1 systemd System and Service Manager
5 * systemd-nspawn will now create /etc/resolv.conf if
6 necessary, before bind-mounting the host's file onto it.
8 * systemd-nspawn will now store meta information about a
9 container on the container's cgroup as extended attribute
10 fields, including the root directory.
12 * The cgroup hierarchy has been reworked in many ways. All
13 objects any of the components systemd creates in the cgroup
14 tree is now suffixed. More specifically, user sessions are
15 now placed in cgroups suffixed with ".session", users in
16 cgroups suffixed with ".user", and nspawn containers in
17 cgroups suffixed with ".nspawn". Furthermore, all cgroup
18 names are now escaped in a simple scheme to avoid collision
19 of userspace object names with kernel filenames. This work
20 is preparation for making these objects relocatable in the
21 cgroup tree, in order to allow easy resource partitioning of
22 these objects without causing naming conflicts.
24 * systemctl list-dependencies gained the new switches
25 --plain, --reverse, --after and --before.
27 * systemd-inhibit now shows the process name of processes that
28 have taken an inhibitor lock.
30 * nss-myhostname will now also resolve "localhost"
31 implicitly. This makes /etc/hosts an optional file and
32 nicely handles that on IPv6 ::1 maps to both "localhost" and
35 * libsystemd-logind.so gained a new call
36 sd_get_machine_names() to enumerate running containers and
37 VMs (currently only supported by very new libvirt and
38 nspawn). sd_login_monitor can now be used to watch
39 VMs/containers coming and going.
41 * .include is not allowed recursively anymore, and only in
42 unit files. Usually it is better to use drop-in snippets in
43 .d/*.conf anyway, as introduced with systemd 198.
45 * systemd-analyze gained a new "critical-chain" command that
46 determines the slowest chain of units run during system
47 boot-up. It is very useful for tracking down where
48 optimizing boot time is the most beneficial.
50 * systemd will no longer allow manipulating service paths in
51 the name=systemd:/system cgroup tree using ControlGroup= in
52 units. (But is still fine with it in all other dirs.)
54 * There's a new systemd-nspawn@.service service file that may
55 be used to easily run nspawn containers as system
56 services. With the container's root directory in
57 /var/lib/container/foobar it is now sufficient to run
58 "systemctl start systemd-nspawn@foobar.service" to boot it.
60 * systemd-cgls gained a new parameter "--machine" to list only
61 the processes within a certain container.
63 * ConditionSecurity= now can check for "apparmor". We still
64 are lacking checks for SMACK and IMA for this condition
65 check though. Patches welcome!
67 * A new configuration file /etc/systemd/sleep.conf has been
68 added that may be used to configure which kernel operation
69 systemd is supposed to execute when "suspend", "hibernate"
70 or "hybrid-sleep" is requested. This makes the new kernel
71 "freeze" state accessible to the user.
73 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
74 the passed argument if applicable.
76 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
77 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
78 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
79 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
80 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
81 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
82 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
87 * The output of 'systemctl list-jobs' got some polishing. The
88 '--type=' argument may now be passed more than once. A new
89 command 'systemctl list-sockets' has been added which shows
90 a list of kernel sockets systemd is listening on with the
91 socket units they belong to, plus the units these socket
94 * The experimental libsystemd-bus library got substantial
95 updates to work in conjunction with the (also experimental)
96 kdbus kernel project. It works well enough to exchange
97 messages with some sophistication. Note that kdbus is not
98 ready yet, and the library is mostly an elaborate test case
99 for now, and not installable.
101 * systemd gained a new unit 'systemd-static-nodes.service'
102 that generates static device nodes earlier during boot, and
103 can run in conjunction with udev.
105 * libsystemd-login gained a new call sd_pid_get_user_unit()
106 to retrieve the user systemd unit a process is running
107 in. This is useful for systems where systemd is used as
110 * systemd-nspawn now places all containers in the new /machine
111 top-level cgroup directory in the name=systemd
112 hierarchy. libvirt will soon do the same, so that we get a
113 uniform separation of /system, /user and /machine for system
114 services, user processes and containers/virtual
115 machines. This new cgroup hierarchy is also useful to stick
116 stable names to specific container instances, which can be
117 recognized later this way (this name may be controlled
118 via systemd-nspawn's new -M switch). libsystemd-login also
119 gained a new call sd_pid_get_machine_name() to retrieve the
120 name of the container/VM a specific process belongs to.
122 * bootchart can now store its data in the journal.
124 * libsystemd-journal gained a new call
125 sd_journal_add_conjunction() for AND expressions to the
126 matching logic. This can be used to express more complex
129 * journactl can now take multiple --unit= and --user-unit=
132 * The cryptsetup logic now understands the "luks.key=" kernel
133 command line switch for specifying a file to read the
134 decryption key from. Also, if a configured key file is not
135 found the tool will now automatically fall back to prompting
138 * Python systemd.journal module was updated to wrap recently
139 added functions from libsystemd-journal. The interface was
140 changed to bring the low level interface in s.j._Reader
141 closer to the C API, and the high level interface in
142 s.j.Reader was updated to wrap and convert all data about
145 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
146 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
147 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
148 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
149 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
150 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
154 * journalctl --update-catalog now understands a new --root=
155 option to operate on catalogs found in a different root
158 * During shutdown after systemd has terminated all running
159 services a final killing loop kills all remaining left-over
160 processes. We will now print the name of these processes
161 when we send SIGKILL to them, since this usually indicates a
164 * If /etc/crypttab refers to password files stored on
165 configured mount points automatic dependencies will now be
166 generated to ensure the specific mount is established first
167 before the key file is attempted to be read.
169 * 'systemctl status' will now show information about the
170 network sockets a socket unit is listening on.
172 * 'systemctl status' will also shown information about any
173 drop-in configuration file for units. (Drop-In configuration
174 files in this context are files such as
175 /etc/systemd/systemd/foobar.service.d/*.conf)
177 * systemd-cgtop now optionally shows summed up CPU times of
178 cgroups. Press '%' while running cgtop to switch between
179 percentage and absolute mode. This is useful to determine
180 which cgroups use up the most CPU time over the entire
181 runtime of the system. systemd-cgtop has also been updated
182 to be 'pipeable' for processing with further shell tools.
184 * 'hostnamectl set-hostname' will now allow setting of FQDN
187 * The formatting and parsing of time span values has been
188 changed. The parser now understands fractional expressions
189 such as "5.5h". The formatter will now output fractional
190 expressions for all time spans under 1min, i.e. "5.123456s"
191 rather than "5s 123ms 456us". For time spans under 1s
192 millisecond values are shown, for those under 1ms
193 microsecond values are shown. This should greatly improve
194 all time-related output of systemd.
196 * libsystemd-login and libsystemd-journal gained new
197 functions for querying the poll() events mask and poll()
198 timeout value for integration into arbitrary event
201 * localectl gained the ability to list available X11 keymaps
202 (models, layouts, variants, options).
204 * 'systemd-analyze dot' gained the ability to filter for
205 specific units via shell-style globs, to create smaller,
206 more useful graphs. I.e. it's now possible to create simple
207 graphs of all the dependencies between only target units, or
208 of all units that Avahi has dependencies with.
210 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
211 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
212 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
213 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
214 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
215 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
216 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
220 * The boot-time readahead implementation for rotating media
221 will now read the read-ahead data in multiple passes which
222 consist of all read requests made in equidistant time
223 intervals. This means instead of strictly reading read-ahead
224 data in its physical order on disk we now try to find a
225 middle ground between physical and access time order.
227 * /etc/os-release files gained a new BUILD_ID= field for usage
228 on operating systems that provide continuous builds of OS
231 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
232 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
233 William Douglas, Zbigniew Jędrzejewski-Szmek
237 * systemd-python gained an API exposing libsystemd-daemon.
239 * The SMACK setup logic gained support for uploading CIPSO
242 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
243 ReadOnlyDirectories= and InaccessibleDirectories= has
244 changed. The private /tmp and /var/tmp directories are now
245 shared by all processes of a service (which means
246 ExecStartPre= may now leave data in /tmp that ExecStart= of
247 the same service can still access). When a service is
248 stopped its temporary directories are immediately deleted
249 (normal clean-up with tmpfiles is still done in addition to
252 * By default, systemd will now set a couple of sysctl
253 variables in the kernel: the safe sysrq options are turned
254 on, IP route verification is turned on, and source routing
255 disabled. The recently added hardlink and softlink
256 protection of the kernel is turned on. These settings should
257 be reasonably safe, and good defaults for all new systems.
259 * The predictable network naming logic may now be turned off
260 with a new kernel command line switch: net.ifnames=0.
262 * A new libsystemd-bus module has been added that implements a
263 pretty complete D-Bus client library. For details see:
265 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
267 * journald will now explicitly flush the journal files to disk
268 at the latest 5min after each write. The file will then also
269 be marked offline until the next write. This should increase
270 reliability in case of a crash. The synchronization delay
271 can be configured via SyncIntervalSec= in journald.conf.
273 * There's a new remote-fs-setup.target unit that can be used
274 to pull in specific services when at least one remote file
275 system is to be mounted.
277 * There are new targets timers.target and paths.target as
278 canonical targets to pull user timer and path units in
279 from. This complements sockets.target with a similar
280 purpose for socket units.
282 * libudev gained a new call udev_device_set_attribute_value()
283 to set sysfs attributes of a device.
285 * The udev daemon now sets the default number of worker
286 processes executed in parallel based on the number of available
287 CPUs instead of the amount of available RAM. This is supposed
288 to provide a more reliable default and limit a too aggressive
289 paralellism for setups with 1000s of devices connected.
291 Contributions from: Auke Kok, Colin Walters, Cristian
292 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
293 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
294 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
295 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
296 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
297 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
298 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
299 Zbigniew Jędrzejewski-Szmek
303 * Configuration of unit files may now be extended via drop-in
304 files without having to edit/override the unit files
305 themselves. More specifically, if the administrator wants to
306 change one value for a service file foobar.service he can
307 now do so by dropping in a configuration snippet into
308 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
309 will load all these snippets and apply them on top of the
310 main unit configuration file, possibly extending or
311 overriding its settings. Using these drop-in snippets is
312 generally nicer than the two earlier options for changing
313 unit files locally: copying the files from
314 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
315 them there; or creating a new file in /etc/systemd/system/
316 that incorporates the original one via ".include". Drop-in
317 snippets into these .d/ directories can be placed in any
318 directory systemd looks for units in, and the usual
319 overriding semantics between /usr/lib, /etc and /run apply
322 * Most unit file settings which take lists of items can now be
323 reset by assigning the empty string to them. For example,
324 normally, settings such as Environment=FOO=BAR append a new
325 environment variable assignment to the environment block,
326 each time they are used. By assigning Environment= the empty
327 string the environment block can be reset to empty. This is
328 particularly useful with the .d/*.conf drop-in snippets
329 mentioned above, since this adds the ability to reset list
330 settings from vendor unit files via these drop-ins.
332 * systemctl gained a new "list-dependencies" command for
333 listing the dependencies of a unit recursively.
335 * Inhibitors are now honored and listed by "systemctl
336 suspend", "systemctl poweroff" (and similar) too, not only
337 GNOME. These commands will also list active sessions by
340 * Resource limits (as exposed by the various control group
341 controllers) can now be controlled dynamically at runtime
342 for all units. More specifically, you can now use a command
343 like "systemctl set-cgroup-attr foobar.service cpu.shares
344 2000" to alter the CPU shares a specific service gets. These
345 settings are stored persistently on disk, and thus allow the
346 administrator to easily adjust the resource usage of
347 services with a few simple commands. This dynamic resource
348 management logic is also available to other programs via the
349 bus. Almost any kernel cgroup attribute and controller is
352 * systemd-vconsole-setup will now copy all font settings to
353 all allocated VTs, where it previously applied them only to
356 * libsystemd-login gained the new sd_session_get_tty() API
359 * This release drops support for a few legacy or
360 distribution-specific LSB facility names when parsing init
361 scripts: $x-display-manager, $mail-transfer-agent,
362 $mail-transport-agent, $mail-transfer-agent, $smtp,
363 $null. Also, the mail-transfer-agent.target unit backing
364 this has been removed. Distributions which want to retain
365 compatibility with this should carry the burden for
366 supporting this themselves and patch support for these back
367 in, if they really need to. Also, the facilities $syslog and
368 $local_fs are now ignored, since systemd does not support
369 early-boot LSB init scripts anymore, and these facilities
370 are implied anyway for normal services. syslog.target has
373 * There are new bus calls on PID1's Manager object for
374 cancelling jobs, and removing snapshot units. Previously,
375 both calls were only available on the Job and Snapshot
378 * systemd-journal-gatewayd gained SSL support.
380 * The various "environment" files, such as /etc/locale.conf
381 now support continuation lines with a backslash ("\") as
382 last character in the line, similar in style (but different)
383 to how this is supported in shells.
385 * For normal user processes the _SYSTEMD_USER_UNIT= field is
386 now implicitly appended to every log entry logged. systemctl
387 has been updated to filter by this field when operating on a
388 user systemd instance.
390 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
391 CAP_AUDIT_CONTROL capabilities to the capabilities set for
392 the container. This makes it easier to boot unmodified
393 Fedora systems in a container, which however still requires
394 audit=0 to be passed on the kernel command line. Auditing in
395 kernel and userspace is unfortunately still too broken in
396 context of containers, hence we recommend compiling it out
397 of the kernel or using audit=0. Hopefully this will be fixed
398 one day for good in the kernel.
400 * nspawn gained the new --bind= and --bind-ro= parameters to
401 bind mount specific directories from the host into the
404 * nspawn will now mount its own devpts file system instance
405 into the container, in order not to leak pty devices from
406 the host into the container.
408 * systemd will now read the firmware boot time performance
409 information from the EFI variables, if the used boot loader
410 supports this, and takes it into account for boot performance
411 analysis via "systemd-analyze". This is currently supported
412 only in conjunction with Gummiboot, but could be supported
413 by other boot loaders too. For details see:
415 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
417 * A new generator has been added that automatically mounts the
418 EFI System Partition (ESP) to /boot, if that directory
419 exists, is empty, and no other file system has been
420 configured to be mounted there.
422 * logind will now send out PrepareForSleep(false) out
423 unconditionally, after coming back from suspend. This may be
424 used by applications as asynchronous notification for
425 system resume events.
427 * "systemctl unlock-sessions" has been added, that allows
428 unlocking the screens of all user sessions at once, similar
429 how "systemctl lock-sessions" already locked all users
430 sessions. This is backed by a new D-Bus call UnlockSessions().
432 * "loginctl seat-status" will now show the master device of a
433 seat. (i.e. the device of a seat that needs to be around for
434 the seat to be considered available, usually the graphics
437 * tmpfiles gained a new "X" line type, that allows
438 configuration of files and directories (with wildcards) that
439 shall be excluded from automatic cleanup ("aging").
441 * udev default rules set the device node permissions now only
442 at "add" events, and do not change them any longer with a
443 later "change" event.
445 * The log messages for lid events and power/sleep keypresses
446 now carry a message ID.
448 * We now have a substantially larger unit test suite, but this
449 continues to be work in progress.
451 * udevadm hwdb gained a new --root= parameter to change the
452 root directory to operate relative to.
454 * logind will now issue a background sync() request to the kernel
455 early at shutdown, so that dirty buffers are flushed to disk early
456 instead of at the last moment, in order to optimize shutdown
459 * A new bootctl tool has been added that is an interface for
460 certain boot loader operations. This is currently a preview
461 and is likely to be extended into a small mechanism daemon
462 like timedated, localed, hostnamed, and can be used by
463 graphical UIs to enumerate available boot options, and
464 request boot into firmware operations.
466 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
467 the rest of the package. It also has been updated to work
468 correctly in initrds.
470 * Policykit previously has been runtime optional, and is now
471 also compile time optional via a configure switch.
473 * systemd-analyze has been reimplemented in C. Also "systemctl
474 dot" has moved into systemd-analyze.
476 * "systemctl status" with no further parameters will now print
477 the status of all active or failed units.
479 * Operations such as "systemctl start" can now be executed
480 with a new mode "--irreversible" which may be used to queue
481 operations that cannot accidentally be reversed by a later
482 job queuing. This is by default used to make shutdown
483 requests more robust.
485 * The Python API of systemd now gained a new module for
486 reading journal files.
488 * A new tool kernel-install has been added that can install
489 kernel images according to the Boot Loader Specification:
491 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
493 * Boot time console output has been improved to provide
494 animated boot time output for hanging jobs.
496 * A new tool systemd-activate has been added which can be used
497 to test socket activation with, directly from the command
498 line. This should make it much easier to test and debug
499 socket activation in daemons.
501 * journalctl gained a new "--reverse" (or -r) option to show
502 journal output in reverse order (i.e. newest line first).
504 * journalctl gained a new "--pager-end" (or -e) option to jump
505 to immediately jump to the end of the journal in the
506 pager. This is only supported in conjunction with "less".
508 * journalctl gained a new "--user-unit=" option, that works
509 similar to "--unit=" but filters for user units rather than
512 * A number of unit files to ease adoption of systemd in
513 initrds has been added. This moves some minimal logic from
514 the various initrd implementations into systemd proper.
516 * The journal files are now owned by a new group
517 "systemd-journal", which exists specifically to allow access
518 to the journal, and nothing else. Previously, we used the
519 "adm" group for that, which however possibly covers more
520 than just journal/log file access. This new group is now
521 already used by systemd-journal-gatewayd to ensure this
522 daemon gets access to the journal files and as little else
523 as possible. Note that "make install" will also set FS ACLs
524 up for /var/log/journal to give "adm" and "wheel" read
525 access to it, in addition to "systemd-journal" which owns
526 the journal files. We recommend that packaging scripts also
527 add read access to "adm" + "wheel" to /var/log/journal, and
528 all existing/future journal files. To normal users and
529 administrators little changes, however packagers need to
530 ensure to create the "systemd-journal" system group at
531 package installation time.
533 * The systemd-journal-gatewayd now runs as unprivileged user
534 systemd-journal-gateway:systemd-journal-gateway. Packaging
535 scripts need to create these system user/group at
538 * timedated now exposes a new boolean property CanNTP that
539 indicates whether a local NTP service is available or not.
541 * systemd-detect-virt will now also detect xen PVs
543 * The pstore file system is now mounted by default, if it is
546 * In addition to the SELinux and IMA policies we will now also
547 load SMACK policies at early boot.
549 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
550 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
551 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
552 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
553 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
554 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
555 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
556 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
557 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
558 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
559 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
560 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
561 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
562 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
566 * Timer units now support calendar time events in addition to
567 monotonic time events. That means you can now trigger a unit
568 based on a calendar time specification such as "Thu,Fri
569 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
570 or fifth day of any month of the year 2013, given that it is
571 a thursday or friday. This brings timer event support
572 considerably closer to cron's capabilities. For details on
573 the supported calendar time specification language see
576 * udev now supports a number of different naming policies for
577 network interfaces for predictable names, and a combination
578 of these policies is now the default. Please see this wiki
579 document for details:
581 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
583 * Auke Kok's bootchart implementation has been added to the
584 systemd tree. It's an optional component that can graph the
585 boot in quite some detail. It's one of the best bootchart
586 implementations around and minimal in its code and
589 * nss-myhostname has been integrated into the systemd source
590 tree. nss-myhostname guarantees that the local hostname
591 always stays resolvable via NSS. It has been a weak
592 requirement of systemd-hostnamed since a long time, and
593 since its code is actually trivial we decided to just
594 include it in systemd's source tree. It can be turned off
595 with a configure switch.
597 * The read-ahead logic is now capable of properly detecting
598 whether a btrfs file system is on SSD or rotating media, in
599 order to optimize the read-ahead scheme. Previously, it was
600 only capable of detecting this on traditional file systems
603 * In udev, additional device properties are now read from the
604 IAB in addition to the OUI database. Also, Bluetooth company
605 identities are attached to the devices as well.
607 * In service files %U may be used as specifier that is
608 replaced by the configured user name of the service.
610 * nspawn may now be invoked without a controlling TTY. This
611 makes it suitable for invocation as its own service. This
612 may be used to set up a simple containerized server system
613 using only core OS tools.
615 * systemd and nspawn can now accept socket file descriptors
616 when they are started for socket activation. This enables
617 implementation of socket activated nspawn
618 containers. i.e. think about autospawning an entire OS image
619 when the first SSH or HTTP connection is received. We expect
620 that similar functionality will also be added to libvirt-lxc
623 * journalctl will now suppress ANSI color codes when
626 * systemctl will no longer show control group information for
627 a unit if a the control group is empty anyway.
629 * logind can now automatically suspend/hibernate/shutdown the
632 * /etc/machine-info and hostnamed now also expose the chassis
633 type of the system. This can be used to determine whether
634 the local system is a laptop, desktop, handset or
635 tablet. This information may either be configured by the
636 user/vendor or is automatically determined from ACPI and DMI
637 information if possible.
639 * A number of PolicyKit actions are now bound together with
640 "imply" rules. This should simplify creating UIs because
641 many actions will now authenticate similar ones as well.
643 * Unit files learnt a new condition ConditionACPower= which
644 may be used to conditionalize a unit depending on whether an
645 AC power source is connected or not, of whether the system
646 is running on battery power.
648 * systemctl gained a new "is-failed" verb that may be used in
649 shell scripts and suchlike to check whether a specific unit
650 is in the "failed" state.
652 * The EnvironmentFile= setting in unit files now supports file
653 globbing, and can hence be used to easily read a number of
654 environment files at once.
656 * systemd will no longer detect and recognize specific
657 distributions. All distribution-specific #ifdeffery has been
658 removed, systemd is now fully generic and
659 distribution-agnostic. Effectively, not too much is lost as
660 a lot of the code is still accessible via explicit configure
661 switches. However, support for some distribution specific
662 legacy configuration file formats has been dropped. We
663 recommend distributions to simply adopt the configuration
664 files everybody else uses now and convert the old
665 configuration from packaging scripts. Most distributions
666 already did that. If that's not possible or desirable,
667 distributions are welcome to forward port the specific
668 pieces of code locally from the git history.
670 * When logging a message about a unit systemd will now always
671 log the unit name in the message meta data.
673 * localectl will now also discover system locale data that is
674 not stored in locale archives, but directly unpacked.
676 * logind will no longer unconditionally use framebuffer
677 devices as seat masters, i.e. as devices that are required
678 to be existing before a seat is considered preset. Instead,
679 it will now look for all devices that are tagged as
680 "seat-master" in udev. By default framebuffer devices will
681 be marked as such, but depending on local systems other
682 devices might be marked as well. This may be used to
683 integrate graphics cards using closed source drivers (such
684 as NVidia ones) more nicely into logind. Note however, that
685 we recommend using the open source NVidia drivers instead,
686 and no udev rules for the closed-source drivers will be
687 shipped from us upstream.
689 Contributions from: Adam Williamson, Alessandro Crismani, Auke
690 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
691 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
692 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
693 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
694 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
695 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
696 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
697 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
698 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
699 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
700 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
705 * udev gained support for loading additional device properties
706 from an indexed database that is keyed by vendor/product IDs
707 and similar device identifiers. For the beginning this
708 "hwdb" is populated with data from the well-known PCI and
709 USB database, but also includes PNP, ACPI and OID data. In
710 the longer run this indexed database shall grow into
711 becoming the one central database for non-essential
712 userspace device metadata. Previously, data from the PCI/USB
713 database was only attached to select devices, since the
714 lookup was a relatively expensive operation due to O(n) time
715 complexity (with n being the number of entries in the
716 database). Since this is now O(1), we decided to add in this
717 data for all devices where this is available, by
718 default. Note that the indexed database needs to be rebuilt
719 when new data files are installed. To achieve this you need
720 to update your packaging scripts to invoke "udevadm hwdb
721 --update" after installation of hwdb data files. For
722 RPM-based distributions we introduced the new
723 %udev_hwdb_update macro for this purpose.
725 * The Journal gained support for the "Message Catalog", an
726 indexed database to link up additional information with
727 journal entries. For further details please check:
729 http://www.freedesktop.org/wiki/Software/systemd/catalog
731 The indexed message catalog database also needs to be
732 rebuilt after installation of message catalog files. Use
733 "journalctl --update-catalog" for this. For RPM-based
734 distributions we introduced the %journal_catalog_update
735 macro for this purpose.
737 * The Python Journal bindings gained support for the standard
738 Python logging framework.
740 * The Journal API gained new functions for checking whether
741 the underlying file system of a journal file is capable of
742 properly reporting file change notifications, or whether
743 applications that want to reflect journal changes "live"
744 need to recheck journal files continuously in appropriate
747 * It is now possible to set the "age" field for tmpfiles
748 entries to 0, indicating that files matching this entry
749 shall always be removed when the directories are cleaned up.
751 * coredumpctl gained a new "gdb" verb which invokes gdb
752 right-away on the selected coredump.
754 * There's now support for "hybrid sleep" on kernels that
755 support this, in addition to "suspend" and "hibernate". Use
756 "systemctl hybrid-sleep" to make use of this.
758 * logind's HandleSuspendKey= setting (and related settings)
759 now gained support for a new "lock" setting to simply
760 request the screen lock on all local sessions, instead of
761 actually executing a suspend or hibernation.
763 * systemd will now mount the EFI variables file system by
766 * Socket units now gained support for configuration of the
767 SMACK security label.
769 * timedatectl will now output the time of the last and next
770 daylight saving change.
772 * We dropped support for various legacy and distro-specific
773 concepts, such as insserv, early-boot SysV services
774 (i.e. those for non-standard runlevels such as 'b' or 'S')
775 or ArchLinux /etc/rc.conf support. We recommend the
776 distributions who still need support this to either continue
777 to maintain the necessary patches downstream, or find a
778 different solution. (Talk to us if you have questions!)
780 * Various systemd components will now bypass PolicyKit checks
781 for root and otherwise handle properly if PolicyKit is not
782 found to be around. This should fix most issues for
783 PolicyKit-less systems. Quite frankly this should have been
784 this way since day one. It is absolutely our intention to
785 make systemd work fine on PolicyKit-less systems, and we
786 consider it a bug if something doesn't work as it should if
787 PolicyKit is not around.
789 * For embedded systems it is now possible to build udev and
790 systemd without blkid and/or kmod support.
792 * "systemctl switch-root" is now capable of switching root
793 more than once. I.e. in addition to transitions from the
794 initrd to the host OS it is now possible to transition to
795 further OS images from the host. This is useful to implement
796 offline updating tools.
798 * Various other additions have been made to the RPM macros
799 shipped with systemd. Use %udev_rules_update() after
800 installing new udev rules files. %_udevhwdbdir,
801 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
802 %_sysctldir are now available which resolve to the right
803 directories for packages to place various data files in.
805 * journalctl gained the new --full switch (in addition to
806 --all, to disable ellipsation for long messages.
808 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
809 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
810 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
811 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
812 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
813 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
814 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
815 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
816 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
820 * journalctl gained new --since= and --until= switches to
821 filter by time. It also now supports nice filtering for
822 units via --unit=/-u.
824 * Type=oneshot services may use ExecReload= and do the
827 * The journal daemon now supports time-based rotation and
828 vacuuming, in addition to the usual disk-space based
831 * The journal will now index the available field values for
832 each field name. This enables clients to show pretty drop
833 downs of available match values when filtering. The bash
834 completion of journalctl has been updated
835 accordingly. journalctl gained a new switch -F to list all
836 values a certain field takes in the journal database.
838 * More service events are now written as structured messages
839 to the journal, and made recognizable via message IDs.
841 * The timedated, localed and hostnamed mini-services which
842 previously only provided support for changing time, locale
843 and hostname settings from graphical DEs such as GNOME now
844 also have a minimal (but very useful) text-based client
845 utility each. This is probably the nicest way to changing
846 these settings from the command line now, especially since
847 it lists available options and is fully integrated with bash
850 * There's now a new tool "systemd-coredumpctl" to list and
851 extract coredumps from the journal.
853 * We now install a README each in /var/log/ and
854 /etc/rc.d/init.d explaining where the system logs and init
855 scripts went. This hopefully should help folks who go to
856 that dirs and look into the otherwise now empty void and
859 * When user-services are invoked (by systemd --user) the
860 $MANAGERPID env var is set to the PID of systemd.
862 * SIGRTMIN+24 when sent to a --user instance will now result
863 in immediate termination of systemd.
865 * gatewayd received numerous feature additions such as a
866 "follow" mode, for live syncing and filtering.
868 * browse.html now allows filtering and showing detailed
869 information on specific entries. Keyboard navigation and
870 mouse screen support has been added.
872 * gatewayd/journalctl now supports HTML5/JSON
873 Server-Sent-Events as output.
875 * The SysV init script compatibility logic will now
876 heuristically determine whether a script supports the
877 "reload" verb, and only then make this available as
880 * "systemctl status --follow" has been removed, use "journalctl
883 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
884 have been removed since they are hardly useful to be
887 * And I'd like to take the opportunity to specifically mention
888 Zbigniew for his great contributions. Zbigniew, you rock!
890 Contributions from: Andrew Eikum, Christian Hesse, Colin
891 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
892 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
893 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
894 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
895 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
896 Jędrzejewski-Szmek, Сковорода Никита Андреевич
900 * If /etc/vconsole.conf is non-existent or empty we will no
901 longer load any console font or key map at boot by
902 default. Instead the kernel defaults will be left
903 intact. This is definitely the right thing to do, as no
904 configuration should mean no configuration, and hard-coding
905 font names that are different on all archs is probably a bad
906 idea. Also, the kernel default key map and font should be
907 good enough for most cases anyway, and mostly identical to
908 the userspace fonts/key maps we previously overloaded them
909 with. If distributions want to continue to default to a
910 non-kernel font or key map they should ship a default
911 /etc/vconsole.conf with the appropriate contents.
913 Contributions from: Colin Walters, Daniel J Walsh, Dave
914 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
915 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
919 * journalctl gained a new --cursor= switch to show entries
920 starting from the specified location in the journal.
922 * We now enforce a size limit on journal entry fields exported
923 with "-o json" in journalctl. Fields larger than 4K will be
924 assigned null. This can be turned off with --all.
926 * An (optional) journal gateway daemon is now available as
927 "systemd-journal-gatewayd.service". This service provides
928 access to the journal via HTTP and JSON. This functionality
929 will be used to implement live log synchronization in both
930 pull and push modes, but has various other users too, such
931 as easy log access for debugging of embedded devices. Right
932 now it is already useful to retrieve the journal via HTTP:
934 # systemctl start systemd-journal-gatewayd.service
935 # wget http://localhost:19531/entries
937 This will download the journal contents in a
938 /var/log/messages compatible format. The same as JSON:
940 # curl -H"Accept: application/json" http://localhost:19531/entries
942 This service is also accessible via a web browser where a
943 single static HTML5 app is served that uses the JSON logic
944 to enable the user to do some basic browsing of the
945 journal. This will be extended later on. Here's an example
946 screenshot of this app in its current state:
948 http://0pointer.de/public/journal-gatewayd
950 Contributions from: Kay Sievers, Lennart Poettering, Robert
951 Milasan, Tom Gundersen
955 * The bash completion logic is now available for journalctl
958 * We don't mount the "cpuset" controller anymore together with
959 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
960 started if no parameters are assigned to it. "cpuset" hence
961 broke code that assumed it it could create "cpu" groups and
964 * journalctl -f will now subscribe to terminal size changes,
965 and line break accordingly.
967 Contributions from: Dave Reisner, Kay Sievers, Lennart
968 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
972 * nspawn will now create a symlink /etc/localtime in the
973 container environment, copying the host's timezone
974 setting. Previously this has been done via a bind mount, but
975 since symlinks cannot be bind mounted this has now been
976 changed to create/update the appropriate symlink.
978 * journalctl -n's line number argument is now optional, and
979 will default to 10 if omitted.
981 * journald will now log the maximum size the journal files may
982 take up on disk. This is particularly useful if the default
983 built-in logic of determining this parameter from the file
984 system size is used. Use "systemctl status
985 systemd-journald.service" to see this information.
987 * The multi-seat X wrapper tool has been stripped down. As X
988 is now capable of enumerating graphics devices via udev in a
989 seat-aware way the wrapper is not strictly necessary
990 anymore. A stripped down temporary stop-gap is still shipped
991 until the upstream display managers have been updated to
992 fully support the new X logic. Expect this wrapper to be
993 removed entirely in one of the next releases.
995 * HandleSleepKey= in logind.conf has been split up into
996 HandleSuspendKey= and HandleHibernateKey=. The old setting
997 is not available anymore. X11 and the kernel are
998 distuingishing between these keys and we should too. This
999 also means the inhibition lock for these keys has been split
1002 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
1003 Poettering, Lukas Nykryn, Václav Pavlín
1007 * Whenever a unit changes state we'll now log this to the
1008 journal and show along the unit's own log output in
1011 * ConditionPathIsMountPoint= can now properly detect bind
1012 mount points too. (Previously, a bind mount of one file
1013 system to another place in the same file system could not be
1014 detected as mount, since they shared struct stat's st_dev
1017 * We will now mount the cgroup controllers cpu, cpuacct,
1018 cpuset and the controllers net_cls, net_prio together by
1021 * nspawn containers will now have a virtualized boot
1022 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
1023 over with a randomized ID at container initialization). This
1024 has the effect of making "journalctl -b" do the right thing
1027 * The JSON output journal serialization has been updated not
1028 to generate "endless" list objects anymore, but rather one
1029 JSON object per line. This is more in line how most JSON
1030 parsers expect JSON objects. The new output mode
1031 "json-pretty" has been added to provide similar output, but
1032 neatly aligned for readability by humans.
1034 * We dropped all explicit sync() invocations in the shutdown
1035 code. The kernel does this implicitly anyway in the kernel
1036 reboot() syscall. halt(8)'s -n option is now a compatibility
1039 * We now support virtualized reboot() in containers, as
1040 supported by newer kernels. We will fall back to exit() if
1041 CAP_SYS_REBOOT is not available to the container. Also,
1042 nspawn makes use of this now and will actually reboot the
1043 container if the containerized OS asks for that.
1045 * journalctl will only show local log output by default
1046 now. Use --merge (-m) to show remote log output, too.
1048 * libsystemd-journal gained the new sd_journal_get_usage()
1049 call to determine the current disk usage of all journal
1050 files. This is exposed in the new "journalctl --disk-usage"
1053 * journald gained a new configuration setting SplitMode= in
1054 journald.conf which may be used to control how user journals
1055 are split off. See journald.conf(5) for details.
1057 * A new condition type ConditionFileNotEmpty= has been added.
1059 * tmpfiles' "w" lines now support file globbing, to write
1060 multiple files at once.
1062 * We added Python bindings for the journal submission
1063 APIs. More Python APIs for a number of selected APIs will
1064 likely follow. Note that we intend to add native bindings
1065 only for the Python language, as we consider it common
1066 enough to deserve bindings shipped within systemd. There are
1067 various projects outside of systemd that provide bindings
1068 for languages such as PHP or Lua.
1070 * Many conditions will now resolve specifiers such as %i. In
1071 addition, PathChanged= and related directives of .path units
1072 now support specifiers as well.
1074 * There's now a new RPM macro definition for the system preset
1077 * journald will now warn if it can't forward a message to the
1078 syslog daemon because it's socket is full.
1080 * timedated will no longer write or process /etc/timezone,
1081 except on Debian. As we do not support late mounted /usr
1082 anymore /etc/localtime always being a symlink is now safe,
1083 and hence the information in /etc/timezone is not necessary
1086 * logind will now always reserve one VT for a text getty (VT6
1087 by default). Previously if more than 6 X sessions where
1088 started they took up all the VTs with auto-spawned gettys,
1089 so that no text gettys were available anymore.
1091 * udev will now automatically inform the btrfs kernel logic
1092 about btrfs RAID components showing up. This should make
1093 simple hotplug based btrfs RAID assembly work.
1095 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
1096 (but not for its children which will stay at the kernel
1097 default). This should allow setups with a lot more listening
1100 * systemd will now always pass the configured timezone to the
1101 kernel at boot. timedated will do the same when the timezone
1104 * logind's inhibition logic has been updated. By default,
1105 logind will now handle the lid switch, the power and sleep
1106 keys all the time, even in graphical sessions. If DEs want
1107 to handle these events on their own they should take the new
1108 handle-power-key, handle-sleep-key and handle-lid-switch
1109 inhibitors during their runtime. A simple way to achiveve
1110 that is to invoke the DE wrapped in an invocation of:
1112 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
1114 * Access to unit operations is now checked via SELinux taking
1115 the unit file label and client process label into account.
1117 * systemd will now notify the administrator in the journal
1118 when he over-mounts a non-empty directory.
1120 * There are new specifiers that are resolved in unit files,
1121 for the host name (%H), the machine ID (%m) and the boot ID
1124 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
1125 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
1126 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
1127 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
1128 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
1129 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
1130 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
1134 * Support for reading structured kernel messages from
1135 /dev/kmsg has now been added and is enabled by default.
1137 * Support for reading kernel messages from /proc/kmsg has now
1138 been removed. If you want kernel messages in the journal
1139 make sure to run a recent kernel (>= 3.5) that supports
1140 reading structured messages from /dev/kmsg (see
1141 above). /proc/kmsg is now exclusive property of classic
1142 syslog daemons again.
1144 * The libudev API gained the new
1145 udev_device_new_from_device_id() call.
1147 * The logic for file system namespace (ReadOnlyDirectory=,
1148 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
1149 require pivot_root() anymore. This means fewer temporary
1150 directories are created below /tmp for this feature.
1152 * nspawn containers will now see and receive all submounts
1153 made on the host OS below the root file system of the
1156 * Forward Secure Sealing is now supported for Journal files,
1157 which provide cryptographical sealing of journal files so
1158 that attackers cannot alter log history anymore without this
1159 being detectable. Lennart will soon post a blog story about
1160 this explaining it in more detail.
1162 * There are two new service settings RestartPreventExitStatus=
1163 and SuccessExitStatus= which allow configuration of exit
1164 status (exit code or signal) which will be excepted from the
1165 restart logic, resp. consider successful.
1167 * journalctl gained the new --verify switch that can be used
1168 to check the integrity of the structure of journal files and
1169 (if Forward Secure Sealing is enabled) the contents of
1172 * nspawn containers will now be run with /dev/stdin, /dev/fd/
1173 and similar symlinks pre-created. This makes running shells
1174 as container init process a lot more fun.
1176 * The fstab support can now handle PARTUUID= and PARTLABEL=
1179 * A new ConditionHost= condition has been added to match
1180 against the hostname (with globs) and machine ID. This is
1181 useful for clusters where a single OS image is used to
1182 provision a large number of hosts which shall run slightly
1183 different sets of services.
1185 * Services which hit the restart limit will now be placed in a
1188 Contributions from: Bertram Poettering, Dave Reisner, Huang
1189 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
1190 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
1194 * When running in --user mode systemd will now become a
1195 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
1196 tree a lot more organized.
1198 * A new PartOf= unit dependency type has been introduced that
1199 may be used to group services in a natural way.
1201 * "systemctl enable" may now be used to enable instances of
1204 * journalctl now prints error log levels in red, and
1205 warning/notice log levels in bright white. It also supports
1206 filtering by log level now.
1208 * cgtop gained a new -n switch (similar to top), to configure
1209 the maximum number of iterations to run for. It also gained
1210 -b, to run in batch mode (accepting no input).
1212 * The suffix ".service" may now be omitted on most systemctl
1213 command lines involving service unit names.
1215 * There's a new bus call in logind to lock all sessions, as
1216 well as a loginctl verb for it "lock-sessions".
1218 * libsystemd-logind.so gained a new call sd_journal_perror()
1219 that works similar to libc perror() but logs to the journal
1220 and encodes structured information about the error number.
1222 * /etc/crypttab entries now understand the new keyfile-size=
1225 * shutdown(8) now can send a (configurable) wall message when
1226 a shutdown is cancelled.
1228 * The mount propagation mode for the root file system will now
1229 default to "shared", which is useful to make containers work
1230 nicely out-of-the-box so that they receive new mounts from
1231 the host. This can be undone locally by running "mount
1232 --make-rprivate /" if needed.
1234 * The prefdm.service file has been removed. Distributions
1235 should maintain this unit downstream if they intend to keep
1236 it around. However, we recommend writing normal unit files
1237 for display managers instead.
1239 * Since systemd is a crucial part of the OS we will now
1240 default to a number of compiler switches that improve
1241 security (hardening) such as read-only relocations, stack
1242 protection, and suchlike.
1244 * The TimeoutSec= setting for services is now split into
1245 TimeoutStartSec= and TimeoutStopSec= to allow configuration
1246 of individual time outs for the start and the stop phase of
1249 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
1250 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
1251 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
1252 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
1253 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
1254 Gundersen, Zbigniew Jędrzejewski-Szmek
1258 * The journal and id128 C APIs are now fully documented as man
1261 * Extra safety checks have been added when transitioning from
1262 the initial RAM disk to the main system to avoid accidental
1265 * /etc/crypttab entries now understand the new keyfile-offset=
1268 * systemctl -t can now be used to filter by unit load state.
1270 * The journal C API gained the new sd_journal_wait() call to
1271 make writing synchronous journal clients easier.
1273 * journalctl gained the new -D switch to show journals from a
1276 * journalctl now displays a special marker between log
1277 messages of two different boots.
1279 * The journal is now explicitly flushed to /var via a service
1280 systemd-journal-flush.service, rather than implicitly simply
1281 by seeing /var/log/journal to be writable.
1283 * journalctl (and the journal C APIs) can now match for much
1284 more complex expressions, with alternatives and
1287 * When transitioning from the initial RAM disk to the main
1288 system we will now kill all processes in a killing spree to
1289 ensure no processes stay around by accident.
1291 * Three new specifiers may be used in unit files: %u, %h, %s
1292 resolve to the user name, user home directory resp. user
1293 shell. This is useful for running systemd user instances.
1295 * We now automatically rotate journal files if their data
1296 object hash table gets a fill level > 75%. We also size the
1297 hash table based on the configured maximum file size. This
1298 together should lower hash collisions drastically and thus
1299 speed things up a bit.
1301 * journalctl gained the new "--header" switch to introspect
1302 header data of journal files.
1304 * A new setting SystemCallFilters= has been added to services
1305 which may be used to apply blacklists or whitelists to
1306 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
1308 * nspawn gained a new --link-journal= switch (and quicker: -j)
1309 to link the container journal with the host. This makes it
1310 very easy to centralize log viewing on the host for all
1311 guests while still keeping the journal files separated.
1313 * Many bugfixes and optimizations
1315 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
1316 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
1317 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
1322 * Several tools now understand kernel command line arguments,
1323 which are only read when run in an initial RAM disk. They
1324 usually follow closely their normal counterparts, but are
1327 * There's a new tool to analyze the readahead files that are
1328 automatically generated at boot. Use:
1330 /usr/lib/systemd/systemd-readahead analyze /.readahead
1332 * We now provide an early debug shell on tty9 if this enabled. Use:
1334 systemctl enable debug-shell.service
1336 * All plymouth related units have been moved into the Plymouth
1337 package. Please make sure to upgrade your Plymouth version
1340 * systemd-tmpfiles now supports getting passed the basename of
1341 a configuration file only, in which case it will look for it
1342 in all appropriate directories automatically.
1344 * udevadm info now takes a /dev or /sys path as argument, and
1345 does the right thing. Example:
1347 udevadm info /dev/sda
1348 udevadm info /sys/class/block/sda
1350 * systemctl now prints a warning if a unit is stopped but a
1351 unit that might trigger it continues to run. Example: a
1352 service is stopped but the socket that activates it is left
1355 * "systemctl status" will now mention if the log output was
1356 shortened due to rotation since a service has been started.
1358 * The journal API now exposes functions to determine the
1359 "cutoff" times due to rotation.
1361 * journald now understands SIGUSR1 and SIGUSR2 for triggering
1362 immediately flushing of runtime logs to /var if possible,
1363 resp. for triggering immediate rotation of the journal
1366 * It is now considered an error if a service is attempted to
1367 be stopped that is not loaded.
1369 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
1371 * systemd-analyze now supports Python 3
1373 * tmpfiles now supports cleaning up directories via aging
1374 where the first level dirs are always kept around but
1375 directories beneath it automatically aged. This is enabled
1376 by prefixing the age field with '~'.
1378 * Seat objects now expose CanGraphical, CanTTY properties
1379 which is required to deal with very fast bootups where the
1380 display manager might be running before the graphics drivers
1381 completed initialization.
1383 * Seat objects now expose a State property.
1385 * We now include RPM macros for service enabling/disabling
1386 based on the preset logic. We recommend RPM based
1387 distributions to make use of these macros if possible. This
1388 makes it simpler to reuse RPM spec files across
1391 * We now make sure that the collected systemd unit name is
1392 always valid when services log to the journal via
1395 * There's a new man page kernel-command-line(7) detailing all
1396 command line options we understand.
1398 * The fstab generator may now be disabled at boot by passing
1399 fstab=0 on the kernel command line.
1401 * A new kernel command line option modules-load= is now understood
1402 to load a specific kernel module statically, early at boot.
1404 * Unit names specified on the systemctl command line are now
1405 automatically escaped as needed. Also, if file system or
1406 device paths are specified they are automatically turned
1407 into the appropriate mount or device unit names. Example:
1409 systemctl status /home
1410 systemctl status /dev/sda
1412 * The SysVConsole= configuration option has been removed from
1413 system.conf parsing.
1415 * The SysV search path is no longer exported on the D-Bus
1418 * The Names= option is been removed from unit file parsing.
1420 * There's a new man page bootup(7) detailing the boot process.
1422 * Every unit and every generator we ship with systemd now
1423 comes with full documentation. The self-explanatory boot is
1426 * A couple of services gained "systemd-" prefixes in their
1427 name if they wrap systemd code, rather than only external
1428 code. Among them fsck@.service which is now
1429 systemd-fsck@.service.
1431 * The HaveWatchdog property has been removed from the D-Bus
1434 * systemd.confirm_spawn= on the kernel command line should now
1437 * There's a new man page crypttab(5) which details all options
1438 we actually understand.
1440 * systemd-nspawn gained a new --capability= switch to pass
1441 additional capabilities to the container.
1443 * timedated will now read known NTP implementation unit names
1444 from /usr/lib/systemd/ntp-units.d/*.list,
1445 systemd-timedated-ntp.target has been removed.
1447 * journalctl gained a new switch "-b" that lists log data of
1448 the current boot only.
1450 * The notify socket is in the abstract namespace again, in
1451 order to support daemons which chroot() at start-up.
1453 * There is a new Storage= configuration option for journald
1454 which allows configuration of where log data should go. This
1455 also provides a way to disable journal logging entirely, so
1456 that data collected is only forwarded to the console, the
1457 kernel log buffer or another syslog implementation.
1459 * Many bugfixes and optimizations
1461 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
1462 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
1463 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
1464 Shawn Landden, Tom Gundersen
1468 * "systemctl help <unit>" now shows the man page if one is
1471 * Several new man pages have been added.
1473 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
1474 MaxLevelConsole= can now be specified in
1475 journald.conf. These options allow reducing the amount of
1476 data stored on disk or forwarded by the log level.
1478 * TimerSlackNSec= can now be specified in system.conf for
1479 PID1. This allows system-wide power savings.
1481 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
1482 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
1487 * logind is now capable of (optionally) handling power and
1488 sleep keys as well as the lid switch.
1490 * journalctl now understands the syntax "journalctl
1491 /usr/bin/avahi-daemon" to get all log output of a specific
1494 * CapabilityBoundingSet= in system.conf now also influences
1495 the capability bound set of usermode helpers of the kernel.
1497 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
1498 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
1499 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
1500 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
1504 * Note that we skipped 139 releases here in order to set the
1505 new version to something that is greater than both udev's
1506 and systemd's most recent version number.
1508 * udev: all udev sources are merged into the systemd source tree now.
1509 All future udev development will happen in the systemd tree. It
1510 is still fully supported to use the udev daemon and tools without
1511 systemd running, like in initramfs or other init systems. Building
1512 udev though, will require the *build* of the systemd tree, but
1513 udev can be properly *run* without systemd.
1515 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
1516 should be used to create dead device nodes as workarounds for broken
1519 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
1520 no longer supported. udev_monitor_new_from_netlink() needs to be
1521 used to subscribe to events.
1523 * udev: when udevd is started by systemd, processes which are left
1524 behind by forking them off of udev rules, are unconditionally cleaned
1525 up and killed now after the event handling has finished. Services or
1526 daemons must be started as systemd services. Services can be
1527 pulled-in by udev to get started, but they can no longer be directly
1528 forked by udev rules.
1530 * udev: the daemon binary is called systemd-udevd now and installed
1531 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
1532 to adapt to that, create symlink, or rename the binary after building
1535 * libudev no longer provides these symbols:
1536 udev_monitor_from_socket()
1537 udev_queue_get_failed_list_entry()
1538 udev_get_{dev,sys,run}_path()
1539 The versions number was bumped and symbol versioning introduced.
1541 * systemd-loginctl and systemd-journalctl have been renamed
1542 to loginctl and journalctl to match systemctl.
1544 * The config files: /etc/systemd/systemd-logind.conf and
1545 /etc/systemd/systemd-journald.conf have been renamed to
1546 logind.conf and journald.conf. Package updates should rename
1547 the files to the new names on upgrade.
1549 * For almost all files the license is now LGPL2.1+, changed
1550 from the previous GPL2.0+. Exceptions are some minor stuff
1551 of udev (which will be changed to LGPL2.1 eventually, too),
1552 and the MIT licensed sd-daemon.[ch] library that is suitable
1553 to be used as drop-in files.
1555 * systemd and logind now handle system sleep states, in
1556 particular suspending and hibernating.
1558 * logind now implements a sleep/shutdown/idle inhibiting logic
1559 suitable for a variety of uses. Soonishly Lennart will blog
1560 about this in more detail.
1562 * var-run.mount and var-lock.mount are no longer provided
1563 (which prevously bind mounted these directories to their new
1564 places). Distributions which have not converted these
1565 directories to symlinks should consider stealing these files
1566 from git history and add them downstream.
1568 * We introduced the Documentation= field for units and added
1569 this to all our shipped units. This is useful to make it
1570 easier to explore the boot and the purpose of the various
1573 * All smaller setup units (such as
1574 systemd-vconsole-setup.service) now detect properly if they
1575 are run in a container and are skipped when
1576 appropriate. This guarantees an entirely noise-free boot in
1577 Linux container environments such as systemd-nspawn.
1579 * A framework for implementing offline system updates is now
1580 integrated, for details see:
1581 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
1583 * A new service type Type=idle is available now which helps us
1584 avoiding ugly interleaving of getty output and boot status
1587 * There's now a system-wide CapabilityBoundingSet= option to
1588 globally reduce the set of capabilities for the
1589 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
1590 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
1591 even CAP_NET_ADMIN system-wide for secure systems.
1593 * There are now system-wide DefaultLimitXXX= options to
1594 globally change the defaults of the various resource limits
1595 for all units started by PID 1.
1597 * Harald Hoyer's systemd test suite has been integrated into
1598 systemd which allows easy testing of systemd builds in qemu
1599 and nspawn. (This is really awesome! Ask us for details!)
1601 * The fstab parser is now implemented as generator, not inside
1604 * systemctl will now warn you if .mount units generated from
1605 /etc/fstab are out of date due to changes in fstab that
1606 haven't been read by systemd yet.
1608 * systemd is now suitable for usage in initrds. Dracut has
1609 already been updated to make use of this. With this in place
1610 initrds get a slight bit faster but primarily are much
1611 easier to introspect and debug since "systemctl status" in
1612 the host system can be used to introspect initrd services,
1613 and the journal from the initrd is kept around too.
1615 * systemd-delta has been added, a tool to explore differences
1616 between user/admin configuration and vendor defaults.
1618 * PrivateTmp= now affects both /tmp and /var/tmp.
1620 * Boot time status messages are now much prettier and feature
1621 proper english language. Booting up systemd has never been
1624 * Read-ahead pack files now include the inode number of all
1625 files to pre-cache. When the inode changes the pre-caching
1626 is not attempted. This should be nicer to deal with updated
1627 packages which might result in changes of read-ahead
1630 * We now temporaritly lower the kernel's read_ahead_kb variable
1631 when collecting read-ahead data to ensure the kernel's
1632 built-in read-ahead does not add noise to our measurements
1633 of necessary blocks to pre-cache.
1635 * There's now RequiresMountsFor= to add automatic dependencies
1636 for all mounts necessary for a specific file system path.
1638 * MountAuto= and SwapAuto= have been removed from
1639 system.conf. Mounting file systems at boot has to take place
1642 * nspawn now learned a new switch --uuid= to set the machine
1643 ID on the command line.
1645 * nspawn now learned the -b switch to automatically search
1648 * vt102 is now the default TERM for serial TTYs, upgraded from
1651 * systemd-logind now works on VT-less systems.
1653 * The build tree has been reorganized. The individual
1654 components now have directories of their own.
1656 * A new condition type ConditionPathIsReadWrite= is now available.
1658 * nspawn learned the new -C switch to create cgroups for the
1659 container in other hierarchies.
1661 * We now have support for hardware watchdogs, configurable in
1664 * The scheduled shutdown logic now has a public API.
1666 * We now mount /tmp as tmpfs by default, but this can be
1667 masked and /etc/fstab can override it.
1669 * Since udisks doesn't make use of /media anymore we are not
1670 mounting a tmpfs on it anymore.
1672 * journalctl gained a new --local switch to only interleave
1673 locally generated journal files.
1675 * We can now load the IMA policy at boot automatically.
1677 * The GTK tools have been split off into a systemd-ui.
1679 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
1680 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
1681 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
1682 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
1683 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
1684 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
1685 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
1686 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
1687 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
1692 * This is mostly a bugfix release
1694 * Support optional initialization of the machine ID from the
1695 KVM or container configured UUID.
1697 * Support immediate reboots with "systemctl reboot -ff"
1699 * Show /etc/os-release data in systemd-analyze output
1701 * Many bugfixes for the journal, including endianness fixes and
1702 ensuring that disk space enforcement works
1704 * sd-login.h is C++ comptaible again
1706 * Extend the /etc/os-release format on request of the Debian
1709 * We now refuse non-UTF8 strings used in various configuration
1710 and unit files. This is done to ensure we don't pass invalid
1711 data over D-Bus or expose it elsewhere.
1713 * Register Mimo USB Screens as suitable for automatic seat
1716 * Read SELinux client context from journal clients in a race
1719 * Reorder configuration file lookup order. /etc now always
1720 overrides /run in order to allow the administrator to always
1721 and unconditionally override vendor supplied or
1722 automatically generated data.
1724 * The various user visible bits of the journal now have man
1725 pages. We still lack man pages for the journal API calls
1728 * We now ship all man pages in HTML format again in the
1731 Contributions from: Dave Reisner, Dirk Eibach, Frederic
1732 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
1733 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
1738 * This is mostly a bugfix release
1740 * systems lacking /etc/os-release are no longer supported.
1742 * Various functionality updates to libsystemd-login.so
1744 * Track class of PAM logins to distuingish greeters from
1747 Contributions from: Kay Sievers, Lennart Poettering, Michael
1752 * This is an important bugfix release for v41.
1754 * Building man pages is now optional which should be useful
1755 for those building systemd from git but unwilling to install
1758 * Watchdog support for supervising services is now usable. In
1759 a future release support for hardware watchdogs
1760 (i.e. /dev/watchdog) will be added building on this.
1762 * Service start rate limiting is now configurable and can be
1763 turned off per service. When a start rate limit is hit a
1764 reboot can automatically be triggered.
1766 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
1768 Contributions from: Benjamin Franzke, Bill Nottingham,
1769 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
1770 Schmidt, Michał Górny, Piotr Drąg
1774 * The systemd binary is installed /usr/lib/systemd/systemd now;
1775 An existing /sbin/init symlink needs to be adapted with the
1778 * The code that loads kernel modules has been ported to invoke
1779 libkmod directly, instead of modprobe. This means we do not
1780 support systems with module-init-tools anymore.
1782 * Watchdog support is now already useful, but still not
1785 * A new kernel command line option systemd.setenv= is
1786 understood to set system wide environment variables
1787 dynamically at boot.
1789 * We now limit the set of capabilities of systemd-journald.
1791 * We now set SIGPIPE to ignore by default, since it only is
1792 useful in shell pipelines, and has little use in general
1793 code. This can be disabled with IgnoreSIPIPE=no in unit
1796 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
1797 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
1802 * This is mostly a bugfix release
1804 * We now expose the reason why a service failed in the
1805 "Result" D-Bus property.
1807 * Rudimentary service watchdog support (will be completed over
1808 the next few releases.)
1810 * When systemd forks off in order execute some service we will
1811 now immediately changes its argv[0] to reflect which process
1812 it will execute. This is useful to minimize the time window
1813 with a generic argv[0], which makes bootcharts more useful
1815 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
1816 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
1817 Mike Kazantsev, Ray Strode
1821 * This is mostly a test release, but incorporates many
1824 * New systemd-cgtop tool to show control groups by their
1827 * Linking against libacl for ACLs is optional again. If
1828 disabled, support tracking device access for active logins
1829 goes becomes unavailable, and so does access to the user
1830 journals by the respective users.
1832 * If a group "adm" exists, journal files are automatically
1833 owned by them, thus allow members of this group full access
1834 to the system journal as well as all user journals.
1836 * The journal now stores the SELinux context of the logging
1837 client for all entries.
1839 * Add C++ inclusion guards to all public headers
1841 * New output mode "cat" in the journal to print only text
1842 messages, without any meta data like date or time.
1844 * Include tiny X server wrapper as a temporary stop-gap to
1845 teach XOrg udev display enumeration. This is used by display
1846 managers such as gdm, and will go away as soon as XOrg
1847 learned native udev hotplugging for display devices.
1849 * Add new systemd-cat tool for executing arbitrary programs
1850 with STDERR/STDOUT connected to the journal. Can also act as
1851 BSD logger replacement, and does so by default.
1853 * Optionally store all locally generated coredumps in the
1854 journal along with meta data.
1856 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
1857 writing short strings to files (for usage for /sys), and for
1858 creating symlinks, character and block device nodes.
1860 * New unit file option ControlGroupPersistent= to make cgroups
1861 persistent, following the mechanisms outlined in
1862 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
1864 * Support multiple local RTCs in a sane way
1866 * No longer monopolize IO when replaying readahead data on
1867 rotating disks, since we might starve non-file-system IO to
1868 death, since fanotify() will not see accesses done by blkid,
1871 * Don't show kernel threads in systemd-cgls anymore, unless
1872 requested with new -k switch.
1874 Contributions from: Dan Horák, Kay Sievers, Lennart
1875 Poettering, Michal Schmidt
1879 * This is mostly a test release, but incorporates many
1882 * The git repository moved to:
1883 git://anongit.freedesktop.org/systemd/systemd
1884 ssh://git.freedesktop.org/git/systemd/systemd
1886 * First release with the journal
1887 http://0pointer.de/blog/projects/the-journal.html
1889 * The journal replaces both systemd-kmsg-syslogd and
1890 systemd-stdout-bridge.
1892 * New sd_pid_get_unit() API call in libsystemd-logind
1894 * Many systemadm clean-ups
1896 * Introduce remote-fs-pre.target which is ordered before all
1897 remote mounts and may be used to start services before all
1900 * Added Mageia support
1902 * Add bash completion for systemd-loginctl
1904 * Actively monitor PID file creation for daemons which exit in
1905 the parent process before having finished writing the PID
1906 file in the daemon process. Daemons which do this need to be
1907 fixed (i.e. PID file creation must have finished before the
1908 parent exits), but we now react a bit more gracefully to them.
1910 * Add colourful boot output, mimicking the well-known output
1911 of existing distributions.
1913 * New option PassCredentials= for socket units, for
1914 compatibility with a recent kernel ABI breakage.
1916 * /etc/rc.local is now hooked in via a generator binary, and
1917 thus will no longer act as synchronization point during
1920 * systemctl list-unit-files now supports --root=.
1922 * systemd-tmpfiles now understands two new commands: z, Z for
1923 relabelling files according to the SELinux database. This is
1924 useful to apply SELinux labels to specific files in /sys,
1927 * Output of SysV services is now forwarded to both the console
1928 and the journal by default, not only just the console.
1930 * New man pages for all APIs from libsystemd-login.
1932 * The build tree got reorganized and a the build system is a
1933 lot more modular allowing embedded setups to specifically
1934 select the components of systemd they are interested in.
1936 * Support for Linux systems lacking the kernel VT subsystem is
1939 * configure's --with-rootdir= got renamed to
1940 --with-rootprefix= to follow the naming used by udev and
1943 * Unless specified otherwise we'll now install to /usr instead
1944 of /usr/local by default.
1946 * Processes with '@' in argv[0][0] are now excluded from the
1947 final shut-down killing spree, following the logic explained
1949 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
1951 * All processes remaining in a service cgroup when we enter
1952 the START or START_PRE states are now killed with
1953 SIGKILL. That means it is no longer possible to spawn
1954 background processes from ExecStart= lines (which was never
1955 supported anyway, and bad style).
1957 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
1958 reloading of units together.
1960 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
1961 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
1962 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
1963 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
1964 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
~ianmdlvl / elogind.git / blob