1 systemd System and Service Manager
5 * The documentation has been updated to cover the various new
6 concepts introduced with 205.
8 * Unit files now understand the new %v specifier which
9 resolves to the kernel version string as returned by "uname
12 * systemctl now supports filtering the unit list output by
13 load state, active state and sub state, using the new
16 * "systemctl status" will now show the results of the
17 condition checks (like ConditionPathExists= and similar) of
18 the last start attempts of the unit. They are also logged to
21 * "journalctl -b" may now be used to look for boot output of a
22 specific boot. Try "journalctl -b -1" for the previous boot,
23 but the syntax is substantially more powerful.
25 * "journalctl --show-cursor" has been added which prints the
26 cursor string the last shown log line. This may then be used
27 with the new "journalctl --after-cursor=" switch to continue
28 browsing logs from that point on.
30 * "journalctl --force" may now be used to force regeneration
33 * Creation of "dead" device nodes has been moved from udev
34 into kmod and tmpfiles. Previously, udev would read the kmod
35 databases to pre-generate dead device nodes based on meta
36 information contained in kernel modules, so that these would
37 be auto-loaded on access rather then at boot. As this
38 doesn't really have much to do with the exposing actual
39 kernel devices to userspace this has always been slightly
40 alien in the udev codebase. Following the new scheme kmod
41 will now generate a runtime snippet for tmpfiles from the
42 module meta information and it now is tmpfiles' job to the
43 create the nodes. This also allows overriding access and
44 other parameters for the nodes using the usual tmpfiles
45 facilities. As side effect this allows us to remove the
46 CAP_SYS_MKNOD capability bit from udevd entirely.
48 * logind's device ACLs may now be applied to these "dead"
49 devices nodes too, thus finally allowing managed access to
50 devices such as /dev/snd/sequencer whithout loading the
51 backing module right-away.
53 * A new RPM macro has been added that may be used to apply
54 tmpfiles configuration during package installation.
56 * systemd-detect-virt and ConditionVirtualization= now can
57 detect User-Mode-Linux machines (UML).
59 * journald will now implicitly log the effective capabilities
60 set of processes in the message metadata.
62 * systemd-cryptsetup has gained support for TrueCrypt volumes.
64 * The initrd interface has been simplified (more specifically,
65 support for passing performance data via environment
66 variables and fsck results via files in /run has been
67 removed). These features were non-essential, and are
68 nowadays available in a much nicer way by having systemd in
69 the initrd serialize its state and have the hosts systemd
72 * The udev "keymap" data files and tools to apply keyboard
73 specific mappings of scan to key codes, and force-release
74 scan code lists have been entirely replaced by a udev
75 "keyboard" builtin and a hwdb data file.
77 * systemd will now honour the kernel's "quiet" command line
78 argument also during late shutdown, resulting in a
79 completely silent shutdown when used.
81 * There's now an option to control the SO_REUSEPORT socket
82 option in .socket units.
84 * Instance units will now automatically get a per-template
85 subslice of system.slice unless something else is explicitly
86 configured. For example, instances of sshd@.service will now
87 implicitly be placed in system-sshd.slice rather than
88 system.slice as before.
90 * Test coverage support may now be enabled at build time.
92 Contributions from: Dave Reisner, Frederic Crozat, Harald
93 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
94 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
95 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
96 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
97 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
98 Giokas, Zbigniew Jędrzejewski-Szmek
102 * Two new unit types have been introduced:
104 Scope units are very similar to service units, however, are
105 created out of pre-existing processes -- instead of PID 1
106 forking off the processes. By using scope units it is
107 possible for system services and applications to group their
108 own child processes (worker processes) in a powerful way
109 which then maybe used to organize them, or kill them
110 together, or apply resource limits on them.
112 Slice units may be used to partition system resources in an
113 hierarchial fashion and then assign other units to them. By
114 default there are now three slices: system.slice (for all
115 system services), user.slice (for all user sessions),
116 machine.slice (for VMs and containers).
118 Slices and scopes have been introduced primarily in
119 context of the work to move cgroup handling to a
120 single-writer scheme, where only PID 1
121 creates/removes/manages cgroups.
123 * There's a new concept of "transient" units. In contrast to
124 normal units these units are created via an API at runtime,
125 not from configuration from disk. More specifically this
126 means it is now possible to run arbitrary programs as
127 independent services, with all execution parameters passed
128 in via bus APIs rather than read from disk. Transient units
129 make systemd substantially more dynamic then it ever was,
130 and useful as a general batch manager.
132 * logind has been updated to make use of scope and slice units
133 for managing user sessions. As a user logs in he will get
134 his own private slice unit, to which all sessions are added
135 as scope units. We also added support for automatically
136 adding an instance of user@.service for the user into the
137 slice. Effectively logind will no longer create cgroup
138 hierarchies on its own now, it will defer entirely to PID 1
139 for this by means of scope, service and slice units. Since
140 user sessions this way become entities managed by PID 1
141 the output of "systemctl" is now a lot more comprehensive.
143 * A new mini-daemon "systemd-machined" has been added which
144 may be used by virtualization managers to register local
145 VMs/containers. nspawn has been updated accordingly, and
146 libvirt will be updated shortly. machined will collect a bit
147 of meta information about the VMs/containers, and assign
148 them their own scope unit (see above). The collected
149 meta-data is then made available via the "machinectl" tool,
150 and exposed in "ps" and similar tools. machined/machinectl
151 is compile-time optional.
153 * As discussed earlier, the low-level cgroup configuration
154 options ControlGroup=, ControlGroupModify=,
155 ControlGroupPersistent=, ControlGroupAttribute= have been
156 removed. Please use high-level attribute settings instead as
159 * A new bus call SetUnitProperties() has been added to alter
160 various runtime parameters of a unit. This is primarily
161 useful to alter cgroup parameters dynamically in a nice way,
162 but will be extended later on to make more properties
163 modifiable at runtime. systemctl gained a new set-properties
164 command that wraps this call.
166 * A new tool "systemd-run" has been added which can be used to
167 run arbitrary command lines as transient services or scopes,
168 while configuring a number of settings via the command
169 line. This tool is currently very basic, however already
170 very useful. We plan to extend this tool to even allow
171 queuing of execution jobs with time triggers from the
172 command line, similar in fashion to "at".
174 * nspawn will now inform the user explicitly that kernels with
175 audit enabled break containers, and suggest the user to turn
178 * Support for detecting the IMA and AppArmor security
179 frameworks with ConditionSecurity= has been added.
181 * journalctl gained a new "-k" switch for showing only kernel
182 messages, mimicking dmesg output; in addition to "--user"
183 and "--system" switches for showing only user's own logs
186 * systemd-delta can now show information about drop-in
187 snippets extending unit files.
189 * libsystemd-bus has been substantially updated but is still
190 not available as public API.
192 * systemd will now look for the "debug" argument on the kernel
193 command line and enable debug logging, similar to
194 "systemd.log_level=debug" already did before.
196 * "systemctl set-default", "systemctl get-default" has been
197 added to configure the default.target symlink, which
198 controls what to boot into by default.
200 * "systemctl set-log-level" has been added as a convenient
201 way to raise and lower systemd logging threshold.
203 * "systemd-analyze plot" will now show the time the various
204 generators needed for execution, as well as information
205 about the unit file loading.
207 * libsystemd-journal gained a new sd_journal_open_files() call
208 for opening specific journal files. journactl also gained a
209 new switch to expose this new functionality. Previously we
210 only supported opening all files from a directory, or all
211 files from the system, as opening individual files only is
212 racy due to journal file rotation.
214 * systemd gained the new DefaultEnvironment= setting in
215 /etc/systemd/system.conf to set environment variables for
218 * If a privileged process logs a journal message with the
219 OBJECT_PID= field set, then journald will automatically
220 augment this with additional OBJECT_UID=, OBJECT_GID=,
221 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
222 system services want to log events about specific client
223 processes. journactl/systemctl has been updated to make use
224 of this information if all log messages regarding a specific
227 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
228 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
229 Reisner, David Coppa, David King, David Strauss, Eelco
230 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
231 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
232 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
233 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
234 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
235 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
236 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
237 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
238 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
243 * The Python bindings gained some minimal support for the APIs
244 exposed by libsystemd-logind.
246 * ConditionSecurity= gained support for detecting SMACK. Since
247 this condition already supports SELinux and AppArmor we only
248 miss IMA for this. Patches welcome!
250 Contributions from: Karol Lewandowski, Lennart Poettering,
251 Zbigniew Jędrzejewski-Szmek
255 * systemd-nspawn will now create /etc/resolv.conf if
256 necessary, before bind-mounting the host's file onto it.
258 * systemd-nspawn will now store meta information about a
259 container on the container's cgroup as extended attribute
260 fields, including the root directory.
262 * The cgroup hierarchy has been reworked in many ways. All
263 objects any of the components systemd creates in the cgroup
264 tree are now suffixed. More specifically, user sessions are
265 now placed in cgroups suffixed with ".session", users in
266 cgroups suffixed with ".user", and nspawn containers in
267 cgroups suffixed with ".nspawn". Furthermore, all cgroup
268 names are now escaped in a simple scheme to avoid collision
269 of userspace object names with kernel filenames. This work
270 is preparation for making these objects relocatable in the
271 cgroup tree, in order to allow easy resource partitioning of
272 these objects without causing naming conflicts.
274 * systemctl list-dependencies gained the new switches
275 --plain, --reverse, --after and --before.
277 * systemd-inhibit now shows the process name of processes that
278 have taken an inhibitor lock.
280 * nss-myhostname will now also resolve "localhost"
281 implicitly. This makes /etc/hosts an optional file and
282 nicely handles that on IPv6 ::1 maps to both "localhost" and
285 * libsystemd-logind.so gained a new call
286 sd_get_machine_names() to enumerate running containers and
287 VMs (currently only supported by very new libvirt and
288 nspawn). sd_login_monitor can now be used to watch
289 VMs/containers coming and going.
291 * .include is not allowed recursively anymore, and only in
292 unit files. Usually it is better to use drop-in snippets in
293 .d/*.conf anyway, as introduced with systemd 198.
295 * systemd-analyze gained a new "critical-chain" command that
296 determines the slowest chain of units run during system
297 boot-up. It is very useful for tracking down where
298 optimizing boot time is the most beneficial.
300 * systemd will no longer allow manipulating service paths in
301 the name=systemd:/system cgroup tree using ControlGroup= in
302 units. (But is still fine with it in all other dirs.)
304 * There's a new systemd-nspawn@.service service file that may
305 be used to easily run nspawn containers as system
306 services. With the container's root directory in
307 /var/lib/container/foobar it is now sufficient to run
308 "systemctl start systemd-nspawn@foobar.service" to boot it.
310 * systemd-cgls gained a new parameter "--machine" to list only
311 the processes within a certain container.
313 * ConditionSecurity= now can check for "apparmor". We still
314 are lacking checks for SMACK and IMA for this condition
315 check though. Patches welcome!
317 * A new configuration file /etc/systemd/sleep.conf has been
318 added that may be used to configure which kernel operation
319 systemd is supposed to execute when "suspend", "hibernate"
320 or "hybrid-sleep" is requested. This makes the new kernel
321 "freeze" state accessible to the user.
323 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
324 the passed argument if applicable.
326 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
327 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
328 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
329 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
330 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
331 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
332 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
337 * The output of 'systemctl list-jobs' got some polishing. The
338 '--type=' argument may now be passed more than once. A new
339 command 'systemctl list-sockets' has been added which shows
340 a list of kernel sockets systemd is listening on with the
341 socket units they belong to, plus the units these socket
344 * The experimental libsystemd-bus library got substantial
345 updates to work in conjunction with the (also experimental)
346 kdbus kernel project. It works well enough to exchange
347 messages with some sophistication. Note that kdbus is not
348 ready yet, and the library is mostly an elaborate test case
349 for now, and not installable.
351 * systemd gained a new unit 'systemd-static-nodes.service'
352 that generates static device nodes earlier during boot, and
353 can run in conjunction with udev.
355 * libsystemd-login gained a new call sd_pid_get_user_unit()
356 to retrieve the user systemd unit a process is running
357 in. This is useful for systems where systemd is used as
360 * systemd-nspawn now places all containers in the new /machine
361 top-level cgroup directory in the name=systemd
362 hierarchy. libvirt will soon do the same, so that we get a
363 uniform separation of /system, /user and /machine for system
364 services, user processes and containers/virtual
365 machines. This new cgroup hierarchy is also useful to stick
366 stable names to specific container instances, which can be
367 recognized later this way (this name may be controlled
368 via systemd-nspawn's new -M switch). libsystemd-login also
369 gained a new call sd_pid_get_machine_name() to retrieve the
370 name of the container/VM a specific process belongs to.
372 * bootchart can now store its data in the journal.
374 * libsystemd-journal gained a new call
375 sd_journal_add_conjunction() for AND expressions to the
376 matching logic. This can be used to express more complex
379 * journactl can now take multiple --unit= and --user-unit=
382 * The cryptsetup logic now understands the "luks.key=" kernel
383 command line switch for specifying a file to read the
384 decryption key from. Also, if a configured key file is not
385 found the tool will now automatically fall back to prompting
388 * Python systemd.journal module was updated to wrap recently
389 added functions from libsystemd-journal. The interface was
390 changed to bring the low level interface in s.j._Reader
391 closer to the C API, and the high level interface in
392 s.j.Reader was updated to wrap and convert all data about
395 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
396 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
397 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
398 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
399 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
400 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
404 * journalctl --update-catalog now understands a new --root=
405 option to operate on catalogs found in a different root
408 * During shutdown after systemd has terminated all running
409 services a final killing loop kills all remaining left-over
410 processes. We will now print the name of these processes
411 when we send SIGKILL to them, since this usually indicates a
414 * If /etc/crypttab refers to password files stored on
415 configured mount points automatic dependencies will now be
416 generated to ensure the specific mount is established first
417 before the key file is attempted to be read.
419 * 'systemctl status' will now show information about the
420 network sockets a socket unit is listening on.
422 * 'systemctl status' will also shown information about any
423 drop-in configuration file for units. (Drop-In configuration
424 files in this context are files such as
425 /etc/systemd/systemd/foobar.service.d/*.conf)
427 * systemd-cgtop now optionally shows summed up CPU times of
428 cgroups. Press '%' while running cgtop to switch between
429 percentage and absolute mode. This is useful to determine
430 which cgroups use up the most CPU time over the entire
431 runtime of the system. systemd-cgtop has also been updated
432 to be 'pipeable' for processing with further shell tools.
434 * 'hostnamectl set-hostname' will now allow setting of FQDN
437 * The formatting and parsing of time span values has been
438 changed. The parser now understands fractional expressions
439 such as "5.5h". The formatter will now output fractional
440 expressions for all time spans under 1min, i.e. "5.123456s"
441 rather than "5s 123ms 456us". For time spans under 1s
442 millisecond values are shown, for those under 1ms
443 microsecond values are shown. This should greatly improve
444 all time-related output of systemd.
446 * libsystemd-login and libsystemd-journal gained new
447 functions for querying the poll() events mask and poll()
448 timeout value for integration into arbitrary event
451 * localectl gained the ability to list available X11 keymaps
452 (models, layouts, variants, options).
454 * 'systemd-analyze dot' gained the ability to filter for
455 specific units via shell-style globs, to create smaller,
456 more useful graphs. I.e. it's now possible to create simple
457 graphs of all the dependencies between only target units, or
458 of all units that Avahi has dependencies with.
460 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
461 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
462 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
463 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
464 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
465 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
466 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
470 * The boot-time readahead implementation for rotating media
471 will now read the read-ahead data in multiple passes which
472 consist of all read requests made in equidistant time
473 intervals. This means instead of strictly reading read-ahead
474 data in its physical order on disk we now try to find a
475 middle ground between physical and access time order.
477 * /etc/os-release files gained a new BUILD_ID= field for usage
478 on operating systems that provide continuous builds of OS
481 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
482 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
483 William Douglas, Zbigniew Jędrzejewski-Szmek
487 * systemd-python gained an API exposing libsystemd-daemon.
489 * The SMACK setup logic gained support for uploading CIPSO
492 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
493 ReadOnlyDirectories= and InaccessibleDirectories= has
494 changed. The private /tmp and /var/tmp directories are now
495 shared by all processes of a service (which means
496 ExecStartPre= may now leave data in /tmp that ExecStart= of
497 the same service can still access). When a service is
498 stopped its temporary directories are immediately deleted
499 (normal clean-up with tmpfiles is still done in addition to
502 * By default, systemd will now set a couple of sysctl
503 variables in the kernel: the safe sysrq options are turned
504 on, IP route verification is turned on, and source routing
505 disabled. The recently added hardlink and softlink
506 protection of the kernel is turned on. These settings should
507 be reasonably safe, and good defaults for all new systems.
509 * The predictable network naming logic may now be turned off
510 with a new kernel command line switch: net.ifnames=0.
512 * A new libsystemd-bus module has been added that implements a
513 pretty complete D-Bus client library. For details see:
515 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
517 * journald will now explicitly flush the journal files to disk
518 at the latest 5min after each write. The file will then also
519 be marked offline until the next write. This should increase
520 reliability in case of a crash. The synchronization delay
521 can be configured via SyncIntervalSec= in journald.conf.
523 * There's a new remote-fs-setup.target unit that can be used
524 to pull in specific services when at least one remote file
525 system is to be mounted.
527 * There are new targets timers.target and paths.target as
528 canonical targets to pull user timer and path units in
529 from. This complements sockets.target with a similar
530 purpose for socket units.
532 * libudev gained a new call udev_device_set_attribute_value()
533 to set sysfs attributes of a device.
535 * The udev daemon now sets the default number of worker
536 processes executed in parallel based on the number of available
537 CPUs instead of the amount of available RAM. This is supposed
538 to provide a more reliable default and limit a too aggressive
539 paralellism for setups with 1000s of devices connected.
541 Contributions from: Auke Kok, Colin Walters, Cristian
542 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
543 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
544 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
545 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
546 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
547 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
548 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
549 Zbigniew Jędrzejewski-Szmek
553 * Configuration of unit files may now be extended via drop-in
554 files without having to edit/override the unit files
555 themselves. More specifically, if the administrator wants to
556 change one value for a service file foobar.service he can
557 now do so by dropping in a configuration snippet into
558 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
559 will load all these snippets and apply them on top of the
560 main unit configuration file, possibly extending or
561 overriding its settings. Using these drop-in snippets is
562 generally nicer than the two earlier options for changing
563 unit files locally: copying the files from
564 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
565 them there; or creating a new file in /etc/systemd/system/
566 that incorporates the original one via ".include". Drop-in
567 snippets into these .d/ directories can be placed in any
568 directory systemd looks for units in, and the usual
569 overriding semantics between /usr/lib, /etc and /run apply
572 * Most unit file settings which take lists of items can now be
573 reset by assigning the empty string to them. For example,
574 normally, settings such as Environment=FOO=BAR append a new
575 environment variable assignment to the environment block,
576 each time they are used. By assigning Environment= the empty
577 string the environment block can be reset to empty. This is
578 particularly useful with the .d/*.conf drop-in snippets
579 mentioned above, since this adds the ability to reset list
580 settings from vendor unit files via these drop-ins.
582 * systemctl gained a new "list-dependencies" command for
583 listing the dependencies of a unit recursively.
585 * Inhibitors are now honored and listed by "systemctl
586 suspend", "systemctl poweroff" (and similar) too, not only
587 GNOME. These commands will also list active sessions by
590 * Resource limits (as exposed by the various control group
591 controllers) can now be controlled dynamically at runtime
592 for all units. More specifically, you can now use a command
593 like "systemctl set-cgroup-attr foobar.service cpu.shares
594 2000" to alter the CPU shares a specific service gets. These
595 settings are stored persistently on disk, and thus allow the
596 administrator to easily adjust the resource usage of
597 services with a few simple commands. This dynamic resource
598 management logic is also available to other programs via the
599 bus. Almost any kernel cgroup attribute and controller is
602 * systemd-vconsole-setup will now copy all font settings to
603 all allocated VTs, where it previously applied them only to
606 * libsystemd-login gained the new sd_session_get_tty() API
609 * This release drops support for a few legacy or
610 distribution-specific LSB facility names when parsing init
611 scripts: $x-display-manager, $mail-transfer-agent,
612 $mail-transport-agent, $mail-transfer-agent, $smtp,
613 $null. Also, the mail-transfer-agent.target unit backing
614 this has been removed. Distributions which want to retain
615 compatibility with this should carry the burden for
616 supporting this themselves and patch support for these back
617 in, if they really need to. Also, the facilities $syslog and
618 $local_fs are now ignored, since systemd does not support
619 early-boot LSB init scripts anymore, and these facilities
620 are implied anyway for normal services. syslog.target has
623 * There are new bus calls on PID1's Manager object for
624 cancelling jobs, and removing snapshot units. Previously,
625 both calls were only available on the Job and Snapshot
628 * systemd-journal-gatewayd gained SSL support.
630 * The various "environment" files, such as /etc/locale.conf
631 now support continuation lines with a backslash ("\") as
632 last character in the line, similar in style (but different)
633 to how this is supported in shells.
635 * For normal user processes the _SYSTEMD_USER_UNIT= field is
636 now implicitly appended to every log entry logged. systemctl
637 has been updated to filter by this field when operating on a
638 user systemd instance.
640 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
641 CAP_AUDIT_CONTROL capabilities to the capabilities set for
642 the container. This makes it easier to boot unmodified
643 Fedora systems in a container, which however still requires
644 audit=0 to be passed on the kernel command line. Auditing in
645 kernel and userspace is unfortunately still too broken in
646 context of containers, hence we recommend compiling it out
647 of the kernel or using audit=0. Hopefully this will be fixed
648 one day for good in the kernel.
650 * nspawn gained the new --bind= and --bind-ro= parameters to
651 bind mount specific directories from the host into the
654 * nspawn will now mount its own devpts file system instance
655 into the container, in order not to leak pty devices from
656 the host into the container.
658 * systemd will now read the firmware boot time performance
659 information from the EFI variables, if the used boot loader
660 supports this, and takes it into account for boot performance
661 analysis via "systemd-analyze". This is currently supported
662 only in conjunction with Gummiboot, but could be supported
663 by other boot loaders too. For details see:
665 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
667 * A new generator has been added that automatically mounts the
668 EFI System Partition (ESP) to /boot, if that directory
669 exists, is empty, and no other file system has been
670 configured to be mounted there.
672 * logind will now send out PrepareForSleep(false) out
673 unconditionally, after coming back from suspend. This may be
674 used by applications as asynchronous notification for
675 system resume events.
677 * "systemctl unlock-sessions" has been added, that allows
678 unlocking the screens of all user sessions at once, similar
679 how "systemctl lock-sessions" already locked all users
680 sessions. This is backed by a new D-Bus call UnlockSessions().
682 * "loginctl seat-status" will now show the master device of a
683 seat. (i.e. the device of a seat that needs to be around for
684 the seat to be considered available, usually the graphics
687 * tmpfiles gained a new "X" line type, that allows
688 configuration of files and directories (with wildcards) that
689 shall be excluded from automatic cleanup ("aging").
691 * udev default rules set the device node permissions now only
692 at "add" events, and do not change them any longer with a
693 later "change" event.
695 * The log messages for lid events and power/sleep keypresses
696 now carry a message ID.
698 * We now have a substantially larger unit test suite, but this
699 continues to be work in progress.
701 * udevadm hwdb gained a new --root= parameter to change the
702 root directory to operate relative to.
704 * logind will now issue a background sync() request to the kernel
705 early at shutdown, so that dirty buffers are flushed to disk early
706 instead of at the last moment, in order to optimize shutdown
709 * A new bootctl tool has been added that is an interface for
710 certain boot loader operations. This is currently a preview
711 and is likely to be extended into a small mechanism daemon
712 like timedated, localed, hostnamed, and can be used by
713 graphical UIs to enumerate available boot options, and
714 request boot into firmware operations.
716 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
717 the rest of the package. It also has been updated to work
718 correctly in initrds.
720 * Policykit previously has been runtime optional, and is now
721 also compile time optional via a configure switch.
723 * systemd-analyze has been reimplemented in C. Also "systemctl
724 dot" has moved into systemd-analyze.
726 * "systemctl status" with no further parameters will now print
727 the status of all active or failed units.
729 * Operations such as "systemctl start" can now be executed
730 with a new mode "--irreversible" which may be used to queue
731 operations that cannot accidentally be reversed by a later
732 job queuing. This is by default used to make shutdown
733 requests more robust.
735 * The Python API of systemd now gained a new module for
736 reading journal files.
738 * A new tool kernel-install has been added that can install
739 kernel images according to the Boot Loader Specification:
741 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
743 * Boot time console output has been improved to provide
744 animated boot time output for hanging jobs.
746 * A new tool systemd-activate has been added which can be used
747 to test socket activation with, directly from the command
748 line. This should make it much easier to test and debug
749 socket activation in daemons.
751 * journalctl gained a new "--reverse" (or -r) option to show
752 journal output in reverse order (i.e. newest line first).
754 * journalctl gained a new "--pager-end" (or -e) option to jump
755 to immediately jump to the end of the journal in the
756 pager. This is only supported in conjunction with "less".
758 * journalctl gained a new "--user-unit=" option, that works
759 similar to "--unit=" but filters for user units rather than
762 * A number of unit files to ease adoption of systemd in
763 initrds has been added. This moves some minimal logic from
764 the various initrd implementations into systemd proper.
766 * The journal files are now owned by a new group
767 "systemd-journal", which exists specifically to allow access
768 to the journal, and nothing else. Previously, we used the
769 "adm" group for that, which however possibly covers more
770 than just journal/log file access. This new group is now
771 already used by systemd-journal-gatewayd to ensure this
772 daemon gets access to the journal files and as little else
773 as possible. Note that "make install" will also set FS ACLs
774 up for /var/log/journal to give "adm" and "wheel" read
775 access to it, in addition to "systemd-journal" which owns
776 the journal files. We recommend that packaging scripts also
777 add read access to "adm" + "wheel" to /var/log/journal, and
778 all existing/future journal files. To normal users and
779 administrators little changes, however packagers need to
780 ensure to create the "systemd-journal" system group at
781 package installation time.
783 * The systemd-journal-gatewayd now runs as unprivileged user
784 systemd-journal-gateway:systemd-journal-gateway. Packaging
785 scripts need to create these system user/group at
788 * timedated now exposes a new boolean property CanNTP that
789 indicates whether a local NTP service is available or not.
791 * systemd-detect-virt will now also detect xen PVs
793 * The pstore file system is now mounted by default, if it is
796 * In addition to the SELinux and IMA policies we will now also
797 load SMACK policies at early boot.
799 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
800 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
801 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
802 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
803 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
804 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
805 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
806 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
807 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
808 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
809 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
810 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
811 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
812 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
816 * Timer units now support calendar time events in addition to
817 monotonic time events. That means you can now trigger a unit
818 based on a calendar time specification such as "Thu,Fri
819 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
820 or fifth day of any month of the year 2013, given that it is
821 a thursday or friday. This brings timer event support
822 considerably closer to cron's capabilities. For details on
823 the supported calendar time specification language see
826 * udev now supports a number of different naming policies for
827 network interfaces for predictable names, and a combination
828 of these policies is now the default. Please see this wiki
829 document for details:
831 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
833 * Auke Kok's bootchart implementation has been added to the
834 systemd tree. It's an optional component that can graph the
835 boot in quite some detail. It's one of the best bootchart
836 implementations around and minimal in its code and
839 * nss-myhostname has been integrated into the systemd source
840 tree. nss-myhostname guarantees that the local hostname
841 always stays resolvable via NSS. It has been a weak
842 requirement of systemd-hostnamed since a long time, and
843 since its code is actually trivial we decided to just
844 include it in systemd's source tree. It can be turned off
845 with a configure switch.
847 * The read-ahead logic is now capable of properly detecting
848 whether a btrfs file system is on SSD or rotating media, in
849 order to optimize the read-ahead scheme. Previously, it was
850 only capable of detecting this on traditional file systems
853 * In udev, additional device properties are now read from the
854 IAB in addition to the OUI database. Also, Bluetooth company
855 identities are attached to the devices as well.
857 * In service files %U may be used as specifier that is
858 replaced by the configured user name of the service.
860 * nspawn may now be invoked without a controlling TTY. This
861 makes it suitable for invocation as its own service. This
862 may be used to set up a simple containerized server system
863 using only core OS tools.
865 * systemd and nspawn can now accept socket file descriptors
866 when they are started for socket activation. This enables
867 implementation of socket activated nspawn
868 containers. i.e. think about autospawning an entire OS image
869 when the first SSH or HTTP connection is received. We expect
870 that similar functionality will also be added to libvirt-lxc
873 * journalctl will now suppress ANSI color codes when
876 * systemctl will no longer show control group information for
877 a unit if a the control group is empty anyway.
879 * logind can now automatically suspend/hibernate/shutdown the
882 * /etc/machine-info and hostnamed now also expose the chassis
883 type of the system. This can be used to determine whether
884 the local system is a laptop, desktop, handset or
885 tablet. This information may either be configured by the
886 user/vendor or is automatically determined from ACPI and DMI
887 information if possible.
889 * A number of PolicyKit actions are now bound together with
890 "imply" rules. This should simplify creating UIs because
891 many actions will now authenticate similar ones as well.
893 * Unit files learnt a new condition ConditionACPower= which
894 may be used to conditionalize a unit depending on whether an
895 AC power source is connected or not, of whether the system
896 is running on battery power.
898 * systemctl gained a new "is-failed" verb that may be used in
899 shell scripts and suchlike to check whether a specific unit
900 is in the "failed" state.
902 * The EnvironmentFile= setting in unit files now supports file
903 globbing, and can hence be used to easily read a number of
904 environment files at once.
906 * systemd will no longer detect and recognize specific
907 distributions. All distribution-specific #ifdeffery has been
908 removed, systemd is now fully generic and
909 distribution-agnostic. Effectively, not too much is lost as
910 a lot of the code is still accessible via explicit configure
911 switches. However, support for some distribution specific
912 legacy configuration file formats has been dropped. We
913 recommend distributions to simply adopt the configuration
914 files everybody else uses now and convert the old
915 configuration from packaging scripts. Most distributions
916 already did that. If that's not possible or desirable,
917 distributions are welcome to forward port the specific
918 pieces of code locally from the git history.
920 * When logging a message about a unit systemd will now always
921 log the unit name in the message meta data.
923 * localectl will now also discover system locale data that is
924 not stored in locale archives, but directly unpacked.
926 * logind will no longer unconditionally use framebuffer
927 devices as seat masters, i.e. as devices that are required
928 to be existing before a seat is considered preset. Instead,
929 it will now look for all devices that are tagged as
930 "seat-master" in udev. By default framebuffer devices will
931 be marked as such, but depending on local systems other
932 devices might be marked as well. This may be used to
933 integrate graphics cards using closed source drivers (such
934 as NVidia ones) more nicely into logind. Note however, that
935 we recommend using the open source NVidia drivers instead,
936 and no udev rules for the closed-source drivers will be
937 shipped from us upstream.
939 Contributions from: Adam Williamson, Alessandro Crismani, Auke
940 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
941 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
942 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
943 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
944 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
945 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
946 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
947 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
948 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
949 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
950 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
955 * udev gained support for loading additional device properties
956 from an indexed database that is keyed by vendor/product IDs
957 and similar device identifiers. For the beginning this
958 "hwdb" is populated with data from the well-known PCI and
959 USB database, but also includes PNP, ACPI and OID data. In
960 the longer run this indexed database shall grow into
961 becoming the one central database for non-essential
962 userspace device metadata. Previously, data from the PCI/USB
963 database was only attached to select devices, since the
964 lookup was a relatively expensive operation due to O(n) time
965 complexity (with n being the number of entries in the
966 database). Since this is now O(1), we decided to add in this
967 data for all devices where this is available, by
968 default. Note that the indexed database needs to be rebuilt
969 when new data files are installed. To achieve this you need
970 to update your packaging scripts to invoke "udevadm hwdb
971 --update" after installation of hwdb data files. For
972 RPM-based distributions we introduced the new
973 %udev_hwdb_update macro for this purpose.
975 * The Journal gained support for the "Message Catalog", an
976 indexed database to link up additional information with
977 journal entries. For further details please check:
979 http://www.freedesktop.org/wiki/Software/systemd/catalog
981 The indexed message catalog database also needs to be
982 rebuilt after installation of message catalog files. Use
983 "journalctl --update-catalog" for this. For RPM-based
984 distributions we introduced the %journal_catalog_update
985 macro for this purpose.
987 * The Python Journal bindings gained support for the standard
988 Python logging framework.
990 * The Journal API gained new functions for checking whether
991 the underlying file system of a journal file is capable of
992 properly reporting file change notifications, or whether
993 applications that want to reflect journal changes "live"
994 need to recheck journal files continuously in appropriate
997 * It is now possible to set the "age" field for tmpfiles
998 entries to 0, indicating that files matching this entry
999 shall always be removed when the directories are cleaned up.
1001 * coredumpctl gained a new "gdb" verb which invokes gdb
1002 right-away on the selected coredump.
1004 * There's now support for "hybrid sleep" on kernels that
1005 support this, in addition to "suspend" and "hibernate". Use
1006 "systemctl hybrid-sleep" to make use of this.
1008 * logind's HandleSuspendKey= setting (and related settings)
1009 now gained support for a new "lock" setting to simply
1010 request the screen lock on all local sessions, instead of
1011 actually executing a suspend or hibernation.
1013 * systemd will now mount the EFI variables file system by
1016 * Socket units now gained support for configuration of the
1017 SMACK security label.
1019 * timedatectl will now output the time of the last and next
1020 daylight saving change.
1022 * We dropped support for various legacy and distro-specific
1023 concepts, such as insserv, early-boot SysV services
1024 (i.e. those for non-standard runlevels such as 'b' or 'S')
1025 or ArchLinux /etc/rc.conf support. We recommend the
1026 distributions who still need support this to either continue
1027 to maintain the necessary patches downstream, or find a
1028 different solution. (Talk to us if you have questions!)
1030 * Various systemd components will now bypass PolicyKit checks
1031 for root and otherwise handle properly if PolicyKit is not
1032 found to be around. This should fix most issues for
1033 PolicyKit-less systems. Quite frankly this should have been
1034 this way since day one. It is absolutely our intention to
1035 make systemd work fine on PolicyKit-less systems, and we
1036 consider it a bug if something doesn't work as it should if
1037 PolicyKit is not around.
1039 * For embedded systems it is now possible to build udev and
1040 systemd without blkid and/or kmod support.
1042 * "systemctl switch-root" is now capable of switching root
1043 more than once. I.e. in addition to transitions from the
1044 initrd to the host OS it is now possible to transition to
1045 further OS images from the host. This is useful to implement
1046 offline updating tools.
1048 * Various other additions have been made to the RPM macros
1049 shipped with systemd. Use %udev_rules_update() after
1050 installing new udev rules files. %_udevhwdbdir,
1051 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
1052 %_sysctldir are now available which resolve to the right
1053 directories for packages to place various data files in.
1055 * journalctl gained the new --full switch (in addition to
1056 --all, to disable ellipsation for long messages.
1058 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
1059 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
1060 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
1061 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
1062 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
1063 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
1064 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
1065 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
1066 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
1070 * journalctl gained new --since= and --until= switches to
1071 filter by time. It also now supports nice filtering for
1072 units via --unit=/-u.
1074 * Type=oneshot services may use ExecReload= and do the
1077 * The journal daemon now supports time-based rotation and
1078 vacuuming, in addition to the usual disk-space based
1081 * The journal will now index the available field values for
1082 each field name. This enables clients to show pretty drop
1083 downs of available match values when filtering. The bash
1084 completion of journalctl has been updated
1085 accordingly. journalctl gained a new switch -F to list all
1086 values a certain field takes in the journal database.
1088 * More service events are now written as structured messages
1089 to the journal, and made recognizable via message IDs.
1091 * The timedated, localed and hostnamed mini-services which
1092 previously only provided support for changing time, locale
1093 and hostname settings from graphical DEs such as GNOME now
1094 also have a minimal (but very useful) text-based client
1095 utility each. This is probably the nicest way to changing
1096 these settings from the command line now, especially since
1097 it lists available options and is fully integrated with bash
1100 * There's now a new tool "systemd-coredumpctl" to list and
1101 extract coredumps from the journal.
1103 * We now install a README each in /var/log/ and
1104 /etc/rc.d/init.d explaining where the system logs and init
1105 scripts went. This hopefully should help folks who go to
1106 that dirs and look into the otherwise now empty void and
1107 scratch their heads.
1109 * When user-services are invoked (by systemd --user) the
1110 $MANAGERPID env var is set to the PID of systemd.
1112 * SIGRTMIN+24 when sent to a --user instance will now result
1113 in immediate termination of systemd.
1115 * gatewayd received numerous feature additions such as a
1116 "follow" mode, for live syncing and filtering.
1118 * browse.html now allows filtering and showing detailed
1119 information on specific entries. Keyboard navigation and
1120 mouse screen support has been added.
1122 * gatewayd/journalctl now supports HTML5/JSON
1123 Server-Sent-Events as output.
1125 * The SysV init script compatibility logic will now
1126 heuristically determine whether a script supports the
1127 "reload" verb, and only then make this available as
1130 * "systemctl status --follow" has been removed, use "journalctl
1133 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
1134 have been removed since they are hardly useful to be
1137 * And I'd like to take the opportunity to specifically mention
1138 Zbigniew for his great contributions. Zbigniew, you rock!
1140 Contributions from: Andrew Eikum, Christian Hesse, Colin
1141 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
1142 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
1143 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
1144 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
1145 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
1146 Jędrzejewski-Szmek, Сковорода Никита Андреевич
1150 * If /etc/vconsole.conf is non-existent or empty we will no
1151 longer load any console font or key map at boot by
1152 default. Instead the kernel defaults will be left
1153 intact. This is definitely the right thing to do, as no
1154 configuration should mean no configuration, and hard-coding
1155 font names that are different on all archs is probably a bad
1156 idea. Also, the kernel default key map and font should be
1157 good enough for most cases anyway, and mostly identical to
1158 the userspace fonts/key maps we previously overloaded them
1159 with. If distributions want to continue to default to a
1160 non-kernel font or key map they should ship a default
1161 /etc/vconsole.conf with the appropriate contents.
1163 Contributions from: Colin Walters, Daniel J Walsh, Dave
1164 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
1165 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1169 * journalctl gained a new --cursor= switch to show entries
1170 starting from the specified location in the journal.
1172 * We now enforce a size limit on journal entry fields exported
1173 with "-o json" in journalctl. Fields larger than 4K will be
1174 assigned null. This can be turned off with --all.
1176 * An (optional) journal gateway daemon is now available as
1177 "systemd-journal-gatewayd.service". This service provides
1178 access to the journal via HTTP and JSON. This functionality
1179 will be used to implement live log synchronization in both
1180 pull and push modes, but has various other users too, such
1181 as easy log access for debugging of embedded devices. Right
1182 now it is already useful to retrieve the journal via HTTP:
1184 # systemctl start systemd-journal-gatewayd.service
1185 # wget http://localhost:19531/entries
1187 This will download the journal contents in a
1188 /var/log/messages compatible format. The same as JSON:
1190 # curl -H"Accept: application/json" http://localhost:19531/entries
1192 This service is also accessible via a web browser where a
1193 single static HTML5 app is served that uses the JSON logic
1194 to enable the user to do some basic browsing of the
1195 journal. This will be extended later on. Here's an example
1196 screenshot of this app in its current state:
1198 http://0pointer.de/public/journal-gatewayd
1200 Contributions from: Kay Sievers, Lennart Poettering, Robert
1201 Milasan, Tom Gundersen
1205 * The bash completion logic is now available for journalctl
1208 * We don't mount the "cpuset" controller anymore together with
1209 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
1210 started if no parameters are assigned to it. "cpuset" hence
1211 broke code that assumed it it could create "cpu" groups and
1214 * journalctl -f will now subscribe to terminal size changes,
1215 and line break accordingly.
1217 Contributions from: Dave Reisner, Kay Sievers, Lennart
1218 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
1222 * nspawn will now create a symlink /etc/localtime in the
1223 container environment, copying the host's timezone
1224 setting. Previously this has been done via a bind mount, but
1225 since symlinks cannot be bind mounted this has now been
1226 changed to create/update the appropriate symlink.
1228 * journalctl -n's line number argument is now optional, and
1229 will default to 10 if omitted.
1231 * journald will now log the maximum size the journal files may
1232 take up on disk. This is particularly useful if the default
1233 built-in logic of determining this parameter from the file
1234 system size is used. Use "systemctl status
1235 systemd-journald.service" to see this information.
1237 * The multi-seat X wrapper tool has been stripped down. As X
1238 is now capable of enumerating graphics devices via udev in a
1239 seat-aware way the wrapper is not strictly necessary
1240 anymore. A stripped down temporary stop-gap is still shipped
1241 until the upstream display managers have been updated to
1242 fully support the new X logic. Expect this wrapper to be
1243 removed entirely in one of the next releases.
1245 * HandleSleepKey= in logind.conf has been split up into
1246 HandleSuspendKey= and HandleHibernateKey=. The old setting
1247 is not available anymore. X11 and the kernel are
1248 distuingishing between these keys and we should too. This
1249 also means the inhibition lock for these keys has been split
1252 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
1253 Poettering, Lukas Nykryn, Václav Pavlín
1257 * Whenever a unit changes state we'll now log this to the
1258 journal and show along the unit's own log output in
1261 * ConditionPathIsMountPoint= can now properly detect bind
1262 mount points too. (Previously, a bind mount of one file
1263 system to another place in the same file system could not be
1264 detected as mount, since they shared struct stat's st_dev
1267 * We will now mount the cgroup controllers cpu, cpuacct,
1268 cpuset and the controllers net_cls, net_prio together by
1271 * nspawn containers will now have a virtualized boot
1272 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
1273 over with a randomized ID at container initialization). This
1274 has the effect of making "journalctl -b" do the right thing
1277 * The JSON output journal serialization has been updated not
1278 to generate "endless" list objects anymore, but rather one
1279 JSON object per line. This is more in line how most JSON
1280 parsers expect JSON objects. The new output mode
1281 "json-pretty" has been added to provide similar output, but
1282 neatly aligned for readability by humans.
1284 * We dropped all explicit sync() invocations in the shutdown
1285 code. The kernel does this implicitly anyway in the kernel
1286 reboot() syscall. halt(8)'s -n option is now a compatibility
1289 * We now support virtualized reboot() in containers, as
1290 supported by newer kernels. We will fall back to exit() if
1291 CAP_SYS_REBOOT is not available to the container. Also,
1292 nspawn makes use of this now and will actually reboot the
1293 container if the containerized OS asks for that.
1295 * journalctl will only show local log output by default
1296 now. Use --merge (-m) to show remote log output, too.
1298 * libsystemd-journal gained the new sd_journal_get_usage()
1299 call to determine the current disk usage of all journal
1300 files. This is exposed in the new "journalctl --disk-usage"
1303 * journald gained a new configuration setting SplitMode= in
1304 journald.conf which may be used to control how user journals
1305 are split off. See journald.conf(5) for details.
1307 * A new condition type ConditionFileNotEmpty= has been added.
1309 * tmpfiles' "w" lines now support file globbing, to write
1310 multiple files at once.
1312 * We added Python bindings for the journal submission
1313 APIs. More Python APIs for a number of selected APIs will
1314 likely follow. Note that we intend to add native bindings
1315 only for the Python language, as we consider it common
1316 enough to deserve bindings shipped within systemd. There are
1317 various projects outside of systemd that provide bindings
1318 for languages such as PHP or Lua.
1320 * Many conditions will now resolve specifiers such as %i. In
1321 addition, PathChanged= and related directives of .path units
1322 now support specifiers as well.
1324 * There's now a new RPM macro definition for the system preset
1327 * journald will now warn if it can't forward a message to the
1328 syslog daemon because it's socket is full.
1330 * timedated will no longer write or process /etc/timezone,
1331 except on Debian. As we do not support late mounted /usr
1332 anymore /etc/localtime always being a symlink is now safe,
1333 and hence the information in /etc/timezone is not necessary
1336 * logind will now always reserve one VT for a text getty (VT6
1337 by default). Previously if more than 6 X sessions where
1338 started they took up all the VTs with auto-spawned gettys,
1339 so that no text gettys were available anymore.
1341 * udev will now automatically inform the btrfs kernel logic
1342 about btrfs RAID components showing up. This should make
1343 simple hotplug based btrfs RAID assembly work.
1345 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
1346 (but not for its children which will stay at the kernel
1347 default). This should allow setups with a lot more listening
1350 * systemd will now always pass the configured timezone to the
1351 kernel at boot. timedated will do the same when the timezone
1354 * logind's inhibition logic has been updated. By default,
1355 logind will now handle the lid switch, the power and sleep
1356 keys all the time, even in graphical sessions. If DEs want
1357 to handle these events on their own they should take the new
1358 handle-power-key, handle-sleep-key and handle-lid-switch
1359 inhibitors during their runtime. A simple way to achiveve
1360 that is to invoke the DE wrapped in an invocation of:
1362 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
1364 * Access to unit operations is now checked via SELinux taking
1365 the unit file label and client process label into account.
1367 * systemd will now notify the administrator in the journal
1368 when he over-mounts a non-empty directory.
1370 * There are new specifiers that are resolved in unit files,
1371 for the host name (%H), the machine ID (%m) and the boot ID
1374 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
1375 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
1376 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
1377 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
1378 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
1379 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
1380 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
1384 * Support for reading structured kernel messages from
1385 /dev/kmsg has now been added and is enabled by default.
1387 * Support for reading kernel messages from /proc/kmsg has now
1388 been removed. If you want kernel messages in the journal
1389 make sure to run a recent kernel (>= 3.5) that supports
1390 reading structured messages from /dev/kmsg (see
1391 above). /proc/kmsg is now exclusive property of classic
1392 syslog daemons again.
1394 * The libudev API gained the new
1395 udev_device_new_from_device_id() call.
1397 * The logic for file system namespace (ReadOnlyDirectory=,
1398 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
1399 require pivot_root() anymore. This means fewer temporary
1400 directories are created below /tmp for this feature.
1402 * nspawn containers will now see and receive all submounts
1403 made on the host OS below the root file system of the
1406 * Forward Secure Sealing is now supported for Journal files,
1407 which provide cryptographical sealing of journal files so
1408 that attackers cannot alter log history anymore without this
1409 being detectable. Lennart will soon post a blog story about
1410 this explaining it in more detail.
1412 * There are two new service settings RestartPreventExitStatus=
1413 and SuccessExitStatus= which allow configuration of exit
1414 status (exit code or signal) which will be excepted from the
1415 restart logic, resp. consider successful.
1417 * journalctl gained the new --verify switch that can be used
1418 to check the integrity of the structure of journal files and
1419 (if Forward Secure Sealing is enabled) the contents of
1422 * nspawn containers will now be run with /dev/stdin, /dev/fd/
1423 and similar symlinks pre-created. This makes running shells
1424 as container init process a lot more fun.
1426 * The fstab support can now handle PARTUUID= and PARTLABEL=
1429 * A new ConditionHost= condition has been added to match
1430 against the hostname (with globs) and machine ID. This is
1431 useful for clusters where a single OS image is used to
1432 provision a large number of hosts which shall run slightly
1433 different sets of services.
1435 * Services which hit the restart limit will now be placed in a
1438 Contributions from: Bertram Poettering, Dave Reisner, Huang
1439 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
1440 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
1444 * When running in --user mode systemd will now become a
1445 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
1446 tree a lot more organized.
1448 * A new PartOf= unit dependency type has been introduced that
1449 may be used to group services in a natural way.
1451 * "systemctl enable" may now be used to enable instances of
1454 * journalctl now prints error log levels in red, and
1455 warning/notice log levels in bright white. It also supports
1456 filtering by log level now.
1458 * cgtop gained a new -n switch (similar to top), to configure
1459 the maximum number of iterations to run for. It also gained
1460 -b, to run in batch mode (accepting no input).
1462 * The suffix ".service" may now be omitted on most systemctl
1463 command lines involving service unit names.
1465 * There's a new bus call in logind to lock all sessions, as
1466 well as a loginctl verb for it "lock-sessions".
1468 * libsystemd-logind.so gained a new call sd_journal_perror()
1469 that works similar to libc perror() but logs to the journal
1470 and encodes structured information about the error number.
1472 * /etc/crypttab entries now understand the new keyfile-size=
1475 * shutdown(8) now can send a (configurable) wall message when
1476 a shutdown is cancelled.
1478 * The mount propagation mode for the root file system will now
1479 default to "shared", which is useful to make containers work
1480 nicely out-of-the-box so that they receive new mounts from
1481 the host. This can be undone locally by running "mount
1482 --make-rprivate /" if needed.
1484 * The prefdm.service file has been removed. Distributions
1485 should maintain this unit downstream if they intend to keep
1486 it around. However, we recommend writing normal unit files
1487 for display managers instead.
1489 * Since systemd is a crucial part of the OS we will now
1490 default to a number of compiler switches that improve
1491 security (hardening) such as read-only relocations, stack
1492 protection, and suchlike.
1494 * The TimeoutSec= setting for services is now split into
1495 TimeoutStartSec= and TimeoutStopSec= to allow configuration
1496 of individual time outs for the start and the stop phase of
1499 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
1500 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
1501 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
1502 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
1503 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
1504 Gundersen, Zbigniew Jędrzejewski-Szmek
1508 * The journal and id128 C APIs are now fully documented as man
1511 * Extra safety checks have been added when transitioning from
1512 the initial RAM disk to the main system to avoid accidental
1515 * /etc/crypttab entries now understand the new keyfile-offset=
1518 * systemctl -t can now be used to filter by unit load state.
1520 * The journal C API gained the new sd_journal_wait() call to
1521 make writing synchronous journal clients easier.
1523 * journalctl gained the new -D switch to show journals from a
1526 * journalctl now displays a special marker between log
1527 messages of two different boots.
1529 * The journal is now explicitly flushed to /var via a service
1530 systemd-journal-flush.service, rather than implicitly simply
1531 by seeing /var/log/journal to be writable.
1533 * journalctl (and the journal C APIs) can now match for much
1534 more complex expressions, with alternatives and
1537 * When transitioning from the initial RAM disk to the main
1538 system we will now kill all processes in a killing spree to
1539 ensure no processes stay around by accident.
1541 * Three new specifiers may be used in unit files: %u, %h, %s
1542 resolve to the user name, user home directory resp. user
1543 shell. This is useful for running systemd user instances.
1545 * We now automatically rotate journal files if their data
1546 object hash table gets a fill level > 75%. We also size the
1547 hash table based on the configured maximum file size. This
1548 together should lower hash collisions drastically and thus
1549 speed things up a bit.
1551 * journalctl gained the new "--header" switch to introspect
1552 header data of journal files.
1554 * A new setting SystemCallFilters= has been added to services
1555 which may be used to apply blacklists or whitelists to
1556 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
1558 * nspawn gained a new --link-journal= switch (and quicker: -j)
1559 to link the container journal with the host. This makes it
1560 very easy to centralize log viewing on the host for all
1561 guests while still keeping the journal files separated.
1563 * Many bugfixes and optimizations
1565 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
1566 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
1567 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
1572 * Several tools now understand kernel command line arguments,
1573 which are only read when run in an initial RAM disk. They
1574 usually follow closely their normal counterparts, but are
1577 * There's a new tool to analyze the readahead files that are
1578 automatically generated at boot. Use:
1580 /usr/lib/systemd/systemd-readahead analyze /.readahead
1582 * We now provide an early debug shell on tty9 if this enabled. Use:
1584 systemctl enable debug-shell.service
1586 * All plymouth related units have been moved into the Plymouth
1587 package. Please make sure to upgrade your Plymouth version
1590 * systemd-tmpfiles now supports getting passed the basename of
1591 a configuration file only, in which case it will look for it
1592 in all appropriate directories automatically.
1594 * udevadm info now takes a /dev or /sys path as argument, and
1595 does the right thing. Example:
1597 udevadm info /dev/sda
1598 udevadm info /sys/class/block/sda
1600 * systemctl now prints a warning if a unit is stopped but a
1601 unit that might trigger it continues to run. Example: a
1602 service is stopped but the socket that activates it is left
1605 * "systemctl status" will now mention if the log output was
1606 shortened due to rotation since a service has been started.
1608 * The journal API now exposes functions to determine the
1609 "cutoff" times due to rotation.
1611 * journald now understands SIGUSR1 and SIGUSR2 for triggering
1612 immediately flushing of runtime logs to /var if possible,
1613 resp. for triggering immediate rotation of the journal
1616 * It is now considered an error if a service is attempted to
1617 be stopped that is not loaded.
1619 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
1621 * systemd-analyze now supports Python 3
1623 * tmpfiles now supports cleaning up directories via aging
1624 where the first level dirs are always kept around but
1625 directories beneath it automatically aged. This is enabled
1626 by prefixing the age field with '~'.
1628 * Seat objects now expose CanGraphical, CanTTY properties
1629 which is required to deal with very fast bootups where the
1630 display manager might be running before the graphics drivers
1631 completed initialization.
1633 * Seat objects now expose a State property.
1635 * We now include RPM macros for service enabling/disabling
1636 based on the preset logic. We recommend RPM based
1637 distributions to make use of these macros if possible. This
1638 makes it simpler to reuse RPM spec files across
1641 * We now make sure that the collected systemd unit name is
1642 always valid when services log to the journal via
1645 * There's a new man page kernel-command-line(7) detailing all
1646 command line options we understand.
1648 * The fstab generator may now be disabled at boot by passing
1649 fstab=0 on the kernel command line.
1651 * A new kernel command line option modules-load= is now understood
1652 to load a specific kernel module statically, early at boot.
1654 * Unit names specified on the systemctl command line are now
1655 automatically escaped as needed. Also, if file system or
1656 device paths are specified they are automatically turned
1657 into the appropriate mount or device unit names. Example:
1659 systemctl status /home
1660 systemctl status /dev/sda
1662 * The SysVConsole= configuration option has been removed from
1663 system.conf parsing.
1665 * The SysV search path is no longer exported on the D-Bus
1668 * The Names= option is been removed from unit file parsing.
1670 * There's a new man page bootup(7) detailing the boot process.
1672 * Every unit and every generator we ship with systemd now
1673 comes with full documentation. The self-explanatory boot is
1676 * A couple of services gained "systemd-" prefixes in their
1677 name if they wrap systemd code, rather than only external
1678 code. Among them fsck@.service which is now
1679 systemd-fsck@.service.
1681 * The HaveWatchdog property has been removed from the D-Bus
1684 * systemd.confirm_spawn= on the kernel command line should now
1687 * There's a new man page crypttab(5) which details all options
1688 we actually understand.
1690 * systemd-nspawn gained a new --capability= switch to pass
1691 additional capabilities to the container.
1693 * timedated will now read known NTP implementation unit names
1694 from /usr/lib/systemd/ntp-units.d/*.list,
1695 systemd-timedated-ntp.target has been removed.
1697 * journalctl gained a new switch "-b" that lists log data of
1698 the current boot only.
1700 * The notify socket is in the abstract namespace again, in
1701 order to support daemons which chroot() at start-up.
1703 * There is a new Storage= configuration option for journald
1704 which allows configuration of where log data should go. This
1705 also provides a way to disable journal logging entirely, so
1706 that data collected is only forwarded to the console, the
1707 kernel log buffer or another syslog implementation.
1709 * Many bugfixes and optimizations
1711 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
1712 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
1713 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
1714 Shawn Landden, Tom Gundersen
1718 * "systemctl help <unit>" now shows the man page if one is
1721 * Several new man pages have been added.
1723 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
1724 MaxLevelConsole= can now be specified in
1725 journald.conf. These options allow reducing the amount of
1726 data stored on disk or forwarded by the log level.
1728 * TimerSlackNSec= can now be specified in system.conf for
1729 PID1. This allows system-wide power savings.
1731 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
1732 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
1737 * logind is now capable of (optionally) handling power and
1738 sleep keys as well as the lid switch.
1740 * journalctl now understands the syntax "journalctl
1741 /usr/bin/avahi-daemon" to get all log output of a specific
1744 * CapabilityBoundingSet= in system.conf now also influences
1745 the capability bound set of usermode helpers of the kernel.
1747 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
1748 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
1749 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
1750 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
1754 * Note that we skipped 139 releases here in order to set the
1755 new version to something that is greater than both udev's
1756 and systemd's most recent version number.
1758 * udev: all udev sources are merged into the systemd source tree now.
1759 All future udev development will happen in the systemd tree. It
1760 is still fully supported to use the udev daemon and tools without
1761 systemd running, like in initramfs or other init systems. Building
1762 udev though, will require the *build* of the systemd tree, but
1763 udev can be properly *run* without systemd.
1765 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
1766 should be used to create dead device nodes as workarounds for broken
1769 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
1770 no longer supported. udev_monitor_new_from_netlink() needs to be
1771 used to subscribe to events.
1773 * udev: when udevd is started by systemd, processes which are left
1774 behind by forking them off of udev rules, are unconditionally cleaned
1775 up and killed now after the event handling has finished. Services or
1776 daemons must be started as systemd services. Services can be
1777 pulled-in by udev to get started, but they can no longer be directly
1778 forked by udev rules.
1780 * udev: the daemon binary is called systemd-udevd now and installed
1781 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
1782 to adapt to that, create symlink, or rename the binary after building
1785 * libudev no longer provides these symbols:
1786 udev_monitor_from_socket()
1787 udev_queue_get_failed_list_entry()
1788 udev_get_{dev,sys,run}_path()
1789 The versions number was bumped and symbol versioning introduced.
1791 * systemd-loginctl and systemd-journalctl have been renamed
1792 to loginctl and journalctl to match systemctl.
1794 * The config files: /etc/systemd/systemd-logind.conf and
1795 /etc/systemd/systemd-journald.conf have been renamed to
1796 logind.conf and journald.conf. Package updates should rename
1797 the files to the new names on upgrade.
1799 * For almost all files the license is now LGPL2.1+, changed
1800 from the previous GPL2.0+. Exceptions are some minor stuff
1801 of udev (which will be changed to LGPL2.1 eventually, too),
1802 and the MIT licensed sd-daemon.[ch] library that is suitable
1803 to be used as drop-in files.
1805 * systemd and logind now handle system sleep states, in
1806 particular suspending and hibernating.
1808 * logind now implements a sleep/shutdown/idle inhibiting logic
1809 suitable for a variety of uses. Soonishly Lennart will blog
1810 about this in more detail.
1812 * var-run.mount and var-lock.mount are no longer provided
1813 (which prevously bind mounted these directories to their new
1814 places). Distributions which have not converted these
1815 directories to symlinks should consider stealing these files
1816 from git history and add them downstream.
1818 * We introduced the Documentation= field for units and added
1819 this to all our shipped units. This is useful to make it
1820 easier to explore the boot and the purpose of the various
1823 * All smaller setup units (such as
1824 systemd-vconsole-setup.service) now detect properly if they
1825 are run in a container and are skipped when
1826 appropriate. This guarantees an entirely noise-free boot in
1827 Linux container environments such as systemd-nspawn.
1829 * A framework for implementing offline system updates is now
1830 integrated, for details see:
1831 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
1833 * A new service type Type=idle is available now which helps us
1834 avoiding ugly interleaving of getty output and boot status
1837 * There's now a system-wide CapabilityBoundingSet= option to
1838 globally reduce the set of capabilities for the
1839 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
1840 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
1841 even CAP_NET_ADMIN system-wide for secure systems.
1843 * There are now system-wide DefaultLimitXXX= options to
1844 globally change the defaults of the various resource limits
1845 for all units started by PID 1.
1847 * Harald Hoyer's systemd test suite has been integrated into
1848 systemd which allows easy testing of systemd builds in qemu
1849 and nspawn. (This is really awesome! Ask us for details!)
1851 * The fstab parser is now implemented as generator, not inside
1854 * systemctl will now warn you if .mount units generated from
1855 /etc/fstab are out of date due to changes in fstab that
1856 haven't been read by systemd yet.
1858 * systemd is now suitable for usage in initrds. Dracut has
1859 already been updated to make use of this. With this in place
1860 initrds get a slight bit faster but primarily are much
1861 easier to introspect and debug since "systemctl status" in
1862 the host system can be used to introspect initrd services,
1863 and the journal from the initrd is kept around too.
1865 * systemd-delta has been added, a tool to explore differences
1866 between user/admin configuration and vendor defaults.
1868 * PrivateTmp= now affects both /tmp and /var/tmp.
1870 * Boot time status messages are now much prettier and feature
1871 proper english language. Booting up systemd has never been
1874 * Read-ahead pack files now include the inode number of all
1875 files to pre-cache. When the inode changes the pre-caching
1876 is not attempted. This should be nicer to deal with updated
1877 packages which might result in changes of read-ahead
1880 * We now temporaritly lower the kernel's read_ahead_kb variable
1881 when collecting read-ahead data to ensure the kernel's
1882 built-in read-ahead does not add noise to our measurements
1883 of necessary blocks to pre-cache.
1885 * There's now RequiresMountsFor= to add automatic dependencies
1886 for all mounts necessary for a specific file system path.
1888 * MountAuto= and SwapAuto= have been removed from
1889 system.conf. Mounting file systems at boot has to take place
1892 * nspawn now learned a new switch --uuid= to set the machine
1893 ID on the command line.
1895 * nspawn now learned the -b switch to automatically search
1898 * vt102 is now the default TERM for serial TTYs, upgraded from
1901 * systemd-logind now works on VT-less systems.
1903 * The build tree has been reorganized. The individual
1904 components now have directories of their own.
1906 * A new condition type ConditionPathIsReadWrite= is now available.
1908 * nspawn learned the new -C switch to create cgroups for the
1909 container in other hierarchies.
1911 * We now have support for hardware watchdogs, configurable in
1914 * The scheduled shutdown logic now has a public API.
1916 * We now mount /tmp as tmpfs by default, but this can be
1917 masked and /etc/fstab can override it.
1919 * Since udisks doesn't make use of /media anymore we are not
1920 mounting a tmpfs on it anymore.
1922 * journalctl gained a new --local switch to only interleave
1923 locally generated journal files.
1925 * We can now load the IMA policy at boot automatically.
1927 * The GTK tools have been split off into a systemd-ui.
1929 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
1930 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
1931 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
1932 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
1933 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
1934 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
1935 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
1936 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
1937 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
1942 * This is mostly a bugfix release
1944 * Support optional initialization of the machine ID from the
1945 KVM or container configured UUID.
1947 * Support immediate reboots with "systemctl reboot -ff"
1949 * Show /etc/os-release data in systemd-analyze output
1951 * Many bugfixes for the journal, including endianness fixes and
1952 ensuring that disk space enforcement works
1954 * sd-login.h is C++ comptaible again
1956 * Extend the /etc/os-release format on request of the Debian
1959 * We now refuse non-UTF8 strings used in various configuration
1960 and unit files. This is done to ensure we don't pass invalid
1961 data over D-Bus or expose it elsewhere.
1963 * Register Mimo USB Screens as suitable for automatic seat
1966 * Read SELinux client context from journal clients in a race
1969 * Reorder configuration file lookup order. /etc now always
1970 overrides /run in order to allow the administrator to always
1971 and unconditionally override vendor supplied or
1972 automatically generated data.
1974 * The various user visible bits of the journal now have man
1975 pages. We still lack man pages for the journal API calls
1978 * We now ship all man pages in HTML format again in the
1981 Contributions from: Dave Reisner, Dirk Eibach, Frederic
1982 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
1983 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
1988 * This is mostly a bugfix release
1990 * systems lacking /etc/os-release are no longer supported.
1992 * Various functionality updates to libsystemd-login.so
1994 * Track class of PAM logins to distuingish greeters from
1997 Contributions from: Kay Sievers, Lennart Poettering, Michael
2002 * This is an important bugfix release for v41.
2004 * Building man pages is now optional which should be useful
2005 for those building systemd from git but unwilling to install
2008 * Watchdog support for supervising services is now usable. In
2009 a future release support for hardware watchdogs
2010 (i.e. /dev/watchdog) will be added building on this.
2012 * Service start rate limiting is now configurable and can be
2013 turned off per service. When a start rate limit is hit a
2014 reboot can automatically be triggered.
2016 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
2018 Contributions from: Benjamin Franzke, Bill Nottingham,
2019 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
2020 Schmidt, Michał Górny, Piotr Drąg
2024 * The systemd binary is installed /usr/lib/systemd/systemd now;
2025 An existing /sbin/init symlink needs to be adapted with the
2028 * The code that loads kernel modules has been ported to invoke
2029 libkmod directly, instead of modprobe. This means we do not
2030 support systems with module-init-tools anymore.
2032 * Watchdog support is now already useful, but still not
2035 * A new kernel command line option systemd.setenv= is
2036 understood to set system wide environment variables
2037 dynamically at boot.
2039 * We now limit the set of capabilities of systemd-journald.
2041 * We now set SIGPIPE to ignore by default, since it only is
2042 useful in shell pipelines, and has little use in general
2043 code. This can be disabled with IgnoreSIPIPE=no in unit
2046 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
2047 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
2052 * This is mostly a bugfix release
2054 * We now expose the reason why a service failed in the
2055 "Result" D-Bus property.
2057 * Rudimentary service watchdog support (will be completed over
2058 the next few releases.)
2060 * When systemd forks off in order execute some service we will
2061 now immediately changes its argv[0] to reflect which process
2062 it will execute. This is useful to minimize the time window
2063 with a generic argv[0], which makes bootcharts more useful
2065 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
2066 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
2067 Mike Kazantsev, Ray Strode
2071 * This is mostly a test release, but incorporates many
2074 * New systemd-cgtop tool to show control groups by their
2077 * Linking against libacl for ACLs is optional again. If
2078 disabled, support tracking device access for active logins
2079 goes becomes unavailable, and so does access to the user
2080 journals by the respective users.
2082 * If a group "adm" exists, journal files are automatically
2083 owned by them, thus allow members of this group full access
2084 to the system journal as well as all user journals.
2086 * The journal now stores the SELinux context of the logging
2087 client for all entries.
2089 * Add C++ inclusion guards to all public headers
2091 * New output mode "cat" in the journal to print only text
2092 messages, without any meta data like date or time.
2094 * Include tiny X server wrapper as a temporary stop-gap to
2095 teach XOrg udev display enumeration. This is used by display
2096 managers such as gdm, and will go away as soon as XOrg
2097 learned native udev hotplugging for display devices.
2099 * Add new systemd-cat tool for executing arbitrary programs
2100 with STDERR/STDOUT connected to the journal. Can also act as
2101 BSD logger replacement, and does so by default.
2103 * Optionally store all locally generated coredumps in the
2104 journal along with meta data.
2106 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
2107 writing short strings to files (for usage for /sys), and for
2108 creating symlinks, character and block device nodes.
2110 * New unit file option ControlGroupPersistent= to make cgroups
2111 persistent, following the mechanisms outlined in
2112 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
2114 * Support multiple local RTCs in a sane way
2116 * No longer monopolize IO when replaying readahead data on
2117 rotating disks, since we might starve non-file-system IO to
2118 death, since fanotify() will not see accesses done by blkid,
2121 * Don't show kernel threads in systemd-cgls anymore, unless
2122 requested with new -k switch.
2124 Contributions from: Dan Horák, Kay Sievers, Lennart
2125 Poettering, Michal Schmidt
2129 * This is mostly a test release, but incorporates many
2132 * The git repository moved to:
2133 git://anongit.freedesktop.org/systemd/systemd
2134 ssh://git.freedesktop.org/git/systemd/systemd
2136 * First release with the journal
2137 http://0pointer.de/blog/projects/the-journal.html
2139 * The journal replaces both systemd-kmsg-syslogd and
2140 systemd-stdout-bridge.
2142 * New sd_pid_get_unit() API call in libsystemd-logind
2144 * Many systemadm clean-ups
2146 * Introduce remote-fs-pre.target which is ordered before all
2147 remote mounts and may be used to start services before all
2150 * Added Mageia support
2152 * Add bash completion for systemd-loginctl
2154 * Actively monitor PID file creation for daemons which exit in
2155 the parent process before having finished writing the PID
2156 file in the daemon process. Daemons which do this need to be
2157 fixed (i.e. PID file creation must have finished before the
2158 parent exits), but we now react a bit more gracefully to them.
2160 * Add colourful boot output, mimicking the well-known output
2161 of existing distributions.
2163 * New option PassCredentials= for socket units, for
2164 compatibility with a recent kernel ABI breakage.
2166 * /etc/rc.local is now hooked in via a generator binary, and
2167 thus will no longer act as synchronization point during
2170 * systemctl list-unit-files now supports --root=.
2172 * systemd-tmpfiles now understands two new commands: z, Z for
2173 relabelling files according to the SELinux database. This is
2174 useful to apply SELinux labels to specific files in /sys,
2177 * Output of SysV services is now forwarded to both the console
2178 and the journal by default, not only just the console.
2180 * New man pages for all APIs from libsystemd-login.
2182 * The build tree got reorganized and a the build system is a
2183 lot more modular allowing embedded setups to specifically
2184 select the components of systemd they are interested in.
2186 * Support for Linux systems lacking the kernel VT subsystem is
2189 * configure's --with-rootdir= got renamed to
2190 --with-rootprefix= to follow the naming used by udev and
2193 * Unless specified otherwise we'll now install to /usr instead
2194 of /usr/local by default.
2196 * Processes with '@' in argv[0][0] are now excluded from the
2197 final shut-down killing spree, following the logic explained
2199 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
2201 * All processes remaining in a service cgroup when we enter
2202 the START or START_PRE states are now killed with
2203 SIGKILL. That means it is no longer possible to spawn
2204 background processes from ExecStart= lines (which was never
2205 supported anyway, and bad style).
2207 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
2208 reloading of units together.
2210 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
2211 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
2212 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
2213 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
2214 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
~ianmdlvl / elogind.git / blob