1 systemd System and Service Manager
5 * Configuration of unit files may now be extended via drop-in
6 files without having to edit/override the unit files
7 themselves. More specifically, if the administrator wants to
8 change one value for a service file foobar.service he can
9 now do so by dropping in a configuration snippet into
10 /etc/systemd/systemd/foobar.service.d/*.conf. The unit logic
11 will load all these snippets and apply them on top of the
12 main unit configuration file, possibly extending or
13 overriding its settings. Using these drop-in snippets is
14 generally nicer than the two earlier options for changing
15 unit files locally: copying the files from
16 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
17 them there; or creating a new file in /etc/systemd/system/
18 that incorporates the original one via ".include". Drop-in
19 snippets into these .d/ directories can be placed in any
20 directory systems looks for units in, and the usual
21 overriding semantics between /usr/lib, /etc and /run apply
24 * Most unit file settings which take lists of items can now be
25 reset by assigning the empty strings to them. For example,
26 normally, settings such as Environment=FOO=BAR append a new
27 environment variable assignment to the environment block,
28 each time they are used. By assigning Environment= the empty
29 string the environment block can be reset to empty. This is
30 particularly useful with the .d/*.conf drop-in snippets
31 mentioned above, since this allows resetting list settings
32 from vendor unit files via these drop-ins.
34 * systemctl gained a new "list-dependencies" command for
35 listing the dependencies of a unit recursively.
37 * Inhibitors are now honored and listed by "systemctl
38 suspend", "systemctl poweroff" (and similar) too, not only
39 GNOME. These commands will also list active sessions by
42 * Resource limits (as exposed by the various control group
43 controllers) can now be controlled dynamically at runtime
44 for all units. More specifically, you can now use a command
45 like "systemctl set-cgroup-attr foobar.service cpu.shares
46 2000" to alter the CPU shares a specific service gets. These
47 settings are stores persistently on disk, and thus allow the
48 administrator to easily adjust the resource usage of
49 services with a few simple commands. This dynamic resource
50 management logic is also available to other services via the
51 bus. Almost any kernel cgroup attribute and controller is
54 * systemd-vconsole-setup will now copy all font settings to
55 all open VTs, where it previously applied them only to the
58 * libsystemd-login gained the new sd_session_get_tty() API
61 * This release drops support with a few
62 legacy/distribution-specific LSB facility names when parsing
63 init scripts: $x-display-manager, $mail-transfer-agent,
64 $mail-transport-agent, $mail-transfer-agent, $smtp,
65 $null. Also, the mail-transfer-agent.target unit backing
66 this has been removed. Distributions which want to retain
67 compatibility with this should carry the burden for support
68 this themselves and patch support for these back in if they
69 really need to. Also, the facilities $syslog and $local_fs
70 are now ignored, since systemd does not support early-boot
71 LSB init scripts anymore, and these facilities are implied
72 anyway for normal services. syslog.target has also been
75 * There are new bus calls on PID1's Manager object for
76 canceling jobs, and removing snapshot units. Previously,
77 both calls were only available on the Job and Snapshot
80 * systemd-journal-gatewayd gained SSL support.
82 * The various "environment" files, such as /etc/locale.conf
83 now support continuation lines with a backslash ("\") as
84 last character in the line, similar in style (but different)
85 to how this is supported in shells.
87 * For normal user processes the _SYSTEMD_USER_UNIT= field is
88 now implicitly appended to every log entry logged. systemctl
89 has been updated to filter by this field when operating on a
90 user systemd instance.
92 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
93 CAP_AUDIT_CONTROL capabilities to the capabilities set for
94 the container. This makes it easier to boot unmodified
95 Fedora systems in a container, which however still requires
96 audit=0 to be passed on the kernel command line. Auditing in
97 kernel and userspace is unfortunately still too broken in
98 context of containers, hence we recommend compiling it out
99 of the kernel or using audit=0. Hopefully this will be fixed
100 one day for good in the kernel.
102 * nspawn gained the new --bind= and --bind-ro= parameters to
103 bind mount specific directories from the host into the
106 * nspawn will now mount its own devpts file system instance
107 into the container, in order not to leake pty devices from
108 the host into the container.
110 * systemd will now read the firmware boot time performance
111 from the EFI variables, if the used boot loader supports
112 this, and take it into account for boot performance analysis
113 via "systemd-analyze". This is currently supported only in
114 conjunction with Gummiboot, but could be supported by other
115 boot loaders too. For details see:
117 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
119 * A new generator has been added that automatically mounts the
120 EFI System Partition (ESP) to /boot, if that directory
121 exists, is empty, and no other partition has been configured
124 * logind will now send out PrepareForSleep(false) out
125 unconditionally, after coming back from suspend. This may be
126 used by applications as asynchronous notification for
127 system resume events.
129 * "systemctl unlock-sessions" has been added, that allows
130 unlocking the screens of all user sessions at once, similar
131 how "systemctl lock-sessions" already locked all users
132 sessions. This is backed by a new D-Bus call UnlockSessions().
134 * "loginctl seat-status" will now show the master device of a
135 seat. (i.e. the device of a seat that needs to be around for
136 the seat to be considered available, usually the graphics
139 * tmpfiles gained a new "X" line type, that allows
140 configuration of files and directories (with wildcards) that
141 shall be excluded from automatic cleanup ("aging").
143 * udev default rules set the device node permissions now only
144 at "add" events, and do not change them any longer with a
145 later "change" event.
147 * The log messages for lid events and power/sleep keypresses
148 now carry a message ID.
150 * We now have a substantially larger unit test suite, but this
151 continues to be work in progress.
153 * udevadm hwdb gained a new --root= parameter to change the
154 root directory to operate relative to.
156 * logind will now issue a background sync() request to the kernel
157 early at shutdown, so that dirty buffers are flushed to disk early
158 instead of at the last moment, in order to optimize shutdown
161 * A new bootctl tool has been added that is an interface for
162 certain boot loader operations. This is currently a preview
163 and is likely to be extended into a small mechanism daemon
164 like timedated, localed, hostnamed, and can be used by
165 graphical UIs to enumerate available boot options, and
166 request boot into firmware operations.
168 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
169 the rest of the package. It also has been updated to work
170 correctly in initrds.
172 * Policykit previously has been runtime optional, and is now
173 also compile time optional via a configure switch.
175 * systemd-analyze has been reimplemented in C. Also "systemctl
176 dot" has moved into systemd-analyze.
178 * "systemctl status" with no further parameters will now print
179 the status of all active or failed units.
181 * Operations such as "systemctl start" can now be executed
182 with a new mode "--irreversible" which may be used to queue
183 operations that cannot accidentally be reversed by a later
184 job queuing. This is by default use to make shutdown
185 requests more robust.
187 * The Python API of systemd now gained a new module for
188 reading journal files.
190 * A new tool kernel-install has been added that can install
191 kernel images according to the Boot Loader Specification:
193 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
195 * Boot time console output has been improved to provide
196 animated boot time output.
198 * A new tool systemd-activate has been added which can be used
199 to test socket activation with, directly from the command
200 line. This should make it much easier to test and debug
201 socket activation in daemons.
203 * journalctl gained a new "--reverse" (or -r) option to show
204 journal output in reverse order (i.e. newest line first).
206 * journalctl gained a new "--user-unit=" option, that works
207 similar to "--unit=" but filters for user units rather than
210 * A number of unit files to ease adoption of systemd in
211 initrds has been added. This moves some minimal logic from
212 the various initrd implementations into systemd proper.
214 * The journal files are now owned by a new group
215 "systemd-journal", which exists specifically to allow access
216 to the journal, and nothing else. Previously, we used the
217 "adm" group for that, which however possible covers more
218 than just journal/log file access. This new group is now
219 already used by systemd-journal-gatewayd to ensure this
220 daemon gets access to the journal files and as little else
221 as possible. Note that "make install" will also set FS ACLs
222 up for /var/log/journal to give "adm" and "wheel" read
223 access to it, in addition to "systemd-journal" which owns
224 the journal files. We recommend that packaging scripts also
225 add read access to "adm", "wheel" to /var/log/journal, and
226 all existing/future journal files. To normal users and
227 administrators little changes, however packagers need to
228 ensure to create the "systemd-journal" system group at
229 package installation time.
231 * The systemd-journal-gatewayd now runs as unprivileged user
232 systemd-journal-gateway:systemd-journal-gateway. Packaging
233 scripts need to create these system user/group at
236 * timedated now exposes a new boolean property CanNTP that
237 indicates whether a local NTP service is available or not.
239 * systemd-detect-virt will now also detect xen PVs
241 * The pstore file system is now mounted by default, if it is
244 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
245 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
246 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
247 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
248 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
249 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
250 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
251 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
252 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
253 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
254 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
255 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
256 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
257 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
261 * Timer units now support calendar time events in addition to
262 monotonic time events. That means you can now trigger a unit
263 based on a calendar time specification such as "Thu,Fri
264 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
265 or fifth day of any month of the year 2013, given that it is
266 a thursday or friday. This brings timer event support
267 considerably closer to cron's capabilities. For details on
268 the supported calendar time specification language see
271 * udev now supports a number of different naming policies for
272 network interfaces for predictable names, and a combination
273 of these policies is now the default. Please see this wiki
274 document for details:
276 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
278 * Auke Kok's bootchart implementation has been added to the
279 systemd tree. It's an optional component that can graph the
280 boot in quite some detail. It's one of the best bootchart
281 implementations around and minimal in its code and
284 * nss-myhostname has been integrated into the systemd source
285 tree. nss-myhostname guarantees that the local hostname
286 always stays resolvable via NSS. It has been a weak
287 requirement of systemd-hostnamed since a long time, and
288 since its code is actually trivial we decided to just
289 include it in systemd's source tree. It can be turned off
290 with a configure switch.
292 * The read-ahead logic is now capable of properly detecting
293 whether a btrfs file system is on SSD or rotating media, in
294 order to optimize the read-ahead scheme. Previously, it was
295 only capable of detecting this on traditional file systems
298 * In udev, additional device properties are now read from the
299 IAB in addition to the OUI database. Also, Bluetooth company
300 identities are attached to the devices as well.
302 * In service files %U may be used as specifier that is
303 replaced by the configured user name of the service.
305 * nspawn may now be invoked without a controlling TTY. This
306 makes it suitable for invocation as its own service. This
307 may be used to set up a simple containerized server system
308 using only core OS tools.
310 * systemd and nspawn can now accept socket file descriptors
311 when they are started for socket activation. This enables
312 implementation of socket activated nspawn
313 containers. i.e. think about autospawning an entire OS image
314 when the first SSH or HTTP connection is received. We expect
315 that similar functionality will also be added to libvirt-lxc
318 * journalctl will now suppress ANSI color codes when
321 * systemctl will no longer show control group information for
322 a unit if a the control group is empty anyway.
324 * logind can now automatically suspend/hibernate/shutdown the
327 * /etc/machine-info and hostnamed now also expose the chassis
328 type of the system. This can be used to determine whether
329 the local system is a laptop, desktop, handset or
330 tablet. This information may either be configured by the
331 user/vendor or is automatically determined from ACPI and DMI
332 information if possible.
334 * A number of PolicyKit actions are now bound together with
335 "imply" rules. This should simplify creating UIs because
336 many actions will now authenticate similar ones as well.
338 * Unit files learnt a new condition ConditionACPower= which
339 may be used to conditionalize a unit depending on whether an
340 AC power source is connected or not, of whether the system
341 is running on battery power.
343 * systemctl gained a new "is-failed" verb that may be used in
344 shell scripts and suchlike to check whether a specific unit
345 is in the "failed" state.
347 * The EnvironmentFile= setting in unit files now supports file
348 globbing, and can hence be used to easily read a number of
349 environment files at once.
351 * systemd will no longer detect and recognize specific
352 distributions. All distribution-specific #ifdeffery has been
353 removed, systemd is now fully generic and
354 distribution-agnostic. Effectively, not too much is lost as
355 a lot of the code is still accessible via explicit configure
356 switches. However, support for some distribution specific
357 legacy configuration file formats has been dropped. We
358 recommend distributions to simply adopt the configuration
359 files everybody else uses now and convert the old
360 configuration from packaging scripts. Most distributions
361 already did that. If that's not possible or desirable,
362 distributions are welcome to forward port the specific
363 pieces of code locally from the git history.
365 * When logging a message about a unit systemd will now always
366 log the unit name in the message meta data.
368 * localectl will now also discover system locale data that is
369 not stored in locale archives, but directly unpacked.
371 * logind will no longer unconditionally use framebuffer
372 devices as seat masters, i.e. as devices that are required
373 to be existing before a seat is considered preset. Instead,
374 it will now look for all devices that are tagged as
375 "seat-master" in udev. By default framebuffer devices will
376 be marked as such, but depending on local systems other
377 devices might be marked as well. This may be used to
378 integrate graphics cards using closed source drivers (such
379 as NVidia ones) more nicely into logind. Note however, that
380 we recommend using the open source NVidia drivers instead,
381 and no udev rules for the closed-source drivers will be
382 shipped from us upstream.
384 Contributions from: Adam Williamson, Alessandro Crismani, Auke
385 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
386 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
387 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
388 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
389 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
390 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
391 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
392 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
393 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
394 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
395 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
400 * udev gained support for loading additional device properties
401 from an indexed database that is keyed by vendor/product IDs
402 and similar device identifiers. For the beginning this
403 "hwdb" is populated with data from the well-known PCI and
404 USB database, but also includes PNP, ACPI and OID data. In
405 the longer run this indexed database shall grow into
406 becoming the one central database for non-essential
407 userspace device metadata. Previously, data from the PCI/USB
408 database was only attached to select devices, since the
409 lookup was a relatively expensive operation due to O(n) time
410 complexity (with n being the number of entries in the
411 database). Since this is now O(1), we decided to add in this
412 data for all devices where this is available, by
413 default. Note that the indexed database needs to be rebuilt
414 when new data files are installed. To achieve this you need
415 to update your packaging scripts to invoke "udevadm hwdb
416 --update" after installation of hwdb data files. For
417 RPM-based distributions we introduced the new
418 %udev_hwdb_update macro for this purpose.
420 * The Journal gained support for the "Message Catalog", an
421 indexed database to link up additional information with
422 journal entries. For further details please check:
424 http://www.freedesktop.org/wiki/Software/systemd/catalog
426 The indexed message catalog database also needs to be
427 rebuilt after installation of message catalog files. Use
428 "journalctl --update-catalog" for this. For RPM-based
429 distributions we introduced the %journal_catalog_update
430 macro for this purpose.
432 * The Python Journal bindings gained support for the standard
433 Python logging framework.
435 * The Journal API gained new functions for checking whether
436 the underlying file system of a journal file is capable of
437 properly reporting file change notifications, or whether
438 applications that want to reflect journal changes "live"
439 need to recheck journal files continously in appropriate
442 * It is now possible to set the "age" field for tmpfiles
443 entries to 0, indicating that files matching this entry
444 shall always be removed when the directories are cleaned up.
446 * coredumpctl gained a new "gdb" verb which invokes gdb
447 right-away on the selected coredump.
449 * There's now support for "hybrid sleep" on kernels that
450 support this, in addition to "suspend" and "hibernate". Use
451 "systemctl hybrid-sleep" to make use of this.
453 * logind's HandleSuspendKey= setting (and related settings)
454 now gained support for a new "lock" setting to simply
455 request the screen lock on all local sessions, instead of
456 actually executing a suspend or hibernation.
458 * systemd will now mount the EFI variables file system by
461 * Socket units now gained support for configuration of the
462 SMACK security label.
464 * timedatectl will now output the time of the last and next
465 daylight saving change.
467 * We dropped support for various legacy and distro-specific
468 concepts, such as insserv, early-boot SysV services
469 (i.e. those for non-standard runlevels such as 'b' or 'S')
470 or ArchLinux /etc/rc.conf support. We recommend the
471 distributions who still need support this to either continue
472 to maintain the necessary patches downstream, or find a
473 different solution. (Talk to us if you have questions!)
475 * Various systemd components will now bypass PolicyKit checks
476 for root and otherwise handle properly if PolicyKit is not
477 found to be around. This should fix most issues for
478 PolicyKit-less systems. Quite frankly this should have been
479 this way since day one. It is absolutely our intention to
480 make systemd work fine on PolicyKit-less systems, and we
481 consider it a bug if something doesn't work as it should if
482 PolicyKit is not around.
484 * For embedded systems it is now possible to build udev and
485 systemd without blkid and/or kmod support.
487 * "systemctl switch-root" is now capable of switching root
488 more than once. I.e. in addition to transitions from the
489 initrd to the host OS it is now possible to transition to
490 further OS images from the host. This is useful to implement
491 offline updating tools.
493 * Various other additions have been made to the RPM macros
494 shipped with systemd. Use %udev_rules_update() after
495 installing new udev rules files. %_udevhwdbdir,
496 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
497 %_sysctldir are now available which resolve to the right
498 directories for packages to place various data files in.
500 * journalctl gained the new --full switch (in addition to
501 --all, to disable ellipsation for long messages.
503 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
504 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
505 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
506 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
507 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
508 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
509 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
510 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
511 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
515 * journalctl gained new --since= and --until= switches to
516 filter by time. It also now supports nice filtering for
517 units via --unit=/-u.
519 * Type=oneshot services may use ExecReload= and do the
522 * The journal daemon now supports time-based rotation and
523 vacuuming, in addition to the usual disk-space based
526 * The journal will now index the available field values for
527 each field name. This enables clients to show pretty drop
528 downs of available match values when filtering. The bash
529 completion of journalctl has been updated
530 accordingly. journalctl gained a new switch -F to list all
531 values a certain field takes in the journal database.
533 * More service events are now written as structured messages
534 to the journal, and made recognizable via message IDs.
536 * The timedated, localed and hostnamed mini-services which
537 previously only provided support for changing time, locale
538 and hostname settings from graphical DEs such as GNOME now
539 also have a minimal (but very useful) text-based client
540 utility each. This is probably the nicest way to changing
541 these settings from the command line now, especially since
542 it lists available options and is fully integrated with bash
545 * There's now a new tool "systemd-coredumpctl" to list and
546 extract coredumps from the journal.
548 * We now install a README each in /var/log/ and
549 /etc/rc.d/init.d explaining where the system logs and init
550 scripts went. This hopefully should help folks who go to
551 that dirs and look into the otherwise now empty void and
554 * When user-services are invoked (by systemd --user) the
555 $MANAGERPID env var is set to the PID of systemd.
557 * SIGRTMIN+24 when sent to a --user instance will now result
558 in immediate termination of systemd.
560 * gatewayd received numerous feature additions such as a
561 "follow" mode, for live syncing and filtering.
563 * browse.html now allows filtering and showing detailed
564 information on specific entries. Keyboard navigation and
565 mouse screen support has been added.
567 * gatewayd/journalctl now supports HTML5/JSON
568 Server-Sent-Events as output.
570 * The SysV init script compatibility logic will now
571 heuristically determine whether a script supports the
572 "reload" verb, and only then make this available as
575 * "systemctl status --follow" has been removed, use "journalctl
578 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
579 have been removed since they are hardly useful to be
582 * And I'd like to take the opportunity to specifically mention
583 Zbigniew for his great contributions. Zbigniew, you rock!
585 Contributions from: Andrew Eikum, Christian Hesse, Colin
586 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
587 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
588 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
589 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
590 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
591 Jędrzejewski-Szmek, Сковорода Никита Андреевич
595 * If /etc/vconsole.conf is non-existent or empty we will no
596 longer load any console font or key map at boot by
597 default. Instead the kernel defaults will be left
598 intact. This is definitely the right thing to do, as no
599 configuration should mean no configuration, and hard-coding
600 font names that are different on all archs is probably a bad
601 idea. Also, the kernel default key map and font should be
602 good enough for most cases anyway, and mostly identical to
603 the userspace fonts/key maps we previously overloaded them
604 with. If distributions want to continue to default to a
605 non-kernel font or key map they should ship a default
606 /etc/vconsole.conf with the appropriate contents.
608 Contributions from: Colin Walters, Daniel J Walsh, Dave
609 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
610 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
614 * journalctl gained a new --cursor= switch to show entries
615 starting from the specified location in the journal.
617 * We now enforce a size limit on journal entry fields exported
618 with "-o json" in journalctl. Fields larger than 4K will be
619 assigned null. This can be turned off with --all.
621 * An (optional) journal gateway daemon is now available as
622 "systemd-journal-gatewayd.service". This service provides
623 access to the journal via HTTP and JSON. This functionality
624 will be used to implement live log synchronization in both
625 pull and push modes, but has various other users too, such
626 as easy log access for debugging of embedded devices. Right
627 now it is already useful to retrieve the journal via HTTP:
629 # systemctl start systemd-journal-gatewayd.service
630 # wget http://localhost:19531/entries
632 This will download the journal contents in a
633 /var/log/messages compatible format. The same as JSON:
635 # curl -H"Accept: application/json" http://localhost:19531/entries
637 This service is also accessible via a web browser where a
638 single static HTML5 app is served that uses the JSON logic
639 to enable the user to do some basic browsing of the
640 journal. This will be extended later on. Here's an example
641 screenshot of this app in its current state:
643 http://0pointer.de/public/journal-gatewayd
645 Contributions from: Kay Sievers, Lennart Poettering, Robert
646 Milasan, Tom Gundersen
650 * The bash completion logic is now available for journalctl
653 * We don't mount the "cpuset" controller anymore together with
654 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
655 started if no parameters are assigned to it. "cpuset" hence
656 broke code that assumed it it could create "cpu" groups and
659 * journalctl -f will now subscribe to terminal size changes,
660 and line break accordingly.
662 Contributions from: Dave Reisner, Kay Sievers, Lennart
663 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
667 * nspawn will now create a symlink /etc/localtime in the
668 container environment, copying the host's timezone
669 setting. Previously this has been done via a bind mount, but
670 since symlinks cannot be bind mounted this has now been
671 changed to create/update the appropriate symlink.
673 * journalctl -n's line number argument is now optional, and
674 will default to 10 if omitted.
676 * journald will now log the maximum size the journal files may
677 take up on disk. This is particularly useful if the default
678 built-in logic of determining this parameter from the file
679 system size is used. Use "systemctl status
680 systemd-journald.service" to see this information.
682 * The multi-seat X wrapper tool has been stripped down. As X
683 is now capable of enumerating graphics devices via udev in a
684 seat-aware way the wrapper is not strictly necessary
685 anymore. A stripped down temporary stop-gap is still shipped
686 until the upstream display managers have been updated to
687 fully support the new X logic. Expect this wrapper to be
688 removed entirely in one of the next releases.
690 * HandleSleepKey= in logind.conf has been split up into
691 HandleSuspendKey= and HandleHibernateKey=. The old setting
692 is not available anymore. X11 and the kernel are
693 distuingishing between these keys and we should too. This
694 also means the inhibition lock for these keys has been split
697 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
698 Poettering, Lukas Nykryn, Václav Pavlín
702 * Whenever a unit changes state we'll now log this to the
703 journal and show along the unit's own log output in
706 * ConditionPathIsMountPoint= can now properly detect bind
707 mount points too. (Previously, a bind mount of one file
708 system to another place in the same file system could not be
709 detected as mount, since they shared struct stat's st_dev
712 * We will now mount the cgroup controllers cpu, cpuacct,
713 cpuset and the controllers net_cls, net_prio together by
716 * nspawn containers will now have a virtualized boot
717 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
718 over with a randomized ID at container initialization). This
719 has the effect of making "journalctl -b" do the right thing
722 * The JSON output journal serialization has been updated not
723 to generate "endless" list objects anymore, but rather one
724 JSON object per line. This is more in line how most JSON
725 parsers expect JSON objects. The new output mode
726 "json-pretty" has been added to provide similar output, but
727 neatly aligned for readability by humans.
729 * We dropped all explicit sync() invocations in the shutdown
730 code. The kernel does this implicitly anyway in the kernel
731 reboot() syscall. halt(8)'s -n option is now a compatibility
734 * We now support virtualized reboot() in containers, as
735 supported by newer kernels. We will fall back to exit() if
736 CAP_SYS_REBOOT is not available to the container. Also,
737 nspawn makes use of this now and will actually reboot the
738 container if the containerized OS asks for that.
740 * journalctl will only show local log output by default
741 now. Use --merge (-m) to show remote log output, too.
743 * libsystemd-journal gained the new sd_journal_get_usage()
744 call to determine the current disk usage of all journal
745 files. This is exposed in the new "journalctl --disk-usage"
748 * journald gained a new configuration setting SplitMode= in
749 journald.conf which may be used to control how user journals
750 are split off. See journald.conf(5) for details.
752 * A new condition type ConditionFileNotEmpty= has been added.
754 * tmpfiles' "w" lines now support file globbing, to write
755 multiple files at once.
757 * We added Python bindings for the journal submission
758 APIs. More Python APIs for a number of selected APIs will
759 likely follow. Note that we intend to add native bindings
760 only for the Python language, as we consider it common
761 enough to deserve bindings shipped within systemd. There are
762 various projects outside of systemd that provide bindings
763 for languages such as PHP or Lua.
765 * Many conditions will now resolve specifiers such as %i. In
766 addition, PathChanged= and related directives of .path units
767 now support specifiers as well.
769 * There's now a new RPM macro definition for the system preset
772 * journald will now warn if it can't foward a message to the
773 syslog daemon because it's socket is full.
775 * timedated will no longer write or process /etc/timezone,
776 except on Debian. As we do not support late mounted /usr
777 anymore /etc/localtime always being a symlink is now safe,
778 and hence the information in /etc/timezone is not necessary
781 * logind will now always reserve one VT for a text getty (VT6
782 by default). Previously if more than 6 X sessions where
783 started they took up all the VTs with auto-spawned gettys,
784 so that no text gettys were available anymore.
786 * udev will now automatically inform the btrfs kernel logic
787 about btrfs RAID components showing up. This should make
788 simple hotplug based btrfs RAID assembly work.
790 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
791 (but not for its children which will stay at the kernel
792 default). This should allow setups with a lot more listening
795 * systemd will now always pass the configured timezone to the
796 kernel at boot. timedated will do the same when the timezone
799 * logind's inhibition logic has been updated. By default,
800 logind will now handle the lid switch, the power and sleep
801 keys all the time, even in graphical sessions. If DEs want
802 to handle these events on their own they should take the new
803 handle-power-key, handle-sleep-key and handle-lid-switch
804 inhibitors during their runtime. A simple way to achiveve
805 that is to invoke the DE wrapped in an invocation of:
807 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
809 * Access to unit operations is now checked via SELinux taking
810 the unit file label and client process label into account.
812 * systemd will now notify the administrator in the journal
813 when he over-mounts a non-empty directory.
815 * There are new specifiers that are resolved in unit files,
816 for the host name (%H), the machine ID (%m) and the boot ID
819 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
820 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
821 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
822 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
823 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
824 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
825 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
829 * Support for reading structured kernel messages from
830 /dev/kmsg has now been added and is enabled by default.
832 * Support for reading kernel messages from /proc/kmsg has now
833 been removed. If you want kernel messages in the journal
834 make sure to run a recent kernel (>= 3.5) that supports
835 reading structured messages from /dev/kmsg (see
836 above). /proc/kmsg is now exclusive property of classic
837 syslog daemons again.
839 * The libudev API gained the new
840 udev_device_new_from_device_id() call.
842 * The logic for file system namespace (ReadOnlyDirectory=,
843 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
844 require pivot_root() anymore. This means fewer temporary
845 directories are created below /tmp for this feature.
847 * nspawn containers will now see and receive all submounts
848 made on the host OS below the root file system of the
851 * Forward Secure Sealing is now supported for Journal files,
852 which provide cryptographical sealing of journal files so
853 that attackers cannot alter log history anymore without this
854 being detectable. Lennart will soon post a blog story about
855 this explaining it in more detail.
857 * There are two new service settings RestartPreventExitStatus=
858 and SuccessExitStatus= which allow configuration of exit
859 status (exit code or signal) which will be excepted from the
860 restart logic, resp. consider successful.
862 * journalctl gained the new --verify switch that can be used
863 to check the integrity of the structure of journal files and
864 (if Forward Secure Sealing is enabled) the contents of
867 * nspawn containers will now be run with /dev/stdin, /dev/fd/
868 and similar symlinks pre-created. This makes running shells
869 as container init process a lot more fun.
871 * The fstab support can now handle PARTUUID= and PARTLABEL=
874 * A new ConditionHost= condition has been added to match
875 against the hostname (with globs) and machine ID. This is
876 useful for clusters where a single OS image is used to
877 provision a large number of hosts which shall run slightly
878 different sets of services.
880 * Services which hit the restart limit will now be placed in a
883 Contributions from: Bertram Poettering, Dave Reisner, Huang
884 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
885 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
889 * When running in --user mode systemd will now become a
890 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
891 tree a lot more organized.
893 * A new PartOf= unit dependency type has been introduced that
894 may be used to group services in a natural way.
896 * "systemctl enable" may now be used to enable instances of
899 * journalctl now prints error log levels in red, and
900 warning/notice log levels in bright white. It also supports
901 filtering by log level now.
903 * cgtop gained a new -n switch (similar to top), to configure
904 the maximum number of iterations to run for. It also gained
905 -b, to run in batch mode (accepting no input).
907 * The suffix ".service" may now be ommited on most systemctl
908 command lines involving service unit names.
910 * There's a new bus call in logind to lock all sessions, as
911 well as a loginctl verb for it "lock-sessions".
913 * libsystemd-logind.so gained a new call sd_journal_perror()
914 that works similar to libc perror() but logs to the journal
915 and encodes structured information about the error number.
917 * /etc/crypttab entries now understand the new keyfile-size=
920 * shutdown(8) now can send a (configurable) wall message when
921 a shutdown is cancelled.
923 * The mount propagation mode for the root file system will now
924 default to "shared", which is useful to make containers work
925 nicely out-of-the-box so that they receive new mounts from
926 the host. This can be undone locally by running "mount
927 --make-rprivate /" if needed.
929 * The prefdm.service file has been removed. Distributions
930 should maintain this unit downstream if they intend to keep
931 it around. However, we recommend writing normal unit files
932 for display managers instead.
934 * Since systemd is a crucial part of the OS we will now
935 default to a number of compiler switches that improve
936 security (hardening) such as read-only relocations, stack
937 protection, and suchlike.
939 * The TimeoutSec= setting for services is now split into
940 TimeoutStartSec= and TimeoutStopSec= to allow configuration
941 of individual time outs for the start and the stop phase of
944 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
945 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
946 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
947 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
948 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
949 Gundersen, Zbigniew Jędrzejewski-Szmek
953 * The journal and id128 C APIs are now fully documented as man
956 * Extra safety checks have been added when transitioning from
957 the initial RAM disk to the main system to avoid accidental
960 * /etc/crypttab entries now understand the new keyfile-offset=
963 * systemctl -t can now be used to filter by unit load state.
965 * The journal C API gained the new sd_journal_wait() call to
966 make writing synchronous journal clients easier.
968 * journalctl gained the new -D switch to show journals from a
971 * journalctl now displays a special marker between log
972 messages of two different boots.
974 * The journal is now explicitly flushed to /var via a service
975 systemd-journal-flush.service, rather than implicitly simply
976 by seeing /var/log/journal to be writable.
978 * journalctl (and the journal C APIs) can now match for much
979 more complex expressions, with alternatives and
982 * When transitioning from the initial RAM disk to the main
983 system we will now kill all processes in a killing spree to
984 ensure no processes stay around by accident.
986 * Three new specifiers may be used in unit files: %u, %h, %s
987 resolve to the user name, user home directory resp. user
988 shell. This is useful for running systemd user instances.
990 * We now automatically rotate journal files if their data
991 object hash table gets a fill level > 75%. We also size the
992 hash table based on the configured maximum file size. This
993 together should lower hash collisions drastically and thus
994 speed things up a bit.
996 * journalctl gained the new "--header" switch to introspect
997 header data of journal files.
999 * A new setting SystemCallFilters= has been added to services
1000 which may be used to apply blacklists or whitelists to
1001 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
1003 * nspawn gained a new --link-journal= switch (and quicker: -j)
1004 to link the container journal with the host. This makes it
1005 very easy to centralize log viewing on the host for all
1006 guests while still keeping the journal files separated.
1008 * Many bugfixes and optimizations
1010 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
1011 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
1012 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
1017 * Several tools now understand kernel command line arguments,
1018 which are only read when run in an initial RAM disk. They
1019 usually follow closely their normal counterparts, but are
1022 * There's a new tool to analyze the readahead files that are
1023 automatically generated at boot. Use:
1025 /usr/lib/systemd/systemd-readahead analyze /.readahead
1027 * We now provide an early debug shell on tty9 if this enabled. Use:
1029 systemctl enable debug-shell.service
1031 * All plymouth related units have been moved into the Plymouth
1032 package. Please make sure to upgrade your Plymouth version
1035 * systemd-tmpfiles now supports getting passed the basename of
1036 a configuration file only, in which case it will look for it
1037 in all appropriate directories automatically.
1039 * udevadm info now takes a /dev or /sys path as argument, and
1040 does the right thing. Example:
1042 udevadm info /dev/sda
1043 udevadm info /sys/class/block/sda
1045 * systemctl now prints a warning if a unit is stopped but a
1046 unit that might trigger it continues to run. Example: a
1047 service is stopped but the socket that activates it is left
1050 * "systemctl status" will now mention if the log output was
1051 shortened due to rotation since a service has been started.
1053 * The journal API now exposes functions to determine the
1054 "cutoff" times due to rotation.
1056 * journald now understands SIGUSR1 and SIGUSR2 for triggering
1057 immediately flushing of runtime logs to /var if possible,
1058 resp. for triggering immediate rotation of the journal
1061 * It is now considered an error if a service is attempted to
1062 be stopped that is not loaded.
1064 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
1066 * systemd-analyze now supports Python 3
1068 * tmpfiles now supports cleaning up directories via aging
1069 where the first level dirs are always kept around but
1070 directories beneath it automatically aged. This is enabled
1071 by prefixing the age field with '~'.
1073 * Seat objects now expose CanGraphical, CanTTY properties
1074 which is required to deal with very fast bootups where the
1075 display manager might be running before the graphics drivers
1076 completed initialization.
1078 * Seat objects now expose a State property.
1080 * We now include RPM macros for service enabling/disabling
1081 based on the preset logic. We recommend RPM based
1082 distributions to make use of these macros if possible. This
1083 makes it simpler to reuse RPM spec files across
1086 * We now make sure that the collected systemd unit name is
1087 always valid when services log to the journal via
1090 * There's a new man page kernel-command-line(7) detailing all
1091 command line options we understand.
1093 * The fstab generator may now be disabled at boot by passing
1094 fstab=0 on the kernel command line.
1096 * A new kernel command line option modules-load= is now understood
1097 to load a specific kernel module statically, early at boot.
1099 * Unit names specified on the systemctl command line are now
1100 automatically escaped as needed. Also, if file system or
1101 device paths are specified they are automatically turned
1102 into the appropriate mount or device unit names. Example:
1104 systemctl status /home
1105 systemctl status /dev/sda
1107 * The SysVConsole= configuration option has been removed from
1108 system.conf parsing.
1110 * The SysV search path is no longer exported on the D-Bus
1113 * The Names= option is been removed from unit file parsing.
1115 * There's a new man page bootup(7) detailing the boot process.
1117 * Every unit and every generator we ship with systemd now
1118 comes with full documentation. The self-explanatory boot is
1121 * A couple of services gained "systemd-" prefixes in their
1122 name if they wrap systemd code, rather than only external
1123 code. Among them fsck@.service which is now
1124 systemd-fsck@.service.
1126 * The HaveWatchdog property has been removed from the D-Bus
1129 * systemd.confirm_spawn= on the kernel command line should now
1132 * There's a new man page crypttab(5) which details all options
1133 we actually understand.
1135 * systemd-nspawn gained a new --capability= switch to pass
1136 additional capabilities to the container.
1138 * timedated will now read known NTP implementation unit names
1139 from /usr/lib/systemd/ntp-units.d/*.list,
1140 systemd-timedated-ntp.target has been removed.
1142 * journalctl gained a new switch "-b" that lists log data of
1143 the current boot only.
1145 * The notify socket is in the abstract namespace again, in
1146 order to support daemons which chroot() at start-up.
1148 * There is a new Storage= configuration option for journald
1149 which allows configuration of where log data should go. This
1150 also provides a way to disable journal logging entirely, so
1151 that data collected is only forwarded to the console, the
1152 kernel log buffer or another syslog implementation.
1154 * Many bugfixes and optimizations
1156 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
1157 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
1158 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
1159 Shawn Landden, Tom Gundersen
1163 * "systemctl help <unit>" now shows the man page if one is
1166 * Several new man pages have been added.
1168 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
1169 MaxLevelConsole= can now be specified in
1170 journald.conf. These options allow reducing the amount of
1171 data stored on disk or forwarded by the log level.
1173 * TimerSlackNSec= can now be specified in system.conf for
1174 PID1. This allows system-wide power savings.
1176 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
1177 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
1182 * logind is now capable of (optionally) handling power and
1183 sleep keys as well as the lid switch.
1185 * journalctl now understands the syntax "journalctl
1186 /usr/bin/avahi-daemon" to get all log output of a specific
1189 * CapabilityBoundingSet= in system.conf now also influences
1190 the capability bound set of usermode helpers of the kernel.
1192 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
1193 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
1194 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
1195 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
1199 * Note that we skipped 139 releases here in order to set the
1200 new version to something that is greater than both udev's
1201 and systemd's most recent version number.
1203 * udev: all udev sources are merged into the systemd source tree now.
1204 All future udev development will happen in the systemd tree. It
1205 is still fully supported to use the udev daemon and tools without
1206 systemd running, like in initramfs or other init systems. Building
1207 udev though, will require the *build* of the systemd tree, but
1208 udev can be properly *run* without systemd.
1210 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
1211 should be used to create dead device nodes as workarounds for broken
1214 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
1215 no longer supported. udev_monitor_new_from_netlink() needs to be
1216 used to subscribe to events.
1218 * udev: when udevd is started by systemd, processes which are left
1219 behind by forking them off of udev rules, are unconditionally cleaned
1220 up and killed now after the event handling has finished. Services or
1221 daemons must be started as systemd services. Services can be
1222 pulled-in by udev to get started, but they can no longer be directly
1223 forked by udev rules.
1225 * udev: the daemon binary is called systemd-udevd now and installed
1226 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
1227 to adapt to that, create symlink, or rename the binary after building
1230 * libudev no longer provides these symbols:
1231 udev_monitor_from_socket()
1232 udev_queue_get_failed_list_entry()
1233 udev_get_{dev,sys,run}_path()
1234 The versions number was bumped and symbol versioning introduced.
1236 * systemd-loginctl and systemd-journalctl have been renamed
1237 to loginctl and journalctl to match systemctl.
1239 * The config files: /etc/systemd/systemd-logind.conf and
1240 /etc/systemd/systemd-journald.conf have been renamed to
1241 logind.conf and journald.conf. Package updates should rename
1242 the files to the new names on upgrade.
1244 * For almost all files the license is now LGPL2.1+, changed
1245 from the previous GPL2.0+. Exceptions are some minor stuff
1246 of udev (which will be changed to LGPL2.1 eventually, too),
1247 and the MIT licensed sd-daemon.[ch] library that is suitable
1248 to be used as drop-in files.
1250 * systemd and logind now handle system sleep states, in
1251 particular suspending and hibernating.
1253 * logind now implements a sleep/shutdown/idle inhibiting logic
1254 suitable for a variety of uses. Soonishly Lennart will blog
1255 about this in more detail.
1257 * var-run.mount and var-lock.mount are no longer provided
1258 (which prevously bind mounted these directories to their new
1259 places). Distributions which have not converted these
1260 directories to symlinks should consider stealing these files
1261 from git history and add them downstream.
1263 * We introduced the Documentation= field for units and added
1264 this to all our shipped units. This is useful to make it
1265 easier to explore the boot and the purpose of the various
1268 * All smaller setup units (such as
1269 systemd-vconsole-setup.service) now detect properly if they
1270 are run in a container and are skipped when
1271 appropriate. This guarantees an entirely noise-free boot in
1272 Linux container environments such as systemd-nspawn.
1274 * A framework for implementing offline system updates is now
1275 integrated, for details see:
1276 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
1278 * A new service type Type=idle is available now which helps us
1279 avoiding ugly interleaving of getty output and boot status
1282 * There's now a system-wide CapabilityBoundingSet= option to
1283 globally reduce the set of capabilities for the
1284 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
1285 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
1286 even CAP_NET_ADMIN system-wide for secure systems.
1288 * There are now system-wide DefaultLimitXXX= options to
1289 globally change the defaults of the various resource limits
1290 for all units started by PID 1.
1292 * Harald Hoyer's systemd test suite has been integrated into
1293 systemd which allows easy testing of systemd builds in qemu
1294 and nspawn. (This is really awesome! Ask us for details!)
1296 * The fstab parser is now implemented as generator, not inside
1299 * systemctl will now warn you if .mount units generated from
1300 /etc/fstab are out of date due to changes in fstab that
1301 haven't been read by systemd yet.
1303 * systemd is now suitable for usage in initrds. Dracut has
1304 already been updated to make use of this. With this in place
1305 initrds get a slight bit faster but primarily are much
1306 easier to introspect and debug since "systemctl status" in
1307 the host system can be used to introspect initrd services,
1308 and the journal from the initrd is kept around too.
1310 * systemd-delta has been added, a tool to explore differences
1311 between user/admin configuration and vendor defaults.
1313 * PrivateTmp= now affects both /tmp and /var/tmp.
1315 * Boot time status messages are now much prettier and feature
1316 proper english language. Booting up systemd has never been
1319 * Read-ahead pack files now include the inode number of all
1320 files to pre-cache. When the inode changes the pre-caching
1321 is not attempted. This should be nicer to deal with updated
1322 packages which might result in changes of read-ahead
1325 * We now temporaritly lower the kernel's read_ahead_kb variable
1326 when collecting read-ahead data to ensure the kernel's
1327 built-in read-ahead does not add noise to our measurements
1328 of necessary blocks to pre-cache.
1330 * There's now RequiresMountsFor= to add automatic dependencies
1331 for all mounts necessary for a specific file system path.
1333 * MountAuto= and SwapAuto= have been removed from
1334 system.conf. Mounting file systems at boot has to take place
1337 * nspawn now learned a new switch --uuid= to set the machine
1338 ID on the command line.
1340 * nspawn now learned the -b switch to automatically search
1343 * vt102 is now the default TERM for serial TTYs, upgraded from
1346 * systemd-logind now works on VT-less systems.
1348 * The build tree has been reorganized. The individual
1349 components now have directories of their own.
1351 * A new condition type ConditionPathIsReadWrite= is now available.
1353 * nspawn learned the new -C switch to create cgroups for the
1354 container in other hierarchies.
1356 * We now have support for hardware watchdogs, configurable in
1359 * The scheduled shutdown logic now has a public API.
1361 * We now mount /tmp as tmpfs by default, but this can be
1362 masked and /etc/fstab can override it.
1364 * Since udisks doesn't make use of /media anymore we are not
1365 mounting a tmpfs on it anymore.
1367 * journalctl gained a new --local switch to only interleave
1368 locally generated journal files.
1370 * We can now load the IMA policy at boot automatically.
1372 * The GTK tools have been split off into a systemd-ui.
1374 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
1375 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
1376 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
1377 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
1378 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
1379 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
1380 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
1381 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
1382 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
1387 * This is mostly a bugfix release
1389 * Support optional initialization of the machine ID from the
1390 KVM or container configured UUID.
1392 * Support immediate reboots with "systemctl reboot -ff"
1394 * Show /etc/os-release data in systemd-analyze output
1396 * Many bugfixes for the journal, including endianess fixes and
1397 ensuring that disk space enforcement works
1399 * sd-login.h is C++ comptaible again
1401 * Extend the /etc/os-release format on request of the Debian
1404 * We now refuse non-UTF8 strings used in various configuration
1405 and unit files. This is done to ensure we don't pass invalid
1406 data over D-Bus or expose it elsewhere.
1408 * Register Mimo USB Screens as suitable for automatic seat
1411 * Read SELinux client context from journal clients in a race
1414 * Reorder configuration file lookup order. /etc now always
1415 overrides /run in order to allow the administrator to always
1416 and unconditionally override vendor supplied or
1417 automatically generated data.
1419 * The various user visible bits of the journal now have man
1420 pages. We still lack man pages for the journal API calls
1423 * We now ship all man pages in HTML format again in the
1426 Contributions from: Dave Reisner, Dirk Eibach, Frederic
1427 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
1428 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
1433 * This is mostly a bugfix release
1435 * systems lacking /etc/os-release are no longer supported.
1437 * Various functionality updates to libsystemd-login.so
1439 * Track class of PAM logins to distuingish greeters from
1442 Contributions from: Kay Sievers, Lennart Poettering, Michael
1447 * This is an important bugfix release for v41.
1449 * Building man pages is now optional which should be useful
1450 for those building systemd from git but unwilling to install
1453 * Watchdog support for supervising services is now usable. In
1454 a future release support for hardware watchdogs
1455 (i.e. /dev/watchdog) will be added building on this.
1457 * Service start rate limiting is now configurable and can be
1458 turned off per service. When a start rate limit is hit a
1459 reboot can automatically be triggered.
1461 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
1463 Contributions from: Benjamin Franzke, Bill Nottingham,
1464 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
1465 Schmidt, Michał Górny, Piotr Drąg
1469 * The systemd binary is installed /usr/lib/systemd/systemd now;
1470 An existing /sbin/init symlink needs to be adapted with the
1473 * The code that loads kernel modules has been ported to invoke
1474 libkmod directly, instead of modprobe. This means we do not
1475 support systems with module-init-tools anymore.
1477 * Watchdog support is now already useful, but still not
1480 * A new kernel command line option systemd.setenv= is
1481 understood to set system wide environment variables
1482 dynamically at boot.
1484 * We now limit the set of capabilities of systemd-journald.
1486 * We now set SIGPIPE to ignore by default, since it only is
1487 useful in shell pipelines, and has little use in general
1488 code. This can be disabled with IgnoreSIPIPE=no in unit
1491 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
1492 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
1497 * This is mostly a bugfix release
1499 * We now expose the reason why a service failed in the
1500 "Result" D-Bus property.
1502 * Rudimentary service watchdog support (will be completed over
1503 the next few releases.)
1505 * When systemd forks off in order execute some service we will
1506 now immediately changes its argv[0] to reflect which process
1507 it will execute. This is useful to minimize the time window
1508 with a generic argv[0], which makes bootcharts more useful
1510 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
1511 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
1512 Mike Kazantsev, Ray Strode
1516 * This is mostly a test release, but incorporates many
1519 * New systemd-cgtop tool to show control groups by their
1522 * Linking against libacl for ACLs is optional again. If
1523 disabled, support tracking device access for active logins
1524 goes becomes unavailable, and so does access to the user
1525 journals by the respective users.
1527 * If a group "adm" exists, journal files are automatically
1528 owned by them, thus allow members of this group full access
1529 to the system journal as well as all user journals.
1531 * The journal now stores the SELinux context of the logging
1532 client for all entries.
1534 * Add C++ inclusion guards to all public headers
1536 * New output mode "cat" in the journal to print only text
1537 messages, without any meta data like date or time.
1539 * Include tiny X server wrapper as a temporary stop-gap to
1540 teach XOrg udev display enumeration. This is used by display
1541 managers such as gdm, and will go away as soon as XOrg
1542 learned native udev hotplugging for display devices.
1544 * Add new systemd-cat tool for executing arbitrary programs
1545 with STDERR/STDOUT connected to the journal. Can also act as
1546 BSD logger replacement, and does so by default.
1548 * Optionally store all locally generated coredumps in the
1549 journal along with meta data.
1551 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
1552 writing short strings to files (for usage for /sys), and for
1553 creating symlinks, character and block device nodes.
1555 * New unit file option ControlGroupPersistent= to make cgroups
1556 persistent, following the mechanisms outlined in
1557 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
1559 * Support multiple local RTCs in a sane way
1561 * No longer monopolize IO when replaying readahead data on
1562 rotating disks, since we might starve non-file-system IO to
1563 death, since fanotify() will not see accesses done by blkid,
1566 * Don't show kernel threads in systemd-cgls anymore, unless
1567 requested with new -k switch.
1569 Contributions from: Dan Horák, Kay Sievers, Lennart
1570 Poettering, Michal Schmidt
1574 * This is mostly a test release, but incorporates many
1577 * The git repository moved to:
1578 git://anongit.freedesktop.org/systemd/systemd
1579 ssh://git.freedesktop.org/git/systemd/systemd
1581 * First release with the journal
1582 http://0pointer.de/blog/projects/the-journal.html
1584 * The journal replaces both systemd-kmsg-syslogd and
1585 systemd-stdout-bridge.
1587 * New sd_pid_get_unit() API call in libsystemd-logind
1589 * Many systemadm clean-ups
1591 * Introduce remote-fs-pre.target which is ordered before all
1592 remote mounts and may be used to start services before all
1595 * Added Mageia support
1597 * Add bash completion for systemd-loginctl
1599 * Actively monitor PID file creation for daemons which exit in
1600 the parent process before having finished writing the PID
1601 file in the daemon process. Daemons which do this need to be
1602 fixed (i.e. PID file creation must have finished before the
1603 parent exits), but we now react a bit more gracefully to them.
1605 * Add colourful boot output, mimicking the well-known output
1606 of existing distributions.
1608 * New option PassCredentials= for socket units, for
1609 compatibility with a recent kernel ABI breakage.
1611 * /etc/rc.local is now hooked in via a generator binary, and
1612 thus will no longer act as synchronization point during
1615 * systemctl list-unit-files now supports --root=.
1617 * systemd-tmpfiles now understands two new commands: z, Z for
1618 relabelling files according to the SELinux database. This is
1619 useful to apply SELinux labels to specific files in /sys,
1622 * Output of SysV services is now forwarded to both the console
1623 and the journal by default, not only just the console.
1625 * New man pages for all APIs from libsystemd-login.
1627 * The build tree got reorganized and a the build system is a
1628 lot more modular allowing embedded setups to specifically
1629 select the components of systemd they are interested in.
1631 * Support for Linux systems lacking the kernel VT subsystem is
1634 * configure's --with-rootdir= got renamed to
1635 --with-rootprefix= to follow the naming used by udev and
1638 * Unless specified otherwise we'll now install to /usr instead
1639 of /usr/local by default.
1641 * Processes with '@' in argv[0][0] are now excluded from the
1642 final shut-down killing spree, following the logic explained
1644 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
1646 * All processes remaining in a service cgroup when we enter
1647 the START or START_PRE states are now killed with
1648 SIGKILL. That means it is no longer possible to spawn
1649 background processes from ExecStart= lines (which was never
1650 supported anyway, and bad style).
1652 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
1653 reloading of units together.
1655 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
1656 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
1657 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
1658 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
1659 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek