1 systemd System and Service Manager
5 * The output of 'systemctl list-jobs' got some polishing. The
6 '--type=' argument may now be passed more than once. A new
7 command 'systemctl list-sockets' has been added which shows
8 a list of kernel sockets systemd is listening on with the
9 socket units they belong to, plus the units these socket
12 * The experimental libsystemd-bus library got substantial
13 updates to work in conjunction with the (also experimental)
14 kdbus kernel project. It works well enough to exchange
15 messages with some sophistication. Note that kdbus is not
16 ready yet, and the library is mostly an elaborate test case
17 for now, and not installable.
19 * systemd gained a new unit 'systemd-static-nodes.service'
20 that generates static device nodes earlier during boot, and
21 can run in conjunction with udev.
23 * libsystemd-login gained a new call sd_pid_get_user_unit()
24 to retrieve the user systemd unit a process is running
25 in. This is useful for systems where systemd is used as
28 * systemd-nspawn now places all containers in the new /machine
29 top-level cgroup directory in the name=systemd
30 hierarchy. libvirt will soon do the same, so that we get a
31 uniform separation of /system, /user and /machine for system
32 services, user processes and containers/virtual
33 machines. This new cgroup hierarchy is also useful to stick
34 stable names to specific container instances, which can be
35 recognized later this way (this name may be controlled
36 via systemd-nspawn's new -M switch). libsystemd-login also
37 gained a new call sd_pid_get_machine_name() to retrieve the
38 name of the container/VM a specific process belongs to.
40 * bootchart can now store its data in the journal.
42 * libsystemd-journal gained a new call
43 sd_journal_add_conjunction() for AND expressions to the
44 matching logic. This can be used to express more complex
47 * journactl can now take multiple --unit= and --user-unit=
50 * The cryptsetup logic now understands the "luks.key=" kernel
51 command line switch for specifying a file to read the
52 decryption key from. Also, if a configured key file is not
53 found the tool will now automatically fall back to prompting
56 * Python systemd.journal module was updated to wrap recently
57 added functions from libsystemd-journal. The interface was
58 changed to bring the low level interface in s.j._Reader
59 closer to the C API, and the high level interface in
60 s.j.Reader was updated to wrap and convert all data about
63 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
64 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
65 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
66 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
67 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
68 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
72 * journalctl --update-catalog now understands a new --root=
73 option to operate on catalogs found in a different root
76 * During shutdown after systemd has terminated all running
77 services a final killing loop kills all remaining left-over
78 processes. We will now print the name of these processes
79 when we send SIGKILL to them, since this usually indicates a
82 * If /etc/crypttab refers to password files stored on
83 configured mount points automatic dependencies will now be
84 generated to ensure the specific mount is established first
85 before the key file is attempted to be read.
87 * 'systemctl status' will now show information about the
88 network sockets a socket unit is listening on.
90 * 'systemctl status' will also shown information about any
91 drop-in configuration file for units. (Drop-In configuration
92 files in this context are files such as
93 /etc/systemd/systemd/foobar.service.d/*.conf)
95 * systemd-cgtop now optionally shows summed up CPU times of
96 cgroups. Press '%' while running cgtop to switch between
97 percentage and absolute mode. This is useful to determine
98 which cgroups use up the most CPU time over the entire
99 runtime of the system. systemd-cgtop has also been updated
100 to be 'pipeable' for processing with further shell tools.
102 * 'hostnamectl set-hostname' will now allow setting of FQDN
105 * The formatting and parsing of time span values has been
106 changed. The parser now understands fractional expressions
107 such as "5.5h". The formatter will now output fractional
108 expressions for all time spans under 1min, i.e. "5.123456s"
109 rather than "5s 123ms 456us". For time spans under 1s
110 millisecond values are shown, for those under 1ms
111 microsecond values are shown. This should greatly improve
112 all time-related output of systemd.
114 * libsystemd-login and libsystemd-journal gained new
115 functions for querying the poll() events mask and poll()
116 timeout value for integration into arbitrary event
119 * localectl gained the ability to list available X11 keymaps
120 (models, layouts, variants, options).
122 * 'systemd-analyze dot' gained the ability to filter for
123 specific units via shell-style globs, to create smaller,
124 more useful graphs. I.e. it's now possible to create simple
125 graphs of all the dependencies between only target units, or
126 of all units that Avahi has dependencies with.
128 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
129 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
130 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
131 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
132 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
133 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
134 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
138 * The boot-time readahead implementation for rotating media
139 will now read the read-ahead data in multiple passes which
140 consist of all read requests made in equidistant time
141 intervals. This means instead of strictly reading read-ahead
142 data in its physical order on disk we now try to find a
143 middle ground between physical and access time order.
145 * /etc/os-release files gained a new BUILD_ID= field for usage
146 on operating systems that provide continuous builds of OS
149 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
150 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
151 William Douglas, Zbigniew Jędrzejewski-Szmek
155 * systemd-python gained an API exposing libsystemd-daemon.
157 * The SMACK setup logic gained support for uploading CIPSO
160 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
161 ReadOnlyDirectories= and InaccessibleDirectories= has
162 changed. The private /tmp and /var/tmp directories are now
163 shared by all processes of a service (which means
164 ExecStartPre= may now leave data in /tmp that ExecStart= of
165 the same service can still access). When a service is
166 stopped its temporary directories are immediately deleted
167 (normal clean-up with tmpfiles is still done in addition to
170 * By default, systemd will now set a couple of sysctl
171 variables in the kernel: the safe sysrq options are turned
172 on, IP route verification is turned on, and source routing
173 disabled. The recently added hardlink and softlink
174 protection of the kernel is turned on. These settings should
175 be reasonably safe, and good defaults for all new systems.
177 * The predictable network naming logic may now be turned off
178 with a new kernel command line switch: net.ifnames=0.
180 * A new libsystemd-bus module has been added that implements a
181 pretty complete D-Bus client library. For details see:
183 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
185 * journald will now explicitly flush the journal files to disk
186 at the latest 5min after each write. The file will then also
187 be marked offline until the next write. This should increase
188 reliability in case of a crash. The synchronization delay
189 can be configured via SyncIntervalSec= in journald.conf.
191 * There's a new remote-fs-setup.target unit that can be used
192 to pull in specific services when at least one remote file
193 system is to be mounted.
195 * There are new targets timers.target and paths.target as
196 canonical targets to pull user timer and path units in
197 from. This complements sockets.target with a similar
198 purpose for socket units.
200 * libudev gained a new call udev_device_set_attribute_value()
201 to set sysfs attributes of a device.
203 * The udev daemon now sets the default number of worker
204 processes executed in parallel based on the number of available
205 CPUs instead of the amount of available RAM. This is supposed
206 to provide a more reliable default and limit a too aggressive
207 paralellism for setups with 1000s of devices connected.
209 Contributions from: Auke Kok, Colin Walters, Cristian
210 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
211 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
212 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
213 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
214 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
215 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
216 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
217 Zbigniew Jędrzejewski-Szmek
221 * Configuration of unit files may now be extended via drop-in
222 files without having to edit/override the unit files
223 themselves. More specifically, if the administrator wants to
224 change one value for a service file foobar.service he can
225 now do so by dropping in a configuration snippet into
226 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
227 will load all these snippets and apply them on top of the
228 main unit configuration file, possibly extending or
229 overriding its settings. Using these drop-in snippets is
230 generally nicer than the two earlier options for changing
231 unit files locally: copying the files from
232 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
233 them there; or creating a new file in /etc/systemd/system/
234 that incorporates the original one via ".include". Drop-in
235 snippets into these .d/ directories can be placed in any
236 directory systemd looks for units in, and the usual
237 overriding semantics between /usr/lib, /etc and /run apply
240 * Most unit file settings which take lists of items can now be
241 reset by assigning the empty string to them. For example,
242 normally, settings such as Environment=FOO=BAR append a new
243 environment variable assignment to the environment block,
244 each time they are used. By assigning Environment= the empty
245 string the environment block can be reset to empty. This is
246 particularly useful with the .d/*.conf drop-in snippets
247 mentioned above, since this adds the ability to reset list
248 settings from vendor unit files via these drop-ins.
250 * systemctl gained a new "list-dependencies" command for
251 listing the dependencies of a unit recursively.
253 * Inhibitors are now honored and listed by "systemctl
254 suspend", "systemctl poweroff" (and similar) too, not only
255 GNOME. These commands will also list active sessions by
258 * Resource limits (as exposed by the various control group
259 controllers) can now be controlled dynamically at runtime
260 for all units. More specifically, you can now use a command
261 like "systemctl set-cgroup-attr foobar.service cpu.shares
262 2000" to alter the CPU shares a specific service gets. These
263 settings are stored persistently on disk, and thus allow the
264 administrator to easily adjust the resource usage of
265 services with a few simple commands. This dynamic resource
266 management logic is also available to other programs via the
267 bus. Almost any kernel cgroup attribute and controller is
270 * systemd-vconsole-setup will now copy all font settings to
271 all allocated VTs, where it previously applied them only to
274 * libsystemd-login gained the new sd_session_get_tty() API
277 * This release drops support for a few legacy or
278 distribution-specific LSB facility names when parsing init
279 scripts: $x-display-manager, $mail-transfer-agent,
280 $mail-transport-agent, $mail-transfer-agent, $smtp,
281 $null. Also, the mail-transfer-agent.target unit backing
282 this has been removed. Distributions which want to retain
283 compatibility with this should carry the burden for
284 supporting this themselves and patch support for these back
285 in, if they really need to. Also, the facilities $syslog and
286 $local_fs are now ignored, since systemd does not support
287 early-boot LSB init scripts anymore, and these facilities
288 are implied anyway for normal services. syslog.target has
291 * There are new bus calls on PID1's Manager object for
292 cancelling jobs, and removing snapshot units. Previously,
293 both calls were only available on the Job and Snapshot
296 * systemd-journal-gatewayd gained SSL support.
298 * The various "environment" files, such as /etc/locale.conf
299 now support continuation lines with a backslash ("\") as
300 last character in the line, similar in style (but different)
301 to how this is supported in shells.
303 * For normal user processes the _SYSTEMD_USER_UNIT= field is
304 now implicitly appended to every log entry logged. systemctl
305 has been updated to filter by this field when operating on a
306 user systemd instance.
308 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
309 CAP_AUDIT_CONTROL capabilities to the capabilities set for
310 the container. This makes it easier to boot unmodified
311 Fedora systems in a container, which however still requires
312 audit=0 to be passed on the kernel command line. Auditing in
313 kernel and userspace is unfortunately still too broken in
314 context of containers, hence we recommend compiling it out
315 of the kernel or using audit=0. Hopefully this will be fixed
316 one day for good in the kernel.
318 * nspawn gained the new --bind= and --bind-ro= parameters to
319 bind mount specific directories from the host into the
322 * nspawn will now mount its own devpts file system instance
323 into the container, in order not to leak pty devices from
324 the host into the container.
326 * systemd will now read the firmware boot time performance
327 information from the EFI variables, if the used boot loader
328 supports this, and takes it into account for boot performance
329 analysis via "systemd-analyze". This is currently supported
330 only in conjunction with Gummiboot, but could be supported
331 by other boot loaders too. For details see:
333 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
335 * A new generator has been added that automatically mounts the
336 EFI System Partition (ESP) to /boot, if that directory
337 exists, is empty, and no other file system has been
338 configured to be mounted there.
340 * logind will now send out PrepareForSleep(false) out
341 unconditionally, after coming back from suspend. This may be
342 used by applications as asynchronous notification for
343 system resume events.
345 * "systemctl unlock-sessions" has been added, that allows
346 unlocking the screens of all user sessions at once, similar
347 how "systemctl lock-sessions" already locked all users
348 sessions. This is backed by a new D-Bus call UnlockSessions().
350 * "loginctl seat-status" will now show the master device of a
351 seat. (i.e. the device of a seat that needs to be around for
352 the seat to be considered available, usually the graphics
355 * tmpfiles gained a new "X" line type, that allows
356 configuration of files and directories (with wildcards) that
357 shall be excluded from automatic cleanup ("aging").
359 * udev default rules set the device node permissions now only
360 at "add" events, and do not change them any longer with a
361 later "change" event.
363 * The log messages for lid events and power/sleep keypresses
364 now carry a message ID.
366 * We now have a substantially larger unit test suite, but this
367 continues to be work in progress.
369 * udevadm hwdb gained a new --root= parameter to change the
370 root directory to operate relative to.
372 * logind will now issue a background sync() request to the kernel
373 early at shutdown, so that dirty buffers are flushed to disk early
374 instead of at the last moment, in order to optimize shutdown
377 * A new bootctl tool has been added that is an interface for
378 certain boot loader operations. This is currently a preview
379 and is likely to be extended into a small mechanism daemon
380 like timedated, localed, hostnamed, and can be used by
381 graphical UIs to enumerate available boot options, and
382 request boot into firmware operations.
384 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
385 the rest of the package. It also has been updated to work
386 correctly in initrds.
388 * Policykit previously has been runtime optional, and is now
389 also compile time optional via a configure switch.
391 * systemd-analyze has been reimplemented in C. Also "systemctl
392 dot" has moved into systemd-analyze.
394 * "systemctl status" with no further parameters will now print
395 the status of all active or failed units.
397 * Operations such as "systemctl start" can now be executed
398 with a new mode "--irreversible" which may be used to queue
399 operations that cannot accidentally be reversed by a later
400 job queuing. This is by default used to make shutdown
401 requests more robust.
403 * The Python API of systemd now gained a new module for
404 reading journal files.
406 * A new tool kernel-install has been added that can install
407 kernel images according to the Boot Loader Specification:
409 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
411 * Boot time console output has been improved to provide
412 animated boot time output for hanging jobs.
414 * A new tool systemd-activate has been added which can be used
415 to test socket activation with, directly from the command
416 line. This should make it much easier to test and debug
417 socket activation in daemons.
419 * journalctl gained a new "--reverse" (or -r) option to show
420 journal output in reverse order (i.e. newest line first).
422 * journalctl gained a new "--pager-end" (or -e) option to jump
423 to immediately jump to the end of the journal in the
424 pager. This is only supported in conjunction with "less".
426 * journalctl gained a new "--user-unit=" option, that works
427 similar to "--unit=" but filters for user units rather than
430 * A number of unit files to ease adoption of systemd in
431 initrds has been added. This moves some minimal logic from
432 the various initrd implementations into systemd proper.
434 * The journal files are now owned by a new group
435 "systemd-journal", which exists specifically to allow access
436 to the journal, and nothing else. Previously, we used the
437 "adm" group for that, which however possibly covers more
438 than just journal/log file access. This new group is now
439 already used by systemd-journal-gatewayd to ensure this
440 daemon gets access to the journal files and as little else
441 as possible. Note that "make install" will also set FS ACLs
442 up for /var/log/journal to give "adm" and "wheel" read
443 access to it, in addition to "systemd-journal" which owns
444 the journal files. We recommend that packaging scripts also
445 add read access to "adm" + "wheel" to /var/log/journal, and
446 all existing/future journal files. To normal users and
447 administrators little changes, however packagers need to
448 ensure to create the "systemd-journal" system group at
449 package installation time.
451 * The systemd-journal-gatewayd now runs as unprivileged user
452 systemd-journal-gateway:systemd-journal-gateway. Packaging
453 scripts need to create these system user/group at
456 * timedated now exposes a new boolean property CanNTP that
457 indicates whether a local NTP service is available or not.
459 * systemd-detect-virt will now also detect xen PVs
461 * The pstore file system is now mounted by default, if it is
464 * In addition to the SELinux and IMA policies we will now also
465 load SMACK policies at early boot.
467 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
468 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
469 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
470 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
471 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
472 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
473 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
474 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
475 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
476 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
477 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
478 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
479 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
480 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
484 * Timer units now support calendar time events in addition to
485 monotonic time events. That means you can now trigger a unit
486 based on a calendar time specification such as "Thu,Fri
487 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
488 or fifth day of any month of the year 2013, given that it is
489 a thursday or friday. This brings timer event support
490 considerably closer to cron's capabilities. For details on
491 the supported calendar time specification language see
494 * udev now supports a number of different naming policies for
495 network interfaces for predictable names, and a combination
496 of these policies is now the default. Please see this wiki
497 document for details:
499 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
501 * Auke Kok's bootchart implementation has been added to the
502 systemd tree. It's an optional component that can graph the
503 boot in quite some detail. It's one of the best bootchart
504 implementations around and minimal in its code and
507 * nss-myhostname has been integrated into the systemd source
508 tree. nss-myhostname guarantees that the local hostname
509 always stays resolvable via NSS. It has been a weak
510 requirement of systemd-hostnamed since a long time, and
511 since its code is actually trivial we decided to just
512 include it in systemd's source tree. It can be turned off
513 with a configure switch.
515 * The read-ahead logic is now capable of properly detecting
516 whether a btrfs file system is on SSD or rotating media, in
517 order to optimize the read-ahead scheme. Previously, it was
518 only capable of detecting this on traditional file systems
521 * In udev, additional device properties are now read from the
522 IAB in addition to the OUI database. Also, Bluetooth company
523 identities are attached to the devices as well.
525 * In service files %U may be used as specifier that is
526 replaced by the configured user name of the service.
528 * nspawn may now be invoked without a controlling TTY. This
529 makes it suitable for invocation as its own service. This
530 may be used to set up a simple containerized server system
531 using only core OS tools.
533 * systemd and nspawn can now accept socket file descriptors
534 when they are started for socket activation. This enables
535 implementation of socket activated nspawn
536 containers. i.e. think about autospawning an entire OS image
537 when the first SSH or HTTP connection is received. We expect
538 that similar functionality will also be added to libvirt-lxc
541 * journalctl will now suppress ANSI color codes when
544 * systemctl will no longer show control group information for
545 a unit if a the control group is empty anyway.
547 * logind can now automatically suspend/hibernate/shutdown the
550 * /etc/machine-info and hostnamed now also expose the chassis
551 type of the system. This can be used to determine whether
552 the local system is a laptop, desktop, handset or
553 tablet. This information may either be configured by the
554 user/vendor or is automatically determined from ACPI and DMI
555 information if possible.
557 * A number of PolicyKit actions are now bound together with
558 "imply" rules. This should simplify creating UIs because
559 many actions will now authenticate similar ones as well.
561 * Unit files learnt a new condition ConditionACPower= which
562 may be used to conditionalize a unit depending on whether an
563 AC power source is connected or not, of whether the system
564 is running on battery power.
566 * systemctl gained a new "is-failed" verb that may be used in
567 shell scripts and suchlike to check whether a specific unit
568 is in the "failed" state.
570 * The EnvironmentFile= setting in unit files now supports file
571 globbing, and can hence be used to easily read a number of
572 environment files at once.
574 * systemd will no longer detect and recognize specific
575 distributions. All distribution-specific #ifdeffery has been
576 removed, systemd is now fully generic and
577 distribution-agnostic. Effectively, not too much is lost as
578 a lot of the code is still accessible via explicit configure
579 switches. However, support for some distribution specific
580 legacy configuration file formats has been dropped. We
581 recommend distributions to simply adopt the configuration
582 files everybody else uses now and convert the old
583 configuration from packaging scripts. Most distributions
584 already did that. If that's not possible or desirable,
585 distributions are welcome to forward port the specific
586 pieces of code locally from the git history.
588 * When logging a message about a unit systemd will now always
589 log the unit name in the message meta data.
591 * localectl will now also discover system locale data that is
592 not stored in locale archives, but directly unpacked.
594 * logind will no longer unconditionally use framebuffer
595 devices as seat masters, i.e. as devices that are required
596 to be existing before a seat is considered preset. Instead,
597 it will now look for all devices that are tagged as
598 "seat-master" in udev. By default framebuffer devices will
599 be marked as such, but depending on local systems other
600 devices might be marked as well. This may be used to
601 integrate graphics cards using closed source drivers (such
602 as NVidia ones) more nicely into logind. Note however, that
603 we recommend using the open source NVidia drivers instead,
604 and no udev rules for the closed-source drivers will be
605 shipped from us upstream.
607 Contributions from: Adam Williamson, Alessandro Crismani, Auke
608 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
609 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
610 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
611 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
612 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
613 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
614 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
615 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
616 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
617 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
618 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
623 * udev gained support for loading additional device properties
624 from an indexed database that is keyed by vendor/product IDs
625 and similar device identifiers. For the beginning this
626 "hwdb" is populated with data from the well-known PCI and
627 USB database, but also includes PNP, ACPI and OID data. In
628 the longer run this indexed database shall grow into
629 becoming the one central database for non-essential
630 userspace device metadata. Previously, data from the PCI/USB
631 database was only attached to select devices, since the
632 lookup was a relatively expensive operation due to O(n) time
633 complexity (with n being the number of entries in the
634 database). Since this is now O(1), we decided to add in this
635 data for all devices where this is available, by
636 default. Note that the indexed database needs to be rebuilt
637 when new data files are installed. To achieve this you need
638 to update your packaging scripts to invoke "udevadm hwdb
639 --update" after installation of hwdb data files. For
640 RPM-based distributions we introduced the new
641 %udev_hwdb_update macro for this purpose.
643 * The Journal gained support for the "Message Catalog", an
644 indexed database to link up additional information with
645 journal entries. For further details please check:
647 http://www.freedesktop.org/wiki/Software/systemd/catalog
649 The indexed message catalog database also needs to be
650 rebuilt after installation of message catalog files. Use
651 "journalctl --update-catalog" for this. For RPM-based
652 distributions we introduced the %journal_catalog_update
653 macro for this purpose.
655 * The Python Journal bindings gained support for the standard
656 Python logging framework.
658 * The Journal API gained new functions for checking whether
659 the underlying file system of a journal file is capable of
660 properly reporting file change notifications, or whether
661 applications that want to reflect journal changes "live"
662 need to recheck journal files continuously in appropriate
665 * It is now possible to set the "age" field for tmpfiles
666 entries to 0, indicating that files matching this entry
667 shall always be removed when the directories are cleaned up.
669 * coredumpctl gained a new "gdb" verb which invokes gdb
670 right-away on the selected coredump.
672 * There's now support for "hybrid sleep" on kernels that
673 support this, in addition to "suspend" and "hibernate". Use
674 "systemctl hybrid-sleep" to make use of this.
676 * logind's HandleSuspendKey= setting (and related settings)
677 now gained support for a new "lock" setting to simply
678 request the screen lock on all local sessions, instead of
679 actually executing a suspend or hibernation.
681 * systemd will now mount the EFI variables file system by
684 * Socket units now gained support for configuration of the
685 SMACK security label.
687 * timedatectl will now output the time of the last and next
688 daylight saving change.
690 * We dropped support for various legacy and distro-specific
691 concepts, such as insserv, early-boot SysV services
692 (i.e. those for non-standard runlevels such as 'b' or 'S')
693 or ArchLinux /etc/rc.conf support. We recommend the
694 distributions who still need support this to either continue
695 to maintain the necessary patches downstream, or find a
696 different solution. (Talk to us if you have questions!)
698 * Various systemd components will now bypass PolicyKit checks
699 for root and otherwise handle properly if PolicyKit is not
700 found to be around. This should fix most issues for
701 PolicyKit-less systems. Quite frankly this should have been
702 this way since day one. It is absolutely our intention to
703 make systemd work fine on PolicyKit-less systems, and we
704 consider it a bug if something doesn't work as it should if
705 PolicyKit is not around.
707 * For embedded systems it is now possible to build udev and
708 systemd without blkid and/or kmod support.
710 * "systemctl switch-root" is now capable of switching root
711 more than once. I.e. in addition to transitions from the
712 initrd to the host OS it is now possible to transition to
713 further OS images from the host. This is useful to implement
714 offline updating tools.
716 * Various other additions have been made to the RPM macros
717 shipped with systemd. Use %udev_rules_update() after
718 installing new udev rules files. %_udevhwdbdir,
719 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
720 %_sysctldir are now available which resolve to the right
721 directories for packages to place various data files in.
723 * journalctl gained the new --full switch (in addition to
724 --all, to disable ellipsation for long messages.
726 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
727 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
728 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
729 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
730 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
731 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
732 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
733 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
734 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
738 * journalctl gained new --since= and --until= switches to
739 filter by time. It also now supports nice filtering for
740 units via --unit=/-u.
742 * Type=oneshot services may use ExecReload= and do the
745 * The journal daemon now supports time-based rotation and
746 vacuuming, in addition to the usual disk-space based
749 * The journal will now index the available field values for
750 each field name. This enables clients to show pretty drop
751 downs of available match values when filtering. The bash
752 completion of journalctl has been updated
753 accordingly. journalctl gained a new switch -F to list all
754 values a certain field takes in the journal database.
756 * More service events are now written as structured messages
757 to the journal, and made recognizable via message IDs.
759 * The timedated, localed and hostnamed mini-services which
760 previously only provided support for changing time, locale
761 and hostname settings from graphical DEs such as GNOME now
762 also have a minimal (but very useful) text-based client
763 utility each. This is probably the nicest way to changing
764 these settings from the command line now, especially since
765 it lists available options and is fully integrated with bash
768 * There's now a new tool "systemd-coredumpctl" to list and
769 extract coredumps from the journal.
771 * We now install a README each in /var/log/ and
772 /etc/rc.d/init.d explaining where the system logs and init
773 scripts went. This hopefully should help folks who go to
774 that dirs and look into the otherwise now empty void and
777 * When user-services are invoked (by systemd --user) the
778 $MANAGERPID env var is set to the PID of systemd.
780 * SIGRTMIN+24 when sent to a --user instance will now result
781 in immediate termination of systemd.
783 * gatewayd received numerous feature additions such as a
784 "follow" mode, for live syncing and filtering.
786 * browse.html now allows filtering and showing detailed
787 information on specific entries. Keyboard navigation and
788 mouse screen support has been added.
790 * gatewayd/journalctl now supports HTML5/JSON
791 Server-Sent-Events as output.
793 * The SysV init script compatibility logic will now
794 heuristically determine whether a script supports the
795 "reload" verb, and only then make this available as
798 * "systemctl status --follow" has been removed, use "journalctl
801 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
802 have been removed since they are hardly useful to be
805 * And I'd like to take the opportunity to specifically mention
806 Zbigniew for his great contributions. Zbigniew, you rock!
808 Contributions from: Andrew Eikum, Christian Hesse, Colin
809 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
810 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
811 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
812 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
813 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
814 Jędrzejewski-Szmek, Сковорода Никита Андреевич
818 * If /etc/vconsole.conf is non-existent or empty we will no
819 longer load any console font or key map at boot by
820 default. Instead the kernel defaults will be left
821 intact. This is definitely the right thing to do, as no
822 configuration should mean no configuration, and hard-coding
823 font names that are different on all archs is probably a bad
824 idea. Also, the kernel default key map and font should be
825 good enough for most cases anyway, and mostly identical to
826 the userspace fonts/key maps we previously overloaded them
827 with. If distributions want to continue to default to a
828 non-kernel font or key map they should ship a default
829 /etc/vconsole.conf with the appropriate contents.
831 Contributions from: Colin Walters, Daniel J Walsh, Dave
832 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
833 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
837 * journalctl gained a new --cursor= switch to show entries
838 starting from the specified location in the journal.
840 * We now enforce a size limit on journal entry fields exported
841 with "-o json" in journalctl. Fields larger than 4K will be
842 assigned null. This can be turned off with --all.
844 * An (optional) journal gateway daemon is now available as
845 "systemd-journal-gatewayd.service". This service provides
846 access to the journal via HTTP and JSON. This functionality
847 will be used to implement live log synchronization in both
848 pull and push modes, but has various other users too, such
849 as easy log access for debugging of embedded devices. Right
850 now it is already useful to retrieve the journal via HTTP:
852 # systemctl start systemd-journal-gatewayd.service
853 # wget http://localhost:19531/entries
855 This will download the journal contents in a
856 /var/log/messages compatible format. The same as JSON:
858 # curl -H"Accept: application/json" http://localhost:19531/entries
860 This service is also accessible via a web browser where a
861 single static HTML5 app is served that uses the JSON logic
862 to enable the user to do some basic browsing of the
863 journal. This will be extended later on. Here's an example
864 screenshot of this app in its current state:
866 http://0pointer.de/public/journal-gatewayd
868 Contributions from: Kay Sievers, Lennart Poettering, Robert
869 Milasan, Tom Gundersen
873 * The bash completion logic is now available for journalctl
876 * We don't mount the "cpuset" controller anymore together with
877 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
878 started if no parameters are assigned to it. "cpuset" hence
879 broke code that assumed it it could create "cpu" groups and
882 * journalctl -f will now subscribe to terminal size changes,
883 and line break accordingly.
885 Contributions from: Dave Reisner, Kay Sievers, Lennart
886 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
890 * nspawn will now create a symlink /etc/localtime in the
891 container environment, copying the host's timezone
892 setting. Previously this has been done via a bind mount, but
893 since symlinks cannot be bind mounted this has now been
894 changed to create/update the appropriate symlink.
896 * journalctl -n's line number argument is now optional, and
897 will default to 10 if omitted.
899 * journald will now log the maximum size the journal files may
900 take up on disk. This is particularly useful if the default
901 built-in logic of determining this parameter from the file
902 system size is used. Use "systemctl status
903 systemd-journald.service" to see this information.
905 * The multi-seat X wrapper tool has been stripped down. As X
906 is now capable of enumerating graphics devices via udev in a
907 seat-aware way the wrapper is not strictly necessary
908 anymore. A stripped down temporary stop-gap is still shipped
909 until the upstream display managers have been updated to
910 fully support the new X logic. Expect this wrapper to be
911 removed entirely in one of the next releases.
913 * HandleSleepKey= in logind.conf has been split up into
914 HandleSuspendKey= and HandleHibernateKey=. The old setting
915 is not available anymore. X11 and the kernel are
916 distuingishing between these keys and we should too. This
917 also means the inhibition lock for these keys has been split
920 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
921 Poettering, Lukas Nykryn, Václav Pavlín
925 * Whenever a unit changes state we'll now log this to the
926 journal and show along the unit's own log output in
929 * ConditionPathIsMountPoint= can now properly detect bind
930 mount points too. (Previously, a bind mount of one file
931 system to another place in the same file system could not be
932 detected as mount, since they shared struct stat's st_dev
935 * We will now mount the cgroup controllers cpu, cpuacct,
936 cpuset and the controllers net_cls, net_prio together by
939 * nspawn containers will now have a virtualized boot
940 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
941 over with a randomized ID at container initialization). This
942 has the effect of making "journalctl -b" do the right thing
945 * The JSON output journal serialization has been updated not
946 to generate "endless" list objects anymore, but rather one
947 JSON object per line. This is more in line how most JSON
948 parsers expect JSON objects. The new output mode
949 "json-pretty" has been added to provide similar output, but
950 neatly aligned for readability by humans.
952 * We dropped all explicit sync() invocations in the shutdown
953 code. The kernel does this implicitly anyway in the kernel
954 reboot() syscall. halt(8)'s -n option is now a compatibility
957 * We now support virtualized reboot() in containers, as
958 supported by newer kernels. We will fall back to exit() if
959 CAP_SYS_REBOOT is not available to the container. Also,
960 nspawn makes use of this now and will actually reboot the
961 container if the containerized OS asks for that.
963 * journalctl will only show local log output by default
964 now. Use --merge (-m) to show remote log output, too.
966 * libsystemd-journal gained the new sd_journal_get_usage()
967 call to determine the current disk usage of all journal
968 files. This is exposed in the new "journalctl --disk-usage"
971 * journald gained a new configuration setting SplitMode= in
972 journald.conf which may be used to control how user journals
973 are split off. See journald.conf(5) for details.
975 * A new condition type ConditionFileNotEmpty= has been added.
977 * tmpfiles' "w" lines now support file globbing, to write
978 multiple files at once.
980 * We added Python bindings for the journal submission
981 APIs. More Python APIs for a number of selected APIs will
982 likely follow. Note that we intend to add native bindings
983 only for the Python language, as we consider it common
984 enough to deserve bindings shipped within systemd. There are
985 various projects outside of systemd that provide bindings
986 for languages such as PHP or Lua.
988 * Many conditions will now resolve specifiers such as %i. In
989 addition, PathChanged= and related directives of .path units
990 now support specifiers as well.
992 * There's now a new RPM macro definition for the system preset
995 * journald will now warn if it can't forward a message to the
996 syslog daemon because it's socket is full.
998 * timedated will no longer write or process /etc/timezone,
999 except on Debian. As we do not support late mounted /usr
1000 anymore /etc/localtime always being a symlink is now safe,
1001 and hence the information in /etc/timezone is not necessary
1004 * logind will now always reserve one VT for a text getty (VT6
1005 by default). Previously if more than 6 X sessions where
1006 started they took up all the VTs with auto-spawned gettys,
1007 so that no text gettys were available anymore.
1009 * udev will now automatically inform the btrfs kernel logic
1010 about btrfs RAID components showing up. This should make
1011 simple hotplug based btrfs RAID assembly work.
1013 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
1014 (but not for its children which will stay at the kernel
1015 default). This should allow setups with a lot more listening
1018 * systemd will now always pass the configured timezone to the
1019 kernel at boot. timedated will do the same when the timezone
1022 * logind's inhibition logic has been updated. By default,
1023 logind will now handle the lid switch, the power and sleep
1024 keys all the time, even in graphical sessions. If DEs want
1025 to handle these events on their own they should take the new
1026 handle-power-key, handle-sleep-key and handle-lid-switch
1027 inhibitors during their runtime. A simple way to achiveve
1028 that is to invoke the DE wrapped in an invocation of:
1030 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
1032 * Access to unit operations is now checked via SELinux taking
1033 the unit file label and client process label into account.
1035 * systemd will now notify the administrator in the journal
1036 when he over-mounts a non-empty directory.
1038 * There are new specifiers that are resolved in unit files,
1039 for the host name (%H), the machine ID (%m) and the boot ID
1042 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
1043 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
1044 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
1045 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
1046 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
1047 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
1048 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
1052 * Support for reading structured kernel messages from
1053 /dev/kmsg has now been added and is enabled by default.
1055 * Support for reading kernel messages from /proc/kmsg has now
1056 been removed. If you want kernel messages in the journal
1057 make sure to run a recent kernel (>= 3.5) that supports
1058 reading structured messages from /dev/kmsg (see
1059 above). /proc/kmsg is now exclusive property of classic
1060 syslog daemons again.
1062 * The libudev API gained the new
1063 udev_device_new_from_device_id() call.
1065 * The logic for file system namespace (ReadOnlyDirectory=,
1066 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
1067 require pivot_root() anymore. This means fewer temporary
1068 directories are created below /tmp for this feature.
1070 * nspawn containers will now see and receive all submounts
1071 made on the host OS below the root file system of the
1074 * Forward Secure Sealing is now supported for Journal files,
1075 which provide cryptographical sealing of journal files so
1076 that attackers cannot alter log history anymore without this
1077 being detectable. Lennart will soon post a blog story about
1078 this explaining it in more detail.
1080 * There are two new service settings RestartPreventExitStatus=
1081 and SuccessExitStatus= which allow configuration of exit
1082 status (exit code or signal) which will be excepted from the
1083 restart logic, resp. consider successful.
1085 * journalctl gained the new --verify switch that can be used
1086 to check the integrity of the structure of journal files and
1087 (if Forward Secure Sealing is enabled) the contents of
1090 * nspawn containers will now be run with /dev/stdin, /dev/fd/
1091 and similar symlinks pre-created. This makes running shells
1092 as container init process a lot more fun.
1094 * The fstab support can now handle PARTUUID= and PARTLABEL=
1097 * A new ConditionHost= condition has been added to match
1098 against the hostname (with globs) and machine ID. This is
1099 useful for clusters where a single OS image is used to
1100 provision a large number of hosts which shall run slightly
1101 different sets of services.
1103 * Services which hit the restart limit will now be placed in a
1106 Contributions from: Bertram Poettering, Dave Reisner, Huang
1107 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
1108 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
1112 * When running in --user mode systemd will now become a
1113 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
1114 tree a lot more organized.
1116 * A new PartOf= unit dependency type has been introduced that
1117 may be used to group services in a natural way.
1119 * "systemctl enable" may now be used to enable instances of
1122 * journalctl now prints error log levels in red, and
1123 warning/notice log levels in bright white. It also supports
1124 filtering by log level now.
1126 * cgtop gained a new -n switch (similar to top), to configure
1127 the maximum number of iterations to run for. It also gained
1128 -b, to run in batch mode (accepting no input).
1130 * The suffix ".service" may now be omitted on most systemctl
1131 command lines involving service unit names.
1133 * There's a new bus call in logind to lock all sessions, as
1134 well as a loginctl verb for it "lock-sessions".
1136 * libsystemd-logind.so gained a new call sd_journal_perror()
1137 that works similar to libc perror() but logs to the journal
1138 and encodes structured information about the error number.
1140 * /etc/crypttab entries now understand the new keyfile-size=
1143 * shutdown(8) now can send a (configurable) wall message when
1144 a shutdown is cancelled.
1146 * The mount propagation mode for the root file system will now
1147 default to "shared", which is useful to make containers work
1148 nicely out-of-the-box so that they receive new mounts from
1149 the host. This can be undone locally by running "mount
1150 --make-rprivate /" if needed.
1152 * The prefdm.service file has been removed. Distributions
1153 should maintain this unit downstream if they intend to keep
1154 it around. However, we recommend writing normal unit files
1155 for display managers instead.
1157 * Since systemd is a crucial part of the OS we will now
1158 default to a number of compiler switches that improve
1159 security (hardening) such as read-only relocations, stack
1160 protection, and suchlike.
1162 * The TimeoutSec= setting for services is now split into
1163 TimeoutStartSec= and TimeoutStopSec= to allow configuration
1164 of individual time outs for the start and the stop phase of
1167 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
1168 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
1169 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
1170 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
1171 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
1172 Gundersen, Zbigniew Jędrzejewski-Szmek
1176 * The journal and id128 C APIs are now fully documented as man
1179 * Extra safety checks have been added when transitioning from
1180 the initial RAM disk to the main system to avoid accidental
1183 * /etc/crypttab entries now understand the new keyfile-offset=
1186 * systemctl -t can now be used to filter by unit load state.
1188 * The journal C API gained the new sd_journal_wait() call to
1189 make writing synchronous journal clients easier.
1191 * journalctl gained the new -D switch to show journals from a
1194 * journalctl now displays a special marker between log
1195 messages of two different boots.
1197 * The journal is now explicitly flushed to /var via a service
1198 systemd-journal-flush.service, rather than implicitly simply
1199 by seeing /var/log/journal to be writable.
1201 * journalctl (and the journal C APIs) can now match for much
1202 more complex expressions, with alternatives and
1205 * When transitioning from the initial RAM disk to the main
1206 system we will now kill all processes in a killing spree to
1207 ensure no processes stay around by accident.
1209 * Three new specifiers may be used in unit files: %u, %h, %s
1210 resolve to the user name, user home directory resp. user
1211 shell. This is useful for running systemd user instances.
1213 * We now automatically rotate journal files if their data
1214 object hash table gets a fill level > 75%. We also size the
1215 hash table based on the configured maximum file size. This
1216 together should lower hash collisions drastically and thus
1217 speed things up a bit.
1219 * journalctl gained the new "--header" switch to introspect
1220 header data of journal files.
1222 * A new setting SystemCallFilters= has been added to services
1223 which may be used to apply blacklists or whitelists to
1224 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
1226 * nspawn gained a new --link-journal= switch (and quicker: -j)
1227 to link the container journal with the host. This makes it
1228 very easy to centralize log viewing on the host for all
1229 guests while still keeping the journal files separated.
1231 * Many bugfixes and optimizations
1233 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
1234 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
1235 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
1240 * Several tools now understand kernel command line arguments,
1241 which are only read when run in an initial RAM disk. They
1242 usually follow closely their normal counterparts, but are
1245 * There's a new tool to analyze the readahead files that are
1246 automatically generated at boot. Use:
1248 /usr/lib/systemd/systemd-readahead analyze /.readahead
1250 * We now provide an early debug shell on tty9 if this enabled. Use:
1252 systemctl enable debug-shell.service
1254 * All plymouth related units have been moved into the Plymouth
1255 package. Please make sure to upgrade your Plymouth version
1258 * systemd-tmpfiles now supports getting passed the basename of
1259 a configuration file only, in which case it will look for it
1260 in all appropriate directories automatically.
1262 * udevadm info now takes a /dev or /sys path as argument, and
1263 does the right thing. Example:
1265 udevadm info /dev/sda
1266 udevadm info /sys/class/block/sda
1268 * systemctl now prints a warning if a unit is stopped but a
1269 unit that might trigger it continues to run. Example: a
1270 service is stopped but the socket that activates it is left
1273 * "systemctl status" will now mention if the log output was
1274 shortened due to rotation since a service has been started.
1276 * The journal API now exposes functions to determine the
1277 "cutoff" times due to rotation.
1279 * journald now understands SIGUSR1 and SIGUSR2 for triggering
1280 immediately flushing of runtime logs to /var if possible,
1281 resp. for triggering immediate rotation of the journal
1284 * It is now considered an error if a service is attempted to
1285 be stopped that is not loaded.
1287 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
1289 * systemd-analyze now supports Python 3
1291 * tmpfiles now supports cleaning up directories via aging
1292 where the first level dirs are always kept around but
1293 directories beneath it automatically aged. This is enabled
1294 by prefixing the age field with '~'.
1296 * Seat objects now expose CanGraphical, CanTTY properties
1297 which is required to deal with very fast bootups where the
1298 display manager might be running before the graphics drivers
1299 completed initialization.
1301 * Seat objects now expose a State property.
1303 * We now include RPM macros for service enabling/disabling
1304 based on the preset logic. We recommend RPM based
1305 distributions to make use of these macros if possible. This
1306 makes it simpler to reuse RPM spec files across
1309 * We now make sure that the collected systemd unit name is
1310 always valid when services log to the journal via
1313 * There's a new man page kernel-command-line(7) detailing all
1314 command line options we understand.
1316 * The fstab generator may now be disabled at boot by passing
1317 fstab=0 on the kernel command line.
1319 * A new kernel command line option modules-load= is now understood
1320 to load a specific kernel module statically, early at boot.
1322 * Unit names specified on the systemctl command line are now
1323 automatically escaped as needed. Also, if file system or
1324 device paths are specified they are automatically turned
1325 into the appropriate mount or device unit names. Example:
1327 systemctl status /home
1328 systemctl status /dev/sda
1330 * The SysVConsole= configuration option has been removed from
1331 system.conf parsing.
1333 * The SysV search path is no longer exported on the D-Bus
1336 * The Names= option is been removed from unit file parsing.
1338 * There's a new man page bootup(7) detailing the boot process.
1340 * Every unit and every generator we ship with systemd now
1341 comes with full documentation. The self-explanatory boot is
1344 * A couple of services gained "systemd-" prefixes in their
1345 name if they wrap systemd code, rather than only external
1346 code. Among them fsck@.service which is now
1347 systemd-fsck@.service.
1349 * The HaveWatchdog property has been removed from the D-Bus
1352 * systemd.confirm_spawn= on the kernel command line should now
1355 * There's a new man page crypttab(5) which details all options
1356 we actually understand.
1358 * systemd-nspawn gained a new --capability= switch to pass
1359 additional capabilities to the container.
1361 * timedated will now read known NTP implementation unit names
1362 from /usr/lib/systemd/ntp-units.d/*.list,
1363 systemd-timedated-ntp.target has been removed.
1365 * journalctl gained a new switch "-b" that lists log data of
1366 the current boot only.
1368 * The notify socket is in the abstract namespace again, in
1369 order to support daemons which chroot() at start-up.
1371 * There is a new Storage= configuration option for journald
1372 which allows configuration of where log data should go. This
1373 also provides a way to disable journal logging entirely, so
1374 that data collected is only forwarded to the console, the
1375 kernel log buffer or another syslog implementation.
1377 * Many bugfixes and optimizations
1379 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
1380 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
1381 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
1382 Shawn Landden, Tom Gundersen
1386 * "systemctl help <unit>" now shows the man page if one is
1389 * Several new man pages have been added.
1391 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
1392 MaxLevelConsole= can now be specified in
1393 journald.conf. These options allow reducing the amount of
1394 data stored on disk or forwarded by the log level.
1396 * TimerSlackNSec= can now be specified in system.conf for
1397 PID1. This allows system-wide power savings.
1399 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
1400 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
1405 * logind is now capable of (optionally) handling power and
1406 sleep keys as well as the lid switch.
1408 * journalctl now understands the syntax "journalctl
1409 /usr/bin/avahi-daemon" to get all log output of a specific
1412 * CapabilityBoundingSet= in system.conf now also influences
1413 the capability bound set of usermode helpers of the kernel.
1415 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
1416 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
1417 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
1418 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
1422 * Note that we skipped 139 releases here in order to set the
1423 new version to something that is greater than both udev's
1424 and systemd's most recent version number.
1426 * udev: all udev sources are merged into the systemd source tree now.
1427 All future udev development will happen in the systemd tree. It
1428 is still fully supported to use the udev daemon and tools without
1429 systemd running, like in initramfs or other init systems. Building
1430 udev though, will require the *build* of the systemd tree, but
1431 udev can be properly *run* without systemd.
1433 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
1434 should be used to create dead device nodes as workarounds for broken
1437 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
1438 no longer supported. udev_monitor_new_from_netlink() needs to be
1439 used to subscribe to events.
1441 * udev: when udevd is started by systemd, processes which are left
1442 behind by forking them off of udev rules, are unconditionally cleaned
1443 up and killed now after the event handling has finished. Services or
1444 daemons must be started as systemd services. Services can be
1445 pulled-in by udev to get started, but they can no longer be directly
1446 forked by udev rules.
1448 * udev: the daemon binary is called systemd-udevd now and installed
1449 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
1450 to adapt to that, create symlink, or rename the binary after building
1453 * libudev no longer provides these symbols:
1454 udev_monitor_from_socket()
1455 udev_queue_get_failed_list_entry()
1456 udev_get_{dev,sys,run}_path()
1457 The versions number was bumped and symbol versioning introduced.
1459 * systemd-loginctl and systemd-journalctl have been renamed
1460 to loginctl and journalctl to match systemctl.
1462 * The config files: /etc/systemd/systemd-logind.conf and
1463 /etc/systemd/systemd-journald.conf have been renamed to
1464 logind.conf and journald.conf. Package updates should rename
1465 the files to the new names on upgrade.
1467 * For almost all files the license is now LGPL2.1+, changed
1468 from the previous GPL2.0+. Exceptions are some minor stuff
1469 of udev (which will be changed to LGPL2.1 eventually, too),
1470 and the MIT licensed sd-daemon.[ch] library that is suitable
1471 to be used as drop-in files.
1473 * systemd and logind now handle system sleep states, in
1474 particular suspending and hibernating.
1476 * logind now implements a sleep/shutdown/idle inhibiting logic
1477 suitable for a variety of uses. Soonishly Lennart will blog
1478 about this in more detail.
1480 * var-run.mount and var-lock.mount are no longer provided
1481 (which prevously bind mounted these directories to their new
1482 places). Distributions which have not converted these
1483 directories to symlinks should consider stealing these files
1484 from git history and add them downstream.
1486 * We introduced the Documentation= field for units and added
1487 this to all our shipped units. This is useful to make it
1488 easier to explore the boot and the purpose of the various
1491 * All smaller setup units (such as
1492 systemd-vconsole-setup.service) now detect properly if they
1493 are run in a container and are skipped when
1494 appropriate. This guarantees an entirely noise-free boot in
1495 Linux container environments such as systemd-nspawn.
1497 * A framework for implementing offline system updates is now
1498 integrated, for details see:
1499 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
1501 * A new service type Type=idle is available now which helps us
1502 avoiding ugly interleaving of getty output and boot status
1505 * There's now a system-wide CapabilityBoundingSet= option to
1506 globally reduce the set of capabilities for the
1507 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
1508 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
1509 even CAP_NET_ADMIN system-wide for secure systems.
1511 * There are now system-wide DefaultLimitXXX= options to
1512 globally change the defaults of the various resource limits
1513 for all units started by PID 1.
1515 * Harald Hoyer's systemd test suite has been integrated into
1516 systemd which allows easy testing of systemd builds in qemu
1517 and nspawn. (This is really awesome! Ask us for details!)
1519 * The fstab parser is now implemented as generator, not inside
1522 * systemctl will now warn you if .mount units generated from
1523 /etc/fstab are out of date due to changes in fstab that
1524 haven't been read by systemd yet.
1526 * systemd is now suitable for usage in initrds. Dracut has
1527 already been updated to make use of this. With this in place
1528 initrds get a slight bit faster but primarily are much
1529 easier to introspect and debug since "systemctl status" in
1530 the host system can be used to introspect initrd services,
1531 and the journal from the initrd is kept around too.
1533 * systemd-delta has been added, a tool to explore differences
1534 between user/admin configuration and vendor defaults.
1536 * PrivateTmp= now affects both /tmp and /var/tmp.
1538 * Boot time status messages are now much prettier and feature
1539 proper english language. Booting up systemd has never been
1542 * Read-ahead pack files now include the inode number of all
1543 files to pre-cache. When the inode changes the pre-caching
1544 is not attempted. This should be nicer to deal with updated
1545 packages which might result in changes of read-ahead
1548 * We now temporaritly lower the kernel's read_ahead_kb variable
1549 when collecting read-ahead data to ensure the kernel's
1550 built-in read-ahead does not add noise to our measurements
1551 of necessary blocks to pre-cache.
1553 * There's now RequiresMountsFor= to add automatic dependencies
1554 for all mounts necessary for a specific file system path.
1556 * MountAuto= and SwapAuto= have been removed from
1557 system.conf. Mounting file systems at boot has to take place
1560 * nspawn now learned a new switch --uuid= to set the machine
1561 ID on the command line.
1563 * nspawn now learned the -b switch to automatically search
1566 * vt102 is now the default TERM for serial TTYs, upgraded from
1569 * systemd-logind now works on VT-less systems.
1571 * The build tree has been reorganized. The individual
1572 components now have directories of their own.
1574 * A new condition type ConditionPathIsReadWrite= is now available.
1576 * nspawn learned the new -C switch to create cgroups for the
1577 container in other hierarchies.
1579 * We now have support for hardware watchdogs, configurable in
1582 * The scheduled shutdown logic now has a public API.
1584 * We now mount /tmp as tmpfs by default, but this can be
1585 masked and /etc/fstab can override it.
1587 * Since udisks doesn't make use of /media anymore we are not
1588 mounting a tmpfs on it anymore.
1590 * journalctl gained a new --local switch to only interleave
1591 locally generated journal files.
1593 * We can now load the IMA policy at boot automatically.
1595 * The GTK tools have been split off into a systemd-ui.
1597 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
1598 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
1599 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
1600 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
1601 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
1602 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
1603 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
1604 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
1605 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
1610 * This is mostly a bugfix release
1612 * Support optional initialization of the machine ID from the
1613 KVM or container configured UUID.
1615 * Support immediate reboots with "systemctl reboot -ff"
1617 * Show /etc/os-release data in systemd-analyze output
1619 * Many bugfixes for the journal, including endianness fixes and
1620 ensuring that disk space enforcement works
1622 * sd-login.h is C++ comptaible again
1624 * Extend the /etc/os-release format on request of the Debian
1627 * We now refuse non-UTF8 strings used in various configuration
1628 and unit files. This is done to ensure we don't pass invalid
1629 data over D-Bus or expose it elsewhere.
1631 * Register Mimo USB Screens as suitable for automatic seat
1634 * Read SELinux client context from journal clients in a race
1637 * Reorder configuration file lookup order. /etc now always
1638 overrides /run in order to allow the administrator to always
1639 and unconditionally override vendor supplied or
1640 automatically generated data.
1642 * The various user visible bits of the journal now have man
1643 pages. We still lack man pages for the journal API calls
1646 * We now ship all man pages in HTML format again in the
1649 Contributions from: Dave Reisner, Dirk Eibach, Frederic
1650 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
1651 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
1656 * This is mostly a bugfix release
1658 * systems lacking /etc/os-release are no longer supported.
1660 * Various functionality updates to libsystemd-login.so
1662 * Track class of PAM logins to distuingish greeters from
1665 Contributions from: Kay Sievers, Lennart Poettering, Michael
1670 * This is an important bugfix release for v41.
1672 * Building man pages is now optional which should be useful
1673 for those building systemd from git but unwilling to install
1676 * Watchdog support for supervising services is now usable. In
1677 a future release support for hardware watchdogs
1678 (i.e. /dev/watchdog) will be added building on this.
1680 * Service start rate limiting is now configurable and can be
1681 turned off per service. When a start rate limit is hit a
1682 reboot can automatically be triggered.
1684 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
1686 Contributions from: Benjamin Franzke, Bill Nottingham,
1687 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
1688 Schmidt, Michał Górny, Piotr Drąg
1692 * The systemd binary is installed /usr/lib/systemd/systemd now;
1693 An existing /sbin/init symlink needs to be adapted with the
1696 * The code that loads kernel modules has been ported to invoke
1697 libkmod directly, instead of modprobe. This means we do not
1698 support systems with module-init-tools anymore.
1700 * Watchdog support is now already useful, but still not
1703 * A new kernel command line option systemd.setenv= is
1704 understood to set system wide environment variables
1705 dynamically at boot.
1707 * We now limit the set of capabilities of systemd-journald.
1709 * We now set SIGPIPE to ignore by default, since it only is
1710 useful in shell pipelines, and has little use in general
1711 code. This can be disabled with IgnoreSIPIPE=no in unit
1714 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
1715 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
1720 * This is mostly a bugfix release
1722 * We now expose the reason why a service failed in the
1723 "Result" D-Bus property.
1725 * Rudimentary service watchdog support (will be completed over
1726 the next few releases.)
1728 * When systemd forks off in order execute some service we will
1729 now immediately changes its argv[0] to reflect which process
1730 it will execute. This is useful to minimize the time window
1731 with a generic argv[0], which makes bootcharts more useful
1733 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
1734 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
1735 Mike Kazantsev, Ray Strode
1739 * This is mostly a test release, but incorporates many
1742 * New systemd-cgtop tool to show control groups by their
1745 * Linking against libacl for ACLs is optional again. If
1746 disabled, support tracking device access for active logins
1747 goes becomes unavailable, and so does access to the user
1748 journals by the respective users.
1750 * If a group "adm" exists, journal files are automatically
1751 owned by them, thus allow members of this group full access
1752 to the system journal as well as all user journals.
1754 * The journal now stores the SELinux context of the logging
1755 client for all entries.
1757 * Add C++ inclusion guards to all public headers
1759 * New output mode "cat" in the journal to print only text
1760 messages, without any meta data like date or time.
1762 * Include tiny X server wrapper as a temporary stop-gap to
1763 teach XOrg udev display enumeration. This is used by display
1764 managers such as gdm, and will go away as soon as XOrg
1765 learned native udev hotplugging for display devices.
1767 * Add new systemd-cat tool for executing arbitrary programs
1768 with STDERR/STDOUT connected to the journal. Can also act as
1769 BSD logger replacement, and does so by default.
1771 * Optionally store all locally generated coredumps in the
1772 journal along with meta data.
1774 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
1775 writing short strings to files (for usage for /sys), and for
1776 creating symlinks, character and block device nodes.
1778 * New unit file option ControlGroupPersistent= to make cgroups
1779 persistent, following the mechanisms outlined in
1780 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
1782 * Support multiple local RTCs in a sane way
1784 * No longer monopolize IO when replaying readahead data on
1785 rotating disks, since we might starve non-file-system IO to
1786 death, since fanotify() will not see accesses done by blkid,
1789 * Don't show kernel threads in systemd-cgls anymore, unless
1790 requested with new -k switch.
1792 Contributions from: Dan Horák, Kay Sievers, Lennart
1793 Poettering, Michal Schmidt
1797 * This is mostly a test release, but incorporates many
1800 * The git repository moved to:
1801 git://anongit.freedesktop.org/systemd/systemd
1802 ssh://git.freedesktop.org/git/systemd/systemd
1804 * First release with the journal
1805 http://0pointer.de/blog/projects/the-journal.html
1807 * The journal replaces both systemd-kmsg-syslogd and
1808 systemd-stdout-bridge.
1810 * New sd_pid_get_unit() API call in libsystemd-logind
1812 * Many systemadm clean-ups
1814 * Introduce remote-fs-pre.target which is ordered before all
1815 remote mounts and may be used to start services before all
1818 * Added Mageia support
1820 * Add bash completion for systemd-loginctl
1822 * Actively monitor PID file creation for daemons which exit in
1823 the parent process before having finished writing the PID
1824 file in the daemon process. Daemons which do this need to be
1825 fixed (i.e. PID file creation must have finished before the
1826 parent exits), but we now react a bit more gracefully to them.
1828 * Add colourful boot output, mimicking the well-known output
1829 of existing distributions.
1831 * New option PassCredentials= for socket units, for
1832 compatibility with a recent kernel ABI breakage.
1834 * /etc/rc.local is now hooked in via a generator binary, and
1835 thus will no longer act as synchronization point during
1838 * systemctl list-unit-files now supports --root=.
1840 * systemd-tmpfiles now understands two new commands: z, Z for
1841 relabelling files according to the SELinux database. This is
1842 useful to apply SELinux labels to specific files in /sys,
1845 * Output of SysV services is now forwarded to both the console
1846 and the journal by default, not only just the console.
1848 * New man pages for all APIs from libsystemd-login.
1850 * The build tree got reorganized and a the build system is a
1851 lot more modular allowing embedded setups to specifically
1852 select the components of systemd they are interested in.
1854 * Support for Linux systems lacking the kernel VT subsystem is
1857 * configure's --with-rootdir= got renamed to
1858 --with-rootprefix= to follow the naming used by udev and
1861 * Unless specified otherwise we'll now install to /usr instead
1862 of /usr/local by default.
1864 * Processes with '@' in argv[0][0] are now excluded from the
1865 final shut-down killing spree, following the logic explained
1867 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
1869 * All processes remaining in a service cgroup when we enter
1870 the START or START_PRE states are now killed with
1871 SIGKILL. That means it is no longer possible to spawn
1872 background processes from ExecStart= lines (which was never
1873 supported anyway, and bad style).
1875 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
1876 reloading of units together.
1878 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
1879 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
1880 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
1881 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
1882 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek