1 systemd System and Service Manager
5 * systemd-python gained an API exposing libsystemd-daemon.
7 * The SMACK setup logic gained support for uploading CIPSO
10 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
11 ReadOnlyDirectories= and InaccessibleDirectories= has
12 changed. The private /tmp and /var/tmp directories are now
13 shared by all processes of a service (which means
14 ExecStartPre= may now leave data in /tmp that ExecStart= of
15 the same service can still access). When a service is
16 stopped its temporary directories are immediately deleted
17 (normal clean-up with tmpfiles still is done in addition to
20 * By default, systemd will now set a couple of sysctl
21 variables in the kernel: the safe sysrq options are turned
22 on, IP route verification is turned on, and source routing
23 disabled. The recently added hardlink and softlink
24 protection of the kernel is turned on. These settings should
25 be reasonably safe, and good defaults for all new systems.
27 * The predictable network naming logic may now be turned off
28 with a new kernel command line switch: net.ifnames=0
30 * A new libsystemd-bus module has been added that implements a
31 pretty complete D-Bus client library. For details see:
33 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
35 * journald will not explicitly flush the journal files to disk
36 the latest 5min after each write. This will also mark the
37 files as offline then until the next read. This should
38 increase reliability. The synchronization delay can be
39 configured via SyncIntervalSec= in journald.conf.
41 * There's a new remote-fs-setup.target unit that can be used
42 to pull in specific services when at least one remote file
43 system is to be mounted.
45 * There are new targets timers.target and paths.target as
46 canonical targets to pull user timer and path units in
47 from. This complements sockets.target with a similar
48 purpose for socket units.
50 Contributions from: Auke Kok, Colin Walters, Cristian
51 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
52 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
53 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
54 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
55 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
56 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
57 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
58 Zbigniew Jędrzejewski-Szmek
62 * Configuration of unit files may now be extended via drop-in
63 files without having to edit/override the unit files
64 themselves. More specifically, if the administrator wants to
65 change one value for a service file foobar.service he can
66 now do so by dropping in a configuration snippet into
67 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
68 will load all these snippets and apply them on top of the
69 main unit configuration file, possibly extending or
70 overriding its settings. Using these drop-in snippets is
71 generally nicer than the two earlier options for changing
72 unit files locally: copying the files from
73 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
74 them there; or creating a new file in /etc/systemd/system/
75 that incorporates the original one via ".include". Drop-in
76 snippets into these .d/ directories can be placed in any
77 directory systemd looks for units in, and the usual
78 overriding semantics between /usr/lib, /etc and /run apply
81 * Most unit file settings which take lists of items can now be
82 reset by assigning the empty string to them. For example,
83 normally, settings such as Environment=FOO=BAR append a new
84 environment variable assignment to the environment block,
85 each time they are used. By assigning Environment= the empty
86 string the environment block can be reset to empty. This is
87 particularly useful with the .d/*.conf drop-in snippets
88 mentioned above, since this adds the ability to reset list
89 settings from vendor unit files via these drop-ins.
91 * systemctl gained a new "list-dependencies" command for
92 listing the dependencies of a unit recursively.
94 * Inhibitors are now honored and listed by "systemctl
95 suspend", "systemctl poweroff" (and similar) too, not only
96 GNOME. These commands will also list active sessions by
99 * Resource limits (as exposed by the various control group
100 controllers) can now be controlled dynamically at runtime
101 for all units. More specifically, you can now use a command
102 like "systemctl set-cgroup-attr foobar.service cpu.shares
103 2000" to alter the CPU shares a specific service gets. These
104 settings are stored persistently on disk, and thus allow the
105 administrator to easily adjust the resource usage of
106 services with a few simple commands. This dynamic resource
107 management logic is also available to other programs via the
108 bus. Almost any kernel cgroup attribute and controller is
111 * systemd-vconsole-setup will now copy all font settings to
112 all allocated VTs, where it previously applied them only to
115 * libsystemd-login gained the new sd_session_get_tty() API
118 * This release drops support for a few legacy or
119 distribution-specific LSB facility names when parsing init
120 scripts: $x-display-manager, $mail-transfer-agent,
121 $mail-transport-agent, $mail-transfer-agent, $smtp,
122 $null. Also, the mail-transfer-agent.target unit backing
123 this has been removed. Distributions which want to retain
124 compatibility with this should carry the burden for
125 supporting this themselves and patch support for these back
126 in, if they really need to. Also, the facilities $syslog and
127 $local_fs are now ignored, since systemd does not support
128 early-boot LSB init scripts anymore, and these facilities
129 are implied anyway for normal services. syslog.target has
132 * There are new bus calls on PID1's Manager object for
133 cancelling jobs, and removing snapshot units. Previously,
134 both calls were only available on the Job and Snapshot
137 * systemd-journal-gatewayd gained SSL support.
139 * The various "environment" files, such as /etc/locale.conf
140 now support continuation lines with a backslash ("\") as
141 last character in the line, similar in style (but different)
142 to how this is supported in shells.
144 * For normal user processes the _SYSTEMD_USER_UNIT= field is
145 now implicitly appended to every log entry logged. systemctl
146 has been updated to filter by this field when operating on a
147 user systemd instance.
149 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
150 CAP_AUDIT_CONTROL capabilities to the capabilities set for
151 the container. This makes it easier to boot unmodified
152 Fedora systems in a container, which however still requires
153 audit=0 to be passed on the kernel command line. Auditing in
154 kernel and userspace is unfortunately still too broken in
155 context of containers, hence we recommend compiling it out
156 of the kernel or using audit=0. Hopefully this will be fixed
157 one day for good in the kernel.
159 * nspawn gained the new --bind= and --bind-ro= parameters to
160 bind mount specific directories from the host into the
163 * nspawn will now mount its own devpts file system instance
164 into the container, in order not to leak pty devices from
165 the host into the container.
167 * systemd will now read the firmware boot time performance
168 information from the EFI variables, if the used boot loader
169 supports this, and takes it into account for boot performance
170 analysis via "systemd-analyze". This is currently supported
171 only in conjunction with Gummiboot, but could be supported
172 by other boot loaders too. For details see:
174 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
176 * A new generator has been added that automatically mounts the
177 EFI System Partition (ESP) to /boot, if that directory
178 exists, is empty, and no other file system has been
179 configured to be mounted there.
181 * logind will now send out PrepareForSleep(false) out
182 unconditionally, after coming back from suspend. This may be
183 used by applications as asynchronous notification for
184 system resume events.
186 * "systemctl unlock-sessions" has been added, that allows
187 unlocking the screens of all user sessions at once, similar
188 how "systemctl lock-sessions" already locked all users
189 sessions. This is backed by a new D-Bus call UnlockSessions().
191 * "loginctl seat-status" will now show the master device of a
192 seat. (i.e. the device of a seat that needs to be around for
193 the seat to be considered available, usually the graphics
196 * tmpfiles gained a new "X" line type, that allows
197 configuration of files and directories (with wildcards) that
198 shall be excluded from automatic cleanup ("aging").
200 * udev default rules set the device node permissions now only
201 at "add" events, and do not change them any longer with a
202 later "change" event.
204 * The log messages for lid events and power/sleep keypresses
205 now carry a message ID.
207 * We now have a substantially larger unit test suite, but this
208 continues to be work in progress.
210 * udevadm hwdb gained a new --root= parameter to change the
211 root directory to operate relative to.
213 * logind will now issue a background sync() request to the kernel
214 early at shutdown, so that dirty buffers are flushed to disk early
215 instead of at the last moment, in order to optimize shutdown
218 * A new bootctl tool has been added that is an interface for
219 certain boot loader operations. This is currently a preview
220 and is likely to be extended into a small mechanism daemon
221 like timedated, localed, hostnamed, and can be used by
222 graphical UIs to enumerate available boot options, and
223 request boot into firmware operations.
225 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
226 the rest of the package. It also has been updated to work
227 correctly in initrds.
229 * Policykit previously has been runtime optional, and is now
230 also compile time optional via a configure switch.
232 * systemd-analyze has been reimplemented in C. Also "systemctl
233 dot" has moved into systemd-analyze.
235 * "systemctl status" with no further parameters will now print
236 the status of all active or failed units.
238 * Operations such as "systemctl start" can now be executed
239 with a new mode "--irreversible" which may be used to queue
240 operations that cannot accidentally be reversed by a later
241 job queuing. This is by default used to make shutdown
242 requests more robust.
244 * The Python API of systemd now gained a new module for
245 reading journal files.
247 * A new tool kernel-install has been added that can install
248 kernel images according to the Boot Loader Specification:
250 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
252 * Boot time console output has been improved to provide
253 animated boot time output for hanging jobs.
255 * A new tool systemd-activate has been added which can be used
256 to test socket activation with, directly from the command
257 line. This should make it much easier to test and debug
258 socket activation in daemons.
260 * journalctl gained a new "--reverse" (or -r) option to show
261 journal output in reverse order (i.e. newest line first).
263 * journalctl gained a new "--pager-end" (or -e) option to jump
264 to immediately jump to the end of the journal in the
265 pager. This is only supported in conjunction with "less".
267 * journalctl gained a new "--user-unit=" option, that works
268 similar to "--unit=" but filters for user units rather than
271 * A number of unit files to ease adoption of systemd in
272 initrds has been added. This moves some minimal logic from
273 the various initrd implementations into systemd proper.
275 * The journal files are now owned by a new group
276 "systemd-journal", which exists specifically to allow access
277 to the journal, and nothing else. Previously, we used the
278 "adm" group for that, which however possibly covers more
279 than just journal/log file access. This new group is now
280 already used by systemd-journal-gatewayd to ensure this
281 daemon gets access to the journal files and as little else
282 as possible. Note that "make install" will also set FS ACLs
283 up for /var/log/journal to give "adm" and "wheel" read
284 access to it, in addition to "systemd-journal" which owns
285 the journal files. We recommend that packaging scripts also
286 add read access to "adm" + "wheel" to /var/log/journal, and
287 all existing/future journal files. To normal users and
288 administrators little changes, however packagers need to
289 ensure to create the "systemd-journal" system group at
290 package installation time.
292 * The systemd-journal-gatewayd now runs as unprivileged user
293 systemd-journal-gateway:systemd-journal-gateway. Packaging
294 scripts need to create these system user/group at
297 * timedated now exposes a new boolean property CanNTP that
298 indicates whether a local NTP service is available or not.
300 * systemd-detect-virt will now also detect xen PVs
302 * The pstore file system is now mounted by default, if it is
305 * In addition to the SELinux and IMA policies we will now also
306 load SMACK policies at early boot.
308 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
309 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
310 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
311 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
312 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
313 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
314 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
315 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
316 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
317 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
318 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
319 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
320 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
321 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
325 * Timer units now support calendar time events in addition to
326 monotonic time events. That means you can now trigger a unit
327 based on a calendar time specification such as "Thu,Fri
328 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
329 or fifth day of any month of the year 2013, given that it is
330 a thursday or friday. This brings timer event support
331 considerably closer to cron's capabilities. For details on
332 the supported calendar time specification language see
335 * udev now supports a number of different naming policies for
336 network interfaces for predictable names, and a combination
337 of these policies is now the default. Please see this wiki
338 document for details:
340 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
342 * Auke Kok's bootchart implementation has been added to the
343 systemd tree. It's an optional component that can graph the
344 boot in quite some detail. It's one of the best bootchart
345 implementations around and minimal in its code and
348 * nss-myhostname has been integrated into the systemd source
349 tree. nss-myhostname guarantees that the local hostname
350 always stays resolvable via NSS. It has been a weak
351 requirement of systemd-hostnamed since a long time, and
352 since its code is actually trivial we decided to just
353 include it in systemd's source tree. It can be turned off
354 with a configure switch.
356 * The read-ahead logic is now capable of properly detecting
357 whether a btrfs file system is on SSD or rotating media, in
358 order to optimize the read-ahead scheme. Previously, it was
359 only capable of detecting this on traditional file systems
362 * In udev, additional device properties are now read from the
363 IAB in addition to the OUI database. Also, Bluetooth company
364 identities are attached to the devices as well.
366 * In service files %U may be used as specifier that is
367 replaced by the configured user name of the service.
369 * nspawn may now be invoked without a controlling TTY. This
370 makes it suitable for invocation as its own service. This
371 may be used to set up a simple containerized server system
372 using only core OS tools.
374 * systemd and nspawn can now accept socket file descriptors
375 when they are started for socket activation. This enables
376 implementation of socket activated nspawn
377 containers. i.e. think about autospawning an entire OS image
378 when the first SSH or HTTP connection is received. We expect
379 that similar functionality will also be added to libvirt-lxc
382 * journalctl will now suppress ANSI color codes when
385 * systemctl will no longer show control group information for
386 a unit if a the control group is empty anyway.
388 * logind can now automatically suspend/hibernate/shutdown the
391 * /etc/machine-info and hostnamed now also expose the chassis
392 type of the system. This can be used to determine whether
393 the local system is a laptop, desktop, handset or
394 tablet. This information may either be configured by the
395 user/vendor or is automatically determined from ACPI and DMI
396 information if possible.
398 * A number of PolicyKit actions are now bound together with
399 "imply" rules. This should simplify creating UIs because
400 many actions will now authenticate similar ones as well.
402 * Unit files learnt a new condition ConditionACPower= which
403 may be used to conditionalize a unit depending on whether an
404 AC power source is connected or not, of whether the system
405 is running on battery power.
407 * systemctl gained a new "is-failed" verb that may be used in
408 shell scripts and suchlike to check whether a specific unit
409 is in the "failed" state.
411 * The EnvironmentFile= setting in unit files now supports file
412 globbing, and can hence be used to easily read a number of
413 environment files at once.
415 * systemd will no longer detect and recognize specific
416 distributions. All distribution-specific #ifdeffery has been
417 removed, systemd is now fully generic and
418 distribution-agnostic. Effectively, not too much is lost as
419 a lot of the code is still accessible via explicit configure
420 switches. However, support for some distribution specific
421 legacy configuration file formats has been dropped. We
422 recommend distributions to simply adopt the configuration
423 files everybody else uses now and convert the old
424 configuration from packaging scripts. Most distributions
425 already did that. If that's not possible or desirable,
426 distributions are welcome to forward port the specific
427 pieces of code locally from the git history.
429 * When logging a message about a unit systemd will now always
430 log the unit name in the message meta data.
432 * localectl will now also discover system locale data that is
433 not stored in locale archives, but directly unpacked.
435 * logind will no longer unconditionally use framebuffer
436 devices as seat masters, i.e. as devices that are required
437 to be existing before a seat is considered preset. Instead,
438 it will now look for all devices that are tagged as
439 "seat-master" in udev. By default framebuffer devices will
440 be marked as such, but depending on local systems other
441 devices might be marked as well. This may be used to
442 integrate graphics cards using closed source drivers (such
443 as NVidia ones) more nicely into logind. Note however, that
444 we recommend using the open source NVidia drivers instead,
445 and no udev rules for the closed-source drivers will be
446 shipped from us upstream.
448 Contributions from: Adam Williamson, Alessandro Crismani, Auke
449 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
450 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
451 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
452 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
453 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
454 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
455 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
456 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
457 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
458 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
459 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
464 * udev gained support for loading additional device properties
465 from an indexed database that is keyed by vendor/product IDs
466 and similar device identifiers. For the beginning this
467 "hwdb" is populated with data from the well-known PCI and
468 USB database, but also includes PNP, ACPI and OID data. In
469 the longer run this indexed database shall grow into
470 becoming the one central database for non-essential
471 userspace device metadata. Previously, data from the PCI/USB
472 database was only attached to select devices, since the
473 lookup was a relatively expensive operation due to O(n) time
474 complexity (with n being the number of entries in the
475 database). Since this is now O(1), we decided to add in this
476 data for all devices where this is available, by
477 default. Note that the indexed database needs to be rebuilt
478 when new data files are installed. To achieve this you need
479 to update your packaging scripts to invoke "udevadm hwdb
480 --update" after installation of hwdb data files. For
481 RPM-based distributions we introduced the new
482 %udev_hwdb_update macro for this purpose.
484 * The Journal gained support for the "Message Catalog", an
485 indexed database to link up additional information with
486 journal entries. For further details please check:
488 http://www.freedesktop.org/wiki/Software/systemd/catalog
490 The indexed message catalog database also needs to be
491 rebuilt after installation of message catalog files. Use
492 "journalctl --update-catalog" for this. For RPM-based
493 distributions we introduced the %journal_catalog_update
494 macro for this purpose.
496 * The Python Journal bindings gained support for the standard
497 Python logging framework.
499 * The Journal API gained new functions for checking whether
500 the underlying file system of a journal file is capable of
501 properly reporting file change notifications, or whether
502 applications that want to reflect journal changes "live"
503 need to recheck journal files continously in appropriate
506 * It is now possible to set the "age" field for tmpfiles
507 entries to 0, indicating that files matching this entry
508 shall always be removed when the directories are cleaned up.
510 * coredumpctl gained a new "gdb" verb which invokes gdb
511 right-away on the selected coredump.
513 * There's now support for "hybrid sleep" on kernels that
514 support this, in addition to "suspend" and "hibernate". Use
515 "systemctl hybrid-sleep" to make use of this.
517 * logind's HandleSuspendKey= setting (and related settings)
518 now gained support for a new "lock" setting to simply
519 request the screen lock on all local sessions, instead of
520 actually executing a suspend or hibernation.
522 * systemd will now mount the EFI variables file system by
525 * Socket units now gained support for configuration of the
526 SMACK security label.
528 * timedatectl will now output the time of the last and next
529 daylight saving change.
531 * We dropped support for various legacy and distro-specific
532 concepts, such as insserv, early-boot SysV services
533 (i.e. those for non-standard runlevels such as 'b' or 'S')
534 or ArchLinux /etc/rc.conf support. We recommend the
535 distributions who still need support this to either continue
536 to maintain the necessary patches downstream, or find a
537 different solution. (Talk to us if you have questions!)
539 * Various systemd components will now bypass PolicyKit checks
540 for root and otherwise handle properly if PolicyKit is not
541 found to be around. This should fix most issues for
542 PolicyKit-less systems. Quite frankly this should have been
543 this way since day one. It is absolutely our intention to
544 make systemd work fine on PolicyKit-less systems, and we
545 consider it a bug if something doesn't work as it should if
546 PolicyKit is not around.
548 * For embedded systems it is now possible to build udev and
549 systemd without blkid and/or kmod support.
551 * "systemctl switch-root" is now capable of switching root
552 more than once. I.e. in addition to transitions from the
553 initrd to the host OS it is now possible to transition to
554 further OS images from the host. This is useful to implement
555 offline updating tools.
557 * Various other additions have been made to the RPM macros
558 shipped with systemd. Use %udev_rules_update() after
559 installing new udev rules files. %_udevhwdbdir,
560 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
561 %_sysctldir are now available which resolve to the right
562 directories for packages to place various data files in.
564 * journalctl gained the new --full switch (in addition to
565 --all, to disable ellipsation for long messages.
567 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
568 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
569 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
570 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
571 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
572 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
573 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
574 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
575 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
579 * journalctl gained new --since= and --until= switches to
580 filter by time. It also now supports nice filtering for
581 units via --unit=/-u.
583 * Type=oneshot services may use ExecReload= and do the
586 * The journal daemon now supports time-based rotation and
587 vacuuming, in addition to the usual disk-space based
590 * The journal will now index the available field values for
591 each field name. This enables clients to show pretty drop
592 downs of available match values when filtering. The bash
593 completion of journalctl has been updated
594 accordingly. journalctl gained a new switch -F to list all
595 values a certain field takes in the journal database.
597 * More service events are now written as structured messages
598 to the journal, and made recognizable via message IDs.
600 * The timedated, localed and hostnamed mini-services which
601 previously only provided support for changing time, locale
602 and hostname settings from graphical DEs such as GNOME now
603 also have a minimal (but very useful) text-based client
604 utility each. This is probably the nicest way to changing
605 these settings from the command line now, especially since
606 it lists available options and is fully integrated with bash
609 * There's now a new tool "systemd-coredumpctl" to list and
610 extract coredumps from the journal.
612 * We now install a README each in /var/log/ and
613 /etc/rc.d/init.d explaining where the system logs and init
614 scripts went. This hopefully should help folks who go to
615 that dirs and look into the otherwise now empty void and
618 * When user-services are invoked (by systemd --user) the
619 $MANAGERPID env var is set to the PID of systemd.
621 * SIGRTMIN+24 when sent to a --user instance will now result
622 in immediate termination of systemd.
624 * gatewayd received numerous feature additions such as a
625 "follow" mode, for live syncing and filtering.
627 * browse.html now allows filtering and showing detailed
628 information on specific entries. Keyboard navigation and
629 mouse screen support has been added.
631 * gatewayd/journalctl now supports HTML5/JSON
632 Server-Sent-Events as output.
634 * The SysV init script compatibility logic will now
635 heuristically determine whether a script supports the
636 "reload" verb, and only then make this available as
639 * "systemctl status --follow" has been removed, use "journalctl
642 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
643 have been removed since they are hardly useful to be
646 * And I'd like to take the opportunity to specifically mention
647 Zbigniew for his great contributions. Zbigniew, you rock!
649 Contributions from: Andrew Eikum, Christian Hesse, Colin
650 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
651 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
652 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
653 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
654 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
655 Jędrzejewski-Szmek, Сковорода Никита Андреевич
659 * If /etc/vconsole.conf is non-existent or empty we will no
660 longer load any console font or key map at boot by
661 default. Instead the kernel defaults will be left
662 intact. This is definitely the right thing to do, as no
663 configuration should mean no configuration, and hard-coding
664 font names that are different on all archs is probably a bad
665 idea. Also, the kernel default key map and font should be
666 good enough for most cases anyway, and mostly identical to
667 the userspace fonts/key maps we previously overloaded them
668 with. If distributions want to continue to default to a
669 non-kernel font or key map they should ship a default
670 /etc/vconsole.conf with the appropriate contents.
672 Contributions from: Colin Walters, Daniel J Walsh, Dave
673 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
674 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
678 * journalctl gained a new --cursor= switch to show entries
679 starting from the specified location in the journal.
681 * We now enforce a size limit on journal entry fields exported
682 with "-o json" in journalctl. Fields larger than 4K will be
683 assigned null. This can be turned off with --all.
685 * An (optional) journal gateway daemon is now available as
686 "systemd-journal-gatewayd.service". This service provides
687 access to the journal via HTTP and JSON. This functionality
688 will be used to implement live log synchronization in both
689 pull and push modes, but has various other users too, such
690 as easy log access for debugging of embedded devices. Right
691 now it is already useful to retrieve the journal via HTTP:
693 # systemctl start systemd-journal-gatewayd.service
694 # wget http://localhost:19531/entries
696 This will download the journal contents in a
697 /var/log/messages compatible format. The same as JSON:
699 # curl -H"Accept: application/json" http://localhost:19531/entries
701 This service is also accessible via a web browser where a
702 single static HTML5 app is served that uses the JSON logic
703 to enable the user to do some basic browsing of the
704 journal. This will be extended later on. Here's an example
705 screenshot of this app in its current state:
707 http://0pointer.de/public/journal-gatewayd
709 Contributions from: Kay Sievers, Lennart Poettering, Robert
710 Milasan, Tom Gundersen
714 * The bash completion logic is now available for journalctl
717 * We don't mount the "cpuset" controller anymore together with
718 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
719 started if no parameters are assigned to it. "cpuset" hence
720 broke code that assumed it it could create "cpu" groups and
723 * journalctl -f will now subscribe to terminal size changes,
724 and line break accordingly.
726 Contributions from: Dave Reisner, Kay Sievers, Lennart
727 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
731 * nspawn will now create a symlink /etc/localtime in the
732 container environment, copying the host's timezone
733 setting. Previously this has been done via a bind mount, but
734 since symlinks cannot be bind mounted this has now been
735 changed to create/update the appropriate symlink.
737 * journalctl -n's line number argument is now optional, and
738 will default to 10 if omitted.
740 * journald will now log the maximum size the journal files may
741 take up on disk. This is particularly useful if the default
742 built-in logic of determining this parameter from the file
743 system size is used. Use "systemctl status
744 systemd-journald.service" to see this information.
746 * The multi-seat X wrapper tool has been stripped down. As X
747 is now capable of enumerating graphics devices via udev in a
748 seat-aware way the wrapper is not strictly necessary
749 anymore. A stripped down temporary stop-gap is still shipped
750 until the upstream display managers have been updated to
751 fully support the new X logic. Expect this wrapper to be
752 removed entirely in one of the next releases.
754 * HandleSleepKey= in logind.conf has been split up into
755 HandleSuspendKey= and HandleHibernateKey=. The old setting
756 is not available anymore. X11 and the kernel are
757 distuingishing between these keys and we should too. This
758 also means the inhibition lock for these keys has been split
761 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
762 Poettering, Lukas Nykryn, Václav Pavlín
766 * Whenever a unit changes state we'll now log this to the
767 journal and show along the unit's own log output in
770 * ConditionPathIsMountPoint= can now properly detect bind
771 mount points too. (Previously, a bind mount of one file
772 system to another place in the same file system could not be
773 detected as mount, since they shared struct stat's st_dev
776 * We will now mount the cgroup controllers cpu, cpuacct,
777 cpuset and the controllers net_cls, net_prio together by
780 * nspawn containers will now have a virtualized boot
781 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
782 over with a randomized ID at container initialization). This
783 has the effect of making "journalctl -b" do the right thing
786 * The JSON output journal serialization has been updated not
787 to generate "endless" list objects anymore, but rather one
788 JSON object per line. This is more in line how most JSON
789 parsers expect JSON objects. The new output mode
790 "json-pretty" has been added to provide similar output, but
791 neatly aligned for readability by humans.
793 * We dropped all explicit sync() invocations in the shutdown
794 code. The kernel does this implicitly anyway in the kernel
795 reboot() syscall. halt(8)'s -n option is now a compatibility
798 * We now support virtualized reboot() in containers, as
799 supported by newer kernels. We will fall back to exit() if
800 CAP_SYS_REBOOT is not available to the container. Also,
801 nspawn makes use of this now and will actually reboot the
802 container if the containerized OS asks for that.
804 * journalctl will only show local log output by default
805 now. Use --merge (-m) to show remote log output, too.
807 * libsystemd-journal gained the new sd_journal_get_usage()
808 call to determine the current disk usage of all journal
809 files. This is exposed in the new "journalctl --disk-usage"
812 * journald gained a new configuration setting SplitMode= in
813 journald.conf which may be used to control how user journals
814 are split off. See journald.conf(5) for details.
816 * A new condition type ConditionFileNotEmpty= has been added.
818 * tmpfiles' "w" lines now support file globbing, to write
819 multiple files at once.
821 * We added Python bindings for the journal submission
822 APIs. More Python APIs for a number of selected APIs will
823 likely follow. Note that we intend to add native bindings
824 only for the Python language, as we consider it common
825 enough to deserve bindings shipped within systemd. There are
826 various projects outside of systemd that provide bindings
827 for languages such as PHP or Lua.
829 * Many conditions will now resolve specifiers such as %i. In
830 addition, PathChanged= and related directives of .path units
831 now support specifiers as well.
833 * There's now a new RPM macro definition for the system preset
836 * journald will now warn if it can't foward a message to the
837 syslog daemon because it's socket is full.
839 * timedated will no longer write or process /etc/timezone,
840 except on Debian. As we do not support late mounted /usr
841 anymore /etc/localtime always being a symlink is now safe,
842 and hence the information in /etc/timezone is not necessary
845 * logind will now always reserve one VT for a text getty (VT6
846 by default). Previously if more than 6 X sessions where
847 started they took up all the VTs with auto-spawned gettys,
848 so that no text gettys were available anymore.
850 * udev will now automatically inform the btrfs kernel logic
851 about btrfs RAID components showing up. This should make
852 simple hotplug based btrfs RAID assembly work.
854 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
855 (but not for its children which will stay at the kernel
856 default). This should allow setups with a lot more listening
859 * systemd will now always pass the configured timezone to the
860 kernel at boot. timedated will do the same when the timezone
863 * logind's inhibition logic has been updated. By default,
864 logind will now handle the lid switch, the power and sleep
865 keys all the time, even in graphical sessions. If DEs want
866 to handle these events on their own they should take the new
867 handle-power-key, handle-sleep-key and handle-lid-switch
868 inhibitors during their runtime. A simple way to achiveve
869 that is to invoke the DE wrapped in an invocation of:
871 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
873 * Access to unit operations is now checked via SELinux taking
874 the unit file label and client process label into account.
876 * systemd will now notify the administrator in the journal
877 when he over-mounts a non-empty directory.
879 * There are new specifiers that are resolved in unit files,
880 for the host name (%H), the machine ID (%m) and the boot ID
883 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
884 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
885 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
886 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
887 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
888 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
889 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
893 * Support for reading structured kernel messages from
894 /dev/kmsg has now been added and is enabled by default.
896 * Support for reading kernel messages from /proc/kmsg has now
897 been removed. If you want kernel messages in the journal
898 make sure to run a recent kernel (>= 3.5) that supports
899 reading structured messages from /dev/kmsg (see
900 above). /proc/kmsg is now exclusive property of classic
901 syslog daemons again.
903 * The libudev API gained the new
904 udev_device_new_from_device_id() call.
906 * The logic for file system namespace (ReadOnlyDirectory=,
907 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
908 require pivot_root() anymore. This means fewer temporary
909 directories are created below /tmp for this feature.
911 * nspawn containers will now see and receive all submounts
912 made on the host OS below the root file system of the
915 * Forward Secure Sealing is now supported for Journal files,
916 which provide cryptographical sealing of journal files so
917 that attackers cannot alter log history anymore without this
918 being detectable. Lennart will soon post a blog story about
919 this explaining it in more detail.
921 * There are two new service settings RestartPreventExitStatus=
922 and SuccessExitStatus= which allow configuration of exit
923 status (exit code or signal) which will be excepted from the
924 restart logic, resp. consider successful.
926 * journalctl gained the new --verify switch that can be used
927 to check the integrity of the structure of journal files and
928 (if Forward Secure Sealing is enabled) the contents of
931 * nspawn containers will now be run with /dev/stdin, /dev/fd/
932 and similar symlinks pre-created. This makes running shells
933 as container init process a lot more fun.
935 * The fstab support can now handle PARTUUID= and PARTLABEL=
938 * A new ConditionHost= condition has been added to match
939 against the hostname (with globs) and machine ID. This is
940 useful for clusters where a single OS image is used to
941 provision a large number of hosts which shall run slightly
942 different sets of services.
944 * Services which hit the restart limit will now be placed in a
947 Contributions from: Bertram Poettering, Dave Reisner, Huang
948 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
949 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
953 * When running in --user mode systemd will now become a
954 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
955 tree a lot more organized.
957 * A new PartOf= unit dependency type has been introduced that
958 may be used to group services in a natural way.
960 * "systemctl enable" may now be used to enable instances of
963 * journalctl now prints error log levels in red, and
964 warning/notice log levels in bright white. It also supports
965 filtering by log level now.
967 * cgtop gained a new -n switch (similar to top), to configure
968 the maximum number of iterations to run for. It also gained
969 -b, to run in batch mode (accepting no input).
971 * The suffix ".service" may now be ommited on most systemctl
972 command lines involving service unit names.
974 * There's a new bus call in logind to lock all sessions, as
975 well as a loginctl verb for it "lock-sessions".
977 * libsystemd-logind.so gained a new call sd_journal_perror()
978 that works similar to libc perror() but logs to the journal
979 and encodes structured information about the error number.
981 * /etc/crypttab entries now understand the new keyfile-size=
984 * shutdown(8) now can send a (configurable) wall message when
985 a shutdown is cancelled.
987 * The mount propagation mode for the root file system will now
988 default to "shared", which is useful to make containers work
989 nicely out-of-the-box so that they receive new mounts from
990 the host. This can be undone locally by running "mount
991 --make-rprivate /" if needed.
993 * The prefdm.service file has been removed. Distributions
994 should maintain this unit downstream if they intend to keep
995 it around. However, we recommend writing normal unit files
996 for display managers instead.
998 * Since systemd is a crucial part of the OS we will now
999 default to a number of compiler switches that improve
1000 security (hardening) such as read-only relocations, stack
1001 protection, and suchlike.
1003 * The TimeoutSec= setting for services is now split into
1004 TimeoutStartSec= and TimeoutStopSec= to allow configuration
1005 of individual time outs for the start and the stop phase of
1008 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
1009 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
1010 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
1011 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
1012 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
1013 Gundersen, Zbigniew Jędrzejewski-Szmek
1017 * The journal and id128 C APIs are now fully documented as man
1020 * Extra safety checks have been added when transitioning from
1021 the initial RAM disk to the main system to avoid accidental
1024 * /etc/crypttab entries now understand the new keyfile-offset=
1027 * systemctl -t can now be used to filter by unit load state.
1029 * The journal C API gained the new sd_journal_wait() call to
1030 make writing synchronous journal clients easier.
1032 * journalctl gained the new -D switch to show journals from a
1035 * journalctl now displays a special marker between log
1036 messages of two different boots.
1038 * The journal is now explicitly flushed to /var via a service
1039 systemd-journal-flush.service, rather than implicitly simply
1040 by seeing /var/log/journal to be writable.
1042 * journalctl (and the journal C APIs) can now match for much
1043 more complex expressions, with alternatives and
1046 * When transitioning from the initial RAM disk to the main
1047 system we will now kill all processes in a killing spree to
1048 ensure no processes stay around by accident.
1050 * Three new specifiers may be used in unit files: %u, %h, %s
1051 resolve to the user name, user home directory resp. user
1052 shell. This is useful for running systemd user instances.
1054 * We now automatically rotate journal files if their data
1055 object hash table gets a fill level > 75%. We also size the
1056 hash table based on the configured maximum file size. This
1057 together should lower hash collisions drastically and thus
1058 speed things up a bit.
1060 * journalctl gained the new "--header" switch to introspect
1061 header data of journal files.
1063 * A new setting SystemCallFilters= has been added to services
1064 which may be used to apply blacklists or whitelists to
1065 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
1067 * nspawn gained a new --link-journal= switch (and quicker: -j)
1068 to link the container journal with the host. This makes it
1069 very easy to centralize log viewing on the host for all
1070 guests while still keeping the journal files separated.
1072 * Many bugfixes and optimizations
1074 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
1075 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
1076 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
1081 * Several tools now understand kernel command line arguments,
1082 which are only read when run in an initial RAM disk. They
1083 usually follow closely their normal counterparts, but are
1086 * There's a new tool to analyze the readahead files that are
1087 automatically generated at boot. Use:
1089 /usr/lib/systemd/systemd-readahead analyze /.readahead
1091 * We now provide an early debug shell on tty9 if this enabled. Use:
1093 systemctl enable debug-shell.service
1095 * All plymouth related units have been moved into the Plymouth
1096 package. Please make sure to upgrade your Plymouth version
1099 * systemd-tmpfiles now supports getting passed the basename of
1100 a configuration file only, in which case it will look for it
1101 in all appropriate directories automatically.
1103 * udevadm info now takes a /dev or /sys path as argument, and
1104 does the right thing. Example:
1106 udevadm info /dev/sda
1107 udevadm info /sys/class/block/sda
1109 * systemctl now prints a warning if a unit is stopped but a
1110 unit that might trigger it continues to run. Example: a
1111 service is stopped but the socket that activates it is left
1114 * "systemctl status" will now mention if the log output was
1115 shortened due to rotation since a service has been started.
1117 * The journal API now exposes functions to determine the
1118 "cutoff" times due to rotation.
1120 * journald now understands SIGUSR1 and SIGUSR2 for triggering
1121 immediately flushing of runtime logs to /var if possible,
1122 resp. for triggering immediate rotation of the journal
1125 * It is now considered an error if a service is attempted to
1126 be stopped that is not loaded.
1128 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
1130 * systemd-analyze now supports Python 3
1132 * tmpfiles now supports cleaning up directories via aging
1133 where the first level dirs are always kept around but
1134 directories beneath it automatically aged. This is enabled
1135 by prefixing the age field with '~'.
1137 * Seat objects now expose CanGraphical, CanTTY properties
1138 which is required to deal with very fast bootups where the
1139 display manager might be running before the graphics drivers
1140 completed initialization.
1142 * Seat objects now expose a State property.
1144 * We now include RPM macros for service enabling/disabling
1145 based on the preset logic. We recommend RPM based
1146 distributions to make use of these macros if possible. This
1147 makes it simpler to reuse RPM spec files across
1150 * We now make sure that the collected systemd unit name is
1151 always valid when services log to the journal via
1154 * There's a new man page kernel-command-line(7) detailing all
1155 command line options we understand.
1157 * The fstab generator may now be disabled at boot by passing
1158 fstab=0 on the kernel command line.
1160 * A new kernel command line option modules-load= is now understood
1161 to load a specific kernel module statically, early at boot.
1163 * Unit names specified on the systemctl command line are now
1164 automatically escaped as needed. Also, if file system or
1165 device paths are specified they are automatically turned
1166 into the appropriate mount or device unit names. Example:
1168 systemctl status /home
1169 systemctl status /dev/sda
1171 * The SysVConsole= configuration option has been removed from
1172 system.conf parsing.
1174 * The SysV search path is no longer exported on the D-Bus
1177 * The Names= option is been removed from unit file parsing.
1179 * There's a new man page bootup(7) detailing the boot process.
1181 * Every unit and every generator we ship with systemd now
1182 comes with full documentation. The self-explanatory boot is
1185 * A couple of services gained "systemd-" prefixes in their
1186 name if they wrap systemd code, rather than only external
1187 code. Among them fsck@.service which is now
1188 systemd-fsck@.service.
1190 * The HaveWatchdog property has been removed from the D-Bus
1193 * systemd.confirm_spawn= on the kernel command line should now
1196 * There's a new man page crypttab(5) which details all options
1197 we actually understand.
1199 * systemd-nspawn gained a new --capability= switch to pass
1200 additional capabilities to the container.
1202 * timedated will now read known NTP implementation unit names
1203 from /usr/lib/systemd/ntp-units.d/*.list,
1204 systemd-timedated-ntp.target has been removed.
1206 * journalctl gained a new switch "-b" that lists log data of
1207 the current boot only.
1209 * The notify socket is in the abstract namespace again, in
1210 order to support daemons which chroot() at start-up.
1212 * There is a new Storage= configuration option for journald
1213 which allows configuration of where log data should go. This
1214 also provides a way to disable journal logging entirely, so
1215 that data collected is only forwarded to the console, the
1216 kernel log buffer or another syslog implementation.
1218 * Many bugfixes and optimizations
1220 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
1221 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
1222 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
1223 Shawn Landden, Tom Gundersen
1227 * "systemctl help <unit>" now shows the man page if one is
1230 * Several new man pages have been added.
1232 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
1233 MaxLevelConsole= can now be specified in
1234 journald.conf. These options allow reducing the amount of
1235 data stored on disk or forwarded by the log level.
1237 * TimerSlackNSec= can now be specified in system.conf for
1238 PID1. This allows system-wide power savings.
1240 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
1241 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
1246 * logind is now capable of (optionally) handling power and
1247 sleep keys as well as the lid switch.
1249 * journalctl now understands the syntax "journalctl
1250 /usr/bin/avahi-daemon" to get all log output of a specific
1253 * CapabilityBoundingSet= in system.conf now also influences
1254 the capability bound set of usermode helpers of the kernel.
1256 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
1257 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
1258 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
1259 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
1263 * Note that we skipped 139 releases here in order to set the
1264 new version to something that is greater than both udev's
1265 and systemd's most recent version number.
1267 * udev: all udev sources are merged into the systemd source tree now.
1268 All future udev development will happen in the systemd tree. It
1269 is still fully supported to use the udev daemon and tools without
1270 systemd running, like in initramfs or other init systems. Building
1271 udev though, will require the *build* of the systemd tree, but
1272 udev can be properly *run* without systemd.
1274 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
1275 should be used to create dead device nodes as workarounds for broken
1278 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
1279 no longer supported. udev_monitor_new_from_netlink() needs to be
1280 used to subscribe to events.
1282 * udev: when udevd is started by systemd, processes which are left
1283 behind by forking them off of udev rules, are unconditionally cleaned
1284 up and killed now after the event handling has finished. Services or
1285 daemons must be started as systemd services. Services can be
1286 pulled-in by udev to get started, but they can no longer be directly
1287 forked by udev rules.
1289 * udev: the daemon binary is called systemd-udevd now and installed
1290 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
1291 to adapt to that, create symlink, or rename the binary after building
1294 * libudev no longer provides these symbols:
1295 udev_monitor_from_socket()
1296 udev_queue_get_failed_list_entry()
1297 udev_get_{dev,sys,run}_path()
1298 The versions number was bumped and symbol versioning introduced.
1300 * systemd-loginctl and systemd-journalctl have been renamed
1301 to loginctl and journalctl to match systemctl.
1303 * The config files: /etc/systemd/systemd-logind.conf and
1304 /etc/systemd/systemd-journald.conf have been renamed to
1305 logind.conf and journald.conf. Package updates should rename
1306 the files to the new names on upgrade.
1308 * For almost all files the license is now LGPL2.1+, changed
1309 from the previous GPL2.0+. Exceptions are some minor stuff
1310 of udev (which will be changed to LGPL2.1 eventually, too),
1311 and the MIT licensed sd-daemon.[ch] library that is suitable
1312 to be used as drop-in files.
1314 * systemd and logind now handle system sleep states, in
1315 particular suspending and hibernating.
1317 * logind now implements a sleep/shutdown/idle inhibiting logic
1318 suitable for a variety of uses. Soonishly Lennart will blog
1319 about this in more detail.
1321 * var-run.mount and var-lock.mount are no longer provided
1322 (which prevously bind mounted these directories to their new
1323 places). Distributions which have not converted these
1324 directories to symlinks should consider stealing these files
1325 from git history and add them downstream.
1327 * We introduced the Documentation= field for units and added
1328 this to all our shipped units. This is useful to make it
1329 easier to explore the boot and the purpose of the various
1332 * All smaller setup units (such as
1333 systemd-vconsole-setup.service) now detect properly if they
1334 are run in a container and are skipped when
1335 appropriate. This guarantees an entirely noise-free boot in
1336 Linux container environments such as systemd-nspawn.
1338 * A framework for implementing offline system updates is now
1339 integrated, for details see:
1340 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
1342 * A new service type Type=idle is available now which helps us
1343 avoiding ugly interleaving of getty output and boot status
1346 * There's now a system-wide CapabilityBoundingSet= option to
1347 globally reduce the set of capabilities for the
1348 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
1349 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
1350 even CAP_NET_ADMIN system-wide for secure systems.
1352 * There are now system-wide DefaultLimitXXX= options to
1353 globally change the defaults of the various resource limits
1354 for all units started by PID 1.
1356 * Harald Hoyer's systemd test suite has been integrated into
1357 systemd which allows easy testing of systemd builds in qemu
1358 and nspawn. (This is really awesome! Ask us for details!)
1360 * The fstab parser is now implemented as generator, not inside
1363 * systemctl will now warn you if .mount units generated from
1364 /etc/fstab are out of date due to changes in fstab that
1365 haven't been read by systemd yet.
1367 * systemd is now suitable for usage in initrds. Dracut has
1368 already been updated to make use of this. With this in place
1369 initrds get a slight bit faster but primarily are much
1370 easier to introspect and debug since "systemctl status" in
1371 the host system can be used to introspect initrd services,
1372 and the journal from the initrd is kept around too.
1374 * systemd-delta has been added, a tool to explore differences
1375 between user/admin configuration and vendor defaults.
1377 * PrivateTmp= now affects both /tmp and /var/tmp.
1379 * Boot time status messages are now much prettier and feature
1380 proper english language. Booting up systemd has never been
1383 * Read-ahead pack files now include the inode number of all
1384 files to pre-cache. When the inode changes the pre-caching
1385 is not attempted. This should be nicer to deal with updated
1386 packages which might result in changes of read-ahead
1389 * We now temporaritly lower the kernel's read_ahead_kb variable
1390 when collecting read-ahead data to ensure the kernel's
1391 built-in read-ahead does not add noise to our measurements
1392 of necessary blocks to pre-cache.
1394 * There's now RequiresMountsFor= to add automatic dependencies
1395 for all mounts necessary for a specific file system path.
1397 * MountAuto= and SwapAuto= have been removed from
1398 system.conf. Mounting file systems at boot has to take place
1401 * nspawn now learned a new switch --uuid= to set the machine
1402 ID on the command line.
1404 * nspawn now learned the -b switch to automatically search
1407 * vt102 is now the default TERM for serial TTYs, upgraded from
1410 * systemd-logind now works on VT-less systems.
1412 * The build tree has been reorganized. The individual
1413 components now have directories of their own.
1415 * A new condition type ConditionPathIsReadWrite= is now available.
1417 * nspawn learned the new -C switch to create cgroups for the
1418 container in other hierarchies.
1420 * We now have support for hardware watchdogs, configurable in
1423 * The scheduled shutdown logic now has a public API.
1425 * We now mount /tmp as tmpfs by default, but this can be
1426 masked and /etc/fstab can override it.
1428 * Since udisks doesn't make use of /media anymore we are not
1429 mounting a tmpfs on it anymore.
1431 * journalctl gained a new --local switch to only interleave
1432 locally generated journal files.
1434 * We can now load the IMA policy at boot automatically.
1436 * The GTK tools have been split off into a systemd-ui.
1438 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
1439 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
1440 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
1441 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
1442 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
1443 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
1444 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
1445 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
1446 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
1451 * This is mostly a bugfix release
1453 * Support optional initialization of the machine ID from the
1454 KVM or container configured UUID.
1456 * Support immediate reboots with "systemctl reboot -ff"
1458 * Show /etc/os-release data in systemd-analyze output
1460 * Many bugfixes for the journal, including endianess fixes and
1461 ensuring that disk space enforcement works
1463 * sd-login.h is C++ comptaible again
1465 * Extend the /etc/os-release format on request of the Debian
1468 * We now refuse non-UTF8 strings used in various configuration
1469 and unit files. This is done to ensure we don't pass invalid
1470 data over D-Bus or expose it elsewhere.
1472 * Register Mimo USB Screens as suitable for automatic seat
1475 * Read SELinux client context from journal clients in a race
1478 * Reorder configuration file lookup order. /etc now always
1479 overrides /run in order to allow the administrator to always
1480 and unconditionally override vendor supplied or
1481 automatically generated data.
1483 * The various user visible bits of the journal now have man
1484 pages. We still lack man pages for the journal API calls
1487 * We now ship all man pages in HTML format again in the
1490 Contributions from: Dave Reisner, Dirk Eibach, Frederic
1491 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
1492 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
1497 * This is mostly a bugfix release
1499 * systems lacking /etc/os-release are no longer supported.
1501 * Various functionality updates to libsystemd-login.so
1503 * Track class of PAM logins to distuingish greeters from
1506 Contributions from: Kay Sievers, Lennart Poettering, Michael
1511 * This is an important bugfix release for v41.
1513 * Building man pages is now optional which should be useful
1514 for those building systemd from git but unwilling to install
1517 * Watchdog support for supervising services is now usable. In
1518 a future release support for hardware watchdogs
1519 (i.e. /dev/watchdog) will be added building on this.
1521 * Service start rate limiting is now configurable and can be
1522 turned off per service. When a start rate limit is hit a
1523 reboot can automatically be triggered.
1525 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
1527 Contributions from: Benjamin Franzke, Bill Nottingham,
1528 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
1529 Schmidt, Michał Górny, Piotr Drąg
1533 * The systemd binary is installed /usr/lib/systemd/systemd now;
1534 An existing /sbin/init symlink needs to be adapted with the
1537 * The code that loads kernel modules has been ported to invoke
1538 libkmod directly, instead of modprobe. This means we do not
1539 support systems with module-init-tools anymore.
1541 * Watchdog support is now already useful, but still not
1544 * A new kernel command line option systemd.setenv= is
1545 understood to set system wide environment variables
1546 dynamically at boot.
1548 * We now limit the set of capabilities of systemd-journald.
1550 * We now set SIGPIPE to ignore by default, since it only is
1551 useful in shell pipelines, and has little use in general
1552 code. This can be disabled with IgnoreSIPIPE=no in unit
1555 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
1556 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
1561 * This is mostly a bugfix release
1563 * We now expose the reason why a service failed in the
1564 "Result" D-Bus property.
1566 * Rudimentary service watchdog support (will be completed over
1567 the next few releases.)
1569 * When systemd forks off in order execute some service we will
1570 now immediately changes its argv[0] to reflect which process
1571 it will execute. This is useful to minimize the time window
1572 with a generic argv[0], which makes bootcharts more useful
1574 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
1575 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
1576 Mike Kazantsev, Ray Strode
1580 * This is mostly a test release, but incorporates many
1583 * New systemd-cgtop tool to show control groups by their
1586 * Linking against libacl for ACLs is optional again. If
1587 disabled, support tracking device access for active logins
1588 goes becomes unavailable, and so does access to the user
1589 journals by the respective users.
1591 * If a group "adm" exists, journal files are automatically
1592 owned by them, thus allow members of this group full access
1593 to the system journal as well as all user journals.
1595 * The journal now stores the SELinux context of the logging
1596 client for all entries.
1598 * Add C++ inclusion guards to all public headers
1600 * New output mode "cat" in the journal to print only text
1601 messages, without any meta data like date or time.
1603 * Include tiny X server wrapper as a temporary stop-gap to
1604 teach XOrg udev display enumeration. This is used by display
1605 managers such as gdm, and will go away as soon as XOrg
1606 learned native udev hotplugging for display devices.
1608 * Add new systemd-cat tool for executing arbitrary programs
1609 with STDERR/STDOUT connected to the journal. Can also act as
1610 BSD logger replacement, and does so by default.
1612 * Optionally store all locally generated coredumps in the
1613 journal along with meta data.
1615 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
1616 writing short strings to files (for usage for /sys), and for
1617 creating symlinks, character and block device nodes.
1619 * New unit file option ControlGroupPersistent= to make cgroups
1620 persistent, following the mechanisms outlined in
1621 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
1623 * Support multiple local RTCs in a sane way
1625 * No longer monopolize IO when replaying readahead data on
1626 rotating disks, since we might starve non-file-system IO to
1627 death, since fanotify() will not see accesses done by blkid,
1630 * Don't show kernel threads in systemd-cgls anymore, unless
1631 requested with new -k switch.
1633 Contributions from: Dan Horák, Kay Sievers, Lennart
1634 Poettering, Michal Schmidt
1638 * This is mostly a test release, but incorporates many
1641 * The git repository moved to:
1642 git://anongit.freedesktop.org/systemd/systemd
1643 ssh://git.freedesktop.org/git/systemd/systemd
1645 * First release with the journal
1646 http://0pointer.de/blog/projects/the-journal.html
1648 * The journal replaces both systemd-kmsg-syslogd and
1649 systemd-stdout-bridge.
1651 * New sd_pid_get_unit() API call in libsystemd-logind
1653 * Many systemadm clean-ups
1655 * Introduce remote-fs-pre.target which is ordered before all
1656 remote mounts and may be used to start services before all
1659 * Added Mageia support
1661 * Add bash completion for systemd-loginctl
1663 * Actively monitor PID file creation for daemons which exit in
1664 the parent process before having finished writing the PID
1665 file in the daemon process. Daemons which do this need to be
1666 fixed (i.e. PID file creation must have finished before the
1667 parent exits), but we now react a bit more gracefully to them.
1669 * Add colourful boot output, mimicking the well-known output
1670 of existing distributions.
1672 * New option PassCredentials= for socket units, for
1673 compatibility with a recent kernel ABI breakage.
1675 * /etc/rc.local is now hooked in via a generator binary, and
1676 thus will no longer act as synchronization point during
1679 * systemctl list-unit-files now supports --root=.
1681 * systemd-tmpfiles now understands two new commands: z, Z for
1682 relabelling files according to the SELinux database. This is
1683 useful to apply SELinux labels to specific files in /sys,
1686 * Output of SysV services is now forwarded to both the console
1687 and the journal by default, not only just the console.
1689 * New man pages for all APIs from libsystemd-login.
1691 * The build tree got reorganized and a the build system is a
1692 lot more modular allowing embedded setups to specifically
1693 select the components of systemd they are interested in.
1695 * Support for Linux systems lacking the kernel VT subsystem is
1698 * configure's --with-rootdir= got renamed to
1699 --with-rootprefix= to follow the naming used by udev and
1702 * Unless specified otherwise we'll now install to /usr instead
1703 of /usr/local by default.
1705 * Processes with '@' in argv[0][0] are now excluded from the
1706 final shut-down killing spree, following the logic explained
1708 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
1710 * All processes remaining in a service cgroup when we enter
1711 the START or START_PRE states are now killed with
1712 SIGKILL. That means it is no longer possible to spawn
1713 background processes from ExecStart= lines (which was never
1714 supported anyway, and bad style).
1716 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
1717 reloading of units together.
1719 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
1720 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
1721 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
1722 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
1723 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek