1 systemd System and Service Manager
5 * The Restart= option for services now understands a new
6 on-watchdog setting, which will restart the service
7 automatically if the service stops sending out watchdog keep
8 alive messages (as configured with WatchdogSec=).
10 * The getty generator (which is responsible for bringing up a
11 getty on configured serial consoles) will no longer only
12 start a getty on the primary kernel console but on all
13 others, too. This makes the order in which console= is
14 specified on the kernel command line less important.
16 * libsystemd-logind gained a new sd_session_get_vt() call to
17 retrieve the VT number of a session.
19 * If the option "tries=0" is set for an entry of /etc/crypttab
20 its passphrase is queried indefinitely instead of any
21 maximum number of tries.
23 * If a service with a configure PID file terminates its PID
24 file will now be removed automatically if it still exists
25 afterwards. This should put an end to stale PID files.
27 * systemd-run will now also take relative binary path names
28 for execution and no longer insists on absolute paths.
30 * InaccessibleDirectories= and ReadOnlyDirectories= now take
31 paths that are optionally prefixed with "-" to indicate that
32 it should not be considered a failure if they don't exist.
34 * journalctl -o (and similar commands) now understands a new
35 output mode "short-precise", it is similar to "short" but
36 shows timestamps with usec accuracy.
38 * The option "discard" (as known from Debian) is now
39 synonymous to "allow-discards" in /etc/crypttab. In fact,
40 "discard" is preferred now (since it is easier to remember
43 * Some licensing clean-ups were made, so that more code is now
44 LGPL-2.1 licensed than before.
46 * A minimal tool to save/restore the display backlight
47 brightness across reboots has been added. It will store the
48 backlight setting as late as possible at shutdown, and
49 restore it as early as possible during reboot.
51 * A logic to automatically discover and enable home and swap
52 partitions on GPT disks has been added. With this in place
53 /etc/fstab becomes optional for many setups as systemd can
54 discover certain partitions located on the root disk
55 automatically. Home partitions are recognized under their
56 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
57 partitions are recognized under their GPT type ID
58 0657fd6da4ab43c484e50933c84b4f4f.
60 * systemd will no longer pass any environment from the kernel
61 or initrd to system services. If you want to set an
62 environment for all services, do so via the kernel command
63 line systemd.setenv= assignment.
65 * The systemd-sysctl tool no longer natively reads the file
66 /etc/sysctl.conf. If desired, the file should be symlinked
67 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
68 legacy support by a symlink rather than built-in code, it
69 also makes the otherwise hidden order of application of the
70 different files visible. (Note that this partly reverts to a
71 pre-198 application order of sysctl knobs!)
73 * The "systemctl set-log-level" and "systemctl dump" commands
74 have been moved to systemd-analyze.
76 * systemd-run learned the new --remain-after-exit switch,
77 which causes the scope unit not to be cleaned up
78 automatically after the process terminated.
80 * tmpfiles learned a new --exclude-prefix= switch to exclude
81 certain paths from operation.
83 * journald will now automatically flush all messages to disk
84 as soon as a message of the log priorities CRIT, ALERT or
87 Contributions from: Andrew Cook, Brandon Philips, Christian
88 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
89 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
90 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
91 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
92 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
93 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
94 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
95 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
96 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
97 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
98 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
99 William Giokas, Zbigniew Jędrzejewski-Szmek
101 -- Berlin, 2013-09-13
105 * The documentation has been updated to cover the various new
106 concepts introduced with 205.
108 * Unit files now understand the new %v specifier which
109 resolves to the kernel version string as returned by "uname
112 * systemctl now supports filtering the unit list output by
113 load state, active state and sub state, using the new
116 * "systemctl status" will now show the results of the
117 condition checks (like ConditionPathExists= and similar) of
118 the last start attempts of the unit. They are also logged to
121 * "journalctl -b" may now be used to look for boot output of a
122 specific boot. Try "journalctl -b -1" for the previous boot,
123 but the syntax is substantially more powerful.
125 * "journalctl --show-cursor" has been added which prints the
126 cursor string the last shown log line. This may then be used
127 with the new "journalctl --after-cursor=" switch to continue
128 browsing logs from that point on.
130 * "journalctl --force" may now be used to force regeneration
133 * Creation of "dead" device nodes has been moved from udev
134 into kmod and tmpfiles. Previously, udev would read the kmod
135 databases to pre-generate dead device nodes based on meta
136 information contained in kernel modules, so that these would
137 be auto-loaded on access rather then at boot. As this
138 doesn't really have much to do with the exposing actual
139 kernel devices to userspace this has always been slightly
140 alien in the udev codebase. Following the new scheme kmod
141 will now generate a runtime snippet for tmpfiles from the
142 module meta information and it now is tmpfiles' job to the
143 create the nodes. This also allows overriding access and
144 other parameters for the nodes using the usual tmpfiles
145 facilities. As side effect this allows us to remove the
146 CAP_SYS_MKNOD capability bit from udevd entirely.
148 * logind's device ACLs may now be applied to these "dead"
149 devices nodes too, thus finally allowing managed access to
150 devices such as /dev/snd/sequencer whithout loading the
151 backing module right-away.
153 * A new RPM macro has been added that may be used to apply
154 tmpfiles configuration during package installation.
156 * systemd-detect-virt and ConditionVirtualization= now can
157 detect User-Mode-Linux machines (UML).
159 * journald will now implicitly log the effective capabilities
160 set of processes in the message metadata.
162 * systemd-cryptsetup has gained support for TrueCrypt volumes.
164 * The initrd interface has been simplified (more specifically,
165 support for passing performance data via environment
166 variables and fsck results via files in /run has been
167 removed). These features were non-essential, and are
168 nowadays available in a much nicer way by having systemd in
169 the initrd serialize its state and have the hosts systemd
170 deserialize it again.
172 * The udev "keymap" data files and tools to apply keyboard
173 specific mappings of scan to key codes, and force-release
174 scan code lists have been entirely replaced by a udev
175 "keyboard" builtin and a hwdb data file.
177 * systemd will now honour the kernel's "quiet" command line
178 argument also during late shutdown, resulting in a
179 completely silent shutdown when used.
181 * There's now an option to control the SO_REUSEPORT socket
182 option in .socket units.
184 * Instance units will now automatically get a per-template
185 subslice of system.slice unless something else is explicitly
186 configured. For example, instances of sshd@.service will now
187 implicitly be placed in system-sshd.slice rather than
188 system.slice as before.
190 * Test coverage support may now be enabled at build time.
192 Contributions from: Dave Reisner, Frederic Crozat, Harald
193 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
194 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
195 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
196 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
197 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
198 Giokas, Zbigniew Jędrzejewski-Szmek
200 -- Berlin, 2013-07-23
204 * Two new unit types have been introduced:
206 Scope units are very similar to service units, however, are
207 created out of pre-existing processes -- instead of PID 1
208 forking off the processes. By using scope units it is
209 possible for system services and applications to group their
210 own child processes (worker processes) in a powerful way
211 which then maybe used to organize them, or kill them
212 together, or apply resource limits on them.
214 Slice units may be used to partition system resources in an
215 hierarchial fashion and then assign other units to them. By
216 default there are now three slices: system.slice (for all
217 system services), user.slice (for all user sessions),
218 machine.slice (for VMs and containers).
220 Slices and scopes have been introduced primarily in
221 context of the work to move cgroup handling to a
222 single-writer scheme, where only PID 1
223 creates/removes/manages cgroups.
225 * There's a new concept of "transient" units. In contrast to
226 normal units these units are created via an API at runtime,
227 not from configuration from disk. More specifically this
228 means it is now possible to run arbitrary programs as
229 independent services, with all execution parameters passed
230 in via bus APIs rather than read from disk. Transient units
231 make systemd substantially more dynamic then it ever was,
232 and useful as a general batch manager.
234 * logind has been updated to make use of scope and slice units
235 for managing user sessions. As a user logs in he will get
236 his own private slice unit, to which all sessions are added
237 as scope units. We also added support for automatically
238 adding an instance of user@.service for the user into the
239 slice. Effectively logind will no longer create cgroup
240 hierarchies on its own now, it will defer entirely to PID 1
241 for this by means of scope, service and slice units. Since
242 user sessions this way become entities managed by PID 1
243 the output of "systemctl" is now a lot more comprehensive.
245 * A new mini-daemon "systemd-machined" has been added which
246 may be used by virtualization managers to register local
247 VMs/containers. nspawn has been updated accordingly, and
248 libvirt will be updated shortly. machined will collect a bit
249 of meta information about the VMs/containers, and assign
250 them their own scope unit (see above). The collected
251 meta-data is then made available via the "machinectl" tool,
252 and exposed in "ps" and similar tools. machined/machinectl
253 is compile-time optional.
255 * As discussed earlier, the low-level cgroup configuration
256 options ControlGroup=, ControlGroupModify=,
257 ControlGroupPersistent=, ControlGroupAttribute= have been
258 removed. Please use high-level attribute settings instead as
261 * A new bus call SetUnitProperties() has been added to alter
262 various runtime parameters of a unit. This is primarily
263 useful to alter cgroup parameters dynamically in a nice way,
264 but will be extended later on to make more properties
265 modifiable at runtime. systemctl gained a new set-properties
266 command that wraps this call.
268 * A new tool "systemd-run" has been added which can be used to
269 run arbitrary command lines as transient services or scopes,
270 while configuring a number of settings via the command
271 line. This tool is currently very basic, however already
272 very useful. We plan to extend this tool to even allow
273 queuing of execution jobs with time triggers from the
274 command line, similar in fashion to "at".
276 * nspawn will now inform the user explicitly that kernels with
277 audit enabled break containers, and suggest the user to turn
280 * Support for detecting the IMA and AppArmor security
281 frameworks with ConditionSecurity= has been added.
283 * journalctl gained a new "-k" switch for showing only kernel
284 messages, mimicking dmesg output; in addition to "--user"
285 and "--system" switches for showing only user's own logs
288 * systemd-delta can now show information about drop-in
289 snippets extending unit files.
291 * libsystemd-bus has been substantially updated but is still
292 not available as public API.
294 * systemd will now look for the "debug" argument on the kernel
295 command line and enable debug logging, similar to
296 "systemd.log_level=debug" already did before.
298 * "systemctl set-default", "systemctl get-default" has been
299 added to configure the default.target symlink, which
300 controls what to boot into by default.
302 * "systemctl set-log-level" has been added as a convenient
303 way to raise and lower systemd logging threshold.
305 * "systemd-analyze plot" will now show the time the various
306 generators needed for execution, as well as information
307 about the unit file loading.
309 * libsystemd-journal gained a new sd_journal_open_files() call
310 for opening specific journal files. journactl also gained a
311 new switch to expose this new functionality. Previously we
312 only supported opening all files from a directory, or all
313 files from the system, as opening individual files only is
314 racy due to journal file rotation.
316 * systemd gained the new DefaultEnvironment= setting in
317 /etc/systemd/system.conf to set environment variables for
320 * If a privileged process logs a journal message with the
321 OBJECT_PID= field set, then journald will automatically
322 augment this with additional OBJECT_UID=, OBJECT_GID=,
323 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
324 system services want to log events about specific client
325 processes. journactl/systemctl has been updated to make use
326 of this information if all log messages regarding a specific
329 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
330 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
331 Reisner, David Coppa, David King, David Strauss, Eelco
332 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
333 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
334 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
335 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
336 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
337 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
338 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
339 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
340 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
345 * The Python bindings gained some minimal support for the APIs
346 exposed by libsystemd-logind.
348 * ConditionSecurity= gained support for detecting SMACK. Since
349 this condition already supports SELinux and AppArmor we only
350 miss IMA for this. Patches welcome!
352 Contributions from: Karol Lewandowski, Lennart Poettering,
353 Zbigniew Jędrzejewski-Szmek
357 * systemd-nspawn will now create /etc/resolv.conf if
358 necessary, before bind-mounting the host's file onto it.
360 * systemd-nspawn will now store meta information about a
361 container on the container's cgroup as extended attribute
362 fields, including the root directory.
364 * The cgroup hierarchy has been reworked in many ways. All
365 objects any of the components systemd creates in the cgroup
366 tree are now suffixed. More specifically, user sessions are
367 now placed in cgroups suffixed with ".session", users in
368 cgroups suffixed with ".user", and nspawn containers in
369 cgroups suffixed with ".nspawn". Furthermore, all cgroup
370 names are now escaped in a simple scheme to avoid collision
371 of userspace object names with kernel filenames. This work
372 is preparation for making these objects relocatable in the
373 cgroup tree, in order to allow easy resource partitioning of
374 these objects without causing naming conflicts.
376 * systemctl list-dependencies gained the new switches
377 --plain, --reverse, --after and --before.
379 * systemd-inhibit now shows the process name of processes that
380 have taken an inhibitor lock.
382 * nss-myhostname will now also resolve "localhost"
383 implicitly. This makes /etc/hosts an optional file and
384 nicely handles that on IPv6 ::1 maps to both "localhost" and
387 * libsystemd-logind.so gained a new call
388 sd_get_machine_names() to enumerate running containers and
389 VMs (currently only supported by very new libvirt and
390 nspawn). sd_login_monitor can now be used to watch
391 VMs/containers coming and going.
393 * .include is not allowed recursively anymore, and only in
394 unit files. Usually it is better to use drop-in snippets in
395 .d/*.conf anyway, as introduced with systemd 198.
397 * systemd-analyze gained a new "critical-chain" command that
398 determines the slowest chain of units run during system
399 boot-up. It is very useful for tracking down where
400 optimizing boot time is the most beneficial.
402 * systemd will no longer allow manipulating service paths in
403 the name=systemd:/system cgroup tree using ControlGroup= in
404 units. (But is still fine with it in all other dirs.)
406 * There's a new systemd-nspawn@.service service file that may
407 be used to easily run nspawn containers as system
408 services. With the container's root directory in
409 /var/lib/container/foobar it is now sufficient to run
410 "systemctl start systemd-nspawn@foobar.service" to boot it.
412 * systemd-cgls gained a new parameter "--machine" to list only
413 the processes within a certain container.
415 * ConditionSecurity= now can check for "apparmor". We still
416 are lacking checks for SMACK and IMA for this condition
417 check though. Patches welcome!
419 * A new configuration file /etc/systemd/sleep.conf has been
420 added that may be used to configure which kernel operation
421 systemd is supposed to execute when "suspend", "hibernate"
422 or "hybrid-sleep" is requested. This makes the new kernel
423 "freeze" state accessible to the user.
425 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
426 the passed argument if applicable.
428 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
429 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
430 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
431 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
432 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
433 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
434 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
439 * The output of 'systemctl list-jobs' got some polishing. The
440 '--type=' argument may now be passed more than once. A new
441 command 'systemctl list-sockets' has been added which shows
442 a list of kernel sockets systemd is listening on with the
443 socket units they belong to, plus the units these socket
446 * The experimental libsystemd-bus library got substantial
447 updates to work in conjunction with the (also experimental)
448 kdbus kernel project. It works well enough to exchange
449 messages with some sophistication. Note that kdbus is not
450 ready yet, and the library is mostly an elaborate test case
451 for now, and not installable.
453 * systemd gained a new unit 'systemd-static-nodes.service'
454 that generates static device nodes earlier during boot, and
455 can run in conjunction with udev.
457 * libsystemd-login gained a new call sd_pid_get_user_unit()
458 to retrieve the user systemd unit a process is running
459 in. This is useful for systems where systemd is used as
462 * systemd-nspawn now places all containers in the new /machine
463 top-level cgroup directory in the name=systemd
464 hierarchy. libvirt will soon do the same, so that we get a
465 uniform separation of /system, /user and /machine for system
466 services, user processes and containers/virtual
467 machines. This new cgroup hierarchy is also useful to stick
468 stable names to specific container instances, which can be
469 recognized later this way (this name may be controlled
470 via systemd-nspawn's new -M switch). libsystemd-login also
471 gained a new call sd_pid_get_machine_name() to retrieve the
472 name of the container/VM a specific process belongs to.
474 * bootchart can now store its data in the journal.
476 * libsystemd-journal gained a new call
477 sd_journal_add_conjunction() for AND expressions to the
478 matching logic. This can be used to express more complex
481 * journactl can now take multiple --unit= and --user-unit=
484 * The cryptsetup logic now understands the "luks.key=" kernel
485 command line switch for specifying a file to read the
486 decryption key from. Also, if a configured key file is not
487 found the tool will now automatically fall back to prompting
490 * Python systemd.journal module was updated to wrap recently
491 added functions from libsystemd-journal. The interface was
492 changed to bring the low level interface in s.j._Reader
493 closer to the C API, and the high level interface in
494 s.j.Reader was updated to wrap and convert all data about
497 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
498 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
499 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
500 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
501 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
502 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
506 * journalctl --update-catalog now understands a new --root=
507 option to operate on catalogs found in a different root
510 * During shutdown after systemd has terminated all running
511 services a final killing loop kills all remaining left-over
512 processes. We will now print the name of these processes
513 when we send SIGKILL to them, since this usually indicates a
516 * If /etc/crypttab refers to password files stored on
517 configured mount points automatic dependencies will now be
518 generated to ensure the specific mount is established first
519 before the key file is attempted to be read.
521 * 'systemctl status' will now show information about the
522 network sockets a socket unit is listening on.
524 * 'systemctl status' will also shown information about any
525 drop-in configuration file for units. (Drop-In configuration
526 files in this context are files such as
527 /etc/systemd/systemd/foobar.service.d/*.conf)
529 * systemd-cgtop now optionally shows summed up CPU times of
530 cgroups. Press '%' while running cgtop to switch between
531 percentage and absolute mode. This is useful to determine
532 which cgroups use up the most CPU time over the entire
533 runtime of the system. systemd-cgtop has also been updated
534 to be 'pipeable' for processing with further shell tools.
536 * 'hostnamectl set-hostname' will now allow setting of FQDN
539 * The formatting and parsing of time span values has been
540 changed. The parser now understands fractional expressions
541 such as "5.5h". The formatter will now output fractional
542 expressions for all time spans under 1min, i.e. "5.123456s"
543 rather than "5s 123ms 456us". For time spans under 1s
544 millisecond values are shown, for those under 1ms
545 microsecond values are shown. This should greatly improve
546 all time-related output of systemd.
548 * libsystemd-login and libsystemd-journal gained new
549 functions for querying the poll() events mask and poll()
550 timeout value for integration into arbitrary event
553 * localectl gained the ability to list available X11 keymaps
554 (models, layouts, variants, options).
556 * 'systemd-analyze dot' gained the ability to filter for
557 specific units via shell-style globs, to create smaller,
558 more useful graphs. I.e. it's now possible to create simple
559 graphs of all the dependencies between only target units, or
560 of all units that Avahi has dependencies with.
562 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
563 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
564 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
565 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
566 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
567 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
568 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
572 * The boot-time readahead implementation for rotating media
573 will now read the read-ahead data in multiple passes which
574 consist of all read requests made in equidistant time
575 intervals. This means instead of strictly reading read-ahead
576 data in its physical order on disk we now try to find a
577 middle ground between physical and access time order.
579 * /etc/os-release files gained a new BUILD_ID= field for usage
580 on operating systems that provide continuous builds of OS
583 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
584 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
585 William Douglas, Zbigniew Jędrzejewski-Szmek
589 * systemd-python gained an API exposing libsystemd-daemon.
591 * The SMACK setup logic gained support for uploading CIPSO
594 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
595 ReadOnlyDirectories= and InaccessibleDirectories= has
596 changed. The private /tmp and /var/tmp directories are now
597 shared by all processes of a service (which means
598 ExecStartPre= may now leave data in /tmp that ExecStart= of
599 the same service can still access). When a service is
600 stopped its temporary directories are immediately deleted
601 (normal clean-up with tmpfiles is still done in addition to
604 * By default, systemd will now set a couple of sysctl
605 variables in the kernel: the safe sysrq options are turned
606 on, IP route verification is turned on, and source routing
607 disabled. The recently added hardlink and softlink
608 protection of the kernel is turned on. These settings should
609 be reasonably safe, and good defaults for all new systems.
611 * The predictable network naming logic may now be turned off
612 with a new kernel command line switch: net.ifnames=0.
614 * A new libsystemd-bus module has been added that implements a
615 pretty complete D-Bus client library. For details see:
617 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
619 * journald will now explicitly flush the journal files to disk
620 at the latest 5min after each write. The file will then also
621 be marked offline until the next write. This should increase
622 reliability in case of a crash. The synchronization delay
623 can be configured via SyncIntervalSec= in journald.conf.
625 * There's a new remote-fs-setup.target unit that can be used
626 to pull in specific services when at least one remote file
627 system is to be mounted.
629 * There are new targets timers.target and paths.target as
630 canonical targets to pull user timer and path units in
631 from. This complements sockets.target with a similar
632 purpose for socket units.
634 * libudev gained a new call udev_device_set_attribute_value()
635 to set sysfs attributes of a device.
637 * The udev daemon now sets the default number of worker
638 processes executed in parallel based on the number of available
639 CPUs instead of the amount of available RAM. This is supposed
640 to provide a more reliable default and limit a too aggressive
641 paralellism for setups with 1000s of devices connected.
643 Contributions from: Auke Kok, Colin Walters, Cristian
644 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
645 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
646 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
647 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
648 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
649 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
650 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
651 Zbigniew Jędrzejewski-Szmek
655 * Configuration of unit files may now be extended via drop-in
656 files without having to edit/override the unit files
657 themselves. More specifically, if the administrator wants to
658 change one value for a service file foobar.service he can
659 now do so by dropping in a configuration snippet into
660 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
661 will load all these snippets and apply them on top of the
662 main unit configuration file, possibly extending or
663 overriding its settings. Using these drop-in snippets is
664 generally nicer than the two earlier options for changing
665 unit files locally: copying the files from
666 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
667 them there; or creating a new file in /etc/systemd/system/
668 that incorporates the original one via ".include". Drop-in
669 snippets into these .d/ directories can be placed in any
670 directory systemd looks for units in, and the usual
671 overriding semantics between /usr/lib, /etc and /run apply
674 * Most unit file settings which take lists of items can now be
675 reset by assigning the empty string to them. For example,
676 normally, settings such as Environment=FOO=BAR append a new
677 environment variable assignment to the environment block,
678 each time they are used. By assigning Environment= the empty
679 string the environment block can be reset to empty. This is
680 particularly useful with the .d/*.conf drop-in snippets
681 mentioned above, since this adds the ability to reset list
682 settings from vendor unit files via these drop-ins.
684 * systemctl gained a new "list-dependencies" command for
685 listing the dependencies of a unit recursively.
687 * Inhibitors are now honored and listed by "systemctl
688 suspend", "systemctl poweroff" (and similar) too, not only
689 GNOME. These commands will also list active sessions by
692 * Resource limits (as exposed by the various control group
693 controllers) can now be controlled dynamically at runtime
694 for all units. More specifically, you can now use a command
695 like "systemctl set-cgroup-attr foobar.service cpu.shares
696 2000" to alter the CPU shares a specific service gets. These
697 settings are stored persistently on disk, and thus allow the
698 administrator to easily adjust the resource usage of
699 services with a few simple commands. This dynamic resource
700 management logic is also available to other programs via the
701 bus. Almost any kernel cgroup attribute and controller is
704 * systemd-vconsole-setup will now copy all font settings to
705 all allocated VTs, where it previously applied them only to
708 * libsystemd-login gained the new sd_session_get_tty() API
711 * This release drops support for a few legacy or
712 distribution-specific LSB facility names when parsing init
713 scripts: $x-display-manager, $mail-transfer-agent,
714 $mail-transport-agent, $mail-transfer-agent, $smtp,
715 $null. Also, the mail-transfer-agent.target unit backing
716 this has been removed. Distributions which want to retain
717 compatibility with this should carry the burden for
718 supporting this themselves and patch support for these back
719 in, if they really need to. Also, the facilities $syslog and
720 $local_fs are now ignored, since systemd does not support
721 early-boot LSB init scripts anymore, and these facilities
722 are implied anyway for normal services. syslog.target has
725 * There are new bus calls on PID1's Manager object for
726 cancelling jobs, and removing snapshot units. Previously,
727 both calls were only available on the Job and Snapshot
730 * systemd-journal-gatewayd gained SSL support.
732 * The various "environment" files, such as /etc/locale.conf
733 now support continuation lines with a backslash ("\") as
734 last character in the line, similar in style (but different)
735 to how this is supported in shells.
737 * For normal user processes the _SYSTEMD_USER_UNIT= field is
738 now implicitly appended to every log entry logged. systemctl
739 has been updated to filter by this field when operating on a
740 user systemd instance.
742 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
743 CAP_AUDIT_CONTROL capabilities to the capabilities set for
744 the container. This makes it easier to boot unmodified
745 Fedora systems in a container, which however still requires
746 audit=0 to be passed on the kernel command line. Auditing in
747 kernel and userspace is unfortunately still too broken in
748 context of containers, hence we recommend compiling it out
749 of the kernel or using audit=0. Hopefully this will be fixed
750 one day for good in the kernel.
752 * nspawn gained the new --bind= and --bind-ro= parameters to
753 bind mount specific directories from the host into the
756 * nspawn will now mount its own devpts file system instance
757 into the container, in order not to leak pty devices from
758 the host into the container.
760 * systemd will now read the firmware boot time performance
761 information from the EFI variables, if the used boot loader
762 supports this, and takes it into account for boot performance
763 analysis via "systemd-analyze". This is currently supported
764 only in conjunction with Gummiboot, but could be supported
765 by other boot loaders too. For details see:
767 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
769 * A new generator has been added that automatically mounts the
770 EFI System Partition (ESP) to /boot, if that directory
771 exists, is empty, and no other file system has been
772 configured to be mounted there.
774 * logind will now send out PrepareForSleep(false) out
775 unconditionally, after coming back from suspend. This may be
776 used by applications as asynchronous notification for
777 system resume events.
779 * "systemctl unlock-sessions" has been added, that allows
780 unlocking the screens of all user sessions at once, similar
781 how "systemctl lock-sessions" already locked all users
782 sessions. This is backed by a new D-Bus call UnlockSessions().
784 * "loginctl seat-status" will now show the master device of a
785 seat. (i.e. the device of a seat that needs to be around for
786 the seat to be considered available, usually the graphics
789 * tmpfiles gained a new "X" line type, that allows
790 configuration of files and directories (with wildcards) that
791 shall be excluded from automatic cleanup ("aging").
793 * udev default rules set the device node permissions now only
794 at "add" events, and do not change them any longer with a
795 later "change" event.
797 * The log messages for lid events and power/sleep keypresses
798 now carry a message ID.
800 * We now have a substantially larger unit test suite, but this
801 continues to be work in progress.
803 * udevadm hwdb gained a new --root= parameter to change the
804 root directory to operate relative to.
806 * logind will now issue a background sync() request to the kernel
807 early at shutdown, so that dirty buffers are flushed to disk early
808 instead of at the last moment, in order to optimize shutdown
811 * A new bootctl tool has been added that is an interface for
812 certain boot loader operations. This is currently a preview
813 and is likely to be extended into a small mechanism daemon
814 like timedated, localed, hostnamed, and can be used by
815 graphical UIs to enumerate available boot options, and
816 request boot into firmware operations.
818 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
819 the rest of the package. It also has been updated to work
820 correctly in initrds.
822 * Policykit previously has been runtime optional, and is now
823 also compile time optional via a configure switch.
825 * systemd-analyze has been reimplemented in C. Also "systemctl
826 dot" has moved into systemd-analyze.
828 * "systemctl status" with no further parameters will now print
829 the status of all active or failed units.
831 * Operations such as "systemctl start" can now be executed
832 with a new mode "--irreversible" which may be used to queue
833 operations that cannot accidentally be reversed by a later
834 job queuing. This is by default used to make shutdown
835 requests more robust.
837 * The Python API of systemd now gained a new module for
838 reading journal files.
840 * A new tool kernel-install has been added that can install
841 kernel images according to the Boot Loader Specification:
843 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
845 * Boot time console output has been improved to provide
846 animated boot time output for hanging jobs.
848 * A new tool systemd-activate has been added which can be used
849 to test socket activation with, directly from the command
850 line. This should make it much easier to test and debug
851 socket activation in daemons.
853 * journalctl gained a new "--reverse" (or -r) option to show
854 journal output in reverse order (i.e. newest line first).
856 * journalctl gained a new "--pager-end" (or -e) option to jump
857 to immediately jump to the end of the journal in the
858 pager. This is only supported in conjunction with "less".
860 * journalctl gained a new "--user-unit=" option, that works
861 similar to "--unit=" but filters for user units rather than
864 * A number of unit files to ease adoption of systemd in
865 initrds has been added. This moves some minimal logic from
866 the various initrd implementations into systemd proper.
868 * The journal files are now owned by a new group
869 "systemd-journal", which exists specifically to allow access
870 to the journal, and nothing else. Previously, we used the
871 "adm" group for that, which however possibly covers more
872 than just journal/log file access. This new group is now
873 already used by systemd-journal-gatewayd to ensure this
874 daemon gets access to the journal files and as little else
875 as possible. Note that "make install" will also set FS ACLs
876 up for /var/log/journal to give "adm" and "wheel" read
877 access to it, in addition to "systemd-journal" which owns
878 the journal files. We recommend that packaging scripts also
879 add read access to "adm" + "wheel" to /var/log/journal, and
880 all existing/future journal files. To normal users and
881 administrators little changes, however packagers need to
882 ensure to create the "systemd-journal" system group at
883 package installation time.
885 * The systemd-journal-gatewayd now runs as unprivileged user
886 systemd-journal-gateway:systemd-journal-gateway. Packaging
887 scripts need to create these system user/group at
890 * timedated now exposes a new boolean property CanNTP that
891 indicates whether a local NTP service is available or not.
893 * systemd-detect-virt will now also detect xen PVs
895 * The pstore file system is now mounted by default, if it is
898 * In addition to the SELinux and IMA policies we will now also
899 load SMACK policies at early boot.
901 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
902 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
903 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
904 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
905 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
906 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
907 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
908 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
909 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
910 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
911 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
912 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
913 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
914 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
918 * Timer units now support calendar time events in addition to
919 monotonic time events. That means you can now trigger a unit
920 based on a calendar time specification such as "Thu,Fri
921 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
922 or fifth day of any month of the year 2013, given that it is
923 a thursday or friday. This brings timer event support
924 considerably closer to cron's capabilities. For details on
925 the supported calendar time specification language see
928 * udev now supports a number of different naming policies for
929 network interfaces for predictable names, and a combination
930 of these policies is now the default. Please see this wiki
931 document for details:
933 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
935 * Auke Kok's bootchart implementation has been added to the
936 systemd tree. It's an optional component that can graph the
937 boot in quite some detail. It's one of the best bootchart
938 implementations around and minimal in its code and
941 * nss-myhostname has been integrated into the systemd source
942 tree. nss-myhostname guarantees that the local hostname
943 always stays resolvable via NSS. It has been a weak
944 requirement of systemd-hostnamed since a long time, and
945 since its code is actually trivial we decided to just
946 include it in systemd's source tree. It can be turned off
947 with a configure switch.
949 * The read-ahead logic is now capable of properly detecting
950 whether a btrfs file system is on SSD or rotating media, in
951 order to optimize the read-ahead scheme. Previously, it was
952 only capable of detecting this on traditional file systems
955 * In udev, additional device properties are now read from the
956 IAB in addition to the OUI database. Also, Bluetooth company
957 identities are attached to the devices as well.
959 * In service files %U may be used as specifier that is
960 replaced by the configured user name of the service.
962 * nspawn may now be invoked without a controlling TTY. This
963 makes it suitable for invocation as its own service. This
964 may be used to set up a simple containerized server system
965 using only core OS tools.
967 * systemd and nspawn can now accept socket file descriptors
968 when they are started for socket activation. This enables
969 implementation of socket activated nspawn
970 containers. i.e. think about autospawning an entire OS image
971 when the first SSH or HTTP connection is received. We expect
972 that similar functionality will also be added to libvirt-lxc
975 * journalctl will now suppress ANSI color codes when
978 * systemctl will no longer show control group information for
979 a unit if a the control group is empty anyway.
981 * logind can now automatically suspend/hibernate/shutdown the
984 * /etc/machine-info and hostnamed now also expose the chassis
985 type of the system. This can be used to determine whether
986 the local system is a laptop, desktop, handset or
987 tablet. This information may either be configured by the
988 user/vendor or is automatically determined from ACPI and DMI
989 information if possible.
991 * A number of PolicyKit actions are now bound together with
992 "imply" rules. This should simplify creating UIs because
993 many actions will now authenticate similar ones as well.
995 * Unit files learnt a new condition ConditionACPower= which
996 may be used to conditionalize a unit depending on whether an
997 AC power source is connected or not, of whether the system
998 is running on battery power.
1000 * systemctl gained a new "is-failed" verb that may be used in
1001 shell scripts and suchlike to check whether a specific unit
1002 is in the "failed" state.
1004 * The EnvironmentFile= setting in unit files now supports file
1005 globbing, and can hence be used to easily read a number of
1006 environment files at once.
1008 * systemd will no longer detect and recognize specific
1009 distributions. All distribution-specific #ifdeffery has been
1010 removed, systemd is now fully generic and
1011 distribution-agnostic. Effectively, not too much is lost as
1012 a lot of the code is still accessible via explicit configure
1013 switches. However, support for some distribution specific
1014 legacy configuration file formats has been dropped. We
1015 recommend distributions to simply adopt the configuration
1016 files everybody else uses now and convert the old
1017 configuration from packaging scripts. Most distributions
1018 already did that. If that's not possible or desirable,
1019 distributions are welcome to forward port the specific
1020 pieces of code locally from the git history.
1022 * When logging a message about a unit systemd will now always
1023 log the unit name in the message meta data.
1025 * localectl will now also discover system locale data that is
1026 not stored in locale archives, but directly unpacked.
1028 * logind will no longer unconditionally use framebuffer
1029 devices as seat masters, i.e. as devices that are required
1030 to be existing before a seat is considered preset. Instead,
1031 it will now look for all devices that are tagged as
1032 "seat-master" in udev. By default framebuffer devices will
1033 be marked as such, but depending on local systems other
1034 devices might be marked as well. This may be used to
1035 integrate graphics cards using closed source drivers (such
1036 as NVidia ones) more nicely into logind. Note however, that
1037 we recommend using the open source NVidia drivers instead,
1038 and no udev rules for the closed-source drivers will be
1039 shipped from us upstream.
1041 Contributions from: Adam Williamson, Alessandro Crismani, Auke
1042 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
1043 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
1044 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
1045 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
1046 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
1047 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
1048 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
1049 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
1050 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
1051 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
1052 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
1057 * udev gained support for loading additional device properties
1058 from an indexed database that is keyed by vendor/product IDs
1059 and similar device identifiers. For the beginning this
1060 "hwdb" is populated with data from the well-known PCI and
1061 USB database, but also includes PNP, ACPI and OID data. In
1062 the longer run this indexed database shall grow into
1063 becoming the one central database for non-essential
1064 userspace device metadata. Previously, data from the PCI/USB
1065 database was only attached to select devices, since the
1066 lookup was a relatively expensive operation due to O(n) time
1067 complexity (with n being the number of entries in the
1068 database). Since this is now O(1), we decided to add in this
1069 data for all devices where this is available, by
1070 default. Note that the indexed database needs to be rebuilt
1071 when new data files are installed. To achieve this you need
1072 to update your packaging scripts to invoke "udevadm hwdb
1073 --update" after installation of hwdb data files. For
1074 RPM-based distributions we introduced the new
1075 %udev_hwdb_update macro for this purpose.
1077 * The Journal gained support for the "Message Catalog", an
1078 indexed database to link up additional information with
1079 journal entries. For further details please check:
1081 http://www.freedesktop.org/wiki/Software/systemd/catalog
1083 The indexed message catalog database also needs to be
1084 rebuilt after installation of message catalog files. Use
1085 "journalctl --update-catalog" for this. For RPM-based
1086 distributions we introduced the %journal_catalog_update
1087 macro for this purpose.
1089 * The Python Journal bindings gained support for the standard
1090 Python logging framework.
1092 * The Journal API gained new functions for checking whether
1093 the underlying file system of a journal file is capable of
1094 properly reporting file change notifications, or whether
1095 applications that want to reflect journal changes "live"
1096 need to recheck journal files continuously in appropriate
1099 * It is now possible to set the "age" field for tmpfiles
1100 entries to 0, indicating that files matching this entry
1101 shall always be removed when the directories are cleaned up.
1103 * coredumpctl gained a new "gdb" verb which invokes gdb
1104 right-away on the selected coredump.
1106 * There's now support for "hybrid sleep" on kernels that
1107 support this, in addition to "suspend" and "hibernate". Use
1108 "systemctl hybrid-sleep" to make use of this.
1110 * logind's HandleSuspendKey= setting (and related settings)
1111 now gained support for a new "lock" setting to simply
1112 request the screen lock on all local sessions, instead of
1113 actually executing a suspend or hibernation.
1115 * systemd will now mount the EFI variables file system by
1118 * Socket units now gained support for configuration of the
1119 SMACK security label.
1121 * timedatectl will now output the time of the last and next
1122 daylight saving change.
1124 * We dropped support for various legacy and distro-specific
1125 concepts, such as insserv, early-boot SysV services
1126 (i.e. those for non-standard runlevels such as 'b' or 'S')
1127 or ArchLinux /etc/rc.conf support. We recommend the
1128 distributions who still need support this to either continue
1129 to maintain the necessary patches downstream, or find a
1130 different solution. (Talk to us if you have questions!)
1132 * Various systemd components will now bypass PolicyKit checks
1133 for root and otherwise handle properly if PolicyKit is not
1134 found to be around. This should fix most issues for
1135 PolicyKit-less systems. Quite frankly this should have been
1136 this way since day one. It is absolutely our intention to
1137 make systemd work fine on PolicyKit-less systems, and we
1138 consider it a bug if something doesn't work as it should if
1139 PolicyKit is not around.
1141 * For embedded systems it is now possible to build udev and
1142 systemd without blkid and/or kmod support.
1144 * "systemctl switch-root" is now capable of switching root
1145 more than once. I.e. in addition to transitions from the
1146 initrd to the host OS it is now possible to transition to
1147 further OS images from the host. This is useful to implement
1148 offline updating tools.
1150 * Various other additions have been made to the RPM macros
1151 shipped with systemd. Use %udev_rules_update() after
1152 installing new udev rules files. %_udevhwdbdir,
1153 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
1154 %_sysctldir are now available which resolve to the right
1155 directories for packages to place various data files in.
1157 * journalctl gained the new --full switch (in addition to
1158 --all, to disable ellipsation for long messages.
1160 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
1161 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
1162 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
1163 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
1164 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
1165 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
1166 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
1167 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
1168 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
1172 * journalctl gained new --since= and --until= switches to
1173 filter by time. It also now supports nice filtering for
1174 units via --unit=/-u.
1176 * Type=oneshot services may use ExecReload= and do the
1179 * The journal daemon now supports time-based rotation and
1180 vacuuming, in addition to the usual disk-space based
1183 * The journal will now index the available field values for
1184 each field name. This enables clients to show pretty drop
1185 downs of available match values when filtering. The bash
1186 completion of journalctl has been updated
1187 accordingly. journalctl gained a new switch -F to list all
1188 values a certain field takes in the journal database.
1190 * More service events are now written as structured messages
1191 to the journal, and made recognizable via message IDs.
1193 * The timedated, localed and hostnamed mini-services which
1194 previously only provided support for changing time, locale
1195 and hostname settings from graphical DEs such as GNOME now
1196 also have a minimal (but very useful) text-based client
1197 utility each. This is probably the nicest way to changing
1198 these settings from the command line now, especially since
1199 it lists available options and is fully integrated with bash
1202 * There's now a new tool "systemd-coredumpctl" to list and
1203 extract coredumps from the journal.
1205 * We now install a README each in /var/log/ and
1206 /etc/rc.d/init.d explaining where the system logs and init
1207 scripts went. This hopefully should help folks who go to
1208 that dirs and look into the otherwise now empty void and
1209 scratch their heads.
1211 * When user-services are invoked (by systemd --user) the
1212 $MANAGERPID env var is set to the PID of systemd.
1214 * SIGRTMIN+24 when sent to a --user instance will now result
1215 in immediate termination of systemd.
1217 * gatewayd received numerous feature additions such as a
1218 "follow" mode, for live syncing and filtering.
1220 * browse.html now allows filtering and showing detailed
1221 information on specific entries. Keyboard navigation and
1222 mouse screen support has been added.
1224 * gatewayd/journalctl now supports HTML5/JSON
1225 Server-Sent-Events as output.
1227 * The SysV init script compatibility logic will now
1228 heuristically determine whether a script supports the
1229 "reload" verb, and only then make this available as
1232 * "systemctl status --follow" has been removed, use "journalctl
1235 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
1236 have been removed since they are hardly useful to be
1239 * And I'd like to take the opportunity to specifically mention
1240 Zbigniew for his great contributions. Zbigniew, you rock!
1242 Contributions from: Andrew Eikum, Christian Hesse, Colin
1243 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
1244 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
1245 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
1246 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
1247 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
1248 Jędrzejewski-Szmek, Сковорода Никита Андреевич
1252 * If /etc/vconsole.conf is non-existent or empty we will no
1253 longer load any console font or key map at boot by
1254 default. Instead the kernel defaults will be left
1255 intact. This is definitely the right thing to do, as no
1256 configuration should mean no configuration, and hard-coding
1257 font names that are different on all archs is probably a bad
1258 idea. Also, the kernel default key map and font should be
1259 good enough for most cases anyway, and mostly identical to
1260 the userspace fonts/key maps we previously overloaded them
1261 with. If distributions want to continue to default to a
1262 non-kernel font or key map they should ship a default
1263 /etc/vconsole.conf with the appropriate contents.
1265 Contributions from: Colin Walters, Daniel J Walsh, Dave
1266 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
1267 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1271 * journalctl gained a new --cursor= switch to show entries
1272 starting from the specified location in the journal.
1274 * We now enforce a size limit on journal entry fields exported
1275 with "-o json" in journalctl. Fields larger than 4K will be
1276 assigned null. This can be turned off with --all.
1278 * An (optional) journal gateway daemon is now available as
1279 "systemd-journal-gatewayd.service". This service provides
1280 access to the journal via HTTP and JSON. This functionality
1281 will be used to implement live log synchronization in both
1282 pull and push modes, but has various other users too, such
1283 as easy log access for debugging of embedded devices. Right
1284 now it is already useful to retrieve the journal via HTTP:
1286 # systemctl start systemd-journal-gatewayd.service
1287 # wget http://localhost:19531/entries
1289 This will download the journal contents in a
1290 /var/log/messages compatible format. The same as JSON:
1292 # curl -H"Accept: application/json" http://localhost:19531/entries
1294 This service is also accessible via a web browser where a
1295 single static HTML5 app is served that uses the JSON logic
1296 to enable the user to do some basic browsing of the
1297 journal. This will be extended later on. Here's an example
1298 screenshot of this app in its current state:
1300 http://0pointer.de/public/journal-gatewayd
1302 Contributions from: Kay Sievers, Lennart Poettering, Robert
1303 Milasan, Tom Gundersen
1307 * The bash completion logic is now available for journalctl
1310 * We don't mount the "cpuset" controller anymore together with
1311 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
1312 started if no parameters are assigned to it. "cpuset" hence
1313 broke code that assumed it it could create "cpu" groups and
1316 * journalctl -f will now subscribe to terminal size changes,
1317 and line break accordingly.
1319 Contributions from: Dave Reisner, Kay Sievers, Lennart
1320 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
1324 * nspawn will now create a symlink /etc/localtime in the
1325 container environment, copying the host's timezone
1326 setting. Previously this has been done via a bind mount, but
1327 since symlinks cannot be bind mounted this has now been
1328 changed to create/update the appropriate symlink.
1330 * journalctl -n's line number argument is now optional, and
1331 will default to 10 if omitted.
1333 * journald will now log the maximum size the journal files may
1334 take up on disk. This is particularly useful if the default
1335 built-in logic of determining this parameter from the file
1336 system size is used. Use "systemctl status
1337 systemd-journald.service" to see this information.
1339 * The multi-seat X wrapper tool has been stripped down. As X
1340 is now capable of enumerating graphics devices via udev in a
1341 seat-aware way the wrapper is not strictly necessary
1342 anymore. A stripped down temporary stop-gap is still shipped
1343 until the upstream display managers have been updated to
1344 fully support the new X logic. Expect this wrapper to be
1345 removed entirely in one of the next releases.
1347 * HandleSleepKey= in logind.conf has been split up into
1348 HandleSuspendKey= and HandleHibernateKey=. The old setting
1349 is not available anymore. X11 and the kernel are
1350 distuingishing between these keys and we should too. This
1351 also means the inhibition lock for these keys has been split
1354 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
1355 Poettering, Lukas Nykryn, Václav Pavlín
1359 * Whenever a unit changes state we'll now log this to the
1360 journal and show along the unit's own log output in
1363 * ConditionPathIsMountPoint= can now properly detect bind
1364 mount points too. (Previously, a bind mount of one file
1365 system to another place in the same file system could not be
1366 detected as mount, since they shared struct stat's st_dev
1369 * We will now mount the cgroup controllers cpu, cpuacct,
1370 cpuset and the controllers net_cls, net_prio together by
1373 * nspawn containers will now have a virtualized boot
1374 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
1375 over with a randomized ID at container initialization). This
1376 has the effect of making "journalctl -b" do the right thing
1379 * The JSON output journal serialization has been updated not
1380 to generate "endless" list objects anymore, but rather one
1381 JSON object per line. This is more in line how most JSON
1382 parsers expect JSON objects. The new output mode
1383 "json-pretty" has been added to provide similar output, but
1384 neatly aligned for readability by humans.
1386 * We dropped all explicit sync() invocations in the shutdown
1387 code. The kernel does this implicitly anyway in the kernel
1388 reboot() syscall. halt(8)'s -n option is now a compatibility
1391 * We now support virtualized reboot() in containers, as
1392 supported by newer kernels. We will fall back to exit() if
1393 CAP_SYS_REBOOT is not available to the container. Also,
1394 nspawn makes use of this now and will actually reboot the
1395 container if the containerized OS asks for that.
1397 * journalctl will only show local log output by default
1398 now. Use --merge (-m) to show remote log output, too.
1400 * libsystemd-journal gained the new sd_journal_get_usage()
1401 call to determine the current disk usage of all journal
1402 files. This is exposed in the new "journalctl --disk-usage"
1405 * journald gained a new configuration setting SplitMode= in
1406 journald.conf which may be used to control how user journals
1407 are split off. See journald.conf(5) for details.
1409 * A new condition type ConditionFileNotEmpty= has been added.
1411 * tmpfiles' "w" lines now support file globbing, to write
1412 multiple files at once.
1414 * We added Python bindings for the journal submission
1415 APIs. More Python APIs for a number of selected APIs will
1416 likely follow. Note that we intend to add native bindings
1417 only for the Python language, as we consider it common
1418 enough to deserve bindings shipped within systemd. There are
1419 various projects outside of systemd that provide bindings
1420 for languages such as PHP or Lua.
1422 * Many conditions will now resolve specifiers such as %i. In
1423 addition, PathChanged= and related directives of .path units
1424 now support specifiers as well.
1426 * There's now a new RPM macro definition for the system preset
1429 * journald will now warn if it can't forward a message to the
1430 syslog daemon because it's socket is full.
1432 * timedated will no longer write or process /etc/timezone,
1433 except on Debian. As we do not support late mounted /usr
1434 anymore /etc/localtime always being a symlink is now safe,
1435 and hence the information in /etc/timezone is not necessary
1438 * logind will now always reserve one VT for a text getty (VT6
1439 by default). Previously if more than 6 X sessions where
1440 started they took up all the VTs with auto-spawned gettys,
1441 so that no text gettys were available anymore.
1443 * udev will now automatically inform the btrfs kernel logic
1444 about btrfs RAID components showing up. This should make
1445 simple hotplug based btrfs RAID assembly work.
1447 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
1448 (but not for its children which will stay at the kernel
1449 default). This should allow setups with a lot more listening
1452 * systemd will now always pass the configured timezone to the
1453 kernel at boot. timedated will do the same when the timezone
1456 * logind's inhibition logic has been updated. By default,
1457 logind will now handle the lid switch, the power and sleep
1458 keys all the time, even in graphical sessions. If DEs want
1459 to handle these events on their own they should take the new
1460 handle-power-key, handle-sleep-key and handle-lid-switch
1461 inhibitors during their runtime. A simple way to achiveve
1462 that is to invoke the DE wrapped in an invocation of:
1464 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
1466 * Access to unit operations is now checked via SELinux taking
1467 the unit file label and client process label into account.
1469 * systemd will now notify the administrator in the journal
1470 when he over-mounts a non-empty directory.
1472 * There are new specifiers that are resolved in unit files,
1473 for the host name (%H), the machine ID (%m) and the boot ID
1476 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
1477 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
1478 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
1479 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
1480 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
1481 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
1482 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
1486 * Support for reading structured kernel messages from
1487 /dev/kmsg has now been added and is enabled by default.
1489 * Support for reading kernel messages from /proc/kmsg has now
1490 been removed. If you want kernel messages in the journal
1491 make sure to run a recent kernel (>= 3.5) that supports
1492 reading structured messages from /dev/kmsg (see
1493 above). /proc/kmsg is now exclusive property of classic
1494 syslog daemons again.
1496 * The libudev API gained the new
1497 udev_device_new_from_device_id() call.
1499 * The logic for file system namespace (ReadOnlyDirectory=,
1500 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
1501 require pivot_root() anymore. This means fewer temporary
1502 directories are created below /tmp for this feature.
1504 * nspawn containers will now see and receive all submounts
1505 made on the host OS below the root file system of the
1508 * Forward Secure Sealing is now supported for Journal files,
1509 which provide cryptographical sealing of journal files so
1510 that attackers cannot alter log history anymore without this
1511 being detectable. Lennart will soon post a blog story about
1512 this explaining it in more detail.
1514 * There are two new service settings RestartPreventExitStatus=
1515 and SuccessExitStatus= which allow configuration of exit
1516 status (exit code or signal) which will be excepted from the
1517 restart logic, resp. consider successful.
1519 * journalctl gained the new --verify switch that can be used
1520 to check the integrity of the structure of journal files and
1521 (if Forward Secure Sealing is enabled) the contents of
1524 * nspawn containers will now be run with /dev/stdin, /dev/fd/
1525 and similar symlinks pre-created. This makes running shells
1526 as container init process a lot more fun.
1528 * The fstab support can now handle PARTUUID= and PARTLABEL=
1531 * A new ConditionHost= condition has been added to match
1532 against the hostname (with globs) and machine ID. This is
1533 useful for clusters where a single OS image is used to
1534 provision a large number of hosts which shall run slightly
1535 different sets of services.
1537 * Services which hit the restart limit will now be placed in a
1540 Contributions from: Bertram Poettering, Dave Reisner, Huang
1541 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
1542 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
1546 * When running in --user mode systemd will now become a
1547 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
1548 tree a lot more organized.
1550 * A new PartOf= unit dependency type has been introduced that
1551 may be used to group services in a natural way.
1553 * "systemctl enable" may now be used to enable instances of
1556 * journalctl now prints error log levels in red, and
1557 warning/notice log levels in bright white. It also supports
1558 filtering by log level now.
1560 * cgtop gained a new -n switch (similar to top), to configure
1561 the maximum number of iterations to run for. It also gained
1562 -b, to run in batch mode (accepting no input).
1564 * The suffix ".service" may now be omitted on most systemctl
1565 command lines involving service unit names.
1567 * There's a new bus call in logind to lock all sessions, as
1568 well as a loginctl verb for it "lock-sessions".
1570 * libsystemd-logind.so gained a new call sd_journal_perror()
1571 that works similar to libc perror() but logs to the journal
1572 and encodes structured information about the error number.
1574 * /etc/crypttab entries now understand the new keyfile-size=
1577 * shutdown(8) now can send a (configurable) wall message when
1578 a shutdown is cancelled.
1580 * The mount propagation mode for the root file system will now
1581 default to "shared", which is useful to make containers work
1582 nicely out-of-the-box so that they receive new mounts from
1583 the host. This can be undone locally by running "mount
1584 --make-rprivate /" if needed.
1586 * The prefdm.service file has been removed. Distributions
1587 should maintain this unit downstream if they intend to keep
1588 it around. However, we recommend writing normal unit files
1589 for display managers instead.
1591 * Since systemd is a crucial part of the OS we will now
1592 default to a number of compiler switches that improve
1593 security (hardening) such as read-only relocations, stack
1594 protection, and suchlike.
1596 * The TimeoutSec= setting for services is now split into
1597 TimeoutStartSec= and TimeoutStopSec= to allow configuration
1598 of individual time outs for the start and the stop phase of
1601 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
1602 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
1603 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
1604 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
1605 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
1606 Gundersen, Zbigniew Jędrzejewski-Szmek
1610 * The journal and id128 C APIs are now fully documented as man
1613 * Extra safety checks have been added when transitioning from
1614 the initial RAM disk to the main system to avoid accidental
1617 * /etc/crypttab entries now understand the new keyfile-offset=
1620 * systemctl -t can now be used to filter by unit load state.
1622 * The journal C API gained the new sd_journal_wait() call to
1623 make writing synchronous journal clients easier.
1625 * journalctl gained the new -D switch to show journals from a
1628 * journalctl now displays a special marker between log
1629 messages of two different boots.
1631 * The journal is now explicitly flushed to /var via a service
1632 systemd-journal-flush.service, rather than implicitly simply
1633 by seeing /var/log/journal to be writable.
1635 * journalctl (and the journal C APIs) can now match for much
1636 more complex expressions, with alternatives and
1639 * When transitioning from the initial RAM disk to the main
1640 system we will now kill all processes in a killing spree to
1641 ensure no processes stay around by accident.
1643 * Three new specifiers may be used in unit files: %u, %h, %s
1644 resolve to the user name, user home directory resp. user
1645 shell. This is useful for running systemd user instances.
1647 * We now automatically rotate journal files if their data
1648 object hash table gets a fill level > 75%. We also size the
1649 hash table based on the configured maximum file size. This
1650 together should lower hash collisions drastically and thus
1651 speed things up a bit.
1653 * journalctl gained the new "--header" switch to introspect
1654 header data of journal files.
1656 * A new setting SystemCallFilters= has been added to services
1657 which may be used to apply blacklists or whitelists to
1658 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
1660 * nspawn gained a new --link-journal= switch (and quicker: -j)
1661 to link the container journal with the host. This makes it
1662 very easy to centralize log viewing on the host for all
1663 guests while still keeping the journal files separated.
1665 * Many bugfixes and optimizations
1667 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
1668 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
1669 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
1674 * Several tools now understand kernel command line arguments,
1675 which are only read when run in an initial RAM disk. They
1676 usually follow closely their normal counterparts, but are
1679 * There's a new tool to analyze the readahead files that are
1680 automatically generated at boot. Use:
1682 /usr/lib/systemd/systemd-readahead analyze /.readahead
1684 * We now provide an early debug shell on tty9 if this enabled. Use:
1686 systemctl enable debug-shell.service
1688 * All plymouth related units have been moved into the Plymouth
1689 package. Please make sure to upgrade your Plymouth version
1692 * systemd-tmpfiles now supports getting passed the basename of
1693 a configuration file only, in which case it will look for it
1694 in all appropriate directories automatically.
1696 * udevadm info now takes a /dev or /sys path as argument, and
1697 does the right thing. Example:
1699 udevadm info /dev/sda
1700 udevadm info /sys/class/block/sda
1702 * systemctl now prints a warning if a unit is stopped but a
1703 unit that might trigger it continues to run. Example: a
1704 service is stopped but the socket that activates it is left
1707 * "systemctl status" will now mention if the log output was
1708 shortened due to rotation since a service has been started.
1710 * The journal API now exposes functions to determine the
1711 "cutoff" times due to rotation.
1713 * journald now understands SIGUSR1 and SIGUSR2 for triggering
1714 immediately flushing of runtime logs to /var if possible,
1715 resp. for triggering immediate rotation of the journal
1718 * It is now considered an error if a service is attempted to
1719 be stopped that is not loaded.
1721 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
1723 * systemd-analyze now supports Python 3
1725 * tmpfiles now supports cleaning up directories via aging
1726 where the first level dirs are always kept around but
1727 directories beneath it automatically aged. This is enabled
1728 by prefixing the age field with '~'.
1730 * Seat objects now expose CanGraphical, CanTTY properties
1731 which is required to deal with very fast bootups where the
1732 display manager might be running before the graphics drivers
1733 completed initialization.
1735 * Seat objects now expose a State property.
1737 * We now include RPM macros for service enabling/disabling
1738 based on the preset logic. We recommend RPM based
1739 distributions to make use of these macros if possible. This
1740 makes it simpler to reuse RPM spec files across
1743 * We now make sure that the collected systemd unit name is
1744 always valid when services log to the journal via
1747 * There's a new man page kernel-command-line(7) detailing all
1748 command line options we understand.
1750 * The fstab generator may now be disabled at boot by passing
1751 fstab=0 on the kernel command line.
1753 * A new kernel command line option modules-load= is now understood
1754 to load a specific kernel module statically, early at boot.
1756 * Unit names specified on the systemctl command line are now
1757 automatically escaped as needed. Also, if file system or
1758 device paths are specified they are automatically turned
1759 into the appropriate mount or device unit names. Example:
1761 systemctl status /home
1762 systemctl status /dev/sda
1764 * The SysVConsole= configuration option has been removed from
1765 system.conf parsing.
1767 * The SysV search path is no longer exported on the D-Bus
1770 * The Names= option is been removed from unit file parsing.
1772 * There's a new man page bootup(7) detailing the boot process.
1774 * Every unit and every generator we ship with systemd now
1775 comes with full documentation. The self-explanatory boot is
1778 * A couple of services gained "systemd-" prefixes in their
1779 name if they wrap systemd code, rather than only external
1780 code. Among them fsck@.service which is now
1781 systemd-fsck@.service.
1783 * The HaveWatchdog property has been removed from the D-Bus
1786 * systemd.confirm_spawn= on the kernel command line should now
1789 * There's a new man page crypttab(5) which details all options
1790 we actually understand.
1792 * systemd-nspawn gained a new --capability= switch to pass
1793 additional capabilities to the container.
1795 * timedated will now read known NTP implementation unit names
1796 from /usr/lib/systemd/ntp-units.d/*.list,
1797 systemd-timedated-ntp.target has been removed.
1799 * journalctl gained a new switch "-b" that lists log data of
1800 the current boot only.
1802 * The notify socket is in the abstract namespace again, in
1803 order to support daemons which chroot() at start-up.
1805 * There is a new Storage= configuration option for journald
1806 which allows configuration of where log data should go. This
1807 also provides a way to disable journal logging entirely, so
1808 that data collected is only forwarded to the console, the
1809 kernel log buffer or another syslog implementation.
1811 * Many bugfixes and optimizations
1813 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
1814 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
1815 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
1816 Shawn Landden, Tom Gundersen
1820 * "systemctl help <unit>" now shows the man page if one is
1823 * Several new man pages have been added.
1825 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
1826 MaxLevelConsole= can now be specified in
1827 journald.conf. These options allow reducing the amount of
1828 data stored on disk or forwarded by the log level.
1830 * TimerSlackNSec= can now be specified in system.conf for
1831 PID1. This allows system-wide power savings.
1833 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
1834 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
1839 * logind is now capable of (optionally) handling power and
1840 sleep keys as well as the lid switch.
1842 * journalctl now understands the syntax "journalctl
1843 /usr/bin/avahi-daemon" to get all log output of a specific
1846 * CapabilityBoundingSet= in system.conf now also influences
1847 the capability bound set of usermode helpers of the kernel.
1849 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
1850 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
1851 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
1852 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
1856 * Note that we skipped 139 releases here in order to set the
1857 new version to something that is greater than both udev's
1858 and systemd's most recent version number.
1860 * udev: all udev sources are merged into the systemd source tree now.
1861 All future udev development will happen in the systemd tree. It
1862 is still fully supported to use the udev daemon and tools without
1863 systemd running, like in initramfs or other init systems. Building
1864 udev though, will require the *build* of the systemd tree, but
1865 udev can be properly *run* without systemd.
1867 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
1868 should be used to create dead device nodes as workarounds for broken
1871 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
1872 no longer supported. udev_monitor_new_from_netlink() needs to be
1873 used to subscribe to events.
1875 * udev: when udevd is started by systemd, processes which are left
1876 behind by forking them off of udev rules, are unconditionally cleaned
1877 up and killed now after the event handling has finished. Services or
1878 daemons must be started as systemd services. Services can be
1879 pulled-in by udev to get started, but they can no longer be directly
1880 forked by udev rules.
1882 * udev: the daemon binary is called systemd-udevd now and installed
1883 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
1884 to adapt to that, create symlink, or rename the binary after building
1887 * libudev no longer provides these symbols:
1888 udev_monitor_from_socket()
1889 udev_queue_get_failed_list_entry()
1890 udev_get_{dev,sys,run}_path()
1891 The versions number was bumped and symbol versioning introduced.
1893 * systemd-loginctl and systemd-journalctl have been renamed
1894 to loginctl and journalctl to match systemctl.
1896 * The config files: /etc/systemd/systemd-logind.conf and
1897 /etc/systemd/systemd-journald.conf have been renamed to
1898 logind.conf and journald.conf. Package updates should rename
1899 the files to the new names on upgrade.
1901 * For almost all files the license is now LGPL2.1+, changed
1902 from the previous GPL2.0+. Exceptions are some minor stuff
1903 of udev (which will be changed to LGPL2.1 eventually, too),
1904 and the MIT licensed sd-daemon.[ch] library that is suitable
1905 to be used as drop-in files.
1907 * systemd and logind now handle system sleep states, in
1908 particular suspending and hibernating.
1910 * logind now implements a sleep/shutdown/idle inhibiting logic
1911 suitable for a variety of uses. Soonishly Lennart will blog
1912 about this in more detail.
1914 * var-run.mount and var-lock.mount are no longer provided
1915 (which prevously bind mounted these directories to their new
1916 places). Distributions which have not converted these
1917 directories to symlinks should consider stealing these files
1918 from git history and add them downstream.
1920 * We introduced the Documentation= field for units and added
1921 this to all our shipped units. This is useful to make it
1922 easier to explore the boot and the purpose of the various
1925 * All smaller setup units (such as
1926 systemd-vconsole-setup.service) now detect properly if they
1927 are run in a container and are skipped when
1928 appropriate. This guarantees an entirely noise-free boot in
1929 Linux container environments such as systemd-nspawn.
1931 * A framework for implementing offline system updates is now
1932 integrated, for details see:
1933 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
1935 * A new service type Type=idle is available now which helps us
1936 avoiding ugly interleaving of getty output and boot status
1939 * There's now a system-wide CapabilityBoundingSet= option to
1940 globally reduce the set of capabilities for the
1941 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
1942 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
1943 even CAP_NET_ADMIN system-wide for secure systems.
1945 * There are now system-wide DefaultLimitXXX= options to
1946 globally change the defaults of the various resource limits
1947 for all units started by PID 1.
1949 * Harald Hoyer's systemd test suite has been integrated into
1950 systemd which allows easy testing of systemd builds in qemu
1951 and nspawn. (This is really awesome! Ask us for details!)
1953 * The fstab parser is now implemented as generator, not inside
1956 * systemctl will now warn you if .mount units generated from
1957 /etc/fstab are out of date due to changes in fstab that
1958 haven't been read by systemd yet.
1960 * systemd is now suitable for usage in initrds. Dracut has
1961 already been updated to make use of this. With this in place
1962 initrds get a slight bit faster but primarily are much
1963 easier to introspect and debug since "systemctl status" in
1964 the host system can be used to introspect initrd services,
1965 and the journal from the initrd is kept around too.
1967 * systemd-delta has been added, a tool to explore differences
1968 between user/admin configuration and vendor defaults.
1970 * PrivateTmp= now affects both /tmp and /var/tmp.
1972 * Boot time status messages are now much prettier and feature
1973 proper english language. Booting up systemd has never been
1976 * Read-ahead pack files now include the inode number of all
1977 files to pre-cache. When the inode changes the pre-caching
1978 is not attempted. This should be nicer to deal with updated
1979 packages which might result in changes of read-ahead
1982 * We now temporaritly lower the kernel's read_ahead_kb variable
1983 when collecting read-ahead data to ensure the kernel's
1984 built-in read-ahead does not add noise to our measurements
1985 of necessary blocks to pre-cache.
1987 * There's now RequiresMountsFor= to add automatic dependencies
1988 for all mounts necessary for a specific file system path.
1990 * MountAuto= and SwapAuto= have been removed from
1991 system.conf. Mounting file systems at boot has to take place
1994 * nspawn now learned a new switch --uuid= to set the machine
1995 ID on the command line.
1997 * nspawn now learned the -b switch to automatically search
2000 * vt102 is now the default TERM for serial TTYs, upgraded from
2003 * systemd-logind now works on VT-less systems.
2005 * The build tree has been reorganized. The individual
2006 components now have directories of their own.
2008 * A new condition type ConditionPathIsReadWrite= is now available.
2010 * nspawn learned the new -C switch to create cgroups for the
2011 container in other hierarchies.
2013 * We now have support for hardware watchdogs, configurable in
2016 * The scheduled shutdown logic now has a public API.
2018 * We now mount /tmp as tmpfs by default, but this can be
2019 masked and /etc/fstab can override it.
2021 * Since udisks doesn't make use of /media anymore we are not
2022 mounting a tmpfs on it anymore.
2024 * journalctl gained a new --local switch to only interleave
2025 locally generated journal files.
2027 * We can now load the IMA policy at boot automatically.
2029 * The GTK tools have been split off into a systemd-ui.
2031 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
2032 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
2033 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
2034 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
2035 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
2036 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
2037 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
2038 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
2039 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
2044 * This is mostly a bugfix release
2046 * Support optional initialization of the machine ID from the
2047 KVM or container configured UUID.
2049 * Support immediate reboots with "systemctl reboot -ff"
2051 * Show /etc/os-release data in systemd-analyze output
2053 * Many bugfixes for the journal, including endianness fixes and
2054 ensuring that disk space enforcement works
2056 * sd-login.h is C++ comptaible again
2058 * Extend the /etc/os-release format on request of the Debian
2061 * We now refuse non-UTF8 strings used in various configuration
2062 and unit files. This is done to ensure we don't pass invalid
2063 data over D-Bus or expose it elsewhere.
2065 * Register Mimo USB Screens as suitable for automatic seat
2068 * Read SELinux client context from journal clients in a race
2071 * Reorder configuration file lookup order. /etc now always
2072 overrides /run in order to allow the administrator to always
2073 and unconditionally override vendor supplied or
2074 automatically generated data.
2076 * The various user visible bits of the journal now have man
2077 pages. We still lack man pages for the journal API calls
2080 * We now ship all man pages in HTML format again in the
2083 Contributions from: Dave Reisner, Dirk Eibach, Frederic
2084 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
2085 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
2090 * This is mostly a bugfix release
2092 * systems lacking /etc/os-release are no longer supported.
2094 * Various functionality updates to libsystemd-login.so
2096 * Track class of PAM logins to distuingish greeters from
2099 Contributions from: Kay Sievers, Lennart Poettering, Michael
2104 * This is an important bugfix release for v41.
2106 * Building man pages is now optional which should be useful
2107 for those building systemd from git but unwilling to install
2110 * Watchdog support for supervising services is now usable. In
2111 a future release support for hardware watchdogs
2112 (i.e. /dev/watchdog) will be added building on this.
2114 * Service start rate limiting is now configurable and can be
2115 turned off per service. When a start rate limit is hit a
2116 reboot can automatically be triggered.
2118 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
2120 Contributions from: Benjamin Franzke, Bill Nottingham,
2121 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
2122 Schmidt, Michał Górny, Piotr Drąg
2126 * The systemd binary is installed /usr/lib/systemd/systemd now;
2127 An existing /sbin/init symlink needs to be adapted with the
2130 * The code that loads kernel modules has been ported to invoke
2131 libkmod directly, instead of modprobe. This means we do not
2132 support systems with module-init-tools anymore.
2134 * Watchdog support is now already useful, but still not
2137 * A new kernel command line option systemd.setenv= is
2138 understood to set system wide environment variables
2139 dynamically at boot.
2141 * We now limit the set of capabilities of systemd-journald.
2143 * We now set SIGPIPE to ignore by default, since it only is
2144 useful in shell pipelines, and has little use in general
2145 code. This can be disabled with IgnoreSIPIPE=no in unit
2148 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
2149 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
2154 * This is mostly a bugfix release
2156 * We now expose the reason why a service failed in the
2157 "Result" D-Bus property.
2159 * Rudimentary service watchdog support (will be completed over
2160 the next few releases.)
2162 * When systemd forks off in order execute some service we will
2163 now immediately changes its argv[0] to reflect which process
2164 it will execute. This is useful to minimize the time window
2165 with a generic argv[0], which makes bootcharts more useful
2167 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
2168 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
2169 Mike Kazantsev, Ray Strode
2173 * This is mostly a test release, but incorporates many
2176 * New systemd-cgtop tool to show control groups by their
2179 * Linking against libacl for ACLs is optional again. If
2180 disabled, support tracking device access for active logins
2181 goes becomes unavailable, and so does access to the user
2182 journals by the respective users.
2184 * If a group "adm" exists, journal files are automatically
2185 owned by them, thus allow members of this group full access
2186 to the system journal as well as all user journals.
2188 * The journal now stores the SELinux context of the logging
2189 client for all entries.
2191 * Add C++ inclusion guards to all public headers
2193 * New output mode "cat" in the journal to print only text
2194 messages, without any meta data like date or time.
2196 * Include tiny X server wrapper as a temporary stop-gap to
2197 teach XOrg udev display enumeration. This is used by display
2198 managers such as gdm, and will go away as soon as XOrg
2199 learned native udev hotplugging for display devices.
2201 * Add new systemd-cat tool for executing arbitrary programs
2202 with STDERR/STDOUT connected to the journal. Can also act as
2203 BSD logger replacement, and does so by default.
2205 * Optionally store all locally generated coredumps in the
2206 journal along with meta data.
2208 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
2209 writing short strings to files (for usage for /sys), and for
2210 creating symlinks, character and block device nodes.
2212 * New unit file option ControlGroupPersistent= to make cgroups
2213 persistent, following the mechanisms outlined in
2214 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
2216 * Support multiple local RTCs in a sane way
2218 * No longer monopolize IO when replaying readahead data on
2219 rotating disks, since we might starve non-file-system IO to
2220 death, since fanotify() will not see accesses done by blkid,
2223 * Don't show kernel threads in systemd-cgls anymore, unless
2224 requested with new -k switch.
2226 Contributions from: Dan Horák, Kay Sievers, Lennart
2227 Poettering, Michal Schmidt
2231 * This is mostly a test release, but incorporates many
2234 * The git repository moved to:
2235 git://anongit.freedesktop.org/systemd/systemd
2236 ssh://git.freedesktop.org/git/systemd/systemd
2238 * First release with the journal
2239 http://0pointer.de/blog/projects/the-journal.html
2241 * The journal replaces both systemd-kmsg-syslogd and
2242 systemd-stdout-bridge.
2244 * New sd_pid_get_unit() API call in libsystemd-logind
2246 * Many systemadm clean-ups
2248 * Introduce remote-fs-pre.target which is ordered before all
2249 remote mounts and may be used to start services before all
2252 * Added Mageia support
2254 * Add bash completion for systemd-loginctl
2256 * Actively monitor PID file creation for daemons which exit in
2257 the parent process before having finished writing the PID
2258 file in the daemon process. Daemons which do this need to be
2259 fixed (i.e. PID file creation must have finished before the
2260 parent exits), but we now react a bit more gracefully to them.
2262 * Add colourful boot output, mimicking the well-known output
2263 of existing distributions.
2265 * New option PassCredentials= for socket units, for
2266 compatibility with a recent kernel ABI breakage.
2268 * /etc/rc.local is now hooked in via a generator binary, and
2269 thus will no longer act as synchronization point during
2272 * systemctl list-unit-files now supports --root=.
2274 * systemd-tmpfiles now understands two new commands: z, Z for
2275 relabelling files according to the SELinux database. This is
2276 useful to apply SELinux labels to specific files in /sys,
2279 * Output of SysV services is now forwarded to both the console
2280 and the journal by default, not only just the console.
2282 * New man pages for all APIs from libsystemd-login.
2284 * The build tree got reorganized and a the build system is a
2285 lot more modular allowing embedded setups to specifically
2286 select the components of systemd they are interested in.
2288 * Support for Linux systems lacking the kernel VT subsystem is
2291 * configure's --with-rootdir= got renamed to
2292 --with-rootprefix= to follow the naming used by udev and
2295 * Unless specified otherwise we'll now install to /usr instead
2296 of /usr/local by default.
2298 * Processes with '@' in argv[0][0] are now excluded from the
2299 final shut-down killing spree, following the logic explained
2301 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
2303 * All processes remaining in a service cgroup when we enter
2304 the START or START_PRE states are now killed with
2305 SIGKILL. That means it is no longer possible to spawn
2306 background processes from ExecStart= lines (which was never
2307 supported anyway, and bad style).
2309 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
2310 reloading of units together.
2312 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
2313 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
2314 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
2315 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
2316 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek