1 systemd System and Service Manager
5 * The Python bindings gained some minimal support for the APIs
6 exposed by libsystemd-logind.
8 * ConditionSecurity= gained support for detecting SMACK. Since
9 this condition already supports SELinux and AppArmor we only
10 miss IMA for this. Patches welcome!
12 Contributions from: Karol Lewandowski, Lennart Poettering,
13 Zbigniew Jędrzejewski-Szmek
17 * systemd-nspawn will now create /etc/resolv.conf if
18 necessary, before bind-mounting the host's file onto it.
20 * systemd-nspawn will now store meta information about a
21 container on the container's cgroup as extended attribute
22 fields, including the root directory.
24 * The cgroup hierarchy has been reworked in many ways. All
25 objects any of the components systemd creates in the cgroup
26 tree are now suffixed. More specifically, user sessions are
27 now placed in cgroups suffixed with ".session", users in
28 cgroups suffixed with ".user", and nspawn containers in
29 cgroups suffixed with ".nspawn". Furthermore, all cgroup
30 names are now escaped in a simple scheme to avoid collision
31 of userspace object names with kernel filenames. This work
32 is preparation for making these objects relocatable in the
33 cgroup tree, in order to allow easy resource partitioning of
34 these objects without causing naming conflicts.
36 * systemctl list-dependencies gained the new switches
37 --plain, --reverse, --after and --before.
39 * systemd-inhibit now shows the process name of processes that
40 have taken an inhibitor lock.
42 * nss-myhostname will now also resolve "localhost"
43 implicitly. This makes /etc/hosts an optional file and
44 nicely handles that on IPv6 ::1 maps to both "localhost" and
47 * libsystemd-logind.so gained a new call
48 sd_get_machine_names() to enumerate running containers and
49 VMs (currently only supported by very new libvirt and
50 nspawn). sd_login_monitor can now be used to watch
51 VMs/containers coming and going.
53 * .include is not allowed recursively anymore, and only in
54 unit files. Usually it is better to use drop-in snippets in
55 .d/*.conf anyway, as introduced with systemd 198.
57 * systemd-analyze gained a new "critical-chain" command that
58 determines the slowest chain of units run during system
59 boot-up. It is very useful for tracking down where
60 optimizing boot time is the most beneficial.
62 * systemd will no longer allow manipulating service paths in
63 the name=systemd:/system cgroup tree using ControlGroup= in
64 units. (But is still fine with it in all other dirs.)
66 * There's a new systemd-nspawn@.service service file that may
67 be used to easily run nspawn containers as system
68 services. With the container's root directory in
69 /var/lib/container/foobar it is now sufficient to run
70 "systemctl start systemd-nspawn@foobar.service" to boot it.
72 * systemd-cgls gained a new parameter "--machine" to list only
73 the processes within a certain container.
75 * ConditionSecurity= now can check for "apparmor". We still
76 are lacking checks for SMACK and IMA for this condition
77 check though. Patches welcome!
79 * A new configuration file /etc/systemd/sleep.conf has been
80 added that may be used to configure which kernel operation
81 systemd is supposed to execute when "suspend", "hibernate"
82 or "hybrid-sleep" is requested. This makes the new kernel
83 "freeze" state accessible to the user.
85 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
86 the passed argument if applicable.
88 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
89 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
90 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
91 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
92 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
93 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
94 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
99 * The output of 'systemctl list-jobs' got some polishing. The
100 '--type=' argument may now be passed more than once. A new
101 command 'systemctl list-sockets' has been added which shows
102 a list of kernel sockets systemd is listening on with the
103 socket units they belong to, plus the units these socket
106 * The experimental libsystemd-bus library got substantial
107 updates to work in conjunction with the (also experimental)
108 kdbus kernel project. It works well enough to exchange
109 messages with some sophistication. Note that kdbus is not
110 ready yet, and the library is mostly an elaborate test case
111 for now, and not installable.
113 * systemd gained a new unit 'systemd-static-nodes.service'
114 that generates static device nodes earlier during boot, and
115 can run in conjunction with udev.
117 * libsystemd-login gained a new call sd_pid_get_user_unit()
118 to retrieve the user systemd unit a process is running
119 in. This is useful for systems where systemd is used as
122 * systemd-nspawn now places all containers in the new /machine
123 top-level cgroup directory in the name=systemd
124 hierarchy. libvirt will soon do the same, so that we get a
125 uniform separation of /system, /user and /machine for system
126 services, user processes and containers/virtual
127 machines. This new cgroup hierarchy is also useful to stick
128 stable names to specific container instances, which can be
129 recognized later this way (this name may be controlled
130 via systemd-nspawn's new -M switch). libsystemd-login also
131 gained a new call sd_pid_get_machine_name() to retrieve the
132 name of the container/VM a specific process belongs to.
134 * bootchart can now store its data in the journal.
136 * libsystemd-journal gained a new call
137 sd_journal_add_conjunction() for AND expressions to the
138 matching logic. This can be used to express more complex
141 * journactl can now take multiple --unit= and --user-unit=
144 * The cryptsetup logic now understands the "luks.key=" kernel
145 command line switch for specifying a file to read the
146 decryption key from. Also, if a configured key file is not
147 found the tool will now automatically fall back to prompting
150 * Python systemd.journal module was updated to wrap recently
151 added functions from libsystemd-journal. The interface was
152 changed to bring the low level interface in s.j._Reader
153 closer to the C API, and the high level interface in
154 s.j.Reader was updated to wrap and convert all data about
157 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
158 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
159 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
160 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
161 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
162 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
166 * journalctl --update-catalog now understands a new --root=
167 option to operate on catalogs found in a different root
170 * During shutdown after systemd has terminated all running
171 services a final killing loop kills all remaining left-over
172 processes. We will now print the name of these processes
173 when we send SIGKILL to them, since this usually indicates a
176 * If /etc/crypttab refers to password files stored on
177 configured mount points automatic dependencies will now be
178 generated to ensure the specific mount is established first
179 before the key file is attempted to be read.
181 * 'systemctl status' will now show information about the
182 network sockets a socket unit is listening on.
184 * 'systemctl status' will also shown information about any
185 drop-in configuration file for units. (Drop-In configuration
186 files in this context are files such as
187 /etc/systemd/systemd/foobar.service.d/*.conf)
189 * systemd-cgtop now optionally shows summed up CPU times of
190 cgroups. Press '%' while running cgtop to switch between
191 percentage and absolute mode. This is useful to determine
192 which cgroups use up the most CPU time over the entire
193 runtime of the system. systemd-cgtop has also been updated
194 to be 'pipeable' for processing with further shell tools.
196 * 'hostnamectl set-hostname' will now allow setting of FQDN
199 * The formatting and parsing of time span values has been
200 changed. The parser now understands fractional expressions
201 such as "5.5h". The formatter will now output fractional
202 expressions for all time spans under 1min, i.e. "5.123456s"
203 rather than "5s 123ms 456us". For time spans under 1s
204 millisecond values are shown, for those under 1ms
205 microsecond values are shown. This should greatly improve
206 all time-related output of systemd.
208 * libsystemd-login and libsystemd-journal gained new
209 functions for querying the poll() events mask and poll()
210 timeout value for integration into arbitrary event
213 * localectl gained the ability to list available X11 keymaps
214 (models, layouts, variants, options).
216 * 'systemd-analyze dot' gained the ability to filter for
217 specific units via shell-style globs, to create smaller,
218 more useful graphs. I.e. it's now possible to create simple
219 graphs of all the dependencies between only target units, or
220 of all units that Avahi has dependencies with.
222 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
223 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
224 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
225 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
226 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
227 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
228 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
232 * The boot-time readahead implementation for rotating media
233 will now read the read-ahead data in multiple passes which
234 consist of all read requests made in equidistant time
235 intervals. This means instead of strictly reading read-ahead
236 data in its physical order on disk we now try to find a
237 middle ground between physical and access time order.
239 * /etc/os-release files gained a new BUILD_ID= field for usage
240 on operating systems that provide continuous builds of OS
243 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
244 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
245 William Douglas, Zbigniew Jędrzejewski-Szmek
249 * systemd-python gained an API exposing libsystemd-daemon.
251 * The SMACK setup logic gained support for uploading CIPSO
254 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
255 ReadOnlyDirectories= and InaccessibleDirectories= has
256 changed. The private /tmp and /var/tmp directories are now
257 shared by all processes of a service (which means
258 ExecStartPre= may now leave data in /tmp that ExecStart= of
259 the same service can still access). When a service is
260 stopped its temporary directories are immediately deleted
261 (normal clean-up with tmpfiles is still done in addition to
264 * By default, systemd will now set a couple of sysctl
265 variables in the kernel: the safe sysrq options are turned
266 on, IP route verification is turned on, and source routing
267 disabled. The recently added hardlink and softlink
268 protection of the kernel is turned on. These settings should
269 be reasonably safe, and good defaults for all new systems.
271 * The predictable network naming logic may now be turned off
272 with a new kernel command line switch: net.ifnames=0.
274 * A new libsystemd-bus module has been added that implements a
275 pretty complete D-Bus client library. For details see:
277 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
279 * journald will now explicitly flush the journal files to disk
280 at the latest 5min after each write. The file will then also
281 be marked offline until the next write. This should increase
282 reliability in case of a crash. The synchronization delay
283 can be configured via SyncIntervalSec= in journald.conf.
285 * There's a new remote-fs-setup.target unit that can be used
286 to pull in specific services when at least one remote file
287 system is to be mounted.
289 * There are new targets timers.target and paths.target as
290 canonical targets to pull user timer and path units in
291 from. This complements sockets.target with a similar
292 purpose for socket units.
294 * libudev gained a new call udev_device_set_attribute_value()
295 to set sysfs attributes of a device.
297 * The udev daemon now sets the default number of worker
298 processes executed in parallel based on the number of available
299 CPUs instead of the amount of available RAM. This is supposed
300 to provide a more reliable default and limit a too aggressive
301 paralellism for setups with 1000s of devices connected.
303 Contributions from: Auke Kok, Colin Walters, Cristian
304 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
305 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
306 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
307 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
308 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
309 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
310 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
311 Zbigniew Jędrzejewski-Szmek
315 * Configuration of unit files may now be extended via drop-in
316 files without having to edit/override the unit files
317 themselves. More specifically, if the administrator wants to
318 change one value for a service file foobar.service he can
319 now do so by dropping in a configuration snippet into
320 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
321 will load all these snippets and apply them on top of the
322 main unit configuration file, possibly extending or
323 overriding its settings. Using these drop-in snippets is
324 generally nicer than the two earlier options for changing
325 unit files locally: copying the files from
326 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
327 them there; or creating a new file in /etc/systemd/system/
328 that incorporates the original one via ".include". Drop-in
329 snippets into these .d/ directories can be placed in any
330 directory systemd looks for units in, and the usual
331 overriding semantics between /usr/lib, /etc and /run apply
334 * Most unit file settings which take lists of items can now be
335 reset by assigning the empty string to them. For example,
336 normally, settings such as Environment=FOO=BAR append a new
337 environment variable assignment to the environment block,
338 each time they are used. By assigning Environment= the empty
339 string the environment block can be reset to empty. This is
340 particularly useful with the .d/*.conf drop-in snippets
341 mentioned above, since this adds the ability to reset list
342 settings from vendor unit files via these drop-ins.
344 * systemctl gained a new "list-dependencies" command for
345 listing the dependencies of a unit recursively.
347 * Inhibitors are now honored and listed by "systemctl
348 suspend", "systemctl poweroff" (and similar) too, not only
349 GNOME. These commands will also list active sessions by
352 * Resource limits (as exposed by the various control group
353 controllers) can now be controlled dynamically at runtime
354 for all units. More specifically, you can now use a command
355 like "systemctl set-cgroup-attr foobar.service cpu.shares
356 2000" to alter the CPU shares a specific service gets. These
357 settings are stored persistently on disk, and thus allow the
358 administrator to easily adjust the resource usage of
359 services with a few simple commands. This dynamic resource
360 management logic is also available to other programs via the
361 bus. Almost any kernel cgroup attribute and controller is
364 * systemd-vconsole-setup will now copy all font settings to
365 all allocated VTs, where it previously applied them only to
368 * libsystemd-login gained the new sd_session_get_tty() API
371 * This release drops support for a few legacy or
372 distribution-specific LSB facility names when parsing init
373 scripts: $x-display-manager, $mail-transfer-agent,
374 $mail-transport-agent, $mail-transfer-agent, $smtp,
375 $null. Also, the mail-transfer-agent.target unit backing
376 this has been removed. Distributions which want to retain
377 compatibility with this should carry the burden for
378 supporting this themselves and patch support for these back
379 in, if they really need to. Also, the facilities $syslog and
380 $local_fs are now ignored, since systemd does not support
381 early-boot LSB init scripts anymore, and these facilities
382 are implied anyway for normal services. syslog.target has
385 * There are new bus calls on PID1's Manager object for
386 cancelling jobs, and removing snapshot units. Previously,
387 both calls were only available on the Job and Snapshot
390 * systemd-journal-gatewayd gained SSL support.
392 * The various "environment" files, such as /etc/locale.conf
393 now support continuation lines with a backslash ("\") as
394 last character in the line, similar in style (but different)
395 to how this is supported in shells.
397 * For normal user processes the _SYSTEMD_USER_UNIT= field is
398 now implicitly appended to every log entry logged. systemctl
399 has been updated to filter by this field when operating on a
400 user systemd instance.
402 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
403 CAP_AUDIT_CONTROL capabilities to the capabilities set for
404 the container. This makes it easier to boot unmodified
405 Fedora systems in a container, which however still requires
406 audit=0 to be passed on the kernel command line. Auditing in
407 kernel and userspace is unfortunately still too broken in
408 context of containers, hence we recommend compiling it out
409 of the kernel or using audit=0. Hopefully this will be fixed
410 one day for good in the kernel.
412 * nspawn gained the new --bind= and --bind-ro= parameters to
413 bind mount specific directories from the host into the
416 * nspawn will now mount its own devpts file system instance
417 into the container, in order not to leak pty devices from
418 the host into the container.
420 * systemd will now read the firmware boot time performance
421 information from the EFI variables, if the used boot loader
422 supports this, and takes it into account for boot performance
423 analysis via "systemd-analyze". This is currently supported
424 only in conjunction with Gummiboot, but could be supported
425 by other boot loaders too. For details see:
427 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
429 * A new generator has been added that automatically mounts the
430 EFI System Partition (ESP) to /boot, if that directory
431 exists, is empty, and no other file system has been
432 configured to be mounted there.
434 * logind will now send out PrepareForSleep(false) out
435 unconditionally, after coming back from suspend. This may be
436 used by applications as asynchronous notification for
437 system resume events.
439 * "systemctl unlock-sessions" has been added, that allows
440 unlocking the screens of all user sessions at once, similar
441 how "systemctl lock-sessions" already locked all users
442 sessions. This is backed by a new D-Bus call UnlockSessions().
444 * "loginctl seat-status" will now show the master device of a
445 seat. (i.e. the device of a seat that needs to be around for
446 the seat to be considered available, usually the graphics
449 * tmpfiles gained a new "X" line type, that allows
450 configuration of files and directories (with wildcards) that
451 shall be excluded from automatic cleanup ("aging").
453 * udev default rules set the device node permissions now only
454 at "add" events, and do not change them any longer with a
455 later "change" event.
457 * The log messages for lid events and power/sleep keypresses
458 now carry a message ID.
460 * We now have a substantially larger unit test suite, but this
461 continues to be work in progress.
463 * udevadm hwdb gained a new --root= parameter to change the
464 root directory to operate relative to.
466 * logind will now issue a background sync() request to the kernel
467 early at shutdown, so that dirty buffers are flushed to disk early
468 instead of at the last moment, in order to optimize shutdown
471 * A new bootctl tool has been added that is an interface for
472 certain boot loader operations. This is currently a preview
473 and is likely to be extended into a small mechanism daemon
474 like timedated, localed, hostnamed, and can be used by
475 graphical UIs to enumerate available boot options, and
476 request boot into firmware operations.
478 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
479 the rest of the package. It also has been updated to work
480 correctly in initrds.
482 * Policykit previously has been runtime optional, and is now
483 also compile time optional via a configure switch.
485 * systemd-analyze has been reimplemented in C. Also "systemctl
486 dot" has moved into systemd-analyze.
488 * "systemctl status" with no further parameters will now print
489 the status of all active or failed units.
491 * Operations such as "systemctl start" can now be executed
492 with a new mode "--irreversible" which may be used to queue
493 operations that cannot accidentally be reversed by a later
494 job queuing. This is by default used to make shutdown
495 requests more robust.
497 * The Python API of systemd now gained a new module for
498 reading journal files.
500 * A new tool kernel-install has been added that can install
501 kernel images according to the Boot Loader Specification:
503 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
505 * Boot time console output has been improved to provide
506 animated boot time output for hanging jobs.
508 * A new tool systemd-activate has been added which can be used
509 to test socket activation with, directly from the command
510 line. This should make it much easier to test and debug
511 socket activation in daemons.
513 * journalctl gained a new "--reverse" (or -r) option to show
514 journal output in reverse order (i.e. newest line first).
516 * journalctl gained a new "--pager-end" (or -e) option to jump
517 to immediately jump to the end of the journal in the
518 pager. This is only supported in conjunction with "less".
520 * journalctl gained a new "--user-unit=" option, that works
521 similar to "--unit=" but filters for user units rather than
524 * A number of unit files to ease adoption of systemd in
525 initrds has been added. This moves some minimal logic from
526 the various initrd implementations into systemd proper.
528 * The journal files are now owned by a new group
529 "systemd-journal", which exists specifically to allow access
530 to the journal, and nothing else. Previously, we used the
531 "adm" group for that, which however possibly covers more
532 than just journal/log file access. This new group is now
533 already used by systemd-journal-gatewayd to ensure this
534 daemon gets access to the journal files and as little else
535 as possible. Note that "make install" will also set FS ACLs
536 up for /var/log/journal to give "adm" and "wheel" read
537 access to it, in addition to "systemd-journal" which owns
538 the journal files. We recommend that packaging scripts also
539 add read access to "adm" + "wheel" to /var/log/journal, and
540 all existing/future journal files. To normal users and
541 administrators little changes, however packagers need to
542 ensure to create the "systemd-journal" system group at
543 package installation time.
545 * The systemd-journal-gatewayd now runs as unprivileged user
546 systemd-journal-gateway:systemd-journal-gateway. Packaging
547 scripts need to create these system user/group at
550 * timedated now exposes a new boolean property CanNTP that
551 indicates whether a local NTP service is available or not.
553 * systemd-detect-virt will now also detect xen PVs
555 * The pstore file system is now mounted by default, if it is
558 * In addition to the SELinux and IMA policies we will now also
559 load SMACK policies at early boot.
561 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
562 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
563 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
564 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
565 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
566 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
567 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
568 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
569 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
570 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
571 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
572 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
573 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
574 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
578 * Timer units now support calendar time events in addition to
579 monotonic time events. That means you can now trigger a unit
580 based on a calendar time specification such as "Thu,Fri
581 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
582 or fifth day of any month of the year 2013, given that it is
583 a thursday or friday. This brings timer event support
584 considerably closer to cron's capabilities. For details on
585 the supported calendar time specification language see
588 * udev now supports a number of different naming policies for
589 network interfaces for predictable names, and a combination
590 of these policies is now the default. Please see this wiki
591 document for details:
593 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
595 * Auke Kok's bootchart implementation has been added to the
596 systemd tree. It's an optional component that can graph the
597 boot in quite some detail. It's one of the best bootchart
598 implementations around and minimal in its code and
601 * nss-myhostname has been integrated into the systemd source
602 tree. nss-myhostname guarantees that the local hostname
603 always stays resolvable via NSS. It has been a weak
604 requirement of systemd-hostnamed since a long time, and
605 since its code is actually trivial we decided to just
606 include it in systemd's source tree. It can be turned off
607 with a configure switch.
609 * The read-ahead logic is now capable of properly detecting
610 whether a btrfs file system is on SSD or rotating media, in
611 order to optimize the read-ahead scheme. Previously, it was
612 only capable of detecting this on traditional file systems
615 * In udev, additional device properties are now read from the
616 IAB in addition to the OUI database. Also, Bluetooth company
617 identities are attached to the devices as well.
619 * In service files %U may be used as specifier that is
620 replaced by the configured user name of the service.
622 * nspawn may now be invoked without a controlling TTY. This
623 makes it suitable for invocation as its own service. This
624 may be used to set up a simple containerized server system
625 using only core OS tools.
627 * systemd and nspawn can now accept socket file descriptors
628 when they are started for socket activation. This enables
629 implementation of socket activated nspawn
630 containers. i.e. think about autospawning an entire OS image
631 when the first SSH or HTTP connection is received. We expect
632 that similar functionality will also be added to libvirt-lxc
635 * journalctl will now suppress ANSI color codes when
638 * systemctl will no longer show control group information for
639 a unit if a the control group is empty anyway.
641 * logind can now automatically suspend/hibernate/shutdown the
644 * /etc/machine-info and hostnamed now also expose the chassis
645 type of the system. This can be used to determine whether
646 the local system is a laptop, desktop, handset or
647 tablet. This information may either be configured by the
648 user/vendor or is automatically determined from ACPI and DMI
649 information if possible.
651 * A number of PolicyKit actions are now bound together with
652 "imply" rules. This should simplify creating UIs because
653 many actions will now authenticate similar ones as well.
655 * Unit files learnt a new condition ConditionACPower= which
656 may be used to conditionalize a unit depending on whether an
657 AC power source is connected or not, of whether the system
658 is running on battery power.
660 * systemctl gained a new "is-failed" verb that may be used in
661 shell scripts and suchlike to check whether a specific unit
662 is in the "failed" state.
664 * The EnvironmentFile= setting in unit files now supports file
665 globbing, and can hence be used to easily read a number of
666 environment files at once.
668 * systemd will no longer detect and recognize specific
669 distributions. All distribution-specific #ifdeffery has been
670 removed, systemd is now fully generic and
671 distribution-agnostic. Effectively, not too much is lost as
672 a lot of the code is still accessible via explicit configure
673 switches. However, support for some distribution specific
674 legacy configuration file formats has been dropped. We
675 recommend distributions to simply adopt the configuration
676 files everybody else uses now and convert the old
677 configuration from packaging scripts. Most distributions
678 already did that. If that's not possible or desirable,
679 distributions are welcome to forward port the specific
680 pieces of code locally from the git history.
682 * When logging a message about a unit systemd will now always
683 log the unit name in the message meta data.
685 * localectl will now also discover system locale data that is
686 not stored in locale archives, but directly unpacked.
688 * logind will no longer unconditionally use framebuffer
689 devices as seat masters, i.e. as devices that are required
690 to be existing before a seat is considered preset. Instead,
691 it will now look for all devices that are tagged as
692 "seat-master" in udev. By default framebuffer devices will
693 be marked as such, but depending on local systems other
694 devices might be marked as well. This may be used to
695 integrate graphics cards using closed source drivers (such
696 as NVidia ones) more nicely into logind. Note however, that
697 we recommend using the open source NVidia drivers instead,
698 and no udev rules for the closed-source drivers will be
699 shipped from us upstream.
701 Contributions from: Adam Williamson, Alessandro Crismani, Auke
702 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
703 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
704 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
705 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
706 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
707 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
708 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
709 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
710 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
711 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
712 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
717 * udev gained support for loading additional device properties
718 from an indexed database that is keyed by vendor/product IDs
719 and similar device identifiers. For the beginning this
720 "hwdb" is populated with data from the well-known PCI and
721 USB database, but also includes PNP, ACPI and OID data. In
722 the longer run this indexed database shall grow into
723 becoming the one central database for non-essential
724 userspace device metadata. Previously, data from the PCI/USB
725 database was only attached to select devices, since the
726 lookup was a relatively expensive operation due to O(n) time
727 complexity (with n being the number of entries in the
728 database). Since this is now O(1), we decided to add in this
729 data for all devices where this is available, by
730 default. Note that the indexed database needs to be rebuilt
731 when new data files are installed. To achieve this you need
732 to update your packaging scripts to invoke "udevadm hwdb
733 --update" after installation of hwdb data files. For
734 RPM-based distributions we introduced the new
735 %udev_hwdb_update macro for this purpose.
737 * The Journal gained support for the "Message Catalog", an
738 indexed database to link up additional information with
739 journal entries. For further details please check:
741 http://www.freedesktop.org/wiki/Software/systemd/catalog
743 The indexed message catalog database also needs to be
744 rebuilt after installation of message catalog files. Use
745 "journalctl --update-catalog" for this. For RPM-based
746 distributions we introduced the %journal_catalog_update
747 macro for this purpose.
749 * The Python Journal bindings gained support for the standard
750 Python logging framework.
752 * The Journal API gained new functions for checking whether
753 the underlying file system of a journal file is capable of
754 properly reporting file change notifications, or whether
755 applications that want to reflect journal changes "live"
756 need to recheck journal files continuously in appropriate
759 * It is now possible to set the "age" field for tmpfiles
760 entries to 0, indicating that files matching this entry
761 shall always be removed when the directories are cleaned up.
763 * coredumpctl gained a new "gdb" verb which invokes gdb
764 right-away on the selected coredump.
766 * There's now support for "hybrid sleep" on kernels that
767 support this, in addition to "suspend" and "hibernate". Use
768 "systemctl hybrid-sleep" to make use of this.
770 * logind's HandleSuspendKey= setting (and related settings)
771 now gained support for a new "lock" setting to simply
772 request the screen lock on all local sessions, instead of
773 actually executing a suspend or hibernation.
775 * systemd will now mount the EFI variables file system by
778 * Socket units now gained support for configuration of the
779 SMACK security label.
781 * timedatectl will now output the time of the last and next
782 daylight saving change.
784 * We dropped support for various legacy and distro-specific
785 concepts, such as insserv, early-boot SysV services
786 (i.e. those for non-standard runlevels such as 'b' or 'S')
787 or ArchLinux /etc/rc.conf support. We recommend the
788 distributions who still need support this to either continue
789 to maintain the necessary patches downstream, or find a
790 different solution. (Talk to us if you have questions!)
792 * Various systemd components will now bypass PolicyKit checks
793 for root and otherwise handle properly if PolicyKit is not
794 found to be around. This should fix most issues for
795 PolicyKit-less systems. Quite frankly this should have been
796 this way since day one. It is absolutely our intention to
797 make systemd work fine on PolicyKit-less systems, and we
798 consider it a bug if something doesn't work as it should if
799 PolicyKit is not around.
801 * For embedded systems it is now possible to build udev and
802 systemd without blkid and/or kmod support.
804 * "systemctl switch-root" is now capable of switching root
805 more than once. I.e. in addition to transitions from the
806 initrd to the host OS it is now possible to transition to
807 further OS images from the host. This is useful to implement
808 offline updating tools.
810 * Various other additions have been made to the RPM macros
811 shipped with systemd. Use %udev_rules_update() after
812 installing new udev rules files. %_udevhwdbdir,
813 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
814 %_sysctldir are now available which resolve to the right
815 directories for packages to place various data files in.
817 * journalctl gained the new --full switch (in addition to
818 --all, to disable ellipsation for long messages.
820 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
821 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
822 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
823 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
824 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
825 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
826 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
827 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
828 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
832 * journalctl gained new --since= and --until= switches to
833 filter by time. It also now supports nice filtering for
834 units via --unit=/-u.
836 * Type=oneshot services may use ExecReload= and do the
839 * The journal daemon now supports time-based rotation and
840 vacuuming, in addition to the usual disk-space based
843 * The journal will now index the available field values for
844 each field name. This enables clients to show pretty drop
845 downs of available match values when filtering. The bash
846 completion of journalctl has been updated
847 accordingly. journalctl gained a new switch -F to list all
848 values a certain field takes in the journal database.
850 * More service events are now written as structured messages
851 to the journal, and made recognizable via message IDs.
853 * The timedated, localed and hostnamed mini-services which
854 previously only provided support for changing time, locale
855 and hostname settings from graphical DEs such as GNOME now
856 also have a minimal (but very useful) text-based client
857 utility each. This is probably the nicest way to changing
858 these settings from the command line now, especially since
859 it lists available options and is fully integrated with bash
862 * There's now a new tool "systemd-coredumpctl" to list and
863 extract coredumps from the journal.
865 * We now install a README each in /var/log/ and
866 /etc/rc.d/init.d explaining where the system logs and init
867 scripts went. This hopefully should help folks who go to
868 that dirs and look into the otherwise now empty void and
871 * When user-services are invoked (by systemd --user) the
872 $MANAGERPID env var is set to the PID of systemd.
874 * SIGRTMIN+24 when sent to a --user instance will now result
875 in immediate termination of systemd.
877 * gatewayd received numerous feature additions such as a
878 "follow" mode, for live syncing and filtering.
880 * browse.html now allows filtering and showing detailed
881 information on specific entries. Keyboard navigation and
882 mouse screen support has been added.
884 * gatewayd/journalctl now supports HTML5/JSON
885 Server-Sent-Events as output.
887 * The SysV init script compatibility logic will now
888 heuristically determine whether a script supports the
889 "reload" verb, and only then make this available as
892 * "systemctl status --follow" has been removed, use "journalctl
895 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
896 have been removed since they are hardly useful to be
899 * And I'd like to take the opportunity to specifically mention
900 Zbigniew for his great contributions. Zbigniew, you rock!
902 Contributions from: Andrew Eikum, Christian Hesse, Colin
903 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
904 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
905 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
906 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
907 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
908 Jędrzejewski-Szmek, Сковорода Никита Андреевич
912 * If /etc/vconsole.conf is non-existent or empty we will no
913 longer load any console font or key map at boot by
914 default. Instead the kernel defaults will be left
915 intact. This is definitely the right thing to do, as no
916 configuration should mean no configuration, and hard-coding
917 font names that are different on all archs is probably a bad
918 idea. Also, the kernel default key map and font should be
919 good enough for most cases anyway, and mostly identical to
920 the userspace fonts/key maps we previously overloaded them
921 with. If distributions want to continue to default to a
922 non-kernel font or key map they should ship a default
923 /etc/vconsole.conf with the appropriate contents.
925 Contributions from: Colin Walters, Daniel J Walsh, Dave
926 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
927 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
931 * journalctl gained a new --cursor= switch to show entries
932 starting from the specified location in the journal.
934 * We now enforce a size limit on journal entry fields exported
935 with "-o json" in journalctl. Fields larger than 4K will be
936 assigned null. This can be turned off with --all.
938 * An (optional) journal gateway daemon is now available as
939 "systemd-journal-gatewayd.service". This service provides
940 access to the journal via HTTP and JSON. This functionality
941 will be used to implement live log synchronization in both
942 pull and push modes, but has various other users too, such
943 as easy log access for debugging of embedded devices. Right
944 now it is already useful to retrieve the journal via HTTP:
946 # systemctl start systemd-journal-gatewayd.service
947 # wget http://localhost:19531/entries
949 This will download the journal contents in a
950 /var/log/messages compatible format. The same as JSON:
952 # curl -H"Accept: application/json" http://localhost:19531/entries
954 This service is also accessible via a web browser where a
955 single static HTML5 app is served that uses the JSON logic
956 to enable the user to do some basic browsing of the
957 journal. This will be extended later on. Here's an example
958 screenshot of this app in its current state:
960 http://0pointer.de/public/journal-gatewayd
962 Contributions from: Kay Sievers, Lennart Poettering, Robert
963 Milasan, Tom Gundersen
967 * The bash completion logic is now available for journalctl
970 * We don't mount the "cpuset" controller anymore together with
971 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
972 started if no parameters are assigned to it. "cpuset" hence
973 broke code that assumed it it could create "cpu" groups and
976 * journalctl -f will now subscribe to terminal size changes,
977 and line break accordingly.
979 Contributions from: Dave Reisner, Kay Sievers, Lennart
980 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
984 * nspawn will now create a symlink /etc/localtime in the
985 container environment, copying the host's timezone
986 setting. Previously this has been done via a bind mount, but
987 since symlinks cannot be bind mounted this has now been
988 changed to create/update the appropriate symlink.
990 * journalctl -n's line number argument is now optional, and
991 will default to 10 if omitted.
993 * journald will now log the maximum size the journal files may
994 take up on disk. This is particularly useful if the default
995 built-in logic of determining this parameter from the file
996 system size is used. Use "systemctl status
997 systemd-journald.service" to see this information.
999 * The multi-seat X wrapper tool has been stripped down. As X
1000 is now capable of enumerating graphics devices via udev in a
1001 seat-aware way the wrapper is not strictly necessary
1002 anymore. A stripped down temporary stop-gap is still shipped
1003 until the upstream display managers have been updated to
1004 fully support the new X logic. Expect this wrapper to be
1005 removed entirely in one of the next releases.
1007 * HandleSleepKey= in logind.conf has been split up into
1008 HandleSuspendKey= and HandleHibernateKey=. The old setting
1009 is not available anymore. X11 and the kernel are
1010 distuingishing between these keys and we should too. This
1011 also means the inhibition lock for these keys has been split
1014 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
1015 Poettering, Lukas Nykryn, Václav Pavlín
1019 * Whenever a unit changes state we'll now log this to the
1020 journal and show along the unit's own log output in
1023 * ConditionPathIsMountPoint= can now properly detect bind
1024 mount points too. (Previously, a bind mount of one file
1025 system to another place in the same file system could not be
1026 detected as mount, since they shared struct stat's st_dev
1029 * We will now mount the cgroup controllers cpu, cpuacct,
1030 cpuset and the controllers net_cls, net_prio together by
1033 * nspawn containers will now have a virtualized boot
1034 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
1035 over with a randomized ID at container initialization). This
1036 has the effect of making "journalctl -b" do the right thing
1039 * The JSON output journal serialization has been updated not
1040 to generate "endless" list objects anymore, but rather one
1041 JSON object per line. This is more in line how most JSON
1042 parsers expect JSON objects. The new output mode
1043 "json-pretty" has been added to provide similar output, but
1044 neatly aligned for readability by humans.
1046 * We dropped all explicit sync() invocations in the shutdown
1047 code. The kernel does this implicitly anyway in the kernel
1048 reboot() syscall. halt(8)'s -n option is now a compatibility
1051 * We now support virtualized reboot() in containers, as
1052 supported by newer kernels. We will fall back to exit() if
1053 CAP_SYS_REBOOT is not available to the container. Also,
1054 nspawn makes use of this now and will actually reboot the
1055 container if the containerized OS asks for that.
1057 * journalctl will only show local log output by default
1058 now. Use --merge (-m) to show remote log output, too.
1060 * libsystemd-journal gained the new sd_journal_get_usage()
1061 call to determine the current disk usage of all journal
1062 files. This is exposed in the new "journalctl --disk-usage"
1065 * journald gained a new configuration setting SplitMode= in
1066 journald.conf which may be used to control how user journals
1067 are split off. See journald.conf(5) for details.
1069 * A new condition type ConditionFileNotEmpty= has been added.
1071 * tmpfiles' "w" lines now support file globbing, to write
1072 multiple files at once.
1074 * We added Python bindings for the journal submission
1075 APIs. More Python APIs for a number of selected APIs will
1076 likely follow. Note that we intend to add native bindings
1077 only for the Python language, as we consider it common
1078 enough to deserve bindings shipped within systemd. There are
1079 various projects outside of systemd that provide bindings
1080 for languages such as PHP or Lua.
1082 * Many conditions will now resolve specifiers such as %i. In
1083 addition, PathChanged= and related directives of .path units
1084 now support specifiers as well.
1086 * There's now a new RPM macro definition for the system preset
1089 * journald will now warn if it can't forward a message to the
1090 syslog daemon because it's socket is full.
1092 * timedated will no longer write or process /etc/timezone,
1093 except on Debian. As we do not support late mounted /usr
1094 anymore /etc/localtime always being a symlink is now safe,
1095 and hence the information in /etc/timezone is not necessary
1098 * logind will now always reserve one VT for a text getty (VT6
1099 by default). Previously if more than 6 X sessions where
1100 started they took up all the VTs with auto-spawned gettys,
1101 so that no text gettys were available anymore.
1103 * udev will now automatically inform the btrfs kernel logic
1104 about btrfs RAID components showing up. This should make
1105 simple hotplug based btrfs RAID assembly work.
1107 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
1108 (but not for its children which will stay at the kernel
1109 default). This should allow setups with a lot more listening
1112 * systemd will now always pass the configured timezone to the
1113 kernel at boot. timedated will do the same when the timezone
1116 * logind's inhibition logic has been updated. By default,
1117 logind will now handle the lid switch, the power and sleep
1118 keys all the time, even in graphical sessions. If DEs want
1119 to handle these events on their own they should take the new
1120 handle-power-key, handle-sleep-key and handle-lid-switch
1121 inhibitors during their runtime. A simple way to achiveve
1122 that is to invoke the DE wrapped in an invocation of:
1124 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
1126 * Access to unit operations is now checked via SELinux taking
1127 the unit file label and client process label into account.
1129 * systemd will now notify the administrator in the journal
1130 when he over-mounts a non-empty directory.
1132 * There are new specifiers that are resolved in unit files,
1133 for the host name (%H), the machine ID (%m) and the boot ID
1136 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
1137 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
1138 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
1139 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
1140 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
1141 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
1142 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
1146 * Support for reading structured kernel messages from
1147 /dev/kmsg has now been added and is enabled by default.
1149 * Support for reading kernel messages from /proc/kmsg has now
1150 been removed. If you want kernel messages in the journal
1151 make sure to run a recent kernel (>= 3.5) that supports
1152 reading structured messages from /dev/kmsg (see
1153 above). /proc/kmsg is now exclusive property of classic
1154 syslog daemons again.
1156 * The libudev API gained the new
1157 udev_device_new_from_device_id() call.
1159 * The logic for file system namespace (ReadOnlyDirectory=,
1160 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
1161 require pivot_root() anymore. This means fewer temporary
1162 directories are created below /tmp for this feature.
1164 * nspawn containers will now see and receive all submounts
1165 made on the host OS below the root file system of the
1168 * Forward Secure Sealing is now supported for Journal files,
1169 which provide cryptographical sealing of journal files so
1170 that attackers cannot alter log history anymore without this
1171 being detectable. Lennart will soon post a blog story about
1172 this explaining it in more detail.
1174 * There are two new service settings RestartPreventExitStatus=
1175 and SuccessExitStatus= which allow configuration of exit
1176 status (exit code or signal) which will be excepted from the
1177 restart logic, resp. consider successful.
1179 * journalctl gained the new --verify switch that can be used
1180 to check the integrity of the structure of journal files and
1181 (if Forward Secure Sealing is enabled) the contents of
1184 * nspawn containers will now be run with /dev/stdin, /dev/fd/
1185 and similar symlinks pre-created. This makes running shells
1186 as container init process a lot more fun.
1188 * The fstab support can now handle PARTUUID= and PARTLABEL=
1191 * A new ConditionHost= condition has been added to match
1192 against the hostname (with globs) and machine ID. This is
1193 useful for clusters where a single OS image is used to
1194 provision a large number of hosts which shall run slightly
1195 different sets of services.
1197 * Services which hit the restart limit will now be placed in a
1200 Contributions from: Bertram Poettering, Dave Reisner, Huang
1201 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
1202 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
1206 * When running in --user mode systemd will now become a
1207 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
1208 tree a lot more organized.
1210 * A new PartOf= unit dependency type has been introduced that
1211 may be used to group services in a natural way.
1213 * "systemctl enable" may now be used to enable instances of
1216 * journalctl now prints error log levels in red, and
1217 warning/notice log levels in bright white. It also supports
1218 filtering by log level now.
1220 * cgtop gained a new -n switch (similar to top), to configure
1221 the maximum number of iterations to run for. It also gained
1222 -b, to run in batch mode (accepting no input).
1224 * The suffix ".service" may now be omitted on most systemctl
1225 command lines involving service unit names.
1227 * There's a new bus call in logind to lock all sessions, as
1228 well as a loginctl verb for it "lock-sessions".
1230 * libsystemd-logind.so gained a new call sd_journal_perror()
1231 that works similar to libc perror() but logs to the journal
1232 and encodes structured information about the error number.
1234 * /etc/crypttab entries now understand the new keyfile-size=
1237 * shutdown(8) now can send a (configurable) wall message when
1238 a shutdown is cancelled.
1240 * The mount propagation mode for the root file system will now
1241 default to "shared", which is useful to make containers work
1242 nicely out-of-the-box so that they receive new mounts from
1243 the host. This can be undone locally by running "mount
1244 --make-rprivate /" if needed.
1246 * The prefdm.service file has been removed. Distributions
1247 should maintain this unit downstream if they intend to keep
1248 it around. However, we recommend writing normal unit files
1249 for display managers instead.
1251 * Since systemd is a crucial part of the OS we will now
1252 default to a number of compiler switches that improve
1253 security (hardening) such as read-only relocations, stack
1254 protection, and suchlike.
1256 * The TimeoutSec= setting for services is now split into
1257 TimeoutStartSec= and TimeoutStopSec= to allow configuration
1258 of individual time outs for the start and the stop phase of
1261 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
1262 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
1263 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
1264 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
1265 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
1266 Gundersen, Zbigniew Jędrzejewski-Szmek
1270 * The journal and id128 C APIs are now fully documented as man
1273 * Extra safety checks have been added when transitioning from
1274 the initial RAM disk to the main system to avoid accidental
1277 * /etc/crypttab entries now understand the new keyfile-offset=
1280 * systemctl -t can now be used to filter by unit load state.
1282 * The journal C API gained the new sd_journal_wait() call to
1283 make writing synchronous journal clients easier.
1285 * journalctl gained the new -D switch to show journals from a
1288 * journalctl now displays a special marker between log
1289 messages of two different boots.
1291 * The journal is now explicitly flushed to /var via a service
1292 systemd-journal-flush.service, rather than implicitly simply
1293 by seeing /var/log/journal to be writable.
1295 * journalctl (and the journal C APIs) can now match for much
1296 more complex expressions, with alternatives and
1299 * When transitioning from the initial RAM disk to the main
1300 system we will now kill all processes in a killing spree to
1301 ensure no processes stay around by accident.
1303 * Three new specifiers may be used in unit files: %u, %h, %s
1304 resolve to the user name, user home directory resp. user
1305 shell. This is useful for running systemd user instances.
1307 * We now automatically rotate journal files if their data
1308 object hash table gets a fill level > 75%. We also size the
1309 hash table based on the configured maximum file size. This
1310 together should lower hash collisions drastically and thus
1311 speed things up a bit.
1313 * journalctl gained the new "--header" switch to introspect
1314 header data of journal files.
1316 * A new setting SystemCallFilters= has been added to services
1317 which may be used to apply blacklists or whitelists to
1318 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
1320 * nspawn gained a new --link-journal= switch (and quicker: -j)
1321 to link the container journal with the host. This makes it
1322 very easy to centralize log viewing on the host for all
1323 guests while still keeping the journal files separated.
1325 * Many bugfixes and optimizations
1327 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
1328 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
1329 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
1334 * Several tools now understand kernel command line arguments,
1335 which are only read when run in an initial RAM disk. They
1336 usually follow closely their normal counterparts, but are
1339 * There's a new tool to analyze the readahead files that are
1340 automatically generated at boot. Use:
1342 /usr/lib/systemd/systemd-readahead analyze /.readahead
1344 * We now provide an early debug shell on tty9 if this enabled. Use:
1346 systemctl enable debug-shell.service
1348 * All plymouth related units have been moved into the Plymouth
1349 package. Please make sure to upgrade your Plymouth version
1352 * systemd-tmpfiles now supports getting passed the basename of
1353 a configuration file only, in which case it will look for it
1354 in all appropriate directories automatically.
1356 * udevadm info now takes a /dev or /sys path as argument, and
1357 does the right thing. Example:
1359 udevadm info /dev/sda
1360 udevadm info /sys/class/block/sda
1362 * systemctl now prints a warning if a unit is stopped but a
1363 unit that might trigger it continues to run. Example: a
1364 service is stopped but the socket that activates it is left
1367 * "systemctl status" will now mention if the log output was
1368 shortened due to rotation since a service has been started.
1370 * The journal API now exposes functions to determine the
1371 "cutoff" times due to rotation.
1373 * journald now understands SIGUSR1 and SIGUSR2 for triggering
1374 immediately flushing of runtime logs to /var if possible,
1375 resp. for triggering immediate rotation of the journal
1378 * It is now considered an error if a service is attempted to
1379 be stopped that is not loaded.
1381 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
1383 * systemd-analyze now supports Python 3
1385 * tmpfiles now supports cleaning up directories via aging
1386 where the first level dirs are always kept around but
1387 directories beneath it automatically aged. This is enabled
1388 by prefixing the age field with '~'.
1390 * Seat objects now expose CanGraphical, CanTTY properties
1391 which is required to deal with very fast bootups where the
1392 display manager might be running before the graphics drivers
1393 completed initialization.
1395 * Seat objects now expose a State property.
1397 * We now include RPM macros for service enabling/disabling
1398 based on the preset logic. We recommend RPM based
1399 distributions to make use of these macros if possible. This
1400 makes it simpler to reuse RPM spec files across
1403 * We now make sure that the collected systemd unit name is
1404 always valid when services log to the journal via
1407 * There's a new man page kernel-command-line(7) detailing all
1408 command line options we understand.
1410 * The fstab generator may now be disabled at boot by passing
1411 fstab=0 on the kernel command line.
1413 * A new kernel command line option modules-load= is now understood
1414 to load a specific kernel module statically, early at boot.
1416 * Unit names specified on the systemctl command line are now
1417 automatically escaped as needed. Also, if file system or
1418 device paths are specified they are automatically turned
1419 into the appropriate mount or device unit names. Example:
1421 systemctl status /home
1422 systemctl status /dev/sda
1424 * The SysVConsole= configuration option has been removed from
1425 system.conf parsing.
1427 * The SysV search path is no longer exported on the D-Bus
1430 * The Names= option is been removed from unit file parsing.
1432 * There's a new man page bootup(7) detailing the boot process.
1434 * Every unit and every generator we ship with systemd now
1435 comes with full documentation. The self-explanatory boot is
1438 * A couple of services gained "systemd-" prefixes in their
1439 name if they wrap systemd code, rather than only external
1440 code. Among them fsck@.service which is now
1441 systemd-fsck@.service.
1443 * The HaveWatchdog property has been removed from the D-Bus
1446 * systemd.confirm_spawn= on the kernel command line should now
1449 * There's a new man page crypttab(5) which details all options
1450 we actually understand.
1452 * systemd-nspawn gained a new --capability= switch to pass
1453 additional capabilities to the container.
1455 * timedated will now read known NTP implementation unit names
1456 from /usr/lib/systemd/ntp-units.d/*.list,
1457 systemd-timedated-ntp.target has been removed.
1459 * journalctl gained a new switch "-b" that lists log data of
1460 the current boot only.
1462 * The notify socket is in the abstract namespace again, in
1463 order to support daemons which chroot() at start-up.
1465 * There is a new Storage= configuration option for journald
1466 which allows configuration of where log data should go. This
1467 also provides a way to disable journal logging entirely, so
1468 that data collected is only forwarded to the console, the
1469 kernel log buffer or another syslog implementation.
1471 * Many bugfixes and optimizations
1473 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
1474 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
1475 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
1476 Shawn Landden, Tom Gundersen
1480 * "systemctl help <unit>" now shows the man page if one is
1483 * Several new man pages have been added.
1485 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
1486 MaxLevelConsole= can now be specified in
1487 journald.conf. These options allow reducing the amount of
1488 data stored on disk or forwarded by the log level.
1490 * TimerSlackNSec= can now be specified in system.conf for
1491 PID1. This allows system-wide power savings.
1493 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
1494 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
1499 * logind is now capable of (optionally) handling power and
1500 sleep keys as well as the lid switch.
1502 * journalctl now understands the syntax "journalctl
1503 /usr/bin/avahi-daemon" to get all log output of a specific
1506 * CapabilityBoundingSet= in system.conf now also influences
1507 the capability bound set of usermode helpers of the kernel.
1509 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
1510 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
1511 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
1512 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
1516 * Note that we skipped 139 releases here in order to set the
1517 new version to something that is greater than both udev's
1518 and systemd's most recent version number.
1520 * udev: all udev sources are merged into the systemd source tree now.
1521 All future udev development will happen in the systemd tree. It
1522 is still fully supported to use the udev daemon and tools without
1523 systemd running, like in initramfs or other init systems. Building
1524 udev though, will require the *build* of the systemd tree, but
1525 udev can be properly *run* without systemd.
1527 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
1528 should be used to create dead device nodes as workarounds for broken
1531 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
1532 no longer supported. udev_monitor_new_from_netlink() needs to be
1533 used to subscribe to events.
1535 * udev: when udevd is started by systemd, processes which are left
1536 behind by forking them off of udev rules, are unconditionally cleaned
1537 up and killed now after the event handling has finished. Services or
1538 daemons must be started as systemd services. Services can be
1539 pulled-in by udev to get started, but they can no longer be directly
1540 forked by udev rules.
1542 * udev: the daemon binary is called systemd-udevd now and installed
1543 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
1544 to adapt to that, create symlink, or rename the binary after building
1547 * libudev no longer provides these symbols:
1548 udev_monitor_from_socket()
1549 udev_queue_get_failed_list_entry()
1550 udev_get_{dev,sys,run}_path()
1551 The versions number was bumped and symbol versioning introduced.
1553 * systemd-loginctl and systemd-journalctl have been renamed
1554 to loginctl and journalctl to match systemctl.
1556 * The config files: /etc/systemd/systemd-logind.conf and
1557 /etc/systemd/systemd-journald.conf have been renamed to
1558 logind.conf and journald.conf. Package updates should rename
1559 the files to the new names on upgrade.
1561 * For almost all files the license is now LGPL2.1+, changed
1562 from the previous GPL2.0+. Exceptions are some minor stuff
1563 of udev (which will be changed to LGPL2.1 eventually, too),
1564 and the MIT licensed sd-daemon.[ch] library that is suitable
1565 to be used as drop-in files.
1567 * systemd and logind now handle system sleep states, in
1568 particular suspending and hibernating.
1570 * logind now implements a sleep/shutdown/idle inhibiting logic
1571 suitable for a variety of uses. Soonishly Lennart will blog
1572 about this in more detail.
1574 * var-run.mount and var-lock.mount are no longer provided
1575 (which prevously bind mounted these directories to their new
1576 places). Distributions which have not converted these
1577 directories to symlinks should consider stealing these files
1578 from git history and add them downstream.
1580 * We introduced the Documentation= field for units and added
1581 this to all our shipped units. This is useful to make it
1582 easier to explore the boot and the purpose of the various
1585 * All smaller setup units (such as
1586 systemd-vconsole-setup.service) now detect properly if they
1587 are run in a container and are skipped when
1588 appropriate. This guarantees an entirely noise-free boot in
1589 Linux container environments such as systemd-nspawn.
1591 * A framework for implementing offline system updates is now
1592 integrated, for details see:
1593 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
1595 * A new service type Type=idle is available now which helps us
1596 avoiding ugly interleaving of getty output and boot status
1599 * There's now a system-wide CapabilityBoundingSet= option to
1600 globally reduce the set of capabilities for the
1601 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
1602 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
1603 even CAP_NET_ADMIN system-wide for secure systems.
1605 * There are now system-wide DefaultLimitXXX= options to
1606 globally change the defaults of the various resource limits
1607 for all units started by PID 1.
1609 * Harald Hoyer's systemd test suite has been integrated into
1610 systemd which allows easy testing of systemd builds in qemu
1611 and nspawn. (This is really awesome! Ask us for details!)
1613 * The fstab parser is now implemented as generator, not inside
1616 * systemctl will now warn you if .mount units generated from
1617 /etc/fstab are out of date due to changes in fstab that
1618 haven't been read by systemd yet.
1620 * systemd is now suitable for usage in initrds. Dracut has
1621 already been updated to make use of this. With this in place
1622 initrds get a slight bit faster but primarily are much
1623 easier to introspect and debug since "systemctl status" in
1624 the host system can be used to introspect initrd services,
1625 and the journal from the initrd is kept around too.
1627 * systemd-delta has been added, a tool to explore differences
1628 between user/admin configuration and vendor defaults.
1630 * PrivateTmp= now affects both /tmp and /var/tmp.
1632 * Boot time status messages are now much prettier and feature
1633 proper english language. Booting up systemd has never been
1636 * Read-ahead pack files now include the inode number of all
1637 files to pre-cache. When the inode changes the pre-caching
1638 is not attempted. This should be nicer to deal with updated
1639 packages which might result in changes of read-ahead
1642 * We now temporaritly lower the kernel's read_ahead_kb variable
1643 when collecting read-ahead data to ensure the kernel's
1644 built-in read-ahead does not add noise to our measurements
1645 of necessary blocks to pre-cache.
1647 * There's now RequiresMountsFor= to add automatic dependencies
1648 for all mounts necessary for a specific file system path.
1650 * MountAuto= and SwapAuto= have been removed from
1651 system.conf. Mounting file systems at boot has to take place
1654 * nspawn now learned a new switch --uuid= to set the machine
1655 ID on the command line.
1657 * nspawn now learned the -b switch to automatically search
1660 * vt102 is now the default TERM for serial TTYs, upgraded from
1663 * systemd-logind now works on VT-less systems.
1665 * The build tree has been reorganized. The individual
1666 components now have directories of their own.
1668 * A new condition type ConditionPathIsReadWrite= is now available.
1670 * nspawn learned the new -C switch to create cgroups for the
1671 container in other hierarchies.
1673 * We now have support for hardware watchdogs, configurable in
1676 * The scheduled shutdown logic now has a public API.
1678 * We now mount /tmp as tmpfs by default, but this can be
1679 masked and /etc/fstab can override it.
1681 * Since udisks doesn't make use of /media anymore we are not
1682 mounting a tmpfs on it anymore.
1684 * journalctl gained a new --local switch to only interleave
1685 locally generated journal files.
1687 * We can now load the IMA policy at boot automatically.
1689 * The GTK tools have been split off into a systemd-ui.
1691 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
1692 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
1693 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
1694 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
1695 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
1696 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
1697 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
1698 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
1699 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
1704 * This is mostly a bugfix release
1706 * Support optional initialization of the machine ID from the
1707 KVM or container configured UUID.
1709 * Support immediate reboots with "systemctl reboot -ff"
1711 * Show /etc/os-release data in systemd-analyze output
1713 * Many bugfixes for the journal, including endianness fixes and
1714 ensuring that disk space enforcement works
1716 * sd-login.h is C++ comptaible again
1718 * Extend the /etc/os-release format on request of the Debian
1721 * We now refuse non-UTF8 strings used in various configuration
1722 and unit files. This is done to ensure we don't pass invalid
1723 data over D-Bus or expose it elsewhere.
1725 * Register Mimo USB Screens as suitable for automatic seat
1728 * Read SELinux client context from journal clients in a race
1731 * Reorder configuration file lookup order. /etc now always
1732 overrides /run in order to allow the administrator to always
1733 and unconditionally override vendor supplied or
1734 automatically generated data.
1736 * The various user visible bits of the journal now have man
1737 pages. We still lack man pages for the journal API calls
1740 * We now ship all man pages in HTML format again in the
1743 Contributions from: Dave Reisner, Dirk Eibach, Frederic
1744 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
1745 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
1750 * This is mostly a bugfix release
1752 * systems lacking /etc/os-release are no longer supported.
1754 * Various functionality updates to libsystemd-login.so
1756 * Track class of PAM logins to distuingish greeters from
1759 Contributions from: Kay Sievers, Lennart Poettering, Michael
1764 * This is an important bugfix release for v41.
1766 * Building man pages is now optional which should be useful
1767 for those building systemd from git but unwilling to install
1770 * Watchdog support for supervising services is now usable. In
1771 a future release support for hardware watchdogs
1772 (i.e. /dev/watchdog) will be added building on this.
1774 * Service start rate limiting is now configurable and can be
1775 turned off per service. When a start rate limit is hit a
1776 reboot can automatically be triggered.
1778 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
1780 Contributions from: Benjamin Franzke, Bill Nottingham,
1781 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
1782 Schmidt, Michał Górny, Piotr Drąg
1786 * The systemd binary is installed /usr/lib/systemd/systemd now;
1787 An existing /sbin/init symlink needs to be adapted with the
1790 * The code that loads kernel modules has been ported to invoke
1791 libkmod directly, instead of modprobe. This means we do not
1792 support systems with module-init-tools anymore.
1794 * Watchdog support is now already useful, but still not
1797 * A new kernel command line option systemd.setenv= is
1798 understood to set system wide environment variables
1799 dynamically at boot.
1801 * We now limit the set of capabilities of systemd-journald.
1803 * We now set SIGPIPE to ignore by default, since it only is
1804 useful in shell pipelines, and has little use in general
1805 code. This can be disabled with IgnoreSIPIPE=no in unit
1808 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
1809 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
1814 * This is mostly a bugfix release
1816 * We now expose the reason why a service failed in the
1817 "Result" D-Bus property.
1819 * Rudimentary service watchdog support (will be completed over
1820 the next few releases.)
1822 * When systemd forks off in order execute some service we will
1823 now immediately changes its argv[0] to reflect which process
1824 it will execute. This is useful to minimize the time window
1825 with a generic argv[0], which makes bootcharts more useful
1827 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
1828 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
1829 Mike Kazantsev, Ray Strode
1833 * This is mostly a test release, but incorporates many
1836 * New systemd-cgtop tool to show control groups by their
1839 * Linking against libacl for ACLs is optional again. If
1840 disabled, support tracking device access for active logins
1841 goes becomes unavailable, and so does access to the user
1842 journals by the respective users.
1844 * If a group "adm" exists, journal files are automatically
1845 owned by them, thus allow members of this group full access
1846 to the system journal as well as all user journals.
1848 * The journal now stores the SELinux context of the logging
1849 client for all entries.
1851 * Add C++ inclusion guards to all public headers
1853 * New output mode "cat" in the journal to print only text
1854 messages, without any meta data like date or time.
1856 * Include tiny X server wrapper as a temporary stop-gap to
1857 teach XOrg udev display enumeration. This is used by display
1858 managers such as gdm, and will go away as soon as XOrg
1859 learned native udev hotplugging for display devices.
1861 * Add new systemd-cat tool for executing arbitrary programs
1862 with STDERR/STDOUT connected to the journal. Can also act as
1863 BSD logger replacement, and does so by default.
1865 * Optionally store all locally generated coredumps in the
1866 journal along with meta data.
1868 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
1869 writing short strings to files (for usage for /sys), and for
1870 creating symlinks, character and block device nodes.
1872 * New unit file option ControlGroupPersistent= to make cgroups
1873 persistent, following the mechanisms outlined in
1874 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
1876 * Support multiple local RTCs in a sane way
1878 * No longer monopolize IO when replaying readahead data on
1879 rotating disks, since we might starve non-file-system IO to
1880 death, since fanotify() will not see accesses done by blkid,
1883 * Don't show kernel threads in systemd-cgls anymore, unless
1884 requested with new -k switch.
1886 Contributions from: Dan Horák, Kay Sievers, Lennart
1887 Poettering, Michal Schmidt
1891 * This is mostly a test release, but incorporates many
1894 * The git repository moved to:
1895 git://anongit.freedesktop.org/systemd/systemd
1896 ssh://git.freedesktop.org/git/systemd/systemd
1898 * First release with the journal
1899 http://0pointer.de/blog/projects/the-journal.html
1901 * The journal replaces both systemd-kmsg-syslogd and
1902 systemd-stdout-bridge.
1904 * New sd_pid_get_unit() API call in libsystemd-logind
1906 * Many systemadm clean-ups
1908 * Introduce remote-fs-pre.target which is ordered before all
1909 remote mounts and may be used to start services before all
1912 * Added Mageia support
1914 * Add bash completion for systemd-loginctl
1916 * Actively monitor PID file creation for daemons which exit in
1917 the parent process before having finished writing the PID
1918 file in the daemon process. Daemons which do this need to be
1919 fixed (i.e. PID file creation must have finished before the
1920 parent exits), but we now react a bit more gracefully to them.
1922 * Add colourful boot output, mimicking the well-known output
1923 of existing distributions.
1925 * New option PassCredentials= for socket units, for
1926 compatibility with a recent kernel ABI breakage.
1928 * /etc/rc.local is now hooked in via a generator binary, and
1929 thus will no longer act as synchronization point during
1932 * systemctl list-unit-files now supports --root=.
1934 * systemd-tmpfiles now understands two new commands: z, Z for
1935 relabelling files according to the SELinux database. This is
1936 useful to apply SELinux labels to specific files in /sys,
1939 * Output of SysV services is now forwarded to both the console
1940 and the journal by default, not only just the console.
1942 * New man pages for all APIs from libsystemd-login.
1944 * The build tree got reorganized and a the build system is a
1945 lot more modular allowing embedded setups to specifically
1946 select the components of systemd they are interested in.
1948 * Support for Linux systems lacking the kernel VT subsystem is
1951 * configure's --with-rootdir= got renamed to
1952 --with-rootprefix= to follow the naming used by udev and
1955 * Unless specified otherwise we'll now install to /usr instead
1956 of /usr/local by default.
1958 * Processes with '@' in argv[0][0] are now excluded from the
1959 final shut-down killing spree, following the logic explained
1961 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
1963 * All processes remaining in a service cgroup when we enter
1964 the START or START_PRE states are now killed with
1965 SIGKILL. That means it is no longer possible to spawn
1966 background processes from ExecStart= lines (which was never
1967 supported anyway, and bad style).
1969 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
1970 reloading of units together.
1972 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
1973 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
1974 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
1975 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
1976 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek