device access for active logins
2088 goes becomes unavailable, and so does access to the user
2089 journals by the respective users.
2091 * If a group "adm" exists, journal files are automatically
2092 owned by them, thus allow members of this group full access
2093 to the system journal as well as all user journals.
2095 * The journal now stores the SELinux context of the logging
2096 client for all entries.
2098 * Add C++ inclusion guards to all public headers
2100 * New output mode "cat" in the journal to print only text
2101 messages, without any meta data like date or time.
2103 * Include tiny X server wrapper as a temporary stop-gap to
2104 teach XOrg udev display enumeration. This is used by display
2105 managers such as gdm, and will go away as soon as XOrg
2106 learned native udev hotplugging for display devices.
2108 * Add new systemd-cat tool for executing arbitrary programs
2109 with STDERR/STDOUT connected to the journal. Can also act as
2110 BSD logger replacement, and does so by default.
2112 * Optionally store all locally generated coredumps in the
2113 journal along with meta data.
2115 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
2116 writing short strings to files (for usage for /sys), and for
2117 creating symlinks, character and block device nodes.
2119 * New unit file option ControlGroupPersistent= to make cgroups
2120 persistent, following the mechanisms outlined in
2121 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
2123 * Support multiple local RTCs in a sane way
2125 * No longer monopolize IO when replaying readahead data on
2126 rotating disks, since we might starve non-file-system IO to
2127 death, since fanotify() will not see accesses done by blkid,
2130 * Don't show kernel threads in systemd-cgls anymore, unless
2131 requested with new -k switch.
2133 Contributions from: Dan Horák, Kay Sievers, Lennart
2134 Poettering, Michal Schmidt
2138 * This is mostly a test release, but incorporates many
2141 * The git repository moved to:
2142 git://anongit.freedesktop.org/systemd/systemd
2143 ssh://git.freedesktop.org/git/systemd/systemd
2145 * First release with the journal
2146 http://0pointer.de/blog/projects/the-journal.html
2148 * The journal replaces both systemd-kmsg-syslogd and
2149 systemd-stdout-bridge.
2151 * New sd_pid_get_unit() API call in libsystemd-logind
2153 * Many systemadm clean-ups
2155 * Introduce remote-fs-pre.target which is ordered before all
2156 remote mounts and may be used to start services before all
2159 * Added Mageia support
2161 * Add bash completion for systemd-loginctl
2163 * Actively monitor PID file creation for daemons which exit in
2164 the parent process before having finished writing the PID
2165 file in the daemon process. Daemons which do this need to be
2166 fixed (i.e. PID file creation must have finished before the
2167 parent exits), but we now react a bit more gracefully to them.
2169 * Add colourful boot output, mimicking the well-known output
2170 of existing distributions.
2172 * New option PassCredentials= for socket units, for
2173 compatibility with a recent kernel ABI breakage.
2175 * /etc/rc.local is now hooked in via a generator binary, and
2176 thus will no longer act as synchronization point during
2179 * systemctl list-unit-files now supports --root=.
2181 * systemd-tmpfiles now understands two new commands: z, Z for
2182 relabelling files according to the SELinux database. This is
2183 useful to apply SELinux labels to specific files in /sys,
2186 * Output of SysV services is now forwarded to both the console
2187 and the journal by default, not only just the console.
2189 * New man pages for all APIs from libsystemd-login.
2191 * The build tree got reorganized and a the build system is a
2192 lot more modular allowing embedded setups to specifically
2193 select the components of systemd they are interested in.
2195 * Support for Linux systems lacking the kernel VT subsystem is
2198 * configure's --with-rootdir= got renamed to
2199 --with-rootprefix= to follow the naming used by udev and
2202 * Unless specified otherwise we'll now install to /usr instead
2203 of /usr/local by default.
2205 * Processes with '@' in argv[0][0] are now excluded from the
2206 final shut-down killing spree, following the logic explained
2208 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
2210 * All processes remaining in a service cgroup when we enter
2211 the START or START_PRE states are now killed with
2212 SIGKILL. That means it is no longer possible to spawn
2213 background processes from ExecStart= lines (which was never
2214 supported anyway, and bad style).
2216 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
2217 reloading of units together.
2219 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
2220 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
2221 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
2222 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
2223 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
~ianmdlvl / elogind.git / blob